article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
energy measurements of giant air showers demonstrated the existence of a distinctive structure , the @xmath4 , in the differential energy spectrum of the cosmic radiation above @xmath10 .
figure 1 shows the energy spectrum measured by four experiments [ 1,2,3,4 ] at these extreme energies .
cosmic ray intensity does not continue to decrease with a spectral index between 3 and 3.2 , observed between @xmath11 and @xmath10 , but an enhancement of the intensity above @xmath3 @xmath2 appears in all experiments .
this enhancement is relative to the extrapolation at high energy with the index of 3 measured at lower energy , below @xmath12 @xmath2 .
the @xmath4 is the distinctive pattern in the spectrum consisting of a minimum of intensity followed by an enhancement observable approximately in the energy band @xmath3-@xmath13 .
the energy at which this enhancement occurs , and its magnitude , of this enhancement do not coincide in different experiments , though precise measurements of the minimum are available from the hires collaboration [ 4,5 ] .
this contribution to the cris conference 2006 follows recent studies [ 6,7 ] ( hereafter paper i and ii , respectively ) reporting a quantitative solution of the long - standing problem of the knee and ankle of the cosmic ray spectrum .
a notable aspect of this solution is that the same mechanisms are responsable for the ankles and the knees of the individual ions .
even purposely it would be impossible to separate the explanation of the ankle from that of the knee .
the intrinsic mechanisms generating the ankles do not depend on the spectral indices of the cosmic rays at the sources nor on any extragalactic component of the cosmic radiation , but is an intrinsic property of the galactic cosmic rays ( sections 3 and 4 ) .
the extragalactic component , if any at the earth around @xmath9 @xmath2 , might add to the galactic component , affecting the magnitude of the ankle , but its necessity and its experimental evidence remain unproved ( section 7 ) .
note that the expected gzk depression or its absence [ 3 ] is beyond the energy band pertinent for the explanation of the ankles presented here .
the method of calculation , used for over a decade , reconstructs ion trajectories by computer simulations and evaluates physical quantities of cosmic rays ( intensities , energy spectra , ion ratios etc ) using the number of trajectories intercepting a given small volume in the galaxy which in the present calculation can be the earth ( also @xmath14 @xmath15 @xmath16 or @xmath17 @xmath18 ) , a sphere concentric to the earth , or the entire disc volume ( section 9 ) .
the computational algorithms , described elsewhere [ 8,9,10 ] take into account the following astronomical , astrophysical and radioastronomy observations : ( 1 ) the spiral magnetic field ; ( 2 ) the field strength ( see figure 3 paper i ) of the spiral magnetic field ; ( 3 ) a chaotic magnetic field with a field strength of about three times that of the regular field ; ( 4 ) the form and the dimension of the galaxy ( see figure 1 in ref.[9 ] ) ; ( 5 ) a uniform distribution of cosmic ray sources in the galactic disk ( see eqn.(3 ) in ref.[9 ] ) ; ( 6 ) the nuclear cross sections ion - hydrogen , @xmath19 ; ( 7 ) the interstellar matter density in the disk , @xmath20 , of 1.24 hydrogen atoms per @xmath21 ; ( 8) the position of the @xmath17 @xmath18 inside the disk , at 14 @xmath22 above the galactic midplane and 8.5 @xmath23 from the galactic center ; ( 9 ) the galactic wind ( section 2 in ref.[10 ] ) . prerequisites of this study regarding the notion of @xmath15 @xmath24 [ 11,10 ] have been previously discussed in detail .
a cosmic ion emanated from a source travels , on average , a mean length , @xmath25 , in the disk volume , primarily determined by @xmath19 and @xmath20 .
typically , this global length @xmath25 is subdivided into thousands and thousands of segments , depending on the particular region of propagation and the energy of the cosmic ion .
the ion propagation normal to the regular spiral field is generated by the chaotic field leading to a transverse - to - longitudinal displacement ratio compatible with the quasi - linear theory of ion propagation in an astrophysical environment .
the chaotic or turbulent magnetic field is materialized by magnetic cloudlets ( see , for instance , ref.[12,13 ] ) with variable dimensions and a field strenght three times greater than that of the regular field .
trajectories at low energy are quite different from those at very high energy .
figures 2 and 3 display a typical low energy trajectory .
the differential energy spectrum of the cosmic rays , @xmath26/@xmath27 , depends on the number of cosmic ray trajectories intercepting the local galactic zone , @xmath28 .
figure 4 shows @xmath28 versus energy for helium , taken as an example .
three energy regions of @xmath28 are clearly distinguishable : ( 1 ) a high plateau ; ( 2 ) a rapid descent ; ( 3 ) a low plateau .
the intensity gap between the high and the low plateau is of capital importance to comprehend the origin of the ankle .
this gap depends on the matter thickness encountered by the cosmic helium in the galaxy and on the nuclear cross section helium - hydrogen , @xmath19(he ) .
the form of the spectrum shown in figure 4 is quite general , and simple mechanisms concur in its creation as clarified below . in the energy band @xmath29-@xmath1
@xmath28 is approximately constant , though slightly decreasing . forcing a power law interpolation on @xmath28 with a constant spectral index , a value of 0.13 for helium results .
this decreasing feature is due to rising cross sections as explained in detail in paper i , sections 3 and 4 . in the high plateau region ,
the indices of @xmath28 for different nuclides tend to decrease with the atomic number .
an elegant derivation of this result is feasible by the notion of @xmath15 @xmath24 [ 10,11 ] .
the second portion of the spectrum , referred to as the knee , is due primarily to the break in the efficiency of the ion bending inherent the galactic magnetic field .
this break has been analyzed in detail elsewhere ( paper i section 4 and paper ii section 5 ) .
the break energy is defined as that particular energy characteristic of a nuclear species , where the bend of @xmath28 occurs and a sharp decrease dominates .
though the effect of the magnetic field on @xmath28 is dominant , the exact form of @xmath28 emerges only after including the position of the solar cavity , the nuclear cross sections and the disc size .
the decrease of the cosmic ray intensity , or equivalently the descent in the bending efficiency and comcomittant effects on @xmath28 of the other parameters , can not continue indefinitely as the energy increases above the break energies .
in fact , the average value of the magnetic field strength is finite ( say , 3 @xmath30 g ) , and because of this finiteness , at some energy , for a specified ion , trajectories remain completely unbent , and consequently , the bending efficiency vanishes .
this condition corresponds to the rectilinear propagation which is at the origin of the @xmath31 as discussed in section 8 .
the existence of the ankle is incubated in the quantity @xmath28 versus energy shown in figure 4 .
subsequent analysis indicates how the low plateau in figure 4 transforms into a physical @xmath4 accessible to experimentation .
figure 5 shows an idealization of the universal spectrum of the cosmic radiation between @xmath29 and @xmath32 @xmath2 .
a notable feature of this spectrum is its linearity in logarithmic scales .
the ankle is a small perturbation of the differential intensity .
in fact the maximum difference between extrapolated and measured intensity is a factor 8 occurring around @xmath33 ( see figure 20 in paper i and the data in ref.[4 ] ) , while in the knee - ankle energy range the differential intensity falls by a round factor @xmath8 , hence the term @xmath34 @xmath35 . a similar conclusion is reached in ref.[15 ] where other empirical arguments are used . the remarkable linearity shown in figure 5 suggests that individual ions may also have approximate linearities . for logical coherence in the explanation of the ankle this linearity
is converted here in a postulate .
notice that this postulate constitutes a useful @xmath36 @xmath37 in the experimental data below @xmath38 and above @xmath39 for a number of reasons [ 16 ] .
also , the quantitative account of the knee and the ankle given here would demand a much softer hypothesis than this powerful postulate .
for instance , only spectral indices in narrow energy bands , before and close the knee , are really useful in the present analysis .
the absence of this postulate does entail other silent hypotheses ( usually unexpressed in the literature ) .
the rejection of the postulate would entail more complex hypotheses , probably with higher arbitrariness .
the experimental data at low energy indicate that spectral indices occasionally are not constant , as reported , for example , by the atic experiment [ 17 ] for a number of ions between @xmath40 and @xmath41 ( c , mg , o , si and probably k ) but a clear spectral break is not experimentally observed below the knee region as pointed out by others [ 18 ] .
the alteration of @xmath28 caused by the spectral indices does not spoil the existence of the ankle already present in @xmath28 versus energy as a low plateau ( see figure 6 ) .
nor could the particular values of the spectral indices of all ions influence the existence the ankle since the sum of the individual ion spectra , on average , should compensate , in order to recover imperatively , the linearity shown in figure 5 .
how the function @xmath28 is altered by the particular spectral indices is illustrated in figure 7 for helium with an index of 2.72 , and also for other ions .
the indices and elemental abundances used in the present calculation are shown in table 1 , inspired from the results of some experiments ( kascade , jacee , atic , runjob , eas - top and others ) .
figure 8 shows the proton energy spectrum with a spectral index of 2.6 , based on the atic experiment [ 17 ] , at very low energy .
this spectrum is normalized to the flux of @xmath42 @xmath43/@xmath2 @xmath44@xmath45 @xmath46 , measured by kascade ( sibyll ) [ 19,20 ] at the energy of @xmath47 @xmath2 .
the computed spectrum agrees fairly well with the measurements ; taking a softer index ( for example , 2.72 instead of 2.6 ) the agreement with the experimental data becomes excellent .
the energy spectrum obtained by the qjsjet algorithms in kascade gives a similar accord , since the resulting spectrum rigidly shifts in intensity .
note that this shift in intensity is absorbed in the normalization and the accord persists .
the computed proton spectrum seems in disagreement with that measured by the tibetas@xmath48 collaboration [ 21 ] which has a lower energy bend and a softer descent compared to these calculations and to the kascade data .
the computed helium and iron spectra also exhibit a good agreement with the kascade and eas - top data as discussed in detail in paper i and ii .
let us remark here that by altering the values of the ion blends , within plausible empirical limits , the accord between computed and measured spectra persists . .
relative abundances of cosmic ions and spectral indices at the arbitrary energy of @xmath49 @xmath2 adopted in the calculation .
the ion groupings cno , ne - s and fe(17 - 26 ) derive from the traditional data analysis in many experiments .
the composition of the blend 3 , not given in the table , is 19.6 , 35.5 , 14.0 , 12.0 , 17.8 and 0.1 with the same spectral indices of the blend 2 except for the proton index which is 2.72 .
[ cols= " < , < , < , < , < , < , < " , ]
the @xmath50 @xmath51 of the cosmic radiation is the sum of the @xmath52 @xmath53 of the individual ions .
of course , it is assumed that electrons , positrons , antiprotons and other elementary particles have negligible fluxes compared to the global nuclide flux . in figure 9 , the computed energy spectra for the ion blend 2 , in the interval @xmath29-@xmath32 @xmath2 ,
are given .
the tibet [ 22 ] and haverah park data [ 1 ] are reported in figure 9 while those of the akeno and agasa experiments [ 3 ] were shown previously ( figure 12 paper ii ) . the computed spectrum is required to agree with the experimental data only at arbitrary energy point of @xmath54 @xmath2 where the intensity ratio between tibet and kascade experiments is taken 1.4 . since no other constraint is imposed the overall accord is fairly good .
the spectral index of 3.06 for this ion blend 2 in the energy interval @xmath55-@xmath56 @xmath2 is in excellent agreement with the data of the quoted experiments and others not shown , to avoid confusion in data superposition .
the ion blends 1 and 3 give 3.05 and 3.0 , respectively , for the global spectral index @xmath48 .
the extragalactic component of cosmic rays at the earth , if it exists , should cluster between @xmath57 ( proton ankle ) and @xmath58 @xmath2 ( iron ankle ) . this fundamental result ,
derived elsewhere ( paper ii , section 7 ) , is almost independent of the method of calculation . taking advantage of this outcome
two components of the cosmic ray flux at the earth may be disentangled : galactic and extragalactic .
the distinction in two components rests on the initial sites ( milky way or outside ) of the acceleration regions once secondaries ( secondary protons , @xmath59he , be , b , etc . )
generated in the wakes of the primaries may be neglected .
the term @xmath60 would denote a number of possibilities : ( a ) debris of normal galaxies in the cosmic vicinity ; ( b ) reentrant particles overflowing from the milky way ; ( c ) debris of peculiar powerful galaxies ; ( d ) extragalactic particles of very high energy accelerated in space between galaxies .
for the case ( b ) the more appropriate term @xmath61 cosmic rays is used .
the ratio between the galactic and extragalactic intensities at earth , @xmath62/@xmath63 , in the band @xmath57 to @xmath58 @xmath2 depends on the ion blend adopted for the galactic component .
detailed calculations with the blend 2 ( paper ii section 6 ) give the energy spectrum of the extradisc component , peaking around @xmath9 @xmath2 , as shown in figure 9 . at this particular energy ,
the difference between the measured intensity and the computed galactic intensity is regarded , by definition , as an extradisc component . for the data shown in figure 9
the ratio @xmath62/@xmath63 is @xmath64 while the agasa data with the same blend 2 give @xmath62/@xmath63=@xmath65 .
the galactic magnetic field is the basic element ( though insufficient ) for the comprehension of the processes generating the ankle .
the form and strength of the galactic magnetic field with its intrinsic turbulence are known with relatively high precision , adequate for this study .
the adequacy is justified elsewhere ( paper ii appendix b ) .
the average magnetic field strength fixes two milestones along the energy axis related to the alteration of the ion trajectories in the milky way : the ankles occur at the energies where the rectilinear propagation of cosmic ions commences , while the knees are placed at energies where the ion bending efficiency departs from an approximately uniform distribution , typical at lower energies . by a uniform bending
is meant an approximately isotropic distribution in the ion directions in the vicinity of the source .
the average distances travelled by cosmic ions in the milky way ( source vicinity ) have been previously calculated [ 10 ] .
one should notice that the quantity @xmath28 versus energy changes by more than two orders of magnitude from the low to the high plateau ( see figure 4 ) .
this computed gap is close to physical reality because of the accord with the data displayed in figures 8 and 9 .
the upper limits of the measured anisotropies in the arrival direction of cosmic rays at the earth around the knee energy are about @xmath66 , and recent measurements [ 23,24 ] indicate the persistence of this figure even above @xmath12 @xmath2 [ 5 ] .
consequently , the intrinsic mechanisms producing the knee dominate by several orders of magnitude those producing anisotropies .
it follows that the accuracy required for knee and ankle calculations is , by far , less demanding that that required for anisotropy calculations . for a given magnetic field configuration in the galaxy ,
the rectilinear propagation of an ion occurs at a distinctive , particular energy .
ion bending expressed by the curvature radius @xmath67 in a small volume permeated by a uniform constant field obeys the equation @xmath67=@xmath68/@xmath69@xmath70 , where @xmath68 is the ion momentum , @xmath69 the charge and the @xmath70 the average field strength in the region where the ion propagates .
for instance , the effect of the ion bending for protons would cease at @xmath71 , while for iron this energy wlould be @xmath72 in an average magnetic field of 3 @xmath30 g if @xmath67 is 100 @xmath22 , the nominal thickness of the gaseous disc .
these numerical games , though recurrent in the literature and textbooks , are both vague and thoroughly elusive to determine the exact positions of the @xmath31 along the energy axis . in figure 10 ,
the onset of the rectilinear propagation for helium is shown as grammage versus energy .
the rectilinear propagation takes place at @xmath6 @xmath2 .
the lack of a unique @xmath67 for all ions impedes the exact determination of the knee energies with the above equation .
to each ion corresponds a @xmath15 @xmath24 around the earth with an intrinsic dimension [ 10,11 ] dependent on the ion , and the correct estimate of the energy threshold of the rectilinear propagation requires the position of the solar cavity , the disc size and the nuclear cross sections .
the mechanisms generating the ankles are the same as for the knees , but the physical conditions at the ankle energy region are less intricate .
dissecting the phenomena which concur in the formation of the @xmath4 and individual @xmath31 we are finally left with the role of the nuclear cross sections .
the results reported in figures 4,6 and 7 , along with those shown in the figure 11 , suffice to highlight this role .
it is obvious from the spectra shown in figure 7 how the function @xmath28 , appearing in figure 4 , contributes to the formation of the ankle ( the spectral indices are incorporated in figure 7 ) .
imagine an ideal , unreal situation where the gap between the low and high plateau is close to zero . neither a change of slope in the spectrum above @xmath73 @xmath2 nor the ankle would have been observed . in this circumstance ,
the universal spectrum would have been entirely bound and regulated by the spectral indices , unaffected by this unreal compression of @xmath28 , with no @xmath31 at all . on the contrary , with a finite gap between the , low and high plateau ,
the appearence of the ankle is a necessity .
the position and magnitude of the ankle and its necessity may be further appreciated by taking into account : ( 1 ) the difference between the high and low plateau for the same ion ; ( 2 ) the difference between the low and high plateau in different ions ; ( 3 ) the difference between low and high plateau for extragalactic cosmic rays .
the high plateau of any ion is fixed by the average matter thickness ( @xmath74 ) encountered by galactic cosmic rays and by @xmath19 . in the energy region of the low plateau
the physical conditions present in the high plateau region are unchanged , except for lower grammages ( see figure 10 ) which ultimately imply low @xmath28 values . to elucidate the point ( 2 ) let us consider two different ions : he and fe , for example .
since @xmath19(fe ) is higher than @xmath19(he ) , galactic iron suffers severe losses in the disc with respect to he losses .
therefore @xmath28 for iron becomes smaller than that for helium , a result displayed in figure 6 ; similarly for other ion pairs with different @xmath19 .
the low plateau of iron is similar to that of helium , and all low plateaux are almost equal for all ions ( see figure 6 ) .
this is simply due to the exponential function governing nuclear collisions via the attenuation length @xmath75 = @xmath76/@xmath19@xmath20@xmath77 .
since @xmath20 is extremely small , less than 0.008 @xmath78/@xmath79 , the difference in the product @xmath19@xmath20 for different ions is quite modest and the quantities @xmath28 for different ions cluster around a minimum value ( see figure 6 ) . by this argument
it follows that the proton gap must be higher than that of helium and the helium gap higher than that of iron etc .
a quite different result pertains to the extragalactic component for which the grammage is higher than that of the galactic component and the quantities @xmath28 in the energy band of the low plateau largely differ . above the energy of the rectilinear propagation the relative intensities of all extradisc ions have to be in the same ratios as do nuclear cross sections : @xmath19(p)/@xmath19(he ) , @xmath19(fe)/@xmath19(he ) etc .
this aspect is beautifully displayed in figure 11 above @xmath12 @xmath2 for he , fe and protons using the number of nuclear collisions in the disc instead of the companion variable @xmath28 .
the origin of the ankle described here may be further appreciated in comparison with other recent studies [ 26 ] where the ankle is a mere effect caused by the extragalactic component , introduced @xmath80 @xmath81 .
codino a. and plouin f. , 2006 , _
galactic basins of helium and iron around the knee energy _ , infn report , infn / tc-06/05 , 20 - 02 - 2006 , frascati , italy .
an updated version of this report will appear soon in astro - ph archive .
|
( _ to appear in the cris 2006 proceedings _ ) the differential intensity of cosmic radiation shows a sequence of depressions referred to as @xmath0 in a large energy band above @xmath1 @xmath2 .
the global depression entailed in the complete spectrum with respect to the extrapolated intensity based on low energy data amounts to a maximum factor of 8 , occurring at @xmath3 @xmath2 , where flux measurements exhibit a relative minimum , referred to as the @xmath4 .
it is demonstrated by a full simulation of cosmic ray trajectories in the galaxy that the intensity minimum around the ankle energy is primarily due to the nuclear interactions of the cosmic ions with the interstellar matter and to the galactic magnetic field .
@xmath5 signal the onset energies of the rectilinear propagation in the milky way at the earth , being for example , @xmath6 @xmath2 for helium and @xmath7 @xmath2 for iron .
the ankle , in spite of its notable importance at the earth , is a local perturbation of the universal spectrum which , between the knee and the ankle , decreases by a round factor @xmath8 , regaining its unperturbed status above @xmath9 @xmath2 .
|
palomar 6 ( @xmath5 = 17@xmath6 44@xmath7 , @xmath8 = @xmath426@xmath9 13@xmath10 ; @xmath11 = 2.1@xmath9 , @xmath12 = 1.8@xmath9 ; j2000 ) is a globular cluster @xmath1 0.8 kpc from the galactic center and @xmath1 0.2 kpc from the plane ( harris 1996 ) . since it lies near the galactic center ,
it is highly reddened .
malkan ( 1981 ) derived the interstellar reddening for palomar 6 , @xmath3 = 1.4 , based on a reddening - free metallicity index @xmath13 .
is defined to be h@xmath14o + 1.5 co , where h@xmath14o and co are the narrow - band infrared photometric absorption indices from h@xmath14o at 2.0 @xmath15 m and co at 2.4 @xmath15 m .
malkan ( 1981 ) claimed that there is a correlation between @xmath13 and a reddening - free ultraviolet line - blanketing parameter @xmath16 ( zinn 1980 ) with , @xmath17 ortolani , bica , & barbuy ( 1995 ) obtained @xmath3 = 1.33 @xmath18 0.10 based on the @xmath19 color magnitude diagram ( cmd ) .
bica et al .
( 1998 ) also studied the cluster employing intermediate resolution ( 5.4 pixel@xmath20 ) integrated spectra and they derived the smallest value of the interstellar reddening value for the cluster , @xmath3 = 1.25 .
the metallicity of the cluster is rather controversial .
malkan ( 1981 ) obtained a metallicity of [ fe / h ] = @xmath41.30 for palomar 6 using the @xmath13 index , ranking it an intermediate metallicity inner halo cluster .
zinn ( 1985 ) derived [ fe / h ] = @xmath40.74 for palomar 6 by reanalyzing malkan s photometry .
ortolani et al . (
1995 ) estimated [ fe / h ] @xmath1 @xmath40.4 based on the @xmath19 cmd .
minniti ( 1995 ) studied the metallicities of seven reddened clusters including palomar 6 .
he obtained spectra of six red - giant branch ( rgb ) stars in palomar 6 with a 2 resolution covering @xmath21 4700 5400 .
he suggested that the sum of mg @xmath22 5175 , fe @xmath22 5270 , and fe @xmath22 5535 lines ( mg + 2fe ) is the most sensitive to metallicity following faber et al .
( 1985 ) , and he derived [ fe / h ] = + 0.2 @xmath18 0.3 for palomar 6 based on the location of the giants in the mg + 2fe _ vs _ @xmath23 diagram .
more recently , bica et al . ( 1998 ) obtained [ @xmath24 = @xmath40.09 by measuring ca ii triplet @xmath21 8498 , 8542 , and 8662 using the intermediate resolution integrated spectra .
as discussed above , palomar 6 suffers a large interstellar reddening .
the interstellar extinction in the visual passband @xmath25 for palomar 6 is expected to be @xmath1 3.9 4.3 mag , depending on its true interstellar reddening value , and the apparent visual magnitude of the horizontal branch ( hb ) @xmath26 is @xmath1 20 mag ( ortolani et al .
thus , photometric study of palomar 6 in the visual passband is very difficult .
it is also likely that differential reddening , which is commonly detected in heavily reddened clusters , is present .
the ir photometry is , therefore , essential for studying palomar 6 , since it is less sensitive to the interstellar reddening and the differential reddening effect . in this paper
, we explore the ground - based @xmath0 photometry of palomar 6 .
a comparison with @xmath27 photometry of minniti , olszewski , rieke ( 1995 ) is made .
the distance modulus using the @xmath28 magnitude of the red horizontal branch ( rhb ) and the metallicity estimate using the rgb slope of the cluster are discussed .
our observations were carried out using the 4 m telescope at kpno on the night of june 15 1997 under photometric conditions .
the detector was a 256@xmath29256 hgcdte nicmos3 array and the jhk filter system was employed .
the image scale was 0.60 arcsec pixel@xmath20 providing a field of view ( fov ) of 2.56 @xmath29 2.56 arcmin .
the journal of observations is presented in table 1 . during our observations , we dithered the telescope pointing to minimize the effects induced by bad pixels and cosmic ray events . for standard star frames ,
the telescope pointing was dithered with an offset of 37 arcsec between successive exposures .
offsets of 6 and 10 arcsec were applied for palomar 6 object frames .
nicmos3 detectors utilize a hybrid architecture in which each pixel has an associated unit cell which controls the biasing and readout of the pixel .
thus , each pixel is essentially independent of the others and charge bleeding or trailing from saturated pixels is not present .
however , this independence also means that such properties as linearity and dark current can vary from pixel to pixel , and it is necessary to calibrate these effects for optimum scientific performance .
the mean dark current of nicmos3 is known to be @xmath1 2 @xmath30/sec .
however it does not scale simply with the exposure time .
thus we obtained dark frames with the same exposure times as object and standard star frames to calibrate the dark currents .
the non - linearity of the detector sensitivity was also calibrated .
we obtained a series of dome flats with exposure times of 3 , 10 , 15 , 20 , and 30 sec maintaining the same illumination intensity .
we also obtained 1 sec exposures before and after each exposure to monitor possible changes in the illumination .
then the fluxes of each exposure were normalized to that of 1 sec exposures .
the measured fluxes for each exposure were plotted as a function of expected fluxes with linear sensitivity .
then we derived a third - order relation , @xmath31,\ ] ] where @xmath32 and @xmath33 are the corrected flux and the observed flux in adu , respectively .
this relation was then applied to each object frame using the iraf task irlincor .
figure 1 shows the non - linear response curve of nicmos3 detector .
the solid line represents the linear relation and filled circles represent the observed fluxes and open circles represent the corrected fluxes using the relation above .
the sky - flats were generated by median - combining object frames of the individual dithering sequences for each passband .
this procedure essentially removes the stellar light and cosmic ray , and provides good sky - flats for less crowded fields , such as standard star frames .
the sky - flats generated using palomar 6 object frames , however , show residuals of stellar images , indicating that this procedure fails when the field is too crowded . applying
sky - flats generated from the other images taken at different times may cause potential problems .
it is known that there are two predominant sources of infrared sky background , which are essentially independent , both physically and spectrally . at short wavelengths ( in particular @xmath34 passband ) , the sky is dominated by emission lines from oh in the upper atmosphere ( typically 90 km altitude ) .
the strength of these lines can vary over the course of a night ; in addition , upper level winds create inhomogeneity and motion of the airglow . as a result
, the intensity of the background can vary unpredictably during the night . at longer wavelengths , thermal emission from the telescope optics and optically thick telluric lines predominates .
the transition between these two regimes occurs at approximately 2.3 microns , so the background with filters other than @xmath35 or @xmath28 is primarily oh airglow .
we produced master sky - flats , which were median - combined sky - flats of the whole set of standard star frames , rather than using sky - flats from a single dithering sequence .
since the standard star frames were taken over the night in our observation , using these master sky - flats will minimize the risk of the abrupt changes of infrared sky background .
figure 2 shows a comparison between the sky - flat generated using the palomar 6 frame ( upper left ) and our master sky - flat ( upper right ) in the @xmath36 passband . as can be seen , the sky - flat using the palomar 6 frames shows residuals of stellar light .
this causes serious problems in the central part of the cluster , where the field is more crowded .
the flat - field corrected images are also shown in the figure .
the lower left panel shows the final image using the sky - flat from the palomar 6 frames and the lower right panel shows that of using the master sky - flat .
the former does not resolve fainter stars very well and even brighter stars were turned out to be affected by the residual flux on the sky flat in the central part of the cluster . therefore , we used the master sky - flat in our analysis .
it is known that there are several hundred bad pixels in the nicmos3 detector used in our observation .
the bad pixel correction was applied after the flat - field correction step .
the bad pixel mask was generated using the standard star frames .
the flat - field corrected standard star frames ( @xmath1 100 frames ) were normalized and then median combined . since the strict definition of a bad pixel is rather subjective except for the non - responsive ( dead ) pixels , the pixels deviating more than 7 @xmath37 around the mean value (= 1.0 ) were defined to be bad pixels .
the pixel values at the bad pixel position were interpolated using those of nearby pixels .
finally , the object frames in each dithering sequence were combined .
our combined image of palomar 6 in @xmath28 passband is shown in figure 3 . during the night , we observed the six ukirt faint standard stars ( casali & hawarden 1992 ) .
it should be noted that the number of ir standard stars is small and the most of the ir standard stars by elias et al .
( 1982 ) are too bright for our camera plus telescope configuration .
all standard frames were analyzed using the photometry task in daophotii ( stetson 1995 ) . with the results of aperture photometry , growth curve analysis was performed using daogrow ( stetson 1990 ) to obtain integrated magnitudes .
the extinction coefficients were derived from one standard star ( fs25 ) .
the first - order @xmath28 passband extinction term @xmath38 is 0.101 mag airmass@xmath20 and the first - order color extinction terms @xmath39 and @xmath40 are @xmath40.004 and 0.045 mag airmass@xmath20 , respectively . to calibrate the photometry , the photometric transformations were assumed to have the following form , @xmath41 the transformation coefficients are @xmath42 , @xmath43 , @xmath44 , @xmath45 , and @xmath46 in the ukirt @xmath0 system .
( the errors are those of the mean . )
figure 4 shows the residuals of @xmath47 , @xmath48 , and @xmath49 using these transformation coefficients .
the mean values are 0.000 @xmath18 0.010 in the @xmath47 passband , 0.000 @xmath18 0.010 in the @xmath48 color , and 0.000 @xmath18 0.007 in the @xmath49 color .
( the errors are those of the mean . )
point - spread function ( psf ) photometry for all palomar 6 frames was performed using daophotii and allstar / allframe ( stetson 1987 , 1994 , 1995 ; turner 1995 ) .
the first step was to perform the aperture photometry with star lists returned from the find routine .
we calculated psfs using isolated bright stars . to obtain good psfs on each frame ,
at least 3 iterations of neighboring - star removal were performed . for the psf calculations
, we adopted a spatially invariable psf since , unlike modern ccds , a nicmos3 array has a small fov , so that the stellar radial profiles are not expected to show large variations over the frame .
also , the number of psf stars is too small ( usually 20 - 30 stars ) to calculate reliable variable psfs in each frame . in case of 2k ccd system ,
the number of psf stars are large enough ( for example 90 - 110 stars in lee & carney 1999 ) to calculate a variable psf in each frame .
after the first allframe pass , we undertook an additional pass to find more stars on the subtracted frames .
since we used a small psf radius ( usually , @xmath50fwhm ) to compute the psf magnitude due to crowding , a psf magnitude must be converted into an aperture magnitude with a larger radius , a process known as _ aperture correction_. the 20 to 30 psf stars on each frame were used for this purpose .
comparisons of the psf magnitude and the aperture magnitude from the growth - curve method yielded aperture corrections . finally airmass corrections and photometric transformations were applied . to convert the ukirt @xmath0 system to the cit @xmath0 system
, we used the following transformation equations given by casali & hawarden ( 1992 ) , @xmath51
our composite cmds of palomar 6 are presented in figure 5 and the sample cmd data are shown in table 2 . in the table , the positions ( columns 2 and 3 ) are presented in pixel units ( 0.60 arcsec pixel@xmath20 ) .
the cmd data are available upon request to the authors or the electronic version of the journal .
off - cluster field populations were not obtained during our observation , therefore , the field star contamination is not removed in our cmds .
since palomar 6 lies near the galactic center , the off - cluster field contamination is expected to be very high ( see below ) , and probably variable due to differential reddening .
our photometric measurements range from the rgb - tip to @xmath1 2 mag below the rhb .
the most prominent feature in our cmds is a well - defined rhb morphology at @xmath52 13.5 mag and @xmath53 13.3 mag .
the scatter in the rgb sequences appear to be larger in ( @xmath54 ) than in ( @xmath55 ) .
this is most likely due to the bright sky background in the @xmath28 passband .
minniti , olszewski , & rieke ( 1995 ) obtained @xmath27 photometry of palomar 6 using the 2.3 m telescope equipped with a 256@xmath29256 hgcdte nicmos3 array at the steward observatory .
their image scale was same as ours ( 0.60 arcsec pixel@xmath20 ) , but their observations covered wider area ( [email protected] arcmin ) . to calibrate their photometry , they observed the standard stars of elias et al .
since standard stars of elias et al .
are too bright for their telescope plus camera configuration , they defocused the telescope to prevent the saturation of bright stars on the detector .
they also observed the globular cluster m22 during the same night and they made comparisons with that of frogel , persson , & cohen ( 1983 ) ; @xmath56 where @xmath57 and @xmath58 refer to minniti et al . minus frogel et al .
( the error is 1 @xmath37 level . )
they noted that the differences are large because their common stars are typically bright and close to the nonlinear regime of the detector array .
we compare our photometry with that of minniti et al . in figure 6 .
the differences are in the sense our photometry minus that of minniti et al .
we compared 134 common bright stars ( @xmath28 @xmath59 12 mag in our magnitude scale ) , and obtained unexpected results .
\(i ) on average , our @xmath28 magnitudes are 1.384 @xmath18 0.024 mag brighter than those of minniti et al .
( the error is that of the mean . )
there exists a gradient in the difference with a slope of @xmath60 @xmath61 0.099@xmath62 , in the sense that our @xmath28 magnitudes are slightly brighter than the mean value for the bright stars .
\(ii ) on average , our @xmath36 magnitudes are 0.705 @xmath18 0.021 mag brighter than those of minniti et al
. there also exists a gradient in the difference with a slope of @xmath63 , in the sense that our @xmath36 magnitudes for the bright stars are slightly brighter than the mean value .
\(iii ) our ( @xmath54 ) colors are 0.679 @xmath18 0.017 mag redder than those of minniti et al . the discrepancies between our work and minniti et al .
are too large to be fully explained by the differences between minniti et al . and
frogel et al . in m22 .
we also show cmds using common stars between our work and minniti et al . in figure 7 .
in the figure , the left panel represents our photometry and right panel represents that of minniti et al.[multiblock footnote omitted ] the off - cluster field star contamination is expected to be very high toward palomar 6 , as discussed in the previous section , and it is necessary to know membership rgb stars in our study . on 23 may 1998 , we obtained high resolution ( @xmath64 40,000 ) ir echelle spectra of 7 bright rgb stars in palomar 6 and arcturus for the @xmath65co 2 - 0 bandhead centered at @xmath1 22045 using the 3.5 m nasa infrared telescope facilities . in order to derive the heliocentric radial velocities ,
we cross - correlated our target spectra with that of arcturus ( @xmath66 = @xmath45.2 km sec@xmath20 , evans 1967 ) .
table 3 shows our heliocentric radial velocity measurements for palomar 6 rgb stars ( see also figure 3 ) .
the third column of the table shows the membership of rgb stars based on their heliocentric radial velocities .
( hereafter , we refer stars a , c , d , and g as membership rgb stars . )
it should be noted that the previous radial velocity of palomar 6 is based on the low resolution spectroscopic study of minniti ( 1995 ) , where the internal measurement error is much larger than ours .
we will discuss the spectroscopic study of palomar 6 rgb stars in a forthcoming paper ( lee , carney , & balanchandran 2002 , in preparation ) . in figure 8
, we show @xmath67 multi - color cmds for palomar 6 , combining our @xmath0 photometry and @xmath68 photometry of ortolani et al .
since we already knew the four cluster membership rgb stars ( filled circles in the figure ) , we drew rgb / rhb fiducial sequences with arbitrary widths in the ( @xmath69 , @xmath68 ) cmd by eye in the extension of the four membership rgb stars .
we then removed outliers by cross - examining @xmath70 and @xmath54 colors .
the remaining stars ( likely the cluster membership stars ) are plotted by crosses in the figure .
our ( @xmath69 , @xmath28 ) and ( @xmath69 , @xmath68 ) cmds may show that our @xmath0 photometry shown in figure 5 suffers a serious field star contamination , which is an expected result since palomar 6 lies near the galactic center .
the rgb stars are clearly separated into at least 3 distinctive branches in ( @xmath69 , @xmath28 ) and ( @xmath69 , @xmath68 ) cmds , in which the cluster s rgb stars are located to the blue indicating that palomar 6 is more metal - poor or less heavily reddened than the surrounding fields . in figure 9 , we show ( @xmath71 , @xmath68 ) cmds for 47 tuc ( kaluzny et al .
1998 ) and palomar 6 ( ortolani et al . 1995 ) . for the 47 tuc cmd , we show a model isochrone for [ fe / h ] = @xmath40.83 and [ @xmath5/fe ] = + 0.30 ( bergbusch & vandenberg 2001 ) and the location of rhb stars in 47 tuc ( see also figure 34 of vandenberg 2000 ) . for
the palomar 6 cmd , filled circles are for palomar 6 membership rgb stars , crosses for stars selected from multi - color cmds ( see figure 8) , and gray dots for stars within 1 arcmin from the cluster center .
since the model isochrone provides an excellent match with 47 tuc , we adopt this model isochrone as a fiducial sequence for 47 tuc .
we then derived the relative distance modulus and interstellar reddening for palomar 6 with respect to those for 47 tuc by adjusting this model isochrone s magnitude to match with those of four palomar 6 rgb stars ( filled circles ) and the model isochrone s color to match with the color that ortolani et al .
( 1995 ) obtained at the intersection between the rhb and the rgb .
the crossing point between the two dashed lines centered at @xmath68 = 19.7 mag and @xmath71 = 2.7 mag indicates the rhb magnitude obtained by ortolani et al .
the color difference @xmath72 = 1.65 mag is corresponding to @xmath73 = 1.27 mag , assuming @xmath74 , and the magnitude difference @xmath75 = 5.1 mag corresponds to @xmath76 = 1.16 mag , in the sense of palomar 6 minus 47 tuc .
it should be noted that the rhb magnitude reported by ortolani et al .
( 1995 ) appears to be @xmath1 0.4 - 0.5 mag fainter than that expected from 47 tuc ( the closed box at @xmath68 @xmath1 19.2 ) . using these relative distance modulus and interstellar reddening values
, we compare @xmath27 photometry for 47 tuc and palomar 6 in figure 10 .
the filled circles are for the 47 tuc rgb / rhb stars ( frogel et al . 1981 ) and
dotted lines at @xmath28 @xmath1 6.5 mag represent the rgb - tip @xmath28 magnitude for 47 tuc ( ferraro et al .
the colors and the magnitudes of the current work ( a ) and those of minniti et al .
( b ) are shifted by @xmath77 = @xmath40.67 mag and @xmath60 = 1.60 mag , assuming @xmath73 = 1.27 and @xmath76 = 1.16 mag between palomar 6 and 47 tuc . in figure 10-(a ) , we also show the mean magnitude and color of the palomar 6 rhb stars ( see below ) .
our rgb - tip @xmath28 magnitude appears to be consistent with that of 47 tuc , while that of minniti et al .
( 1995 ) appears to be @xmath1 1.5 mag fainter than 47 tuc . a comparison with 2 micron all - sky survey ( 2mass , curtri et al .
2000 ) may also provide an opportunity to assess our photometry , although 2mass may have a potential problem in crowded fields due to a larger pixel scale ( 2 arcsec pixel@xmath20 ) .
figure 11 shows a comparison of our photometry with that of 2mass . since 2mass employed the @xmath78 passband , we converted our magnitudes and colors into those of the 2mass photometric system using the transformation relations given by carpenter ( 2001 ) .
the results are @xmath79 where the difference are in the sense our work minus 2mass in the 2mass photometric system .
( the errors are those of the mean . )
the difference in the @xmath80 color is rather large , in the sense that our photometry is @xmath1 0.15 mag bluer than that of 2mass , but the differences are much smaller than 1.4 mag in the @xmath78 passband or 0.7 mag in the @xmath80 color . finally , we observed the four high proper motion stars ( lhs 514 , 2584 , 2824 , and 3094 ) during the same night
. table 4 shows a comparison of the unpublished results of carney and our measurements .
as can be seen in the table , our measurements are in good agreement with that of carney to within 0.1 mag .
therefore we conclude that the discrepancy between our work and that of minniti et al .
( 1995 ) is most likely due to the inaccurate photometric measurement by minniti et al .
( 1995 ) .
we derive the distance modulus and the interstellar reddening for palomar 6 using the method described by kuchinski et al .
( 1995 ) , who recommended to use the magnitude and the color at the intersection between the rgb and the rhb in ir photometry . in figure 12
, we show the @xmath27 cmd for 47 tuc of frogel et al .
we also show the slope of the rgb and the mean rhb magnitude and color .
kuchinski et al .
( 1995 ) obtained @xmath81 = 0.55 mag and @xmath82 = 12.02 mag at the intersection point between the rgb and rhb for 47 tuc .
following the similar method , we derived these values for palomar 6 , using the possible membership stars obtained from figure 8 .
first , we derive the slope of the palomar 6 rgb stars @xmath83 we then define a rectangle whose sides are parallel to @xmath54 and @xmath28 by eye and calculate the mean color and magnitude .
we obtain @xmath84 = 1.09 @xmath18 0.04 mag and @xmath85 = 13.53 @xmath18 0.15 mag for palomar 6 .
( the error is 1 @xmath37 level . ) in figure 12 , we show the slope of the rgb , the rectangle that we adopted and the mean rhb color and magnitude for palomar 6 .
we obtain @xmath81 = 1.19 mag for palomar 6 using equation [ pal6slope ] . by comparing the magnitudes @xmath82 and the colors @xmath81 at the rgb / rhb intersection points between 47 tuc and palomar 6
, we obtained @xmath73 = 1.21 mag and @xmath76 = 1.09 mag , assuming @xmath86 and @xmath87 ( rieke & lebofsky 1985 ) . using the distance modulus @xmath2 = 13.25 mag and the interstellar reddening @xmath3 = 0.04 mag for 47 tuc ( harris 1996 )
, we obtained the interstellar reddening @xmath3 = 1.25 mag and the distance modulus @xmath2 = 14.34 mag for palomar 6 .
we also derive these values using the @xmath88 cmd .
we obtained @xmath89 = 0.52 mag and @xmath90 = 12.10 mag for 47 tuc , and @xmath89 = 0.95 mag and @xmath90 = 13.77 mag for palomar 6 . assuming @xmath91 and @xmath92 ( rieke & lebofsky 1985 )
, we obtained @xmath3 = 1.34 mag and @xmath2 = 14.22 mag for palomar 6 . for our study
, we adopt the mean values , @xmath3 = 1.30 mag and @xmath2 = 14.28 mag for palomar 6 .
our interstellar reddening estimate for palomar 6 shows good agreement with previous values .
however , our distance modulus estimate is @xmath1 0.5 mag smaller than that of ortolani et al .
( 1995 ) who obtained @xmath2 = 14.76 using the hb magnitude shown in figure 8 . using our distance modulus ,
the distance of palomar 6 from the sun is @xmath1 7.2 kpc .
the galactocentric distance @xmath93 becomes 0.9 kpc if @xmath94 = 8.0 kpc ( reid 1993 ) .
palomar 6 is clearly located in the galaxy s central regions .
finally , a caution is advised for using rhb magnitudes to estimate the distance modulus of the metal - rich globular clusters .
sandage ( 1990 ) studied the vertical height of the horizontal branch of the globular clusters as a function of metallicity .
his study suggested that the vertical height of the hb increases with metallicity .
for example , the vertical height of the 47 tuc rhb is @xmath1 0.7 mag in the visual passband .
palomar 6 is expected to have the similar value of the rhb vertical height ( see his figure 16 ) . in our cmds , the palomar 6 rhb vertical height is @xmath1 0.5 - 0.6 mag in the @xmath28 passband . it should also be noted that the @xmath95 magnitude of field red clump stars in the galactic bulge is @xmath1 13 mag ( alves 2000 ) , having similar magnitude to palomar 6 rhb stars depending on the interstellar reddening toward field red clump stars . thus accidental
inclusion of field red clump stars in our palomar 6 cmd could be high and future photometric study of the off - cluster field would be desirable .
kuchinski et al . (
1995 ) studied the slope of the rgb in the ( @xmath96 ) cmd as a reddening- and distance - independent metallicity index for metal - rich globular cluster systems .
they derived a linear relation between metallicity and the slope of the rgb @xmath97 } = -3.09(\pm 0.90 ) - 24.85(\pm 8.90)\times(rgb~slope ) , \label{rgbslope1}\ ] ] where the slope of the rgb is defined to be @xmath98 .
kuchinski & frogel ( 1995 ) recalculated this linear relation with the addition of two more clusters and they obtained , @xmath97 } = -2.98(\pm 0.70 ) - 23.84(\pm 6.83)\times(rgb~slope ) .
\label{rgbslope2}\ ] ] we derived the slope of rgb of palomar 6 and our result is shown in figure 11 and equation 6 . as kuchinski et al .
( 1995 ) and kuchinski & frogel ( 1995 ) noted , the slope of the rgb does not depend on the interstellar reddening and , therefore , conversion to dereddened colors and magnitudes is not necessary . with our slope for palomar 6 ,
we obtained [ fe / h ] = @xmath41.22 @xmath18 0.18 using equation [ rgbslope1 ] ( kuchinski et al .
1995 ) and @xmath41.19 @xmath18 0.18 using equation [ rgbslope2 ] ( kuchinski & frogel 1995 ) .
it should be noted that our results are from selected palomar 6 rgb stars as discussed in the previous section .
if we use all rgb stars , [ fe / h ] = @xmath41.05 @xmath18 0.19 using equation [ rgbslope1 ] and @xmath41.03 @xmath18 0.19 using equation [ rgbslope2 ] and they are slightly lower than those from selected rgb stars .
= @xmath40.6 ) and ngc 6553 ( [ fe / h ] = @xmath40.3 ) from figure 1 of ferraro et al .
it should be noted that , however , the hb morphology is vulnerable to other effects besides metallicity ( see for example lee , demarque , & zinn 1994 ) . ]
our metallicity estimate for palomar 6 is consistent with that of malkan ( 1981 ) , but does not agree with those of zinn ( 1985 ) , ortolani et al .
( 1995 ) and minniti ( 1995 ) .
the metallicity of palomar 6 will be discussed in the subsequent paper .
we obtained [ fe / h ] of the three rgb stars in palomar 6 using high - resolution ( @xmath64 40,000 ) ir echelle spectra .
our metallicity of palomar 6 using ir spectra is [ fe / h ] = @xmath41.08 @xmath18 0.06 , consistent with those derived from the slope of the rgb .
we have presented the @xmath0 photometry of palomar 6 .
our photometric measurements range from the rgb - tip to @xmath1 2 mag below the rhb .
our photometry does not agree with that of minniti et al .
( 1995 ) and an independent study using the southern telescope facilities would be very desirable in the future . we have discussed the distance modulus and interstellar reddening of palomar 6 by comparing the mean magnitude and color of the rhb stars with respect to those of 47 tuc . our interstellar reddening estimate is consistent with previous values by others , while our distance modulus is slightly smaller than that of ortolani et al .
( 1995 ) , who obtained the distance modulus of palomar 6 by using the similar method in the visual passband .
nevertheless , our study has suggested that palomar 6 is clearly located in the galaxy s central regions .
we have also discussed the metallicity of palomar 6 using the slope of the rgb .
our metallicity estimate is in good agreement with that of malkan ( 1981 ) , but does not agree with those of zinn ( 1985 ) , ortolani et al .
( 1995 ) and minniti et al .
( 1995 ) .
this is part of ph.d .
thesis work of j. -w .
lee at the university of north carolina at chapel hill .
j. -w .
lee thanks dr .
dante minniti for providing his photometric table .
we also thank an anonymous referee for useful comments and a careful review of the paper .
this research was supported by the national aeronautics and space administration ( nasa ) grant number go-07318.04 - 96a from the space telescope science institute , which is operated by the association of universities for research in astronomy ( aura ) , inc .
, under nasa contract nas 5 - 26555 and the national science foundation grants ast96@xmath419381 and ast99@xmath488156 .
support for this work was also provided in part by the creative research initiative program of korean ministry of science and technology .
cccc j & 0.5 @xmath29 5 sec & 1.0 & 6 , 10 + & 5.0 @xmath29 5 sec & 1.0 & 6 , 10 + h & 0.5 @xmath29 5 sec & 1.0 & 6 , 10 + & 5.0 @xmath29 5 sec & 1.0 & 6 , 10 + k & 0.5 @xmath29 5 sec & 1.0 & 6 , 10 + & 5.0 @xmath29 5 sec & 1.0 & 6 , 10 + rrrrrr 47 & 186.657 & 16.844 & 10.204 & 1.426 & 0.244 + 71 & 237.766 & 25.472 & 10.717 & 1.353 & 0.278 + 4 & 224.406 & 28.821 & 8.015 & 1.441 & 0.279 + 64 & 196.970 & 46.016 & 10.482 & 1.379 & 0.243 + 8 & 258.721 & 49.039 & 8.204 & 1.634 & 0.388 + 69 & 29.972 & 52.933 & 10.506 & 1.565 & 0.334 + 5 & 249.247 & 62.583 & 7.786 & 1.701 & 0.476 + 48 & 165.293 & 69.201 & 10.202 & 1.461 & 0.250 + 31 & 174.410 & 73.902 & 9.614 & 1.462 & 0.257 + 1 & 221.751 & 80.466 & 7.995 & 1.402 & 0.329 +
|
we present @xmath0 photometry of palomar 6 .
our photometric measurements range from the rgb - tip to @xmath1 2 mag below the rhb and our cmds show that palomar 6 appears to have a well - defined rhb population .
the distance modulus and interstellar reddening of the cluster are estimated by comparing the magnitude and color of palomar 6 rhb stars with respect to those of 47 tuc .
we obtain @xmath2 = 14.28 mag and @xmath3 = 1.30 mag for the cluster and our study suggests that palomar 6 is clearly located in the galaxy s central regions .
we also discuss the metallicity of the cluster using the slope of the rgb .
we obtain [ fe / h ] @xmath1 @xmath41.2 for palomar 6 and our metallicity estimate is @xmath1 0.5 - 1.0 dex lower than previous estimates by others .
|
in order to exploit the new generation of massively parallel supercomputers that are becoming available , weather and climate models will require good parallel scalability .
this requirement has driven the development of numerical methods that do not depend on the orthogonal coordinate system and quadrilateral structure of the longitude - latitude grid , whose polar resolution clustering is predicted to lead to a scalability bottleneck .
a significant challenge is to obtain good scalability without sacrificing accuracy ; in particular conservation , balance , and wave propagation are important for accurate modelling of the atmosphere @xcite .
building on earlier work @xcite , @xcite presented a finite volume scheme for the shallow water equations on polygonal meshes .
they start from the continuous shallow water equations in the so - called vector invariant form : @xmath0 where @xmath1 , the geopotential , is equal to the fluid depth times the gravitational acceleration , @xmath2 is the total geopotential at the fluid s upper surface including the contribution from orography , @xmath3 is the velocity , @xmath4 is the mass flux , and @xmath5 .
the @xmath6 symbol is defined by @xmath7 where @xmath8 is the unit vertical vector .
finally , @xmath9 is the potential vorticity ( pv ) , where @xmath10 is the absolute vorticity , with @xmath11 the coriolis parameter and @xmath12 the relative vorticity , and @xmath13 is the pv flux . by the use of a c - grid placement of prognostic variables , and by ensuring that the numerical method mimics key mathematical properties of the continuous governing equations ( hence the term ` mimetic ' )
, the scheme was designed to have good conservation and balance properties .
these good properties were verified in numerical tests on hexagonal and cubed sphere spherical meshes
. however , their scheme has a number of drawbacks .
most seriously , the coriolis operator , whose discrete form is essential to obtaining good geostrophic balance , is numerically inconsistent and fails to converge in the @xmath14 norm @xcite . also ,
although the gradient and divergence operators are consistent , their combination to form the discrete laplacian operator also fails to converge in the @xmath14 norm in some cases .
these inaccuracies are clearly visible in idealized convergence tests , and give rise to marked ` grid imprinting ' for initially symmetrical flows . although they are less conspicuous in more complex flows , they are clearly undesirable .
@xcite ( see also * ? ? ?
* ; * ? ? ?
* ) showed that the same mimetic properties can be obtained using a certain class of mixed finite element method .
the mimetic properties follow from the choice of an appropriate hierarchy of function spaces for the prognostic and diagnostic variables ( e.g. section [ sec_compound ] below ) , which also provides a finite element analogue of the c - grid placement of variables , or a higher - order generalization .
( the use of such a hierarchy goes by various names in the literature , including ` mimetic finite elements ' , ` compatible finite elements ' , and ` finite element exterior calculus ' ; see @xcite for a discussion of the shallow water equation case in the language of exterior calculus . )
importantly , the resulting schemes are numerically consistent . while the mimetic finite element approach appears very attractive , it is not yet clear which particular choice of mesh and function spaces is most suitable .
standard finite element methods use triangular or quadrilateral elements . for the lowest - order mimetic finite element scheme on triangles ,
the dispersion relation for the linearized shallow water equations suffers from extra branches of inertio - gravity waves , which are badly behaved numerical artefacts @xcite , analogous to the problem that occurs on the triangular c - grid @xcite .
higher - order finite element methods also typically exhibit anomalous features in their wave dispersion relations , such as extra branches , frequency gaps , or zero group velocity modes .
some progress has been made in reducing these problems , at least on quadrilateral meshes , through the inclusion of dissipation or modification of the mass matrix ( e.g. * ? ? ?
* ; * ? ? ?
* ) , though the remedies are somewhat heuristic except in the most idealized cases .
finally , coupling to subgrid models of physical processes such as cumulus convection or cloud microphysics may be less straightforward with higher - order elements ( p. lauritzen , pers . comm . ) .
these factors suggest that it may still be worthwhile investigating lowest - order schemes on quadrilateral and hexagonal meshes .
the above arguments raise two related questions .
can the mimetic finite element method inspire a development to fix the inconsistency of the mimetic finite volume method ?
alternatively , can the mimetic finite element method at lowest order be adapted to work on polygonal meshes such as hexagons ?
below we answer the second question by showing that the mimetic finite element method can indeed be adapted .
in fact , from a certain viewpoint the mimetic finite volume and mimetic finite element schemes have very similar mathematical structure .
the notation below is chosen to emphasize this similarity any vector , such as @xmath15 in ( [ uperp ] ) . ] . moreover ,
the similarity is sufficiently strong that much of the code of the mimetic finite volume model of @xcite could be re - used in the model presented below .
this , in turn , facilitates the cleanest possible comparison of the two approaches .
the adaptation of the mimetic finite element method employs two novel features .
the first is the definition of a suitable hierarchy of finite element function spaces on polygonal meshes .
this is achieved by defining compound elements built out of triangular subelements , and is described in section [ sec_compound ] .
the second ingredient is the introduction of a dual family of function spaces that are defined on the dual of the original mesh .
this permits the definition of a spatially averaged mass field that lives in the same function space as the vorticity and potential vorticity fields ; this , in turn , enables the use of an accurate finite volume scheme on the dual mesh for advection of potential vorticity , and keeps the formulation of the finite element model as close as possible to that of the finite volume model .
the scheme described here is suitable for arbitrary two - dimensional polygonal meshes on flat domains or , as used here , curved surfaces approximated by planar facets .
two particular meshes are used to obtain the results in section [ sec_results ] , namely the same variants of the hexagonal - icosahedral mesh and the cubed sphere mesh used by @xcite , in order to facilitate comparison with their results .
coarse - resolutions versions are shown in fig .
[ fig_grids ] .
any polygonal mesh has a corresponding dual mesh .
( we will refer to the original mesh as the ` primal ' mesh where necessary to distinguish it from the dual . )
each primal cell contains one dual vertex ; each dual cell contains one primal vertex ; each primal edge corresponds to one dual edge and these usually cross each other . figure [ fig_grids ] shows both primal and dual edges for the two meshes .
the mimetic properties of the scheme arise from the relationships between the finite element function spaces .
three function spaces are used on the primal mesh ( @xmath16 , @xmath17 , and @xmath18 ) , and three on the dual mesh ( @xmath19 , @xmath20 , and @xmath21 ) . figure [ fig_spaces ]
indicates that @xmath22 ( i.e. @xmath23 ) maps from @xmath16 to @xmath17 and @xmath24 maps from @xmath17 to @xmath18 and @xmath25 ( like @xmath26 and @xmath24 ) can both be defined as intrinsic operations on a curved surface , without reference to @xmath8 or a third dimension . ] .
more precisely , the primal function spaces satisfy the following properties
. * property list 1 * * @xmath27 . *
@xmath28 with @xmath29 .
* @xmath30 . *
@xmath31 , and @xmath32 .
that is , @xmath22 maps onto the kernel of @xmath24 .
the second condition assumes spherical geometry so that there are no lateral boundaries .
the same assumption will be made throughout this paper ; in particular , no boundary terms will arise when integrating by parts in 2d is @xmath33 where @xmath1 is a potential , @xmath34 is a stream function , and @xmath35 is a harmonic vector field , i.e. one satisfying @xmath36 the fourth condition in property list 1 implies that all nondivergent fields @xmath3 can be written as @xmath37 , which rules out the possibility of harmonic vector fields .
this is appropriate for spherical geometry , since there exist no non - zero harmonic vector fields on the sphere .
however , for a doubly period plane , for example , for which a constant vector field is harmonic , we would have to extend the fourth condition to allow for harmonic vector fields .
this issue does not affect any of the discussion below except for the discrete helmholtz decomposition ( section [ deriv2 ] ) , which would only need to be extended in the obvious way to allow for harmonic vector fields . ] . in a similar way , fig .
[ fig_spaces ] indicates that @xmath26 maps from @xmath21 to @xmath20 and @xmath25 maps from @xmath20 to @xmath19 .
more precisely , the dual function spaces satisfy the following properties
. * property list 2 * * @xmath38 . * @xmath39 with @xmath40 .
* @xmath41 .
* @xmath42 , and @xmath43 .
that is , @xmath26 maps onto the kernel of @xmath25 .
as noted earlier , standard finite element schemes in two dimensions typically use triangular or quadrilateral elements .
several families of mixed finite elements that satisfy property list 1 on such meshes are known . however , in order to apply our scheme on more general polygonal meshes we will need to define families of mixed finite elements satisfying property list 1 on those meshes .
one way to do this is to use _
compound elements_. any polygonal element can be subdivided into a number of triangular subelements . a basis function on the polygonal element
can then be defined as a suitable linear combination of basis functions on the subelements .
the allowed linear combinations are determined by the requirement to satisfy property list 1 or 2 ; see below .
the desire to use a dual mesh increases the need for finite element spaces on polygons , and hence for compound elements . only in special cases ( such as the cubed sphere , fig .
[ fig_grids ] ) can both the primal and dual meshes be built of triangles and quadrilaterals ; other cases require higher degree polygons for either the primal or dual mesh ( or both ) . for a triangular primal mesh , @xcite describe a scheme for the construction of a dual hierarchy of function spaces .
the dual mesh elements are compound elements , similar , though not identical , to those used here .
however , their scheme is limited to the case of a triangular primal mesh and a barycentric refinement for the construction of the dual . in a complementary study
, @xcite describes how finite element basis functions satifying property list 1 may be constructed on arbitrary polygonal elements , without the need to divide into subelements , through a process of _ harmonic extension_. for example , let @xmath44 be a basis function for @xmath16 associated with primal vertex @xmath45 .
define @xmath44 to equal @xmath46 at vertex @xmath45 and zero at all other vertices .
next extend @xmath44 harmonically along primal mesh edges ; that is , its second derivative should vanish so that its gradient is constant along each edge
. then extend @xmath44 harmonically into the interior of each element ; that is , solve @xmath47 subject to the dirichlet boundary conditions given by the known values of @xmath44 on element edges . in a similar way ,
let @xmath48 be a basis function for @xmath17 associated with edge @xmath49 .
define the normal component of @xmath48 to be a nonzero constant along edge @xmath49 ( some arbitrary sign convention must be chosen to define the positive direction ) and zero at all other edges .
then extend @xmath48 harmonically into the interior of each element ; that is , solve @xmath50 and @xmath51 subject to the known values of the normal component at element edges , for example by writing @xmath52 , implying @xmath53 and @xmath54 for constants @xmath55 and @xmath56 .
the boundary conditions determine the value of @xmath55 , but not @xmath56 .
however , condition ( [ harmonicv0 ] ) along with the fourth property in list 1 implies that we must choose @xmath57 , so that ( [ harmonicv1b ] ) reduces to @xmath58 for the last function space @xmath18 the basis function associated with cell @xmath59 is defined to be a nonzero constant in cell @xmath59 and zero in all other cells . it
may then be verified that the properties in list 1 do indeed hold for the spaces spanned by these basis functions .
although the harmonic extension approach provides a general method for constructing the lowest order mimetic finite element spaces on polygonal meshes , its drawback is that , except for the simplest element shapes , the basis functions can not be found analytically .
even if they are found numerically , the inner products required for the finite element method can not be computed exactly , either analytically or by numerical quadrature .
here we take inspiration from both @xcite and @xcite to construct spaces of compound finite elements for arbitrary polygonal primal and dual meshes , by a process that might be called _ discrete harmonic extension_. for the function spaces on the primal mesh , in effect , we solve a finite element discretization of ( [ harmonicv0 ] ) , ( [ harmonicv1a ] ) , and ( [ harmonicv1c ] ) on the mesh of triangular subelements in order to construct the compound basis elements for the original polygonal mesh . for this discretization
we use the lowest order mimetic finite element spaces on the triangular subelements , in which @xmath16 comprises continuous piecewise linear elements , @xmath17 comprises the lowest order raviart - thomas elements , and @xmath18 comprises piecewise constant elements ; @xmath60 in standard shorthand .
although only discrete versions of ( [ harmonicv0 ] ) , ( [ harmonicv1a ] ) , and ( [ harmonicv1c ] ) are solved , it may be verified that the properties in list 1 hold exactly .
the basis functions on the triangular subelements are known analytically , and the compound elements are linear combinations of these ; therefore , integrals of products of basis functions , for example to compute entries of a mass matrix , can all be computed exactly .
the resulting compound elements provide a generalization to polygonal meshes of the @xmath60 hierarchy of spaces , so we will refer to them as compound @xmath60 elements . like the non - compound spaces described by @xcite , the expansion coefficients for @xmath16 correspond to mesh vertices , for @xmath17 to edges , and for @xmath18 to cells .
thus , this hierarchy provides a finite element analogue of the polygonal c - grid if we choose to represent velocity in @xmath17 and the mass variable in @xmath18 . the construction of basis elements for the dual spaces proceeds in a very similar way , except that the basis function for @xmath61 is given by @xmath62 the solution of ( [ harmonicv1a ] ) and ( [ harmonicv1c ] ) .
this gives rise to a compound @xmath63 hierarchy of spaces , where n0 refers to the lowest order two - dimensional ndlec elements .
an important detail concerns the number of subelements needed .
it may appear natural to subdivide an @xmath64-gon cell into @xmath64 triangular subelements .
however , it will be necessary to calculate integrals on the overlap between primal and dual elements ( section [ massmatrices ] ) . in order to be able to do this
when the domain is a curved surface approximated by plane triangular subelement facets , both the primal and dual compound element meshes must be built from triangular subelements of the the same supermesh .
to achieve this we divide @xmath64-gon cells ( whether primal or dual ) into @xmath65 subelements ( fig .
[ fig_supermesh ] ) .
it is convenient to normalize the basis functions as follows : @xmath66 here @xmath67 is the unit normal vector to primal edge @xmath49 and @xmath68 is the unit tangent vector to dual edge @xmath49 , with @xmath68 and @xmath67 pointing in the same sense ( i.e. @xmath69 , though they need not be parallel if the dual edges are not orthogonal to the primal edges ) , as in @xcite . the normalization is chosen so that degrees of freedom for fields in @xmath18 and @xmath19 correspond to area integrals of scalars over primal cells and dual cells , respectively , degrees of freedom for a field in @xmath17 correspond to normal fluxes integrated along primal edges , degrees of freedom in @xmath20 correspond to circulations integrated along dual edges , and degrees of freedom for fields in @xmath16 and @xmath21 correspond to nodal values of scalars at primal vertices and dual vertices respectively .
again , this corresponds closely to the framework of @xcite .
@xcite have analyzed the wave dispersion properties for finite element discretizations of the linear shallow water equations using these compound elements . that paper gives explicit expressions for the @xmath17 and @xmath18 compound element basis functions for the cases of a square mesh and a regular hexagonal mesh on a plane .
for more general meshes it is straightforward and convenient to construct the compound element basis functions numerically .
figure [ fig_elements ] shows typical basis elements for the three primal mesh function spaces for quadrilateral and hexagonal cells . ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous . , title="fig:",width=188 ] ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous . , title="fig:",width=188 ] ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous . ,
title="fig:",width=226 ] ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous .
, title="fig:",width=264 ] ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous .
, title="fig:",width=226 ] ; middle row @xmath70 ; bottom row @xmath71 . in the middle row , at subelement edges the normal components of the basis vectors are continuous .
, title="fig:",width=226 ] the fields used in the computation are represented as expansions in terms of these basis elements .
for example , @xmath72 for the prognostic geopotential and velocity fields , and @xmath73 for the relative vorticity field . here the sums are global sums over all basis elements in the relevant spaces . in some cases it will be useful to introduce dual space representations of fields ; these will be indicated by a hat symbol where necessary to distinguish them from the corresponding primal space representations .
for example , @xmath74 the fields @xmath1 and @xmath75 have the same number of degrees of freedom , and it is possible construct a well - conditioned and reversible map between them by demanding that they agree when integrated against any test function in the primal space @xmath18 .
similarly , the fields @xmath3 and @xmath76 have the same number of degrees of freedom , and it is possible construct a well - conditioned and reversible map between them by demanding that they agree when integrated against any test function in the primal space @xmath17 .
it will also be useful to introduce spatially averaged versions of some fields .
for example , @xmath77 here , @xmath78 and @xmath79 have the same number of degrees of freedom , and it is possible construct a well - conditioned and reversible map between them by demanding that they agree when integrated against any test function in the primal space @xmath16 .
@xmath78 or @xmath79 can be obtained from @xmath1 by demanding that they agree when integrated against any test function in @xmath16 ; in effect this provides an averaging operation from @xmath18 to @xmath16 or @xmath19 .
( however , we should not expect to be able to obtain @xmath1 from @xmath78 or @xmath79 , as this would require an un - averaging operation , which will be ill - conditioned if it exists at all . )
it will be convenient to be able to refer to the vector of degrees of freedom for any field . to do this
, we will use the same letter ( with hat , tilde or bar if needed ) but in upper case .
thus , for example , @xmath80 will be the vector of values @xmath81 , @xmath82 will be the vector of values @xmath83 , etc .
finite element schemes solve the governing equations by approximating the solution in the chosen function spaces , written as expansions in terms of basis functions ( e.g. ( [ eqn_exp_v2 ] ) , ( [ eqn_exp_v1 ] ) ) , and demanding that the equations be satisfied in weak form , that is , when multiplied by any test function in the appropriate space and integrated over the domain . in this approach ( [ ctsmass ] ) becomes @xmath84 or , regarding the integral as an inner product for which we introduce angle backet notation , @xmath85 similarly , ( [ ctsvel ] ) becomes @xmath86 or @xmath87 ( the construction of the nonlinear terms @xmath88 , @xmath89 and @xmath90 is discussed in section [ nonlinearswe ] below . )
the method generally leads to a system of algebraic equations for the unknown coefficients in the expansion of the solution .
the following subsections show how the mimetic finite element method can be re - expressed in terms of certain matrix operators acting on the coefficient vectors @xmath80 , @xmath91 , etc .
the notation is chosen to highlight the similarity to the finite volume scheme of @xcite .
the velocity basis elements are constructed and normalized so as to have constant divergence over the cell upwind of the edge where the degree of freedom resides , with area integral equal to @xmath46 , and constant divergence over the cell downwind of this edge , with area integral equal to @xmath92 , with zero velocity and hence zero divergence in all other cells .
thus @xmath93 where @xmath94 is equal to @xmath46 when the normal at edge @xmath49 points out of cell @xmath59 , equal to @xmath92 when the normal at edge @xmath49 points into cell @xmath59 , and is zero otherwise .
we will write @xmath95 for the matrix whose transpose has components @xmath94 .
@xmath95 is called an incidence matrix because it describes some aspects of the grid topology .
hence , the divergence @xmath96 of an arbitrary velocity field @xmath3 is @xmath97 equating coefficients of @xmath98 gives @xmath99 or , in matrix - vector notation @xmath100 note we could have demanded that ( [ eqn_div ] ) should hold when integrated against any test function in @xmath18 , to obtain the same result .
however , this would obscure the fact that ( [ eqn_div ] ) actually holds at every point in the domain ( except on cell edges where all terms are discontinuous ) , not just when integrated against a test function . in this sense
, @xmath101 is a _
strong _ derivative operator .
similarly , the basis elements in @xmath16 are constructed so that @xmath102 where @xmath103 is defined to equal @xmath46 if edge @xmath49 is incident on vertex @xmath45 and the unit tangent vector @xmath104 at edge @xmath49 points towards vertex @xmath45 , @xmath92 if it points away from vertex @xmath45 , and zero otherwise
. the unit normal and unit tangent at any edge are related by @xmath105 .
hence , a stream function @xmath34 is related to the corresponding rotational velocity field @xmath3 by @xmath106 equating coefficients of @xmath48 and defining @xmath107 to be the matrix whose entries are @xmath103 gives the matrix - vector form @xmath108 equation ( [ eqn_gradperp ] ) holds pointwise ( again with the exception of discontinuities ) , so @xmath109 is a strong derivative operator .
the matrices @xmath107 and @xmath95 are exactly the same as in the finite volume framework of @xcite . in particular , they have the property that @xmath110 giving a discrete analogue of the continuous property @xmath111 .
analogous relations hold on the dual spaces .
@xmath112 implies that the discrete analogue of @xmath113 is @xmath114 where @xmath115 .
similarly @xmath116 implies that the discrete analogue of @xmath117 is @xmath118 where @xmath119 .
again , these are strong derivative operators . the matrices @xmath120 and @xmath121 have the property @xmath122 giving a discrete analogue in the dual space of the continuous relation @xmath123 .
define the following mass matrices for the primal function spaces : @xmath124 the expressions in parentheses indicate that @xmath125 maps @xmath18 to itself , etc .
( analogous mass matrices may be defined for the dual spaces ; however , they will not be needed here . )
the following matrices are also needed .
@xmath126 for completeness we may also define @xmath127 though we will not need to employ this matrix in the shallow water scheme . one further operator will be needed to construct the kinetic energy per unit mass .
it is @xmath128 where @xmath129 is the area of primal cell @xmath59 .
all of these matrices can be precomputed , so that no quadrature needs to be done at run time .
moreover , they are all sparse , so they can be efficiently stored as lists of stencils and coefficients .
let @xmath15 be the coefficients of the expansion of the projection of @xmath130 into @xmath17 : @xmath131 using ( [ massm ] ) and ( [ operw ] ) gives the discrete version of the @xmath6 operator : @xmath132 demanding agreement between ( [ eqn_exp_v2 ] ) and ( [ eqn_exp_v0 ] ) when integrated against any test function in @xmath18 leads to @xmath133 similarly , demanding agreement between ( [ eqn_exp_v1 ] ) and ( [ eqn_exp_v1 ] ) when integrated against any test function in @xmath17 gives @xmath134 while demanding agreement between ( [ eqn_exp_v0 ] ) and ( [ eqn_exp_v2 ] ) when integrated against any test function in @xmath16 gives @xmath135 the relations ( [ eqn_hodge02 ] ) , ( [ eqn_hodge11 ] ) , ( [ eqn_hodge20 ] ) provide invertible maps between the primal and dual function spaces .
thus , they are examples of discrete hodge star operators ( e.g. * ? ? ?
they may be contrasted with the analogous relations employed by @xcite and @xcite for the finite volume case , which do not involve mass matrices . demanding agreement between ( [ eqn_exp_v2 ] ) , ( [ eqn_exp_v0_bar ] ) , and ( [ eqn_exp_v2_tilde ] ) when integrated against any test function in @xmath16 leads to @xmath136 this is the matrix representation of the averaging operator discussed in section [ sec_compound ] .
a field in @xmath18 is discontinuous , so its gradient in @xmath17 can only be defined in a _ weak _ sense , by integrating against all test functions in @xmath17 .
for example , @xmath137 must be approximated as @xmath138 where @xmath28 , @xmath139 .
expanding both @xmath1 and @xmath140 in terms of basis elements and integrating by parts then leads to the matrix form @xmath141 similarly , the curl of a vector field in @xmath17 must be defined by integration against all test functions in @xmath16 .
for example , the discrete analogue of @xmath142 after expanding in basis functions and integrating by parts , is @xmath143 combining these two results , the discrete analogue of @xmath144 is @xmath145 which is identically zero . these derivative operators can be combined to obtain the laplacian of a scalar . for a scalar @xmath28 ,
the discrete laplacian is @xmath146 . for a scalar @xmath147 ,
the discrete laplacian is @xmath148 .
the operators introduced above lead to a discrete version of the helmholtz decomposition , in which an arbitrary vector field is decomposed into its divergent and rotational parts : @xmath149 figure [ opermap ] summarizes how the operators introduced here map between the different function spaces .
the operators defined above satisfy some key relations that underpin the mimetic properties of the scheme .
we have already seen that @xmath150 and @xmath122 leading to discrete analogues of @xmath111 and @xmath123 .
next , note that the basis elements @xmath44 give a partition of unity , that is @xmath151 at every point in the domain .
consequently @xmath152 and @xmath153 now let @xmath154 by considering the projection of @xmath155 into @xmath17 @xmath156 and integrating by parts and using the matrices defined in sections [ deriv1 ] and [ massmatrices ] , we obtain @xmath157 this identity is key to obtaining the steady geostrophic mode property ( section [ geostrophic ] below ) .
a rough interpretation is that averaging velocities to construct the coriolis terms ( @xmath158 ) then taking their divergence ( @xmath121 ) gives the same result as computing the velocity divergence ( @xmath95 ) followed by averaging to @xmath16 ( @xmath159 ) .
one consequence is @xmath160 .
an identical formula to ( [ trisk ] ) relating @xmath159 and @xmath158 was obtained by @xcite for the finite volume case .
the result was originally derived for the construction of the coriolis terms on orthogonal grids by @xcite , and @xcite showed that it could be embedded in a more general framework applicable to nonorthogonal grids .
moreover , @xcite showed that , for any given @xmath159 with the appropriate stencil ( which we have here ) and satisfying ( [ sumr ] ) , there is a unique antisymmetric @xmath158 satisfying ( [ trisk ] ) , and gave an explicit construction for @xmath158 in terms of @xmath159 .
thus , although the context and interpretation are slightly different here , we can , nevertheless , use the thuburn et al .
construction in implementing the mixed finite - element version of the @xmath158 operator !
now consider the two representations of any vector field @xmath161 , @xmath162 related by @xmath163 so that @xmath164 since @xmath165 , @xmath166 and integrating by parts gives @xmath167 hence @xmath168 finally , substituting from ( [ uuhat ] ) and noting that @xmath82 is arbitrary gives @xmath169 the interpretation of this identity is that , for a velocity field in @xmath20 , taking the curl followed by mapping to the primal space is equivalent to mapping the velocity field to the primal space then taking its curl .
one consequence is that @xmath170 .
finally , let @xmath171 and @xmath172 be two discrete representations of a scalar field related by @xmath173 so that @xmath174 since @xmath175 for any @xmath70 , we have @xmath176 or , using ( [ xxhat ] ) and noting that @xmath177 is arbitrary , @xmath178 using these identities and the hodge star operators , it can be seen that taking a weak derivative in the primal space is equivalent to applying a hodge star to map to the dual space , taking a strong derivative in the dual space , and applying another hodge star to map back to the primal space : @xmath179 @xcite .
thus , certain paths in fig . [ opermap ] commute .
weak derivative operators in the dual space can be defined by demanding a similar equivalence with primal space strong derivatives ; however , the resulting formulas are less elegant and , in any case , will not be needed .
we first examine the spatial discretization of the linear shallow water equations to illustrate how some key conservation and balance properties arise .
the rotating shallow water equations ( [ ctsmass ] ) , ( [ ctsvel ] ) when linearized about a resting basic state with constant geopotential @xmath180 and with constant coriolis parameter @xmath11 become @xmath181 by writing these in weak form ( analogous to ( [ weakmass ] ) and ( [ weakvel ] ) ) , expanding @xmath1 and @xmath3 in terms of basis functions , and using the notation and operators defined above , we obtain @xmath182 mass conservation is trivially satisfied ( for both the linear and nonlinear equations ) because the discrete divergence is a strong operator , so the domain integral of the discrete divergence of any vector field vanishes . for the linearized equations the total energy is given by @xmath183 hence
, the rate of change of total energy is @xmath184 where we have used the fact that @xmath125 and @xmath185 are symmetric , @xmath158 is antisymmetric , and @xmath186 .
the linear shallow water equations support steady non - divergent flows in geostrophic balance . a numerical method must respect this property in order to be able to represent geostrophic balance .
however , it is non - trivial to achieve this property because several ingredients must fall into place .
* the geopotential @xmath1 must be steady .
the steadiness of @xmath1 follows immediately from the assumption that @xmath187 .
* the relative vorticity @xmath188 must be steady ; neither the pressure gradient nor the coriolis term should generate vorticity .
first note that , from property list 1 , @xmath189 for some @xmath190 . taking the curl of the momentum equation
then gives @xmath191 the pressure gradient term does not contribute because @xmath192 , and the coriolis term does not contribute because @xmath193 .
* there must exist a geopotential @xmath1 that balances the coriolis term so that the divergence is steady .
taking the divergence of the momentum equation gives the divergence tendency @xmath194 if we define @xmath195 and use the transpose of ( [ trisk ] ) we find that @xmath196 does indeed vanish ; thus the required @xmath1 does exist .
consequently , the scheme does support steady geostrophic modes for the linearized equations .
( note , it is not necessarily true that any given @xmath1 field can be balanced by some non - divergent velocity field . on some meshes , particularly those with triangular primal cells , there might not be enough velocity degrees of freedom to balance all possible @xmath1 fields . )
a generalization of the steady geostrophic mode property is that the scheme should have a suitable pv equation . in this section
we consider the linear case ; the nonlinear case is dealt with in section [ nonlinearpv ] .
the mass field @xmath1 and the vorticity field @xmath188 live in different spaces . to construct a suitable discrete pv
we need an averaged mass field @xmath78 that lives in the same space as @xmath188 .
the linearized pv should be independent of time .
for this to hold , @xmath188 and @xmath78 must see the same divergence field . for a general ( possibly divergent ) velocity field @xmath3
, the vorticity equation ( [ linvort ] ) becomes @xmath197 define @xmath78 using ( [ eqn_avephi ] ) .
then the evolution of @xmath78 is given by @xmath198 ( using ( [ trisk ] ) ) .
thus , @xmath199 and @xmath200 see the same divergence @xmath201 ; consequently the linearized pv @xmath202 is independent of time .
the nonlinear rotating shallow water equations are ( [ ctsmass ] ) and ( [ ctsvel ] ) . writing these in weak form ( [ weakmass ] ) and ( [ weakvel ] ) , and letting @xmath203 , @xmath204 , and @xmath205 be the vectors of coefficients for the discrete representations of the mass flux @xmath4 , the pv flux @xmath13 , and the kinetic energy per unit mass @xmath206 , the nonlinear discretization becomes @xmath207 the remaining issue is how to construct suitable values of the three nonlinear terms @xmath205 , @xmath203 , and @xmath208 .
the discretization of @xmath90 follows the standard finite element construction , which is to project @xmath209 into @xmath17 .
it may easily be verified that this is equivalent to projecting @xmath90 into @xmath18 before taking the weak gradient . using the @xmath210 operator defined in section [ massmatrices ] ,
the expansion coefficients @xmath205 of the projected @xmath90 are given by @xmath211 because the @xmath1 field is approximated as piecewise constant , its degrees of freedom can be interpreted as primal cell integrals .
similarly , the degrees of freedom of the @xmath3 field are the integrals of the normal velocity fluxes across primal cell edges , and the @xmath95 operator looks exactly like a finite volume divergence operator .
thus , it is straightforward to use a finite volume advection scheme for advection of @xmath1 .
the mass flux is constructed using a forward in time advection scheme , identical to that used by @xcite , using the fluxes @xmath91 and the mass field @xmath80 as input .
we write this symbolically as @xmath212 the subscript @xmath46 indicates that this version of the advection scheme operates on the primal mesh and works with densities or concentrations .
so far we have not needed to use the dual mesh representation of any field . however , in order to use the same finite volume advection scheme as @xcite to compute pv fluxes , we need a piecewise constant representation of the pv field on dual cells , and a representation of the mass flux field in terms of components normal to dual cell edges .
these are naturally given by the dual function spaces : @xmath213 and @xmath214 applying ( [ perp ] ) followed by ( [ eqn_hodge11 ] ) to the mass flux gives @xmath215 now consider the evolution of the dual mass field @xmath216 .
@xmath217 i.e. @xmath218 since @xmath121 acts exactly like a finite volume divergence operator on the dual mesh , @xmath216 behaves exactly as if it were evolving according to a finite volume advection scheme .
next , in order for pv to evolve in a way consistent with the mass field @xmath216 , we construct pv fluxes in @xmath20 using the dual mesh finite volume advection scheme : @xmath219 the subscript 2 indicates that this version of the advection scheme operates on the dual mesh and works with quantities analogous to mixing ratios ( such as pv @xmath220 ) .
finally , these dual mesh pv fluxes are mapped to the primal mesh for use in the momentum equation : @xmath221 it may be verified that the resulting vorticity equation for @xmath222 is indeed analogous to ( [ phibar ] ) , involving the potential vorticity flux @xmath223 . using ( [ eqn_hodge20 ] ) , ( [ xi ] ) , ( [ mudot ] ) and ( [ jdeqdh ] ) , we have @xmath224 hence , @xmath225 which is of the desired form .
the similarity of ( [ dualpv ] ) and ( [ phibar ] ) means that it is possible to construct pv fluxes from the dual mass fluxes @xmath226 such that the evolution of the pv is consistent with the evolution of @xmath227 .
the same time integration scheme as in @xcite is used .
@xmath228 here , @xmath229 indicates the usual ( possibly off - centred ) crank - nicolson approximation to the integral over one time interval : @xmath230 ( for any field @xmath34 ) where @xmath231 .
all results presented below use @xmath232 .
@xmath233 is an approximation to the time integral of the mass flux across primal cell edges computed using the advection scheme .
the velocity field used for the advection is @xmath234 .
we write this symbolically as @xmath235 ( the notation @xmath236 , as distinct from @xmath237 in ( [ adv1 ] ) , indicates that here we are working with time integrals @xmath234 and @xmath233 . ) finally , @xmath238 is an approximation to the time integral of the pv flux across dual edges computed using the advection scheme .
dual grid time integrated mass fluxes are calculated from the primal grid time integrated mass fluxes as @xmath239 these are then used in the dual grid advection scheme to compute the time integrated pv fluxes : @xmath240 the system ( [ timephi ] ) , ( [ timeu ] ) is nonlinear in the unknowns @xmath241 , @xmath242 .
it can be solved efficiently using an incremental method ; this may be viewed as a newton method with an approximate jacobian .
after @xmath243 iterations ( [ timephi ] ) and ( [ timeu ] ) will not be satisfied exactly but will have some residuals @xmath244 , @xmath245 defined by : @xmath246 here @xmath247 and @xmath248 are the approximations after @xmath243 iterations to @xmath241 and @xmath242 and it is understood that these have been used in evaluating @xmath233 , @xmath238 , and @xmath249 .
we then seek updated values @xmath250 that will reduce the residuals , where the increments @xmath251 , @xmath252 satisfy @xmath253 @xmath254 here , @xmath255 is a reference value of @xmath1 ; in the current implementation it is given by @xmath256 interpolated to cell edges . to avoid the appearance of the non - sparse matrix @xmath257 in the helmholtz problem below ,
a sparse approximation @xmath258 has been introduced . the construction of @xmath258 is briefly discussed in the appendix .
eliminating @xmath252 leaves a helmholtz problem for @xmath251 : @xmath259 in the current implementation , the helmholtz problem is solved using a single sweep of a full multigrid algorithm .
this gives sufficient accuracy to avoid harming the convergence rate of the newton iteration .
once @xmath251 is found , @xmath252 is obtained by backsubstitution in ( [ uinc ] ) .
finally , ( [ increment ] ) is used to obtain improved estimates for the unknowns .
testing to date has given satisfactory results with 4 newton iterations .
the algorithm requires the inversion of several of the linear operators represented as matrices above .
the appendix describes how this is done .
the same tests were applied to the finite element shallow water model as were applied to the finite volume model of @xcite .
only a subset of results are shown here to emphasize the differences between the two models .
other aspects are the following .
* * stability . *
all experimentation to date suggests the two models have the same stability limit : with no temporal off - centring ( @xmath232 ) the models are stable for large gravity wave courant numbers and advective courant numbers less than 1 . * * advection . *
the same advection scheme is used in the two models to compute mass , pv , and tracer fluxes on primal and dual meshes . in particular , the models share the consistency between mass and pv , between mass and tracers , and between primal mass and dual mass discussed by @xcite . *
* balance .
* the balance test discussed in section 6.8 of @xcite was repeated for the finite element shallow water model .
the results on both the hexagonal and cubed sphere meshes were very similar to those for the finite volume model and the endgame semi - implicit semi - lagrangian model @xcite , implying that any spurious numerical generation of imbalance is extremely weak . * * computational rossby modes . *
the experiment to test the ability of the scheme on hexagonal meshes to handle grid - scale vorticity features was not repeated here .
however , given the general arguments in @xcite ( see also * ? ? ?
* ) , and the similarities between the numerics of the finite volume and finite element models , the results are expected to be very similar for the finite element model . the discrete laplacian defined in section [ deriv2 ] was applied to the @xmath18 representation of the field @xmath260 on the unit sphere , where @xmath261 is latitude and @xmath262 is longitude , and the @xmath14 and @xmath263 errors computed on different resolution meshes .
the results are shown in table [ tab : laplacian ] .
for the inverses that occur once per time step , 10 jacobi iterations are used . for those that occur at every newton iteration ,
2 jacobi iterations are used taking the solution obtained at the previous newton iteration as the first guess ( or ( [ jacobi_fg ] ) on the first newton iteration ) .
a sparse approximate inverse @xmath258 of the @xmath17 mass matrix is needed for the helmholtz problem . on the hexagonal mesh
it is sufficient to use a diagonal approximation @xmath264 however , on the cubed sphere mesh , whose dual and primal edges are not mutually orthogonal , such a diagonal approximation is less accurate and limits the convergence of the newton iterations
. therefore we use instead an approximation based on a single jacobi iteration towards the inverse of @xmath185 : @xmath265 where @xmath266 is the identity matrix .
this approximate inverse is not diagonal but has the same stencil as @xmath185 itself .
cotter , c. j. and thuburn , j. : a finite element exterior calculus framework for the rotating shallow - water equations , j. comp .
, 257 , 15061526 , 2014 .
danilov , s. : on the utility of triangular c - grid type discretization for numerical modeling of large - scale ocean flows , ocean dyn . , 60 , 13611369 , 2010 .
heikes , r. and randall , d. : numerical integration of the shallow - water equations on a twisted icosahedral grid , part ii : a detailed description of the grid and analysis of numerical accuracy , mon .
weather rev .
, 123 , 18811997 , 1995b
. le roux , d. y. , rostand , v. , and pouliot , b. : analysis of numerically induced oscillations in 2d finite - element shallow - water models .
part i : inertia - gravity waves , siam j. sci .
, 29 , 331360 , 2007 .
lock , s .- j . ,
bitzer , h .- w . ,
coals , a. , gadian , a. , and mobbs , s. : demonstration of a cut - cell representation of 3d orography for studies of atmospheric flows over very steep hills , mon .
weather rev .
, 140 , 411424 .
mcrae , a. t. t. and cotter , c. : energy- and enstrophy - conserving schemes for the shallow - water equations , based on mimetic finite elements , submitted to q. j. r. meteorol .
( available at http://dx.doi.org/10.1002/qj.2291 ) .
ringler , t. d. , thuburn , j. , klemp , j. b. , and skamarock , w. c. : a unified approach to energy conservation and potential vorticity dynamics for arbitrarily structured c - grids , j. comput .
phys . , 229 , 30653090 , 2010 .
thuburn j. , cotter , c. j. , and dubos , t. : a mimetic , semi - implicit , forward - in - time , finite volume shallow water model : comparison of hexagonal icosahedral and cubed sphere grids , geosci . model dev . ,
7 , 909929 , 2014 .
ullrich , p.a .
: understanding the treatment of waves in atmospheric models .
part 1 : the shortest resolved waves of the 1d linearized shallow - water equations , quart .
meteorol . soc . , 140 , 14261440 , 2014 . williamson , d. l. , drake , j. b. , hack , j. j. , jakob , r. , and swartztrauber , p. n. : a standard test set for numerical approximations to the shallow water equations in spherical geometry , j. comput .
phys . , 102 , 211224 , 1992 .
zerroukat , m. , wood , n. , staniforth , a. , white , a. a. , and thuburn , j. : an inherently mass - conserving semi - implicit semi - lagrangian discretization of the shallow water equations on the sphere , quart .
soc . , 135 , 11041116 , 2009 .
|
a new numerical method is presented for solving the rotating shallow water equations on a rotating sphere using quasi - uniform polygonal meshes .
the method uses special families of finite element function spaces to mimic key mathematical properties of the continuous equations and thereby capture several desirable physical properties related to balance and conservation .
the method relies on two novel features .
the first is the use of _ compound finite elements _ to provide suitable finite element spaces on general polygonal meshes .
the second is the use of _ dual finite element spaces _ on the dual of the original mesh , along with suitably defined discrete hodge star operators to map between the primal and dual meshes , enabling the use of a finite volume scheme on the dual mesh to compute potential vorticity fluxes .
the resulting method has the same mimetic properties as a finite volume method presented previously , but is more accurate on a number of standard test cases .
compound finite element , dual finite element , mimetic , shallow water
|
in the special theory of relativity ( srt ) the rotation of the reference frame , contrary to the translation , has an absolute character and can be locally measured by the foucault s pendulum or by the sagnac experiment .
indeed , this peculiarity of rotation , inherited by newtonian physics , is difficult to understand in a relativistic context . as a matter of fact
, many authors who were contrary to srt , had found , in the relativistic approach to rotation , important arguments against the self - consistency of the theory . already in 1909
ehrenfest@xcite pointed out an internal contradiction in srt , applied to the case of a rotating disk ; few years later , in 1913 , sagnac@xcite,@xcite evidenced an apparent contradiction in srt with respect to the experimental data . since those years
, these seminal papers had influenced discussions on the foundations of srt , even if these `` paradoxes '' disappear when a careful analysis is undertaken , following the very axioms of the theory .
however , it may appear surprising that , even after one century , in some papers or in textbooks , misleading or even uncorrect arguments are given to explain the ehrenfest s paradox and the sagnac effect .
for instance , klauber in 1998 @xcite,@xcite proposed a new theory of rotating frames , in order to amend the contradictions which , according to him , appear in srt when applied to rotating frames .
elsewhere@xcite we carefully studied the ehrenfest s paradox , showing that it can be solved on the bases of purely kinematical arguments in srt .
to this end , we adopted a geometrical approach , based on ( i ) a precise definition of the concept of space of the disk , ( ii ) a precise choice of the standard rods used by the observers on the platform .
the space of the disk has been formally identified with what we called the `` relative space '' ; its geometry has been recognized to be non - euclidean and its metric coincides with the one which is found in classic textbooks of relativity , in spite of a shift of the context , that we have stressed .
we adopted the projection technique introduced by cattaneo@xcite , @xcite , @xcite , @xcite , @xcite to approach the problem , since this allows an elegant and straightforward description of the geometry of any reference frame .
even though a global _ isotropic _ @xmath0 splitting of space - time is not possible when we deal with rotating observers , the introduction of the relative space allows well defined procedures for space measurements that can be actually performed globally by the observers in rotating frames and which reduce to the standard measurements in any locally co - moving inertial frames .
here we are going to show that these procedures allow a systematic study of the space measurements , which makes the ehrenfest s paradox disappear .
moreover , we are going to illustrate in a clear way ( with the aid of a pictorial representation ) the kinematical origin of the dilation which leads to the solution of the ehrenfest s paradox . according to us the concept of relative space can help to make rid of lots of misunderstandings , often caused by the lack of proper definitions of some crucial concepts in the theoretical apparatus used to describe these physical problems : indeed , they may appear puzzling or contradictory in srt only when ambiguous entities and procedures are adopted .
according to ehrenfest@xcite , the formulation of the paradox is the following one : let @xmath1 be the radii of the rotating disk , as measured , respectively , by the inertial and rotating observer ; @xmath2 is the constant angular velocity of the disk , as measured in the inertial frame .
the paradox arises when the following contradictory statements are taken into account : \(a ) the circumference of the disk must show a contraction relative to its rest state , @xmath3 , since each element of the circumference moves in its own direction with instantaneous speed @xmath4 .
( b ) if one considers an element of a radius , its instantaneous velocity is perpendicular to its length ; thus , an element of the radius can not show a contraction with respect to the rest state .
therefore @xmath5 on the bases of this contradiction , ehrenfest pointed out the apparent inconsistence of the kinematics of bodies which are rigid , according to the definition of rigidity given by born ( see below ) . + despite the great number of authors who tried to explain the paradox , the most popular attempts of solution can be summarized in this way : * neither the rods along the rim nor the circumference do contract ; neither the rods along the radius nor the radius itself do contract ; as a consequence , the space of the rotating disk is euclidean ; + _ ( see f.i .
tartaglia@xcite , klauber@xcite,@xcite ) _ * both the rods along the rim and the circumference contract ; neither the rods along the radius nor the radius itself do contract ; as a consequence the surface of the disk bends , because of rotation . + _ ( sted - donaldson@xcite , galli@xcite ) _ * rods along the rim do contract , while the circumference does not ; neither the rods along the radius nor the radius itself do contract . as a consequence
the space of the disk is not euclidean : @xmath6 , where @xmath7 is the lorentz factor ; + _ ( einstein@xcite , berenda@xcite , arzelis@xcite , grn@xcite,@xcite , mller @xcite , landau - lifshitz@xcite ) _
let us say few words about why these solution are not completely satisfactory .
+ the approach ( 1 ) is , indeed , the one which is in agreement with the common sense , since people can hardly figure out how non - euclidean features may appear on a disk because of its rotation
. however , to maintain the statement `` nothing happens on a rotating disk '' a remarkable price must be paid . for instance , klauber analysis is based on a deep criticism of the foundations of srt : he claims that srt can not be applied to rotating frames , but it must be deeply amended to take into account the non inertial motion of rotating observers . in particular , he maintains that the `` hypothesis of locality''(see mashhoon@xcite ) , which states the local equivalence of an accelerated observer with a momentarily co - moving inertial observer , is not valid in rotating frames .
accordingly , his approach appears a challenge to the very foundations of relativity , because the hypothesis of locality is one of the most important axioms of theory .
dropping this axiom ( which ultimately is justified by the experimental observations ) forces us to abandon the idea that the theory of relativity can describe the physical world , since , actually , there are no perfectly inertial frames in the real world acceleration '' . ] . on the other hands ,
tartaglia s result can be justified and accepted if and only if a specific choice of the measuring rods is done . however
, this choice is questionable when applied to the case of a rotating disk : tartaglia s measuring rods are not the standard rods of srt . as we shall show in sec .
[ sec : georot ] below , in srt we can locally substitute the light rays for the standard rigid rods ; furthermore , we shall show in sec .
[ sec : lengths ] that a careful study of the effects of an acceleration process explains the kinematical origin of the ehrenfest paradox .
the approach ( 2 ) introduces difficulties in the explanation of the paradox ( for instance the dynamical properties of the rotating disk must be taken into account ) and inconsistencies : a non symmetric deformation with respect to the plane of the non - rotating disk should determine a screw sense in space , thus violating spatial parity _ in a purely kinematical context_. finally , the approach ( 3 ) formally agrees with ours : however , those authors do not define explicitly the geometric context in which measurements are made .
moreover , some of them refer to the `` space of the disk '' as if it were a submanifold or a subspace embedded in space - time : this is not the case , since the lack of time - orthogonality ( see below ) does not allow a global @xmath0 splitting . here
, we are going to show that the relative space is the only extended space , having a clear operational meaning , that can be formally defined as the physical space of the disk .
the physical space - time is a ( pseudo)riemannian manifold @xmath8 , that is a pair @xmath9 , where @xmath10 is a connected 4-dimensional haussdorf manifold and @xmath11 is the metric tensor is endowed with an affine connection compatible with the metric , i.e. the standard levi - civita connection . ] .
let the signature of the manifold be @xmath12 .
suitably differentiability condition , on both @xmath10 and @xmath13 , are assumed .
let @xmath14 be the inertial laboratory frame , in which the platform ( with its measuring apparatus ) rotates with a constant angular velocity @xmath2 .
let @xmath14 be parameterized by a set of coordinates @xmath15 , where @xmath16 is the inertial time of @xmath14 and @xmath17 are the cylindrical spatial coordinates . in this frame ,
let us consider the equations @xmath18 if @xmath19\,,$ ] these equations describe the points of a cylinder with radius @xmath20 , rotating with constant angular velocity @xmath21 .
when @xmath22 is the same for each point of the system , we deal with a rotating disk , whose points have cylindrical coordinates @xmath23 , at the initial time @xmath24 .
+ the world - lines of each point of the disk are time - like helixes ( whose pitch , depending on @xmath21 , is constant ) , wrapping around the cylindrical surface @xmath25 , with @xmath26 $ ] .
these helixes fill , without intersecting , the whole space - time region defined by @xmath27 ; is usually imposed : this simply means that the velocity of the points of the disk can not reach the velocity of light . ]
they constitute a time - like `` congruence '' @xmath28 which defines the rotating frame @xmath29 , at rest with respect to the disk .
+ let us introduce the coordinate transformation @xmath30 this coordinate transformation defines the passage from a chart @xmath31 adapted to the inertial frame @xmath14 to a chart @xmath32 adapted to the rotating frame @xmath29 is a system of coordinates in a subset @xmath33 , these coordinates are said to be `` adapted '' to the physical frame if @xmath34 ( greek indices run from 0 to 3 , latin indices run from 1 to 3 ) .
accordingly , the couple @xmath35 is said to be a `` chart '' adapted to the physical frame .
the subset @xmath36 is the coordinate domain of this chart ; in our case , @xmath36 is defined by the condition @xmath370 , r]$ ] . in what follows
, we shall always refer to this domain , even if it will not be explicitly declared . ] .
in the chart @xmath32 the metric tensor is written in the form@xcite : @xmath38 this is the so called `` born metric '' . in the chart
@xmath39 the time @xmath40 is equal to the coordinate time @xmath16 of the inertial frame @xmath14 . in this way
, we label each event @xmath41 in @xmath29 using the time of a clock at rest in @xmath14 , whose world - line ( a straight line parallel to the time axis ) intersects @xmath41 , and not by means of a clock at rest on the disk .
the transformation ( [ catt ] ) has a galilean character , and this is due to the peculiarity of angular velocity which , contrary to translational velocity , has an absolute value , that can be locally measured . + * remark * the parameterization of the rotating frame @xmath29 by the coordinate @xmath16 of the inertial frame @xmath14 is the only way to synchronize the clocks ( globally ) on the platform : in fact their proper times can not be synchronized by einstein s convention , because of rotation .
we recall here that a physical frame is said to be _ time - orthogonal _ when there exists at least one adapted chart in which @xmath42 : this means that the lines @xmath43 are orthogonal to the 3-manifold @xmath44 .
this is an intrinsic property of the physical frame , i.e. it does not depend on the coordinates used to parameterize the frame . a way to characterize
this feature is the introduction of the `` spatial vortex tensor '' , which is a tensor for coordinate transformation `` internal '' to the physical frame .
the spatial vortex tensor of the rotating frame @xmath29 is not null : hence this is not a time - orthogonal frame . in particular ,
a global clocks synchronization is impossible in these non time - orthogonal frames@xcite . using the metric ( [ born ] ) the line element , in the chart @xmath39 ,
is written in the general axis - symmetric form : @xmath45 we can introduce the local spatial geometry of the disk , which defines the proper spatial line element , on the basis of the local optical geometry . to this end
we can use the radar method@xcite , @xcite .
let @xmath46 be a point in the rotating frame , where a light source , a light absorber and a clock are lodged ; let @xmath47 be a near point where a reflector is lodged .
the world - lines of these points are the time like helices @xmath48 and @xmath49 ( see figure [ fig : radar ] ) .
a light signal is emitted by the source in @xmath46 and propagates along the null world - line @xmath50 toward @xmath47 ; here it is reflected back to @xmath46 ( along the null world - line @xmath51 ) , where it is finally absorbed .
let @xmath52 be the proper time , read by a clock in @xmath46 , between the emission and absorption events : then , according to the radar method , the proper distance between @xmath46 and @xmath47 is defined by @xmath53 now , we are going to parameterize these events , using the coordinates adapted to the rotating frame , in order to obtain the explicit expression of the proper spatial line element .
let @xmath54 and @xmath55 be , respectively , the spatial coordinates of @xmath46 and @xmath47 .
the space - time intervals between the events of emission @xmath56 and reflection @xmath20 , and between the events of reflection @xmath20 and absorption @xmath57 , are null .
hence , by setting @xmath58 in ( [ eq : bornline ] ) , we can solve for @xmath59 , and obtain the two solutions : @xmath60 which correspond to the propagation along the two directions between @xmath46 and @xmath47 . and @xmath51 .
] so , if @xmath61 is the coordinate time of the reflection event , the coordinate times of the emission and absorption events are , respectively , @xmath62 and @xmath63 .
consequently , the coordinate time elapsed between these two events turns out to be @xmath64 the corresponding proper time difference is @xmath65 hence , using ( [ eq : dtaudtprimo ] ) , ( [ eq : deltati ] ) and the definition ( [ eq : spacegeo1 ] ) of the radar spatial proper distance , we obtain @xmath66 explicitly , by inserting the elements of the metric tensor ( [ born ] ) , the proper spatial line element is written as @xmath67 where @xmath68 .
in general this method leads to defining the proper spatial line element in the form : @xmath69 where @xmath70 can be thought of as a `` spatial metric tensor '' . according to eq .
( [ eq : spaceline ] ) the components of this tensor are @xmath71 if we concern with a disk , where the @xmath72 coordinate is constant , we get @xmath73 to complete the description of the rotating frame , it is easy to check that this frame is rigid according to born s definition of rigidity : a body moves rigidly if the spatial distance @xmath74 between neighbouring points of the body , as measured in their successive ( locally inertial ) rest frames , is constant in time . this is the case of the rotating frame @xmath29 , since the spatial metric tensor @xmath70 does not depend on time .
and @xmath49 are the world - lines of the two neighbouring points @xmath46 and @xmath47 in the rotating frame , between which a light signal propagates .
the event @xmath56 corresponds to the emission of the light signal ( at time @xmath75 ) which propagates along the null world - line @xmath50 .
the events @xmath20 corresponds to the reflection of the signal ( at time @xmath76 ) which , after propagating along the null world line @xmath51 , is absorbed at the space - time event @xmath57 ( at time @xmath77).,width=264,height=264 ] in the previous section we have outlined the local spatial geometry of a rotating disk on the bases of the local optical geometry .
namely , when light rays are used , locally , as standard rods , the line element which allows measurements of lengths is given by ( [ eq : spaceline ] ) .
this local spatial geometry is defined at each point of the rotating frame . however , in order to have the possibility of confronting measurements performed at different points in the frame , a procedure to extend all over the disk the local spatial geometry is required .
but this can not be done in a straightforward way , because a rotating frame is not time - orthogonal and hence it is not possible to choose an adapted charted in which , globally , the lines @xmath78 are orthogonal to the 3-manifold @xmath44 ( each of which is described by the metric ( [ eq : metricagamma ] ) ) . in other words , a global foliation , which would lead `` naturally '' to the definition of the space of the disk , is not allowed . nevertheless ,
if we shift the context , from the ill defined notion of space of the disk thought of as a subspace or a submanifold embedded in space - time , to a definition which has a well defined and operational meaning , we are lead to the concept of the `` relative space '' . to this end ,
first of all let us start from a more formal description of the local splitting of space - time that allows us to write the local spatial metric ( [ eq : metricagamma ] ) .
let @xmath79 be a time - like helix , describing the evolution of a point of the disk , and let @xmath41 be an event which belongs to @xmath79 .
we can identify a 1-dimensional vector space @xmath80 ( `` time direction '' ) , which is spanned by the vector tangent to @xmath79 in @xmath41 , and a 3-dimensional vector space @xmath81 ( `` space platform '' ) , normal to @xmath80 ; @xmath81 is endowed with the metric ( [ eq : metricagamma ] ) .
hence the space tangent to @xmath41 has a splitting in the form @xmath82 .
this local splitting has an important property : * * * * _ _ the splitting _
_ @xmath83 _ _ and the spatial metric tensor _ _ @xmath84 are invariant along the lines of congruence @xmath28 , i.e. they are dragged along the world evolution .
this property is strictly related to the rigidity of the rotating frame , namely it depends on the fact that the metric ( [ born ] ) is globally stationary and the metric ( [ eq : metricagamma ] ) is locally static .
in particular , it can be explained in terms of isometries , i.e. killing fields in the submanifold @xmath85 of the space - time(see @xcite ) . as a consequence it is possible to define a one - parameter group of diffeomorphisms with respect to which both the splitting @xmath82 and the space metric tensor @xmath84 are invariant .
the lines of @xmath28 constitute the trajectories of this `` space @xmath86 time isometry '' .
+ this fact suggests a procedure to define an _ extended 3-space _ , which we shall call _ relative space_. naively , it can be thought of as the union of the infinitesimal ( local ) space platforms , but a more rigorous definition can be given : + * definition . *
each element of the relative space is an equivalence class of _ points and of space platforms _ , which verify this equivalence relation : re[relequiv ] : _ `` two points ( two space platforms ) are equivalent if they belong to the same line of the congruence ' ' _ .
that is , the _ relative space _ is the `` quotient space '' of the world tube of the disk , with respect to the equivalence relation re , among points and space platforms belonging to the lines of the congruence @xmath28 .
this definition simply means that the relative space is the manifold whose points _ are _ the lines of the congruence .
our definition emphasizes the role of the space platforms : the reference frame defined ( as above ) by the relative space coincides _ everywhere _ with the local rest frame of the rotating disk .
in other words , the relative space is a formal tool which allows a connection among all the local optical geometries that are defined in the neighbourhood of each point of the space . as a consequence ,
space measurements globally defined in the relative space reduce , immediately , to standard measurements in any local frame , co - moving with the disk . in this way , a natural procedure to make a comparison between observations performed by observers at different points is available .
the physical context in which these distant observations are made , is defined , both from a mathematical and operational point of view , by the relative space .
we want to stress , again , that it is not possible to describe the relative space in terms of space - time foliation , i.e. in the form @xmath87 , where @xmath88 is an appropriate coordinate time , because the space of the disk , as we saw before , is not time - orthogonal .
hence , thinking of the space of the disk as a sub - manifold or a subspace embedded in the space - time is misleading : instead , the space of the disk , defined by the relative space , must be thought of as a quotient space .
if we long for some kind of visualization , we can think of the relative space as the union of the infinitesimal space platforms , each of which is associated , by means of the request of m - orthogonality , to one and only one of lines of the congruence .
after having introduced the relative space and described its geometry , we come back to our original purposes , i.e. the solution of the ehrenfest s paradox .
space measurements are locally performed by a rotating observer by means of the spatial metric tensor @xmath89 : @xmath90 as a consequence , the length of an infinitesimal segment on the rim of the circumference is @xmath91 from ( [ catt])@xmath92 it follows that at fixed coordinate time of the inertial frame @xmath14 , @xmath93 .
consequently , the angle all around the periphery of the disk , measured on it , is equal to @xmath94 : @xmath95 $ ] .
+ hence , the measurement of the circumference on the rim of the disk , performed by the rotating observer , turns out to be : @xmath96 we want to stress that the dilation appearing in ( [ eq : lsigma ] ) has a pure kinematical origin and by no means it ought to be interpreted as a dynamical process , involving the structure of the disk , as some authors claimed in the past@xcite .
the meaning of this dilation will be made clear in next section .
for the observer in the inertial frame the length given in eq .
( [ eq : dlsigma ] ) appears contracted by the standard factor @xmath97 : @xmath98 since @xmath99 we obtain that , in correspondence to the measure of the circumference given in eq .
( [ eq : lsigma ] ) , performed by the rotating observer , the inertial observer measures a length @xmath100 , as expected , since the space of the inertial frame @xmath14 is euclidean . the expression ( [ eq : lsigma ] )
is in agreement with the fact that the space geometry of the disk is not euclidean : a curvature tensor can be computed and it is not null . + * remark 1 .
* in the past , different authors , like berenda@xcite , arzelis@xcite , grn@xcite calculated the curvature of the space of the disk .
it is worthwhile to notice that they did not define the proper geometrical context in which their calculation were performed ; moreover , their calculations do not rely upon the use of a splitting technique : they just computed the components of the curvature tensor of the _ space - time _ which have all spatial indices ( @xmath101 ) , and they referred to it as the _ space _ curvature tensor . on the other hands , we showed that the relative space is the natural context in which space measurements are performed and we computed its curvature ( see @xcite ) using the cattaneo s splitting technique .
nevertheless their results are equal to ours , and this is due to the fact that , for the rotating disk in uniform motion , the physical frame is stationary ; however , things are different for those physical reference frames which lack symmetries , such as the axis - symmetry and the stationarity of the rotating disk . +
we want to clarify , here , the origin of the dilation appearing in ( [ eq : lsigma ] ) , which , as we showed , is ultimately responsible for the solution of the ehrenfest s paradox . in order to do that
, it is useful to analyze what happens to `` standard lengths '' when they undergo an acceleration process .
+ let us consider the world - strip of an infinitesimal piece of the rim of the disk , which is at rest until @xmath24 in the inertial frame @xmath14 ( see figure [ fig : figura ] ) . when @xmath24 , the disk starts being accelerated in such a way that all points of its rim have identical motion , as observed in @xmath14 .
if @xmath102 $ ] is the interval representing the period of time during which acceleration acts , @xmath103 the acceleration distribution of all points of the rim is the same , as observed in the inertial frame @xmath14 . from a pictorial point of view
, this means that the world lines of all points of the rim are congruent ( i.e. superimposable ) . during the acceleration period ,
the disk is not born - rigid although it appears rigid in @xmath14 .
this means that , depending on the simultaneity criterium in the inertial frame , the length of the infinitesimal piece of the rim is always congruent with the starting segment @xmath104 ; in particular , when @xmath105 , it is represented by the segment @xmath106 . on the other hands , from the point of view of the local observer at rest on the rim , whose world line passes through @xmath57 when @xmath24 , the simultaneity criterium is not defined by the family of straight lines parallel to @xmath104 , but it varies at each instant , depending on the velocity ( in @xmath14 ) of the rim itself .
namely , when the acceleration period finishes , the piece of the rim is represented by the segment @xmath107 , in the local co - moving frame associated with @xmath108 . let us put @xmath109 , where @xmath110 is the wavelength of a monochromatic radiation emitted by a source at rest in @xmath14 is infinitesimal , the monochromatic radiation must be chosen in such a way that @xmath110 is very small when compared with the length of the circumference . ] .
the @xmath111-circumference of radius @xmath110 , with center in @xmath108 , whose equation is @xmath112 can be built by considering , in each reference frame , the wavelength of the given radiation , emitted by a source at rest in that frame .
this @xmath111-circumference , which is a hyperbola in the minkowskian plane , intersects the segment @xmath113 in @xmath114 , and we obtain @xmath115 this relation means that the world - strip @xmath116 of a length @xmath110 , at rest on the rim , does not cover entirely the world - strip @xmath117 of this length , as measured in @xmath14 ( see figure [ fig : figura ] ) . from a physical point of view ,
equation ( [ eq : mcirc ] ) shows that each element of the periphery of the disk , of proper length @xmath110 , is stretched during the acceleration period .
this a purely kinematical result of our acceleration program .
however , this result remains valid if one takes into account the interactions among the physical points of the rim ( f.i . those interactions which ensure rigidity in the phase of stationary motion ) .
in particular , during and after the acceleration period , each point of the disk is subject to both radial and tangential stresses ; the former maintain each point on the circumference @xmath118 , while the latter ought to give zero resultant , because of the axial symmetry : _ each point is pulled in the same way by its near points , in both directions_. as a consequence , the elongation of every element of the rim , due to tidal forces experienced during the acceleration period , remains even when acceleration finishes .
+ of a length @xmath119 , at rest on the rim , does not cover entirely the world - strip @xmath117 of this length , as measured in the inertial frame.,width=302,height=302 ] * remark * the arguments given treat the disk as a set of non interacting particles .
the only constraint is that every particle must move along a circular trajectory , with a given law of motion , according to a kinematical definition of rigidity in @xmath14 .
+ from the considerations above , it follows that the dilation which is responsible for the solution of the ehrenfest s paradox has a pure kinematical origin .
the enlargement of the rod ( assumed as a standard rod ) , in the rest frame at the end of acceleration phase , is due to the change of the simultaneity criterium . in figure
[ fig : figura ] this is represented by the change in the slope of the infinitesimal space platforms which are associated , by means of the request of @xmath111-orthogonality , to the lines of the congruence @xmath28 .
we introduced the concept of relative space in order to make clear the fundamental processes of measurements that take place on a rotating platform , or more generally , in a rotating frame .
in particular , space measurements are involved in the solution of the ehrenfest s paradox .
often , in the literature on these subjects , misunderstandings arise , and we believe that they rely on the lack of clear and self consistent definitions of the fundamental concepts used . according to us , the concept of relative space makes clear , in a mathematical way , the physical context in which measurements are performed by rotating observers .
even though a global _ isotropic _ @xmath0 splitting of the space - time is not allowed when dealing with rotating observers , space measurements are defined globally in the relative space , and reduce in a straightforward way to the standard measurements in any local frame , co - moving with the disk . in the relative space
we have been able to outline the solution of the ehrenfest s paradox , evidencing its kinematical origin , which is related to the way of measuring lengths in a rotating platform .
furthermore , the geometry of the space of the disk , which follows from our assumptions , turns out to be non euclidean , according to einstein s early intuition .
the solution of the ehrenfest s paradox , that we outlined in this paper , is strongly dependent both on a proper definition of the physical space of the disk , i.e. the relative space , and on a proper choice of the congruence adopted to perform the measurements in such a space , i.e. the ( local ) optical congruence . in conclusion
, it appears that srt , even when applied to rotating platforms , is self - consistent and does not raise paradoxes , provided that proper definitions of geometrical and kinematical entities are adopted .
we believe that our approach to the study of these apparently paradoxical problems leads to a deeper understandings of the very foundations of the theory and evidences , in a clear way , some operational aspects of the measurement processes involved . + * acknowledgements * the author is grateful to professor guido rizzi , for valuable suggestions and comments .
|
we introduce here the concept of relative space , an extended 3-space which is recognized as the only space having an operational meaning in the study of the space geometry of a rotating disk . accordingly , we illustrate how space measurements are performed in the relative space , and we show that an old - aged puzzling problem , that is the ehrenfest s paradox , is explained in this purely relativistic context .
furthermore , we illustrate the kinematical origin of the tangential dilation which is responsible for the solution of the ehrenfest s paradox .
|
the influential paper of navarro , frenk & white @xcite proposed a universal , broken power - law profile for dark matter halos in cold dark matter ( cdm ) cosmologies .
the dependence of @xmath3 on distance from the halo center was found to be well described by a two - parameter family of distributions , @xmath4 the central density of the nfw profile diverges as @xmath5 , similar to what was found in earlier @xmath6-body studies ( e.g. @xcite ) .
however the small - radius dependence was little more than an ansatz since the relevant scales were barely resolved in the simulations .
a debate ensued as to whether the profiles are indeed universal , and if so , what power of radius describes the dark matter density in the limit @xmath7 .
a number of subsequent studies @xcite found steeper central profiles , e.g. @xmath8 , the `` moore '' profile . at about the same time , the luminous ( stellar ) densities in galaxies were found to be centrally divergent @xcite ; in contrast with cdm halos , the power law index of stellar nuclei varies widely but correlates well with galaxy properties , in the sense that fainter galaxies have steeper cusps @xcite .
while the power - law nature of stellar cusps suggests a gravitational origin , gas dynamics probably also played a role in their formation , including processes like radiative dissipation , star formation , and energy infusion from supernova explosions .
the power - law nature of density stratification in virialized objects arises naturally in the standard cosmological paradigm where the growth of objects proceeds from smallest to largest scales @xcite starting from a featureless spectrum of primordial fluctuations @xmath9 .
building blocks of structure can be identified with peaks in the field of primordial gaussian fluctuations .
the ensemble - averaged `` profile '' of an individual density peak is then a power law with slope related to the tilt @xmath10 of the primordial spectrum on a given scale @xcite .
furthermore , evolution of virialized systems through mergers tends to preserve the power - law slope of the central cusp @xcite .
outside of the central cusp , the profile will be modified by the tidal field of neighboring peaks and by the redistribution of material in unequal - mass mergers .
it was noticed rather quickly following the advent of realistic large - scale cosmological @xmath6-body simulations that after the slope and normalization of the density profiles are taken into account , cusps predicted by the simulations are too concentrated to accommodate dynamical constraints derived from the rotation curves of ( especially low surface brightness ) galaxies @xcite .
better fits were obtained with ad hoc , constant - central - density profiles , e.g. the non - singular `` isothermal sphere '' @xcite .
core sizes are not well constrained but recent model fits to the rotation curve data give @xmath11 kpc for most galaxies ( e.g. @xcite ) . other observations on galaxy scales ( such as strong lensing @xcite )
have led to more equivocal assessments of the compatibility of dark matter models with observations . meanwhile
, the most detailed study to date of resolution effects in the cdm simulations @xcite argues that the innermost resolved cusp slope is marginally steeper than nfw , @xmath12 , and that it exhibits a shallowing trend with decreasing radius .
the cusp problem in cdm cosmologies can be alleviated either by postulating changes in the basic physics , or by adding baryons that couple to the dark matter and modify its profile .
proposed changes in fundamental physics include weakly - interacting particles with large scattering cross - sections @xcite , `` warm '' dark matter @xcite , broken scale - invariance @xcite , or modified gravity @xcite .
standard - physics solutions include outflow of gas during an early phase of galaxy formation @xcite , torques from a barlike perturbation @xcite or infall of a population of condensed objects which disrupt the cusp @xcite .
current notions of galaxy formation are not precise enough to allow testing of most of these ideas . here
we focus on the last mechanism , displacement of dark matter by condensed objects , because the dynamics are straightforward and the predicted level of the effect is quantifiable .
a lower limit on the displaced mass is set by the known masses of compact objects at the centers of galaxies , the supermassive black holes ( sbhs ) @xcite , some of which are known to have been present since @xmath13 @xcite .
regardless of their detailed formation history , these black holes must have grown partly through the coalescence of less - massive black holes during galaxy mergers , displacing and ejecting dark ( and baryonic ) matter .
predicted core sizes are @xmath14 hundreds of parsecs , consistent with the sizes of _ stellar _ cores in moderate- to high - luminosity galaxies , which probably formed in the same way . while it remains to be seen whether this model can explain all of the discrepancies between predicted and observed densities at the centers of halos , particularly in low - surface - brightness and dwarf galaxies , the model is almost certainly relevant to the dark matter distribution within a few tens of parsecs of the sbhs where dark matter annihilation radiation would be peaked
accumulation of baryons at the center of a dark - matter halo will deepen the gravitational potential and steepen the cusp , but the net effect depends on whether the baryons also deposit energy into the dark matter as they fall in . the simplest model ( fig .
1a ) is adiabatic growth of a central point mass due to radial infall of gas @xcite .
let the initial mass of the central lump be zero and its final mass @xmath1 . for simplicity ,
assume that the orbits of the dark matter particles are circles ; if growth of the central lump is slow , the orbits will remain circular due to adiabatic invariance .
conservation of angular momentum implies @xmath15 where @xmath16 and @xmath17 are the initial and final radii of a dark - matter particle , @xmath18 is the dark mass within @xmath16 and @xmath1 is assumed to dominate the potential within @xmath17 .
conservation of dark matter further implies @xmath19 . if the initial dark matter density profile is a power law , @xmath20 , the two conservation laws imply @xmath21 , @xmath22 .
a more realistic , isotropic orbital distribution predicts the same relation between @xmath23 and @xmath24 @xcite . for @xmath25 , the index of the power - law
`` spike ''
so formed lies between @xmath26 and @xmath27 ; the spike extends roughly to the outer radius of influence of the central object , @xmath28 , where @xmath29 is the velocity dispersion of the dark - matter particles in the pre - existing halo . in physical units , @xmath30 where the central mass has been normalized to @xmath31 and the velocity dispersion to @xmath32 km s@xmath33
this model has been widely used to predict the flux of annihilation radiation from dark matter particles around the milky way black hole ; most of the radiation would come from the spike @xcite .
the outcome is very different if the lumps form off - center , or are deposited in the halo following a merger ( fig .
1b ) . the lumps spiral inward due to dynamical friction , transferring their energy into the background and _ lowering _ the central density @xcite .
one well - studied case is the formation of a binary sbh following a merger . the binary first forms at a separation of @xmath34 , with @xmath35 the binary mass .
continued evolution of the binary occurs through the * gravitational slingshot * @xcite : dark matter particles that approach the binary within a distance @xmath36 are kicked out at velocities of @xmath37 .
slingshot ejection occurs at a rate determined by the coupled equations @xmath38 @xcite . here
@xmath39 is the mass ejected by the binary and @xmath3 and @xmath29 refer to the background .
the dimensionless numbers @xmath40 and @xmath41 are functions of @xmath42 and the binary separartion ; in the limit of a `` hard '' binary , @xmath43 , and for @xmath44 , @xmath45 and @xmath46 @xcite .
solving the coupled equations for @xmath47 is not straightforward however since @xmath3 changes with time as the dark matter is ejected .
simple models that take into account this time dependence @xcite give @xmath48 , consistent with what is seen in detailed @xmath6-body simulations @xcite . fig .
2 shows the results of one such simulation @xcite .
the merger was between equal - mass spheroids with steep central cusps , @xmath49 , and point particles representing the two sbhs .
the @xmath50 cusp is gradually converted into an @xmath51 cusp by the gravitational slingshot ; at the end of the simulation , the ejected mass is roughly twice the mass of the binary sbh . ; solid lines show the evolution of @xmath3 starting roughly at the time when a hard binary is formed .
( b ) surface density profiles . ]
the total mass ejected by a point - mass binary depends on how long the binary continues to shrink .
an upper limit in the case of binary sbhs is given by the mass ejected in decaying to a separation of @xmath52 , the distance at which emission of gravitational radiation induces a rapid coalescence .
the ejected mass is @xmath53 with @xmath54 the mass of the smaller black hole and @xmath55 @xcite . for binary sbhs ,
@xmath56 , @xmath57 pc and the total mass ejected in decaying to @xmath52 is @xmath58 . however
a binary may `` stall '' after ejecting less mass if the density in its vicinity drops too low @xcite .
if the number of massive infalling objects exceeds two , heating of the background by binaries can occur indirectly as well : massive particles are given kicks , then lose their kinetic energy to the background as they spiral in and are kicked out again etc . as a crude model of this process , consider the evolution of an ensemble of point particles of mass @xmath59 orbiting in a dark - matter halo with @xmath49 slightly steeper than the cusps found in cdm simulations but easier to treat .
the energy released as one lump spirals in from radius @xmath16 to @xmath17 is @xmath60 .
decay will halt when the lumps form a self - gravitating system of radius @xmath61 with @xmath1 the total baryonic mass .
equating the energy released during infall with the energy of dark matter initially within @xmath62 , the `` core radius , '' gives @xmath63 most of the particles that deposit their energy within @xmath62 will come from radii @xmath64 , implying @xmath65 and a displaced mass of @xmath66 .
evolution will continue as the massive particles form binaries and begin to kick out other massive particles .
assume that ejection occurs via many small kicks , such that almost all of the binding energy so released can find its way into the dark matter as the ejected particles spiral back into the core .
the energy released by a single binary in shrinking to a separation such that its orbital velocity equals the escape velocity from the core is @xmath67 .
if all of the massive particles find themselves in such binaries before their final ejection and if most of their energy is deposited near the center of the halo , the core mass will be @xmath68 e. g. @xmath69 for @xmath70 .
a detailed @xmath6-body simulation illustrates this process ( fig .
the initial model for the dark - matter halo has @xmath8 at small radii and total mass @xmath71 .
the condensed objects were represented by 20 point masses , each weighing 10 times as much as a dark matter particle and with combined mass @xmath72 .
evolution occurs in two stages , as discussed above .
as the orbits of the massive particles initially decay , they transfer energy to the background and lower the central density .
once the massive particles form a bound subsystem , further evolution is driven by the formation and disruption of binaries .
the result is a slower growth of the dark matter core , as massive particles are ejected from the central regions and fall back in . at the end of the simulation
the dark mass ejected from the cusp is @xmath73 . only a single massive binary is left at the end
most of the particles that displaced the dark matter have been ejected from the system .
this model is the dynamical analog of wind - driven outflow models @xcite : in both schemes , the baryons that created the core are no longer in evidence after the core is in place .
central density cusps .
axis scalings are based on an assumed black hole mass of @xmath31 , a halo mass of @xmath74 and a virial radius of @xmath75 kpc .
] is this a realistic model ?
multiple baryonic clumps are often postulated to form via gaseous dissipation in the early stages of galaxy formation ( e.g. @xcite ) but their masses , numbers and degree of central concentration are poorly known . another possible candidate for the lumps
is an early population of black holes , e.g. the remnants of population iii stars ( @xmath76 ) @xcite .
but current theory suggests that only one such black hole forms per mini - halo at a redshift of @xmath77 @xcite , and furthermore the inspiral time for an object of this mass would be of order @xmath78 yr , unless it came attached to a larger mass .
black holes of intermediate mass ( @xmath79 ) are a third possibility but their existence is debated and their time of formation uncertain .
sbhs have been present in at least some spheroids since a redshift of @xmath80 @xcite ; many of the sbhs that we see now should therefore be the products of multiple mergers .
typically those mergers will have been `` minor , '' @xmath81 .
in such an interaction the ejected mass scales with @xmath54 , the mass of the smaller hole ; the total mass ejected in a sequence of such mergers would then depend only on @xmath1 , the final mass . however if a black hole grows via a merger hierarchy of comparably - massive black holes , the effect on the density cusp should be to a certain extent * cumulative * : a merger of two galaxies whose cusps had previously been destroyed by binary black holes , will produce a shallower central profile than a merger between two galaxies with initially steep cusps .
this is verified via an @xmath6-body experiment in fig .
4 . the ejected mass scales both with @xmath1 , the final black hole mass , and @xmath6 , the number of stages in the merger hierarchy . there is some quantitative empirical support for idea that cores are produced from the binding energy released by binary black holes .
the luminous ( stellar ) matter in elliptical galaxies and bulges exhibits steep power - law cusps in faint galaxies and cores in bright galaxies @xcite ; among the `` core '' galaxies , the core mass is of order @xmath0 with @xmath1 the current black hole mass @xcite .
the largest cores observed , in galaxies like m87 , have sizes of @xmath82 kpc ; it seems circumstantially reasonable to suppose that dark - matter cores could be comparable in size .
the persistence of steep cusps in faint elliptical galaxies is puzzling but may indicate that the stellar cusps formed by dissipative processes during or after the last significant merger @xcite ; in any case , presence of a stellar cusp is not necessarily evidence for a dark matter cusp .
sbhs represent an extreme in terms of compactness , but galaxies contain other condensed components that could in principle act like `` black holes '' during mergers , displacing dark matter as they spiral inward .
bulges are an obvious candidate ; with masses @xmath83 times those of sbhs @xcite they could in principle displace a much larger quantity of dark matter . in order for two bulges to act like a `` binary black hole ,
'' they must be compact enough to not overlap appreciably at the separation where they first form a bound subsystem .
this separation is @xmath84 with @xmath29 the dark matter velocity dispersion . requiring this separation to be greater than the bulge scale length @xmath85 gives @xmath86 or @xmath87 , with @xmath88 the bulge velocity dispersion .
thus a `` hot '' bulge is a `` hard '' bulge .
in fact @xmath88 appears to be comparable to , though somewhat larger than , @xmath29 @xcite so it is unlikely that bulges at least as we currently observe them would be effective at displacing dark matter during mergers .
if dark matter consists of massive weakly interacting particles like neutralinos , the particles should self - annihilate producing a potentially observable signal .
the annihilation rate in a spherical halo is @xmath89 with @xmath90 the neutralino mass , @xmath91 the annihilation cross section , and @xmath3 the dark matter density @xcite .
annihilation products include gamma rays , from the decay of neutral pions ; high - energy neutrinos ; and synchrotron emission from high energy positrons and electrons moving in the ambient magnetic field .
since the annihilation flux depends on the squared density of neutralinos , the signal would be greatly enhanced along lines of sight where the dark matter is clumped .
these include the center of the milky way galaxy , where the density in a smooth halo would be maximum , as well as directions that intersect the centers of relic halos orbiting as subclumps in the milky way halo @xcite .
the annihilation signal from the galactic center would be further enhanced if there is a power - law spike associated with the central sbh , as discussed above .
gondolo & silk @xcite considered this case and showed that the flux is increased by @xmath92 orders of magnitude compared with models lacking a central spike .
this prediction has been claimed to be inconsistent with measured fluxes of annihilation products , including synchrotron emission @xcite and neutrinos @xcite ( but see @xcite ) .
the central slope of the milky way s dark matter distribution could have been substantially lowered if the halo experienced a merger after formation of the spike , assuming that the infalling halo also contained a sbh .
at least one such merger , with a mass ratio of @xmath93 or less , is statistically likely to have been experienced by a halo as massive as that of the milky way since a redshift of @xmath94 @xcite .
4 shows density profiles of merged dark matter halos , in simulations where the initial profile of the larger halo was normalized to match the milky way , with a central @xmath95 density cusp @xcite .
the spike is efficiently destroyed by the binary sbh in all of the mergers , greatly reduced the predicted annihilation flux .
cusp destruction may have been avoided in the case of our galaxy @xcite .
little is known about the detailed merger history of the milky way but it has been argued that the ages of stars in the galaxy s `` thick disk '' imply a time of @xmath96 gyr since the last significant merger @xcite . the milky way sbh may have acquired most of its mass during or even after this merger event . however ,
if the sbh grew significantly after a merger that had destroyed a pre - existing dark matter spike , it is unlikely that the new spike would be as steep .
the reason is that the dark matter density would probably be non - singular following the merger ( e.g. fig .
4 ) , and growth of a point mass in a non - singular background produces a shallow spike , @xmath8 @xcite
. the annihilation flux from such a spike would be only slightly greater than the expected background in the direction of the galactic center @xcite . the @xmath6-body simulation illustrated in fig .
3 was carried out by m. hemsendorf , who kindly gave us permission to show his results here . the merger tree shown in fig .
4 was generated from a computer code written by l. verde .
we thank r. jimenez , c. power and l. verde for comments on the manuscript .
this work was supported by nsf grant ast 00 - 71099 and nasa grants nag5 - 6037 and nag5 - 9046 .
j. f. navarro , c. s. frenk , s. d. m. white : astrophys .
j. * 462 * , 563 ( 1996 ) j. dubinski , r. g. carlberg : astrophys .
j. * 378 * , 496 ( 1991 ) t. fukushige , j. makino : astrophys .
j. * 477 * , l9 ( 1997 ) s. ghigna et al .
: mon . not .
. soc . * 300 * , 146 ( 1998 ) b. moore et al .
: astrophys .
j. * 499 * , l5 ( 1998 ) b. moore et al .
: mon . not .
r. astron .
soc . * 310 * , 1147 ( 1999 ) a. klypin , a. v. kravtsov , j. s. bullock , j. r. primack : astrophys .
j. * 554 * , 903 ( 2001 ) l. ferrarese et al .
j. * 108 * , 1598 ( 1994 ) t. lauer et al .
j. * 110 * , 2622 ( 1995 ) k. gebhardt et al .
j. * 112 * , 105 ( 1997 ) w. h. press , p. schechter : astrophys .
j. * 187 * , 425 ( 1974 ) y. hoffman , j. shaham : astrophys . j. * 297 * , 16 ( 1985 ) d. syer , s. d. m. white : mon . not .
. soc . * 293 * , 337 ( 1998 ) e. l. okas : mon . not .
r. astron .
soc . * 311 * , 423 ( 2000 ) j. e. barnes : in iau symp .
186 , galaxy interactions at low and high redshift , ed .
j. e. barnes & d. b. sanders ( san francisco : asp ) , 137 ( 1999 ) w. j. g. de blok , s. s. mcgaugh , j. m. van der hulst : mon . not .
. soc . * 283 * , 18 ( 1996 ) s. s. mcgaugh , w. j. g. de blok : astrophys . j. * 499 * , 41 ( 1998 ) w. j. g. de blok , s. s. mcgaugh , v. c. rubin : astrophys . j. * 122 * , 2396 ( 2001 ) i. t. iliev , p. r. shapiro : astrophys . j. * 546 * , l1 ( 2001 ) r. jimenez , l. verde , s. p.
oh : arxiv : astro - ph/0201352 d. rusin , c. ma : astrophys .
j. , l33 ( 2001 ) c. r. keeton : astrophys .
j. , 46 ( 2001 ) c. power et al .
: arxiv : astro - ph/0201544 e. carlson , m. machacek , l. hall : astrophys .
j. * 398 * , 43 ( 1992 ) d. n. spergel , p. steinhardt : phys .
84 * , 3760 ( 2000 ) j. sommer - larsen , a. dolgov : astrophys .
j. * 551 * , 608 ( 2001 ) m. kamionkowski , a. liddle : phys .
* 84 * , 4525 ( 1999 ) r. h. sanders , s. mcgaugh : ann . rev .
, in press ( 2002 ) n. arkani - hamed et al .
: jhep * 12 * , 10 ( 2000 ) j. f. navarro , v. r. eke , c. s. frenk : mon . not .
r. astron .
soc . * 283 * , l72 ( 1996 ) j. binney , o. gerhard , j. silk : mon . not .
321 * , 471 ( 2001 ) m. d. weinberg , n. katz : astro - ph/0110632 ( 2001 ) d. merritt , f. cruz : astrophys .
j. * 551 * , l41 ( 2001 ) a. el - zant , i. shlosman , & hoffman , y. : astrophys .
j. * 560 * , 636 ( 2001 ) d. merritt , l. ferrarese : ` relationship of black holes and bulges ' . in : _ the central kpc of starbursts and agn _ , ed . by j. h. knapen et al .
( asp , chelsea , michigan , 2002 ) pp . 335 - 362 x. fan et al .
j. * 120 * 1167 ( 2000 ) p.
j. e. peebles : gen .
. grav . * 3 * , 63 ( 1972 ) j. r. ipser , p. sikivie : phys .
d. * 35 * , 3695 ( 1987 ) p. gondolo , j. silk : phys .
lett * 83 * , 1719 ( 1999 ) p.
gondolo : phys .
b * 494 * , 191 ( 2000 ) d. merritt : astrophys . j. * 264 * , 24 ( 1983 ) t. nakano , j. makino : astrophys .
j. * 510 * 155 ( 1999 ) * 560 * , 636 ( 2001 ) p. ullio , h. zhao , m. kamionkowski : phys .
d * 64 * 043504 - 1 ( 2001 ) w. c. saslaw , m. valtonen , s. aarseth : astrophys .
j. * 190 * 253 ( 1974 ) s. mikkola , m. valtonen : mon . not .
. soc . * 259 * 115 ( 1972 ) g. d. quinlan : new astronomy * 1 * 35 ( 1996 ) d. merritt : ` black holes and galaxy evolution ' . in : _
xvth iap meeting dynamics of galaxies : from the early universe to the present , asp conf .
197 _ , ed . by f. combes , g. a. mamon , v. charmandaris ( asp , chelsea , michigan , 2000 ) pp .
221230 m. milosavljevi ' c , d. merritt : astrophys .
j. * 563 * 34 ( 2001 ) m. milosavljevi ' c et al . : in preparation ( 2002 ) v. bromm , p. s. coppi , r. b. larson : astrophys .
j. * 527 * , l5 ( 1999 ) t. abel , g. l. bryan , m. l. norman : science * 295 * , 93 ( 2002 ) m. milosavljevi ' c , d. merritt , a. rest , f. c. van den bosch : mon . not .
r. astron .
soc . * 331 * l51 ( 2002 ) e. serabyn , m. morris : nature * 382 * , 602 ( 1996 ) d. merritt , l. ferrarese : mon . not .
* 320 * , 30 ( 2001 ) l. ferrarese : astro - ph/0203469 l. berstrm , p. ullio , j. h. buckley : astropart .
* 9 * , 137 ( 1998 ) l. bergstrm , j edsj , p. gondolo , p. ullio : phys .
d * 59 * , 043506 ( 1999 ) c. calcneo - roldn , b. moore : phys . rev .
d * 62 * , 123005 ( 1999 ) g. auriemma : astro - ph/0203331 ( 2002 ) g. bertone , g. sigl , j. silk : mon . not .
r. astron .
soc . * 326 * , 799 ( 2001 ) g. bertone , g. sigl , j. silk : astro - ph/0203488 ( 2002 ) d. merritt , m. milosavljevi ' c , l. verde , r. jimenez : phys .
lett . * 88 * , 191301 ( 2002 ) g. wyse , ` the merging history of the milky way disk ' . in : _
galaxy disks and disk galaxies _ , astron . soc .
230 , ed . by j. g. funes , e. m. corsini .
( asp , san francisco 2001 )
|
formation and disruption of dark - matter cusps are reviewed .
accumulation of baryons at the center of a halo can displace the dark matter , converting singular density cusps into low - density cores .
the displaced mass can be of order @xmath0 with @xmath1 the mass of the infalling population .
if @xmath1 is identified with the masses of the black holes currently observed at the centers of bright galaxies , predicted core radii are @xmath2 pc .
other mechanisms , such as early mass outflow , may explain the large dark - matter cores in dwarf and low - surface - brightness galaxies .
predictions of dark matter annihilation radiation from the center of the milky way galaxy are shown to be strongly dependent on the galaxy s merger history .
|
a 3 + 1 decomposition of einstein s equations results in a set of constraint equations , which constrain the gravitational fields at all instants of coordinate time , and a set of evolution equations , which propagate the fields forward in time ( e.g. ) .
the four constraint equations can constrain only a subset of the gravitational fields . therefore , the constraint equations can be solved , for example for the construction of initial data , only after the constrained variables have been separated from freely specifiable ones , and after suitable choices have been made for the latter ( see e.g. @xcite for reviews ) .
the constrained variables are separated from the freely specifiable ones by choosing a decomposition of the constraint equations . the conformal thin - sandwich decomposition @xcite has been particularly popular for the construction of quasiequilibrium data ; it has been used extensively , for example , to model compact binaries containing black holes or neutron stars ( see , e.g. , for reviews ) . in the conformal thin - sandwich formalism , the spatial metric is conformally decomposed into a conformal factor and the conformally related metric , and the extrinsic curvature into its trace and a traceless part . in the so - called extended version @xcite , the freely specifiable variables are the conformally related metric and the trace of the extrinsic curvature together with their time derivatives ( which we may set to zero to construct equilibrium data ) , and the constrained variables are the lapse , the shift , and the conformal factor .
black holes may be constructed within the conformal thin - sandwich formalism by excising the black hole interior , and imposing suitable inner boundary conditions . in particular
, these boundary conditions may be chosen so that the black hole is momentarily isolated , or in equilibrium ( see @xcite , also compare the isolated horizon formalism laid out in @xcite and references therein ) . as discussed in detail in @xcite ,
these geometric conditions lead to boundary conditions on some of the constrained variables in the thin - sandwich formalism , namely the conformal factor and the shift vector .
the boundary condition for the lapse , however , remains arbitrary .
some of the choices in this formalism will clearly have an effect on the physical content of the solution .
we can expect to find equilibrium solutions only if we set the time derivatives of the conformally related metric and the trace of the extrinsic curvature to zero .
also , a conformally flat solution is physically distinct from solutions that are not conformally flat .
the choice of the trace of the extrinsic curvature , or mean curvature , is usually associated with an initial temporal gauge , and the lapse plays a similar role .
it is less clear , then , whether or how the mean curvature and the boundary condition on the lapse affect the solutions . in @xcite
, the authors found that sequences of binary black holes , and in particular their innermost stable circular orbit , do depend on the horizon lapse for their example of a non - maximal slicing , i.e. non - zero mean curvature .
this finding , however , may be an artifact of their particular choice of the mean curvature , namely a superposition of two copies of its analytical value for a single schwarzschild black hole expressed in kerr - schild coordinates ( see ( [ k_ks ] ) below ) , one for each companion in the binary .
as the authors caution , the resulting background geometry then depends on binary separation , making the physical meaning of these sequences somewhat arguable .
we consider a very simple physical system in order to analyze whether , at least in this context , the choice of the mean curvature and the horizon lapse affect the physical content of the solutions .
specifically , we solve the constraint equations in the thin - sandwich decomposition to construct thin shells of non - interacting , isotropic particles in circular orbit about a schwarzschild black hole ( compare @xcite , whose results we generalize to account for the black hole ) .
for one particular choice of the mean curvature and the horizon lapse the equations can be solved analytically ( see appendix [ appa ] ) , and we construct numerical solutions for many others . we find that these different choices have no effect on our solutions when they are expressed in terms of gauge - invariant quantities .
we write the spacetime metric @xmath0 in the form @xmath1 where @xmath2 is the lapse function , @xmath3 the shift vector , and @xmath4 the spatial metric .
we further decompose the latter as @xmath5 where @xmath6 is a conformal factor and @xmath7 a conformally related metric .
we then solve einstein s constraint equations in the conformal thin - sandwich decomposition ( see , as well as @xcite for reviews ) .
specifically , the hamiltonian constraint becomes @xmath8 here @xmath9 is the energy density as measured by a normal observer , @xmath10 , and @xmath11 and @xmath12 are the covariant derivative and the ricci scalar associated with the metric @xmath7 .
we have also split the extrinsic curvature @xmath13 into its trace @xmath14 and a traceless part @xmath15 according to @xmath16 from the evolution equation for the spatial metric we can express @xmath17 as @xmath18 here @xmath19 and @xmath20 , and the conformal killing operator @xmath21 is defined as @xmath22 the momentum constraint can now be written as @xmath23 where @xmath24 is a vector laplacian , and @xmath25 is the mass current as measured by a normal observer . finally , the trace of the evolution equation for @xmath13 , combined with the hamiltonian constraint , results in @xmath26 where @xmath27 the trace of the spatial stress .
the above equations form a set of equations for the lapse @xmath2 , the shift @xmath3 and the conformal factor @xmath6 .
before these equations can be solved , however , we have to make choices for the freely specifiable quantities @xmath7 , @xmath28 , @xmath14 and @xmath29 . for the construction of quasiequilibrium data
it is natural to choose @xmath30 and @xmath31 .
we will also restrict our analysis to spherical symmetry , where we may assume conformal flatness , @xmath32 , without loss of generality . here
@xmath33 is the flat metric in whatever coordinate system .
we will , however , experiment with different choices for @xmath14 ( see equations ( [ k ] ) below ) , as well as with different boundary conditions for the lapse @xmath2 ( see ( [ alphahorbound ] ) below ) . with these choices , and in spherical symmetry ,
the above equations simplify dramatically .
we write the spatial metric as @xmath34 where @xmath35 is the isotropic radial coordinate .
the shift @xmath3 is now purely radial , and we abbreviate @xmath36
. we may evaluate ( [ lbar ] ) to find @xmath37 so that @xmath38 becomes @xmath39 the hamiltonian constraint ( [ psiconstraint ] ) can then be written as [ diffeq ] @xmath40 the momentum constraint ( [ betaconstraint ] ) as @xmath41 and the lapse equation ( [ alphaconstraint ] ) as @xmath42 here we have expressed all quantities in terms of those variables that are used in our code . in the above equations the trace of the extrinsic curvature @xmath14 can still be chosen arbitrarily .
following @xcite we consider two different possibilities , namely maximal slicing [ k ] @xmath43 and kerr - schild slicing @xmath44 here @xmath45 is the areal radius , @xmath46 , and we identify @xmath47 with the black hole s irreducible mass ( see ( [ mirr ] ) below ) .
kerr - schild coordinates are identical to ingoing eddington - finkelstein coordinates . throughout this paper
, we use the subscript @xmath48 to refer to the black hole , and @xmath49 to indicate a property of the shell . at spatial infinity
we impose asymptotic flatness , which results in the boundary conditions @xmath50 as @xmath51 .
we excise the black hole interior inside an isotropic radius @xmath52 and impose the black hole equilibrium boundary conditions of @xcite on the resulting excision surface @xmath53 ( compare the notion of isolated horizons laid out in ) . in particular
, the condition @xmath54 where @xmath55 is the outward - pointing unit normal to the horizon , ensures that this surface corresponds to a marginally trapped surface ( apparent horizon ) , while the condition @xmath56 ensures that the coordinate system tracks the horizon .
( the tangential components of the shift vanish identically in spherical symmetry . ) in our case , ( [ psihorbound ] ) becomes [ inboundcond ] @xmath57 and ( [ betahorbound ] ) @xmath58 as discussed in @xcite , the boundary condition for the lapse @xmath59 is arbitrary .
we will experiment with the dirichlet boundary condition @xmath60 and will compare results for values of @xmath61 ranging from zero to unity in increments of 0.1 .
we consider a spherically symmetric shell of isotropic , non - interacting particles in circular orbit about the black hole ( compare @xcite ) .
the rest energy ( baryon mass ) of a spherically symmetric matter source may be written as @xmath62 where @xmath63 is the lorentz factor between a normal observer @xmath64 and an observer comoving with the fluid @xmath65 .
for an infinitesimally thin shell we can then identify the rest energy density ( baryon density ) as @xmath66 where @xmath67 is the ( isotropic ) radius of the shell . since the particles are non - interacting
, their stress - energy tensor is that of a pressureless fluid ( dust ) : @xmath68 .
the matter sources @xmath69 , @xmath70 and @xmath71 in equations ( [ psiconstraintmod ] ) , ( [ betaconstraintmod ] ) and ( [ alphaconstraintmod ] ) can therefore be expressed as @xmath72 the delta function in these matter sources leads to a discontinuity in the derivatives of the solutions .
we can find the jump in these derivatives by integrating equations ( [ betaconstraint ] ) , ( [ psiconstraint ] ) and ( [ alphaconstraint ] ) from @xmath73 to @xmath74 , which , in the limit @xmath75 , results in the jump conditions @xmath76 since the particles are non - interacting , their 4-velocity @xmath65 obeys the geodesic equation @xmath77 assuming circular orbits with @xmath78 and @xmath79 and , without loss of generality , focussing on a particle in the equatorial plane , we find @xmath80 using the normalization condition @xmath81 , which is equivalent to @xmath82 , we evaluate ( [ geodesicmod ] ) to find the geodesic condition @xmath83 for circular orbits . to compute the christoffel symbols we averaged the derivatives of the gravitational field variables inside and outside the shell ( compare @xcite ) .
we compute the adm and komar masses using the expressions ( see e.g. chap . 7 of @xcite ) @xmath84 and @xmath85 where @xmath86 is the outward pointing unit surface element of a closed surface at infinity . in our case , these expressions reduce to @xmath87 and @xmath88 for all configurations considered in this paper , the two mass expressions are found to be equivalent to within the accuracy of our code when the geodesic equation ( [ geodesiccondition ] ) is used to force the particles into circular orbit ( compare @xcite ) .
this is in agreement with a general theorem about the equality of adm and komar masses established by beig @xcite and ashtekar & magnon - ashtekar @xcite .
we also define the binding energy as @xmath89 either terms of the adm or komar mass .
finally , we compute the black hole s irreducible mass from the area @xmath90 of the apparent horizon , @xmath91
we developed a pseudo - spectral code to solve the differential equations ( [ diffeq ] ) subject to the boundary conditions ( [ infbound ] ) and ( [ inboundcond ] ) as well as the jump conditions ( [ jumpconditions ] ) , using chebyshev polynomials as basis functions ( see @xcite for a recent review of spectral methods ) . equations ( [ betaconstraintmod ] ) and ( [ alphaconstraintmod ] ) can be solved directly , while equation ( [ psiconstraintmod ] ) has to be linearized and then solved iteratively .
one complication arises as a consequence of the jump conditions ( [ jumpconditions ] ) .
representing the solution functions across these jumps as a linear combination of the continuous chebyshev polynomials would result in undesirable gibbs phenomena . to avoid this problem
, we solve the equations in two separate domains inside and outside the shell , each one represented by @xmath92 chebyshev polynomials .
the @xmath92 coefficients can then be determined by evaluating the equation at @xmath92 collocation points in each domain , and the jump conditions ( [ jumpconditions ] ) can then be imposed exactly as matching conditions between the two sets of chebyshev polynomials .
each set of chebyshev polynomials is @xmath93 with @xmath94 and @xmath95 $ ] .
we map the inner region into the interval @xmath96 $ ] with the transformation @xmath97 and the outer region with @xmath98 our computational domain therefore extends to @xmath99 , and we can evaluate the masses ( [ adm ] ) and ( [ komar ] ) exactly .
in addition to the choices for @xmath61 and @xmath14 , our solution depends on the parameters @xmath100 and @xmath67 in ( [ restenergydensity ] ) , and the excision radius @xmath101 . to construct a solution of given black hole mass @xmath102 , we need to iterate over @xmath101 until the resulting irreducible black hole mass ( [ mirr ] ) agrees with the desired black hole mass @xmath103 to within a certain pre - determined tolerance . for a given shell radius
@xmath67 a further iteration is needed to fix the lorentz factor @xmath104 in such a way that the solution satisfies the geodesic condition ( [ geodesiccondition ] ) , and the particles are in circular orbit . in practice , we instead fix @xmath104 and then iterate over @xmath67 until ( [ geodesiccondition ] ) is satisfied . for each case
we start the iteration from the analytical solution for @xmath105 and @xmath106 provided in appendix [ appa ] , and continue until the solution has converged .
specifically , our convergence criterion requires that the relative change between iteration steps in any of the fields is less then @xmath107 at all collocation points .
we tested our program for a number of different known vacuum solutions expressing the schwarzschild geometry in different coordinate systems , as well as an analytical solution describing thin shells around black holes that we derive in appendix [ appa ] .
as a first vacuum test ( for which we set @xmath108 and @xmath109 ) , we considered schwarzschild in standard " isotropic coordinates , representing the symmetry plane in a carter - penrose diagram . in our code , we can produce the solution @xmath110 by choosing @xmath111 and @xmath106 . after applying the transformations ( [ transi ] ) and ( [ transo ] ) ,
these solutions become linear in our code s coordinates @xmath112 and @xmath113 , meaning that the solution can be represented exactly in terms of the first two chebyshev polynomials . for any @xmath114
, our code therefore converges to the correct solution within the predetermined tolerances . as a function of the number of collocation points @xmath92 for two analytic representations of the schwarzschild geometry in isotropic coordinates .
the label @xmath115 denotes the analytic puncture solution presented in @xcite , while the label @xmath116 denotes the kerr- schild solution in @xcite .
the errors drop off exponentially as a function of the number of collocation points @xmath92 , until a ` floor ' of specified tolerance has been reached.,width=288 ] more interesting are the two isotropic representations of the schwarzschild geometry presented in @xcite and @xcite .
the former is a transformation of kerr - schild ( eddington - finkelstein ) coordinates to isotropic coordinates , keeping the same time slicing , which we can produce by choosing @xmath117 and @xmath118 in our code .
the latter is an isotropic representation of a maximal slice ( with the critical parameter @xmath119 , see @xcite ) , which has recently attracted interest as an analytic puncture " solution ( compare ) .
we can produce this solution by choosing @xmath120 and @xmath121 in our code . to measure the deviation from an analytic solution
, we compute the average of the absolute error @xmath122 at all @xmath92 collocation points in each of the two domains , @xmath123 where the collocation points @xmath124 are in the region outside of the shell . in fig .
[ apkserror ] we graph this error as a function of the number of collocation points @xmath92 . as expected , the errors fall off exponentially for all variables , until they reach a floor corresponding to the predetermined tolerance .
in addition to these vacuum solutions we also consider an analytic solution describing thin shells of non - interacting particles around a static black hole .
as we demonstrate in appendix [ appa ] , we can solve the differential equations ( [ diffeq ] ) subject to the boundary conditions ( [ infbound ] ) and ( [ inboundcond ] ) as well as the jump conditions ( [ jumpconditions ] ) analytically for maximal slicing and @xmath106 .
as for the solution ( [ ssiso ] ) , the field variables become linear in our code s variables @xmath112 and @xmath125 ( see ( [ analyticpsish ] ) and ( [ analyticalphash ] ) below ) , so that they can be represented exactly for any @xmath114 .
in addition to testing the solution of the field equations , however , this test also verifies that our code correctly solves the jump conditions at the shell . as in example ,
we show in fig .
[ shellplot ] the analytic and numerical solutions for the lapse @xmath2 and the conformal factor @xmath6 as a function of areal radius @xmath45 for a lorentz factor of @xmath126 and a mass ratio @xmath127 .
our numerical solutions agree with the analytical ones to within better than @xmath107 , making them indistinguishable in the plot . and the lapse @xmath2 as a function of areal radius for the analytic shell solution ( @xmath105 and @xmath128 ) for a lorentz factor of @xmath126 and a mass ratio of @xmath129 , corresponding to an areal shell radius of about @xmath130 .
the field variables are continuous across the shell , but , according to the jump conditions ( [ jumpconditions ] ) their derivatives are not .
our numerical solutions agree with the analytical ones to within better than @xmath107 , making the error far smaller than the line width in the graph.,width=288 ]
) @xmath131 as a function of the shell s areal radius @xmath45 for an extreme - mass - ratio sequence with @xmath132 .
values of the horizon lapse @xmath133 range from zero to unity in increments of @xmath134 .
the graph includes eleven solid lines ( representing an evaluation of the binding energy ( [ mbind ] ) in maximal slicing ) and eleven dashed lines ( for kerr - schild slicing ) .
all 22 lines coincide within our numerical error .
we computed these sequences using @xmath135 collocation points in each domain.,width=288 ] we now construct constant - mass sequences , meaning sequences of varying shell radius @xmath67 but constant shell rest mass @xmath136 and black hole irreducible mass @xmath102 . specifically , we focus on extreme - mass - ratio " sequences with @xmath137 and equal - mass sequences @xmath138 . for both choices of the mass ratio
we construct sequences for our two choices of the extrinsic curvature ( [ k ] ) , maximal slicing and kerr - schild slicing , and for the horizon lapse ( [ alphahorbound ] ) ranging from @xmath106 to @xmath139 in increments of 0.1 . in the following ,
we will graph the binding energy ( [ mbind ] ) as a function of the areal radius .
typically , the binding energy @xmath140 is only a small fraction of the involved masses , meaning that the relative error in the binding energy is larger than that for the masses and hence the fields themselves .
we found that , to achieve similar accuracy , the extreme - mass - ratio sequences ( for which the binding energy is much smaller than for the equal - mass sequences ) required slightly more collocation points than the equal - mass sequences . in fig .
[ largeratio ] we show the binding energy for extreme - mass- ratio sequences with @xmath141 .
the graph represents 22 plots , corresponding to the eleven different values of @xmath133 and to evaluating the adm binding energy ( [ mbind ] ) for both maximal slicing and kerr - schild slicing . to within the accuracy of our code ,
all 22 lines agree with each other , so that they all lie on top of each other and appear as one line in fig .
[ largeratio ] .
the minimum of the binding energy corresponds to the innermost stable circular orbit ( isco ) . in the extreme - mass - ratio limit
we may neglect the particles self - gravity , so that we are effectively solving for a test - particle in the schwarzschild geometry .
as expected , we find that the isco is located at @xmath142 , representing another independent test of our code . as a function of the shell s areal radius @xmath45 for an equal mass sequence , @xmath143 . as in fig .
[ largeratio ] , values of @xmath61 again range from zero to unity in increments of @xmath134 , and the graph again contains eleven lines each for maximal and kerr - schild slicing .
all 22 lines again coincide within our numerical error .
we obtained these results with @xmath144 collocation points in each domain.,width=288 ] in fig .
[ equalratio ] we show the equivalent graphs for equal - mass sequences with @xmath145 . as for the extreme - mass- ratio sequences in fig .
[ largeratio ] all 22 graphs coincide to within our numerical accuracy .
we note that the isco is now located at a larger radius of about @xmath146 . even in this case
, we do not find any evidence that the choice of the slicing condition ( [ k ] ) , or the choice of the boundary condition for the lapse on the horizon ( [ alphahorbound ] ) , have any effect on the physical content of our solutions .
clearly , these different choices lead to different solutions for the conformal factor @xmath6 , the lapse @xmath2 and the shift @xmath147 . when expressed in terms of gauge - invariant quantities , however , all our solutions become indistinguishable to within the accuracy of our numerical code .
we solve einstein s constraint equations in the extended conformal thin - sandwich decomposition to construct spherical shells of non - interacting , isotropic particles in circular orbit about a non - rotating black hole .
we construct these solutions for both maximal slicing and kerr - schild slicing ( see equations ( [ k ] ) ) , and for a number of different choices for the horizon lapse .
these different choices lead to very different solutions for the lapse @xmath2 , the shift @xmath3 and the conformal factor @xmath6 .
however , when expressed in terms of gauge - invariant quantities for example the binding energy as a function of the shell s areal radius for given shell and black hole masses our solutions become indistinguishable .
at least in the limited context of our spherically symmetric solutions , these findings provide no evidence that the choices for the mean curvature and horizon lapse affect the physical content of the solutions .
km gratefully acknowledges support from the coles undergraduate research fellowship fund at bowdoin college and from the maine space grant consortium .
km and twb would like to thank the laboratoire univers et theories at the observatoire de paris , meudon , for its hospitality .
this work was supported in part by nsf grant phy-0456917 to bowdoin college and by the anr grant 06 - 2 - 134423 _ mthodes mathmatiques pour la relativit gnrale_.
for maximal slicing ( @xmath105 ) and the lapse boundary condition @xmath148 , we can find an analytic solution to the differential equations ( [ diffeq ] ) , subject to the boundary conditions ( [ infbound ] ) and ( [ inboundcond ] ) as well as the jump conditions ( [ jumpconditions ] ) .
the solution depends on the input parameters @xmath100 , @xmath104 , @xmath149 , and @xmath67 .
we may then enforce the geodesic condition ( [ geodesiccondition ] ) to eliminate one of these variables , and compute the mass of the black hole from ( [ mirr ] ) .
the solution presented here represents a generalization of the solutions of @xcite , who considered the same system but without a black hole .
we begin with the momentum constraint ( [ betaconstraintmod ] ) . for @xmath105 and @xmath106
we find that @xmath150 is a self - consistent solution to both the equation and its boundary conditions .
this implies that eqs .
( [ psiconstraintmod ] ) and ( [ alphaconstraintmod ] ) for respectively the conformal factor @xmath6 and the combination @xmath151 reduce to flat laplace equations in the vacuum regions away from the shell . in spherical symmetry ,
the only possible solutions are of the form @xmath152 , where @xmath153 and @xmath154 are arbitrary constants that have to be determined from the boundary conditions . for each function
we need four conditions to determine these constants both in the interior and the exterior of the shell .
these four conditions arise from the outer boundary conditions ( [ infbound ] ) , the inner boundary conditions ( [ inboundcond ] ) , continuity of the functions at the shell , and the jump conditions ( [ jumpconditions ] ) for their first derivatives . using these conditions , we find @xmath155 1 + \dfrac{a \left(r_{\rm sh}+r_{\rm bh } \right ) - r_{\rm sh } } { r } \ , , & r \geq r_{\rm sh } \end{array } \right.\ ] ] for the conformal factor and @xmath156 \displaystyle \dfrac{1+\left ( c \left ( r_{\rm bh } - r_{\rm sh } \right ) - r_{\rm sh } \right)/r}{1+\left(a \left(r_{\rm sh } + r_{\rm bh } \right)-r_{\rm sh } \right)/r } \ , , & r \geq r_{\rm sh } \end{array } \right.\ ] ] for the lapse . here
the constants @xmath157 and @xmath158 are given by @xmath159 and @xmath160 inserting these constants , which themselves depend on the values of the conformal factor and the lapse at the shell , into ( [ analyticpsish ] ) and ( [ analyticalphash ] ) , we find @xmath161\ ] ] and @xmath162 here we have abbreviated @xmath163 .
so far these solutions depend on all four parameters @xmath100 , @xmath164 , @xmath101 , and @xmath67 .
we now find a relation between these parameters by inserting ( [ analyticpsish ] ) and ( [ analyticalphash ] ) into the geodesic equation ( [ geodesicmod ] ) , which yields @xmath165 unfortunately , this expression is not very useful for our purposes in this form .
we find an alternative form by evaluating ( [ geodesiccondition ] ) for our solution ( [ analyticpsish ] ) and ( [ analyticalphash ] ) , which results in a fifth order polynomial for @xmath104 , @xmath166 in the limit of a zero - mass black hole we have @xmath167 and ( [ geoanalytic ] ) reduces to @xmath168 which agrees with the corresponding equation ( 26 ) of @xcite . instead of parameterizing the solution by @xmath101 , it is more desirable to fix the black hole mass @xmath169 . towards this end
we combine ( [ geoanalytic ] ) with ( [ analyticpsish ] ) and ( [ mirr ] ) , which results in a quadratic equation for @xmath67 .
the solutions to this equation are @xmath170/ \nonumber \\ & & \big [ 2(m_{\rm bh}^2 - 2m_{\rm bh } r_{\rm bh } ) \big ] .\end{aligned}\ ] ] we henceforth ignore the @xmath171 " solution , as only the
@xmath172 " solution refers to stable solutions .
finally , we may insert ( [ constanalytic ] ) back into ( [ geoanalytic ] ) , which results into a polynomial for @xmath101 that can be factored into two cubic polynomials @xmath173 here we have abbreviated @xmath174 , and the coefficients @xmath175 and @xmath176 are , in terms of the mass ratio @xmath177 , @xmath178 \nonumber \\ b_2 & = & -4q(-2 + 3w^2)^2(8q^2 - 8qw+5w^2 ) \nonumber \\ a_1 & = & 2[16 q^4 - 48 q^3 w + ( 44 q^2 - 48 q^4 ) w^2 \nonumber \\ & & + ( 144 q^3 - 12 q ) w^3+(36q^4 - 136q^2 \nonumber \\ & & + 1)w^4 + ( 38 q-108
q^3)w^5+(105 q^2 \nonumber \\ & & - 2 ) w^6 -30qw^7+w^8 ] \nonumber \\ b_1 & = & 2(-2 + 3w^2)[-8q^4 + 16q^3w \nonumber \\ & & + ( 12q^4 - 20q^2 ) w^2 + ( 10q-24q^3)w^3 \nonumber \\ & & + ( 26q^2 - 2 ) w^4 - 12qw^5 + 3w^6 ] \nonumber \\ a_0 & = & b_0 = -qw^2(-2q+w+3qw^2 - 2w^3)^2 \nonumber\end{aligned}\ ] ] we can now construct a solution for given masses @xmath102 and @xmath136 and lorentz factor @xmath104 as follows . given the mass ratio @xmath179 and @xmath104 we first find the six solutions for @xmath180 from the two polynomials in ( [ cubicpol1 ] ) .
we then insert the corresponding values @xmath101 into ( [ constanalytic ] ) , which yields six solutions for @xmath67 . the largest real solution is the solution of interest ; we keep only this solution as well as its corresponding value of @xmath101 and disregard all others .
these values can then be inserted into ( [ analyticpsish ] ) and ( [ analyticalphash ] ) , which determines the solution completely .
|
we solve einstein s constraint equations in the conformal thin - sandwich decomposition to model thin shells of non - interacting particles in circular orbit about a non - rotating black hole .
we use these simple models to explore the effects of some of the freely specifiable quantities in this decomposition on the physical content of the solutions . specifically , we adopt either maximal slicing or kerr - schild slicing , and make different choices for the value of the lapse on the black hole horizon .
for one particular choice of these quantities the resulting equations can be solved analytically ; for all others we construct numerical solutions .
we find that these different choices have no effect on our solutions when they are expressed in terms of gauge - invariant quantities .
|
understanding the mechanism underlying quark confinement from the first principle of qcd is still a challenging problem in theoretical particle physics @xcite . as a possible step towards this goal
, it will be efficient to extract the dominant field mode @xmath2 responsible for confinement from the yang - mills field @xmath3 to clarify the physics behind the phenomena of confinement .
the well - known mathematical identity called the cartan decomposition @xcite is used to decompose the field variable @xmath3 valued in the lie algebra @xmath4 of a gauge group @xmath5 into the simultaneously diagonalizable part in the cartan subalgebra @xmath6 and the remaining off - diagonal part in the orthogonal complement of @xmath7 .
however , the cartan decomposition is not suited for studying the non - perturbative features of the gauge field theory with local gauge invariance , since the cartan decomposition can not retain the original form after the gauge transformation , namely , the local rotation of the cartan - weyl basis for the lie algebra . in view of these , the novel decomposition called the cho - duan - ge - faddeev - niemi ( cdgfn ) decomposition @xcite is quite attractive , since the cdgfn decomposition given in the form @xmath8 is gauge covariant , namely , it keeps its form under the gauge transformation or the local color rotation . in the cdgfn decomposition
, the unit lie algebra valued field @xmath9 called the color direction field or the color field for short plays the crucial role for retaining the local gauge covariance of the decomposition . for @xmath10 , the color field @xmath11
is constructed from the maximally commuting generators @xmath12 @xmath13 in the cartan subalgebra @xmath6 according to the local adjoint rotation by a group element @xmath14 of the gauge group @xmath5 at every point @xmath15 of spacetime : @xmath16 the color direction field belongs to the subset of @xmath17 which is topologically equivalent to @xmath18 , where a subgroup @xmath19 called the maximal stability subgroup of @xmath5 is specified from the degeneracy among the eigenvalues of a representation matrix of @xmath12 .
in other words , the color field @xmath9 is regarded as the local embedding of the cartan direction @xmath12 in the internal space of the non - abelian group @xmath5 . from this viewpoint ,
the cartan decomposition is identified with a global limit of the cdgfn decomposition , which urges us to consider that the abelian projection method @xcite is nothing but a gauge - fixed version of the gauge - covariant cdgfn decomposition .
the application of the novel decomposition to the yang - mills non - abelian gauge field paves the way for understanding quark confinement in a gauge - independent manner .
in fact , this method has been extensively used to investigate quark confinement in the @xmath1 yang - mills theory in the last decade , see e.g. , @xcite for a review . a promising mechanism for understanding quark confinement is well known as the dual superconductivity @xcite .
it is a hypothesis based on the electro - magnetic dual analog of the type ii superconductor in which the magnetic field applied to the bulk of the superconductor is squeezed to form the magnetic vortex due to the meissner effect of excluding the magnetic field from the superconductor @xcite .
the color electric field created by a pair of a quark and an antiquark would be squeezed to form an electric flux - tube or a hadron string with its ends on a quark and an antiquark .
for the dual superconductivity to work , therefore , one needs magnetic objects , say magnetic monopoles to be condensed in the vacuum of the yang - mills theory , which is supposed to be dual to the ordinary superconductivity caused by condensation of electric objects , a pair of electrons called the cooper pairs @xcite .
the magnetic monopole in the pure yang - mills theory has been mostly constructed by the abelian projection method , which @xcite breaks explicitly the original non - abelian gauge group @xmath5 to the maximal torus subgroup @xmath20 . however , this is not the only way to define magnetic monopoles in pure yang - mills theory without the higgs field .
in fact , one can define gauge - invariant magnetic monopoles in the pure yang - mills theory without breaking the original gauge symmetry by using the non - abelian stokes theorem @xcite for the wilson loop operator which is in itself gauge invariant @xcite .
the gauge - invariant magnetic monopole is specified by the maximal stability subgroup @xmath21 which is uniquely determined for the highest - weight state of a given representation for a quark source . for quarks in the fundamental representation of @xmath1 , especially , the maximal stability subgroup @xmath21 is given by @xmath22 , which is distinct from the maximal torus group @xmath23 for @xmath24 . the contribution of magnetic monopoles to the wilson loop average can be calculated by using the path - integral framework using the reformulation of the yang - mills theory based on change of variables in accord with the field decomposition .
we find that there are some options for reformulating the @xmath1 yang - mills theory ( @xmath24 ) corresponding to the different choice of the color direction field compatible with the maximal stability group , although there is only one way to reformulate the @xmath25 yang - mills theory .
whichever options we use , we need the field decomposition formula , which allows us to decompose an arbitrary element @xmath26 of a lie algebra @xmath27 to the part @xmath28 in the lie algebra of @xmath19 and the remaining part @xmath29 .
see e.g. , @xcite and also @xcite for a review .
another popular object of topological nature which is believed to be responsible for confinement is the center vortex @xcite , which is associated to the center subgroup of @xmath5 .
in fact , the confinement / deconfinement phase transition at finite temperature in the @xmath1 yang - mills theory is associated with the restoration / spontaneous breaking of the center symmetry @xmath30 , which is signaled by vanishing / nonvanishing of the polyakov loop average .
see e.g. , @xcite for reviews .
we suppose , however , that magnetic monopoles and magnetic vortices can not be independent topological objects .
they could be different views of a single physical object , just like two sides of a coin , to be simultaneously defined in a self - consistent way @xcite .
however , this statement remains still a conjecture to be proved .
the purpose of this paper is to extend the gauge - covariant field decomposition of the yang - mills field and the non - abelian stokes theorem for the wilson loop operator developed so far for @xmath1 to the exceptional group @xmath0 , which is a preliminary step toward reformulating the @xmath0 yang - mills theory @xcite to discuss the mechanism for confinement / deconfinement .
our interests of the exceptional group @xmath0 lie in a fact that the @xmath0 yang - mills theory has the linear potential @xcite and that the center vortex confinement mechanism is argued to work for @xmath0 in @xcite , although @xmath0 has a trivial center subgroup , consisting only of the identity element @xcite .
we want to define the gauge - invariant magnetic monopole in the @xmath0 yang - mills theory and then examine whether the magnetic monopole defined in our framework can be a universal topological object responsible for confinement , irrespective of the gauge group .
this investigation will help us to prove or disprove the above conjecture on the interrelation between magnetic monopoles and magnetic vortices .
the present paper is organized as follows . in sec .
ii , we first determine the maximal stability subgroup for a given representation of @xmath0 , after presenting some basic properties of @xmath0 . in sec .
iii , we subsequently derive the gauge - covariant decomposition formula for the @xmath0 yang - mills field corresponding to each maximal stability subgroup .
we show that the @xmath0 yang - mills field has different gauge - covariant decompositions depending on the maximal stability subgroup , which are more complicated than those obtained for the @xmath1 yang - mills field .
in fact , it turns out that the decomposition formula for @xmath0 can not be obtained as a simple extension of that for @xmath1 .
this is because the fact that all roots of @xmath1 have the same norm was used in deriving the decomposition formula for @xmath1 .
however , some roots of @xmath0 have different norm from the other roots .
consequently , the relevant decomposition for @xmath0 can not be obtained by using the double commutators , in sharp contrast to @xmath31 .
nevertheless , the multiple commutators with the cartan generators enable us to obtain the desired decomposition .
remarkably , we have found that the decomposition formula for @xmath0 can be obtained using sextuple commutators with the cartan generators or the color fields .
moreover , the method presented in this paper for obtaining the decomposition formula for @xmath0 can be applied to any semi - simple lie algebra , and therefore the reformulation of the yang - mills theory would be possible for an arbitrary semi - simple gauge group . in sec .
iv , we derive a non - abelian stokes theorem for the wilson loop operator of the @xmath0 yang - mills field , which is written using the sextuple commutators with the color direction fields , as an application of the decomposition formula .
this enables us to define gauge - invariant magnetic monopoles in the @xmath0 yang - mills theory in sec .
v. we find that which kind of magnetic monopoles can be defined is determined by the stability subgroup of @xmath5 .
we show that the magnetic charge derived from the gauge - invariant magnetic monopole is subject to a novel quantization condition , which is similar to , but different from the quantization condition for the dirac magnetic monopole and thooft - polyakov magnetic monopole .
the final section is devoted to conclusion and discussion .
some technical derivations are collected in appendices a , b and c.
in this section , we give some basic properties of an exceptional group g(2 ) .
we begin with the dynkin diagram of @xmath0 given by fig . [
fig : dynkin ] .
it indicates that @xmath0 has two simple roots ( i.e. , rank 2 ) with the opening angle @xmath32 . in this paper
, we use @xmath33 as simple roots .
we see that the other positive roots are obtained as @xmath34 fig .
[ fig : root ] is the root diagram of @xmath0 .
hence , there are two cartan generators @xmath35 ( @xmath36 ) and twelve shift operators @xmath37 ( @xmath38 ) , where @xmath39 is the root system , i.e. , the set of positive and negative root vectors .
they satisfy the commutation relation called the cartan standard form : @xmath40 & = 0 \quad ( j , k=1,2 ) , \notag\\ [ h_k , e_\alpha ] & = \alpha_k e_\alpha , \notag\\ [ e_\alpha , e_{-\alpha } ] & = \alpha\cdot h \notag\\ [ e_\alpha , e_\beta ] & \propto \begin{cases } e_{\alpha+\beta } & ( \alpha+\beta\in \mathcal{r})\\ 0 & ( { \rm otherwise } ) \end{cases } \label{3}\end{aligned}\ ] ] where @xmath41 denotes the @xmath42th component of the root vector @xmath43 and @xmath44 is the inner product defined by @xmath45 . in this paper
we consider a unitary representation .
therefore representation matrices satisfy the hermiticity : @xmath46 and the normalization : @xmath47 where the value of @xmath48 depends on the representation . using this property
, we can define the inner product in the lie algebra as @xmath49 which is independent of the representation @xmath50 . .
] . ]
the weight vector @xmath51 of a representation specified by the dynkin index @xmath52 $ ] of the lie group with the rank @xmath53 is obtained from the relation : @xmath54 for @xmath0 , the highest - weight vector @xmath55 ( @xmath56 ) of a representation with the dynkin index @xmath57 $ ] or @xmath58 $ ] satisfies @xmath59 and hence determined as @xmath60 the weight diagrams are determined by @xmath55 ( @xmath56 ) , as given in fig .
[ fig : diag ] .
the highest - weight vector @xmath61 corresponds to the 14-dimensional adjoint representation @xmath62 with the dynkin index @xmath57 $ ] , while the highest - weight vector @xmath63 corresponds to the 7-dimensional fundamental representation @xmath64 with the dynkin index @xmath58 $ ] .
an arbitrary irreducible representation of @xmath0 is labeled by the two dynkin indices @xmath65 $ ] and its highest weight @xmath66 can be written as @xmath67 notice that @xmath0 contains the lie group @xmath31 as a subgroup .
we can see from the root diagram that the lie algebra @xmath68 of @xmath31 , denoted as @xmath69 , is generated by a set of elements in @xmath68 : @xmath70 therefore , a representations of @xmath0 is written as direct sums of representations of @xmath31 .
for example , the fundamental representations of @xmath0 are written as @xmath71 it is known @xcite that one can construct a number of the reformulations of the yang - mills theory which are discriminated by the maximal stability subgroup .
therefore it is important to know which subgroup is identified with the maximal stability subgroup for each representation . in view of this , we first derive a certain property to be satisfied by the generators belonging to the lie algebra of the maximal stability subgroup of @xmath0 . by using this property , then , we determine the maximal stability subgroup for each representation of @xmath0 . the maximal stability subgroup @xmath19 for the representation @xmath50 of a group @xmath5 is defined to be a subgroup whose element @xmath72 leaves the highest - weight state @xmath73 of the representation @xmath50 invariant up to a phase factor ] but if we do so the presentation become rather cumbersome . therefore we omit @xmath74 throughout this subsection . ] : @xmath75 hence , an element of its lie algebra @xmath76 can be written as a linear combination of the cartan generators and shift - up and -down operators @xmath37 and @xmath77 ( where @xmath43 is a positive root ) such that @xmath78 and @xmath79 .
here notice that , if there is @xmath80 in the linear combination , then there is also @xmath77 , that is to say , the mutually hermitian - conjugate generators @xmath80 and @xmath77 must appear in pairs in the linear combination , since all matrices in a unitary representation of the lie algebra are hermitian .
we show in the following that @xmath81 and @xmath82 if and only if @xmath83 . here , we should remember that @xmath84 and @xmath85 satisfy the commutation relations of @xmath86 .
we see from this fact that if @xmath87 and @xmath88 then @xmath89 belongs to the space of the trivial representation of @xmath25 and hence @xmath90 . because @xmath73 is highest weight state , @xmath91 .
hence if @xmath92 then @xmath82 . in the same way
, the converse can be proven .
thus we arrive at the conclusion that @xmath93 can be written as a linear combination of the cartan generators @xmath12 and shift operators @xmath94 with positive root vectors @xmath43 that are orthogonal to the highest - weight vector @xmath66 : @xmath95 thus , it is easy to see that all representations of @xmath0 are classified into the following three categories . 1 . for the highest weight @xmath96 , the positive root orthogonal to the highest weight is @xmath97 alone .
hence , the maximal stability subgroup is a @xmath98 with the generators @xmath99 , @xmath100 , @xmath101 and @xmath102 : @xmath103 which agrees with a subset of @xmath31 specified by ( [ 8 ] ) .
2 . for the highest weight @xmath104 ,
the positive root orthogonal to the highest weight is @xmath105 alone .
hence the maximal stability subgroup is another @xmath98 with the generators @xmath99 , @xmath100 , @xmath106 and @xmath107 : @xmath108 which differs from a subset of @xmath31 specified by ( [ 8 ] ) .
3 . for the highest weight @xmath109 ( @xmath110 ) ,
the maximal stability subgroup is equal to the maximal torus subgroup @xmath111 generated by the cartan subalgebra @xmath112 : @xmath113 this fact is confirmed as follows .
we can write any positive root as @xmath114 , where @xmath42 and @xmath115 are non - negative integers that are not zero simultaneously .
hence , the relation , @xmath116 implies that all positive roots are not orthogonal to the highest weight when @xmath117 and @xmath118 .
thus , in this case , the generators of the maximal stability subgroup are given by @xmath99 and @xmath100 .
let @xmath26 be an arbitrary element of the lie algebra . to write the wilson loop using the color direction fields , and to reformulate the @xmath0 yang - mills theory
, we have to decompose @xmath26 into the part @xmath28 belonging to @xmath76 , and the remaining part @xmath29 using its commutators with @xmath35 .
this is achieved by using double commutators in the case of @xmath1 , see @xcite .
but , in the case of @xmath0 , we have to use sextuple commutators . its proof is given in appendix [ sec : a ] . in this section ,
we give the explicit form of such a decomposition for any representation . before proceeding to the @xmath0 case
, we reconsider the @xmath31 case from the viewpoint of this paper . for @xmath31 ,
it is known @xcite that the maximal stability subgroup is @xmath98 or @xmath111 . in the case of the maximal stability subgroup @xmath98 with generators @xmath99 , @xmath100 , @xmath119 and @xmath120 ,
the decomposition formula is written as @xmath121 , \end{aligned}\ ] ] while in the case of the maximal stability subgroup @xmath111 , the decomposition formula is written as @xmath122.\end{aligned}\ ] ] the derivation of these formulas is written in appendix c and d in @xcite .
we rederive them using another method which can be applied also to @xmath0 .
first , we consider the commutator of an arbitrary element of the cartan subalgebra with @xmath26 .
an arbitrary element of the cartan subalgebra can be written as @xmath123 , where @xmath124 is an arbitrary 2-dimensional vector . using the cartan decomposition , @xmath125 and the commutation relation ( [ 3 ] ) ,
we can write the commutator as @xmath126 = \sum_{\alpha\in\mathcal{r}_+}(\nu\cdot\alpha)(\mathscr{f}_\alpha^ * e_\alpha - \mathscr{f}_\alpha e_{-\alpha}).\end{aligned}\ ] ] here , we choose @xmath127 as @xmath124 , which is orthogonal to @xmath128 , i.e. , @xmath129 .
then the commutator reads @xmath130 & = ( \gamma^1\cdot\alpha^{(2)})(\mathscr{f}_{\alpha^{(2)}}^ * e_{\alpha^{(2 ) } } - \mathscr{f}_{\alpha^{(2 ) } } e_{-\alpha^{(2 ) } } ) + ( \gamma^1\cdot\alpha^{(3)})(\mathscr{f}_{\alpha^{(3)}}^ * e_{\alpha^{(3 ) } } - \mathscr{f}_{\alpha^{(3 ) } } e_{-\alpha^{(3)}}),\end{aligned}\ ] ] where the terms corresponding to @xmath128 disappear .
by taking the commutator once more , we can eliminate another term .
for the vector @xmath131 , which is orthogonal to @xmath132 , i.e. , @xmath133 , the double commutator is written as @xmath134 & = ( \gamma^2\cdot\alpha^{(3)})(\gamma^1\cdot\alpha^{(3)})(\mathscr{f}_{\alpha^{(3)}}^ * e_{\alpha^{(3 ) } } + \mathscr{f}_{\alpha^{(3 ) } } e_{-\alpha^{(3)}})\notag\\ & = \frac34(\mathscr{f}_{\alpha^{(3)}}^ * e_{\alpha^{(3 ) } } + \mathscr{f}_{\alpha^{(3 ) } } e_{-\alpha^{(3)}}).\end{aligned}\ ] ] thus we obtain @xmath135.\end{aligned}\ ] ] in this way , we can extract the element of @xmath26 corresponding to a particular positive root by taking the double commutator .
for the other positive roots , the similar identity holds : @xmath136,\\ \label{26 } \mathscr{f}_{\alpha^{(2)}}^ * e_{\alpha^{(2 ) } } + \mathscr{f}_{\alpha^{(2 ) } } e_{-\alpha^{(2 ) } } & = \frac43[\gamma^1\cdot h,[\gamma^3\cdot h , \mathscr{f}]],\end{aligned}\ ] ] where we have introduced @xmath137 , which is orthogonal to @xmath138 . using these expressions , we can write the decomposition formula for any case of the maximal stability subgroup . for @xmath139 , the decomposition formula
is written as @xmath140 - \frac13 [ h_2,[h_2,\mathscr{f } ] ] \notag\\ \mathscr{f}_{g/\tilde{h } } & = ( \mathscr{f}_{\alpha^{(1)}}^ * e_{\alpha^{(1 ) } } + \mathscr{f}_{\alpha^{(1 ) } } e_{-\alpha^{(1 ) } } ) + ( \mathscr{f}_{\alpha^{(3)}}^ * e_{\alpha^{(3 ) } } + \mathscr{f}_{\alpha^{(3 ) } } e_{-\alpha^{(3 ) } } ) \notag\\ & = \frac43 [ h_2,[h_2,\mathscr{f } ] ] , \end{aligned}\ ] ] while for @xmath141 , the decomposition formula is written as @xmath142,\end{aligned}\ ] ] where we have used the commuting property : @xmath143 = [ [ h_j , h_k],\mathscr{f}]+[h_k,[h_j,\mathscr{f}]]=[h_k,[h_j,\mathscr{f}]],\end{aligned}\ ] ] following from @xmath144 = 0 $ ] .
now we consider the @xmath0 case .
we want to write the coset part @xmath29 of @xmath26 as a linear combination of multiple commutators with the cartan generators : @xmath145\cdots],\end{aligned}\ ] ] where the sum is over independent terms by taking account of the commuting property ( [ 18a ] ) . we can obtain ( [ 17 ] ) for any choice of @xmath146 if , for every positive root @xmath147 , the relevant shift part @xmath148 is written in the form : @xmath149\cdots].\end{aligned}\ ] ] in the following , we give a derivation of this fact ( [ 18 ] ) . in the way similar to that written in the above for @xmath31 , indeed , we can obtain ( [ 18 ] ) . by using the commutation relation ( [ 3 ] ) , the commutator is calculated as @xmath150 = \sum_{\alpha\in\mathcal{r}_+ } ( \nu\cdot \alpha)(\mathscr{f}_\alpha^*e_{\alpha}- \mathscr{f}_\alpha e_{-\alpha}).\end{aligned}\ ] ] if @xmath124 is chosen to be orthogonal to a particular @xmath43 , then the corresponding terms of @xmath37 and @xmath77 disappear from this expression .
thus , by taking the commutator repeatedly , we can eliminate all shift terms except one shift term @xmath148 that corresponds to a particular positive root @xmath147 .
since there are six positive roots in @xmath0 , we have to eliminate five shift terms .
we can do so using the quintuple commutator : @xmath151 ] ] ] = ( \nu^1\cdot\beta)(\nu^2\cdot\beta)(\nu^3\cdot\beta)(\nu^4\cdot\beta)(\nu^5\cdot\beta)(\mathscr{f}_\beta^*e_{\beta}- \mathscr{f}_\beta e_{-\beta}),\end{aligned}\ ] ] where @xmath152 are appropriate 2-dimensional vectors .
is 2-dimensional .
if the dimension of the root vectors is larger , we can eliminate more than one term by taking the commutator once . ] in this expression , the sign of the term of @xmath153 is opposite to that of @xmath154 . to make both signs equal
, we need to take the commutator once more .
we choose an 2-dimensional vector @xmath124 which is non - orthogonal to @xmath147 to obtain the non - vanishing commutator of @xmath123 and ( [ 19 ] ) : @xmath155 = ( \nu\cdot\beta)(\nu^1\cdot\beta)(\nu^2\cdot\beta)(\nu^3\cdot\beta)(\nu^4\cdot\beta)(\nu^5\cdot\beta)(\mathscr{f}_\beta^*e_{\beta}+ \mathscr{f}_\beta e_{-\beta}).\end{aligned}\ ] ] thus we obtain the key relation : @xmath156 ] ] ] ] , \notag\\ & n : = ( \nu\cdot\beta)(\nu^1\cdot\beta)(\nu^2\cdot\beta)(\nu^3\cdot\beta)(\nu^4\cdot\beta)(\nu^5\cdot\beta ) .
\label{21}\end{aligned}\ ] ] although this expression is nothing but the desired one ( [ 18 ] ) , it should be remarked that the coefficients @xmath157 is not uniquely determined .
if we multiply @xmath124 by a constant , the coefficients @xmath157 do not change .
this point will be observed more concretely shortly . to obtain the expression ( [ 21 ] ) for each positive root concretely ,
we introduce six unit vectors @xmath158 ( @xmath159 ) such that @xmath158 is positive and orthogonal to one of the positive roots , say @xmath160 ( @xmath159 ) : @xmath161 see fig .
[ fig : orvec ] .
consequently these vectors satisfy the following conditions : @xmath162 for example , @xmath163 is obtained as @xmath164 ] ] ] ] , \notag\\ & n = ( \gamma\cdot\alpha^{(1)})(\gamma^2\cdot\alpha^{(1)})(\gamma^3\cdot\alpha^{(1)})(\gamma^4\cdot\alpha^{(1)})(\gamma^5\cdot\alpha^{(1)})(\gamma^6\cdot\alpha^{(1 ) } ) , \end{aligned}\ ] ] where @xmath165 is an arbitrary 2-dimensional vector that is not orthogonal to @xmath128 .
to obtain more explicit form , we put an arbitrary 2-dimensional vector @xmath165 in the form : @xmath166 ( orthogonal decomposition of @xmath165 ) where @xmath167 is orthogonal to @xmath128 and @xmath168 is parallel to @xmath128 : @xmath169 and @xmath170 . here
@xmath171 to avoid @xmath172 .
using @xmath173 , we have @xmath174 , @xmath175 , @xmath176 , @xmath177 , @xmath178 , and @xmath179 , which yields @xmath180 combining the result @xmath181 with @xmath182 , therefore , we obtain @xmath183 ] ] ] ] \notag\\ & \quad + c_1[\gamma^1\cdot h,[\gamma^2\cdot h,[\gamma^3\cdot h,[\gamma^4\cdot h,[\gamma^5\cdot h,[\gamma^6\cdot h,\mathscr{f } ] ] ] ] ] ] , \end{aligned}\ ] ] where @xmath184 .
we can see from this expression that the non - uniqueness of an expression of @xmath163 comes from the fact that the following sextuple commutator is identically vanishing : @xmath185 ] ] ] ] \notag \\ & \quad= \sum_{j_1,\dots , j_6\in{1,2}}\bar{\zeta}_{j_1\cdots j_6}[h_{j_1},[h_{j_2},[h_{j_3},[h_{j_4},[h_{j_5},[h_{j_6},\mathscr{f}]]]]]]= 0 , \notag\\ & \bar{\zeta}_{11112 } = \frac3{16},\ ; \bar{\zeta}_{111222}=-\frac58 , \ ; \bar{\zeta}_{122222}=\frac3{16 } .\end{aligned}\ ] ] thus , the non - uniqueness of the decomposition formula is attributed to degree of freedom due to one parameter @xmath186 . in the same way as the above
, we obtain @xmath187 ] ] ] ] + c_2 \mathcal{z } , \\
\mathscr{r}_{\alpha^{(3 ) } } & = -\frac{16}3[\gamma^6\cdot h,[\gamma^1\cdot h,[\gamma^2\cdot h,[\gamma^4\cdot h,[\gamma^5\cdot h,[\gamma^6\cdot h,\mathscr{f } ] ] ] ] ] ] + c_3 \mathcal{z } , \\
\mathscr{r}_{\alpha^{(4 ) } } & = \frac{16}3(\sqrt{3})^6[\gamma^1\cdot h,[\gamma^1\cdot h,[\gamma^2\cdot h,[\gamma^3\cdot h,[\gamma^5\cdot h,[\gamma^6\cdot h,\mathscr{f}]]]]]]+ c_4 \mathcal{z } , \\
\mathscr{r}_{\alpha^{(5 ) } } & = \frac{16}3(\sqrt3)^6[\gamma^2\cdot h,[\gamma^1\cdot h,[\gamma^2\cdot h,[\gamma^3\cdot h,[\gamma^4\cdot h,[\gamma^6\cdot h,\mathscr{f } ] ] ] ] ] ] + c_5\mathcal{z } , \\
\mathscr{r}_{\alpha^{(6 ) } } & = -\frac{16}3(\sqrt3)^6[\gamma^3\cdot h,[\gamma^1\cdot h,[\gamma^2\cdot h,[\gamma^3\cdot h,[\gamma^4\cdot h,[\gamma^5\cdot h,\mathscr{f } ] ] ] ] ] ] + c_6 \mathcal{z } , \end{aligned}\ ] ] where @xmath188 are arbitrary constants . here
, we have used the commuting property ( [ 18a ] ) .
collecting an appropriate set of @xmath189 , we obtain the desired decomposition formula corresponding to each category of representations given in ( [ 12])([13])([14 ] ) : 1 . for @xmath190 with the generators @xmath99 , @xmath100 , @xmath101 and @xmath102 ,
the highest weight is @xmath191 . here , we redefine the highest weight as @xmath192 in order to obtain simpler form .
this is possible because new one is obtained by acting a weyl group element on an old one .
thus @xmath19 has the generators @xmath99 , @xmath100 , @xmath119 and @xmath120 .
the @xmath19-commutative part @xmath28 and the coset part @xmath29 of @xmath26 are given by @xmath193 the explicit form is given by @xmath194 ] ] ] ] + c_2\mathcal{z } , \notag\\ \tilde{\zeta}_{111111 } & = 1,\ ; \tilde{\zeta}_{111122 } = -\frac{10}3,\ ; \tilde{\zeta}_{112222 } = 1,\notag\\ \mathscr{f}_{g/\tilde{h } } & = \sum_{j_1,\dots , j_6}\zeta_{j_1\cdots j_6}[h_{j_1},[h_{j_2},[h_{j_3},[h_{j_4},[h_{j_5},[h_{j_6},\mathscr{f } ] ] ] ] ] ] + c\mathcal{z } , \notag\\ \zeta_{111122 } & = \frac{721}3,\ ; \zeta_{112222 } = -154,\ ; \zeta_{222222}=27 , \label{zeta}\end{aligned}\ ] ] where the other @xmath195s and @xmath196 are zero and @xmath197 .
the simplest choice is @xmath198 and @xmath199 .
2 . for @xmath200 with the generators @xmath99 , @xmath100 , @xmath106 and @xmath107 , @xmath201 we obtain @xmath202 ] ] ] ] + c_5\mathcal{z } , \notag\\ \tilde{\zeta}'_{111122 } & = 27,\ ; \tilde{\zeta}'_{112222 } = -90,\ ; \tilde{\zeta}'_{222222 } = 27,\notag\\ \mathscr{f}_{g/\tilde{h } } & = \sum_{j_1,\dots , j_6}\zeta'_{j_1\cdots j_6}[h_{j_1},[h_{j_2},[h_{j_3},[h_{j_4},[h_{j_5},[h_{j_6},\mathscr{f } ] ] ] ] ] ] + c'\mathcal{z } , \notag\\ \zeta'_{111111 } & = 1,\ ; \zeta'_{111122 } = 210 , \ ; \zeta'_{112222}=-63 , \label{zeta'}\end{aligned}\ ] ] where the other @xmath203s and @xmath196 are zero .
we can take the simplest choice @xmath204 and @xmath205 .
3 . for @xmath206 ,
@xmath207 we obtain @xmath208 ] ] ] ] + c''\mathcal{z } , \notag\\ \zeta''_{111111 } & = 1,\ ; \zeta''_{111122 } = 237 , \ ; \zeta''_{112222}=-153 , \ ; \zeta''_{222222}=27 , \label{zeta''}\end{aligned}\ ] ] where the other @xmath209s are zero .
we can take the simplest choice @xmath210 . using the decomposition formula
, we can define the field decomposition in the similar way to the case of the gauge group @xmath1 .
for this purpose , we define the color direction field for @xmath0 as @xmath211 where @xmath212 is the adjoint representation of @xmath213 , where @xmath213 is an arbitrary group - valued field . because @xmath214 here we have used the same notation @xmath50 to denote the group representation and the corresponding algebra representation , which does not cause the confusion because the domains are different . ] for any lie algebra valued field @xmath215 , by applying the decomposition formula to @xmath216 and operating @xmath212 on the both sides , we can decompose @xmath215 into the part @xmath217 belonging to @xmath218 and the remaining part @xmath219 : @xmath220\cdots ] , \notag\\ & \mathscr{f}_{g/\tilde{h}}(x ) = \sum_{j_1,\dots , j_6}\eta_{j_1\cdots j_6}[\bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x),[\bm{n}_{j_6}(x),\mathscr{f}(x)]]\cdots ] , \label{f - decomp}\end{aligned}\ ] ] where @xmath221 and @xmath222 are appropriate coefficients specified by the maximal stability subgroup .
we decompose the yang - mills field @xmath223 into two pieces , @xmath224 and @xmath225 : @xmath226 where the decomposed fields @xmath224 and @xmath225 are obtained as the solution of the defining equations : @xmath227\bm{n}_j(x ) : = \partial_\mu\bm{n}_j(x ) - ig_{{}_{\rm ym}}[\mathscr{v}_\mu(x ) , \bm{n}_j(x ) ] , \label{def-1 } \\ & 0 = \mathscr{x}_\mu(x)_{\tilde{h } } \leftrightarrow \mathscr{x}_\mu(x ) = \sum_{j_1,\dots , j_6}\eta_{j_1\cdots j_6}[\bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x),[\bm{n}_{j_6}(x),\mathscr{x}_\mu(x)]]\cdots ] .
\label{def-2}\end{aligned}\ ] ] using the first defining equation ( [ def-1 ] ) , we find @xmath228\bm{n}_j(x ) = & \mathscr{d}_\mu[\mathscr{v}]\bm{n}_j(x ) - ig_{{}_{\rm ym}}[\mathscr{x}_\mu(x ) , \bm{n}_j(x ) ] = ig_{{}_{\rm ym } } [ \bm{n}_j(x ) , \mathscr{x}_\mu(x ) ] .
\end{aligned}\ ] ] by substituting this relation into the second defining equation ( [ def-2 ] ) , @xmath225 is rewritten as @xmath229\bm{n}_{j_6}(x ) ] \cdots ] .\end{aligned}\ ] ] then @xmath224 is written as @xmath230\bm{n}_{j_6}(x ) ] \cdots ] .\end{aligned}\ ] ] thus @xmath224 and @xmath225 are written in terms of the original yang - mills field @xmath223 and the color fields @xmath231 .
notice that @xmath224 is further cast into @xmath232 ] \cdots ] \nonumber\\ & + ig_{{}_{\rm ym}}^{-1 } \sum_{j_1,\dots , j_6}\eta_{j_1\cdots j_6}[\bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x ) , \partial_\mu \bm{n}_{j_6}(x ) ] \cdots ] \nonumber\\ = & \sum_{j=1,2 } ( \mathscr{a}_\mu(x),\bm{n}_{j}(x))\bm{n}_{j}(x ) + \sum_{j_1,\dots , j_6 } \xi_{j_1\dots j_6 } [ \bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x),[\bm{n}_{j_6}(x),\mathscr{a}_\mu(x)]]\cdots ] \nonumber\\ & + ig_{{}_{\rm ym}}^{-1 } \sum_{j_1,\dots , j_6}\eta_{j_1\cdots j_6}[\bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x ) , \partial_\mu \bm{n}_{j_6}(x ) ] \cdots ] , \end{aligned}\ ] ] where we have applied the the formula ( [ f - decomp ] ) to @xmath223 in the last step .
therefore , @xmath224 is decomposed into @xmath233 and @xmath234 : @xmath235\cdots ] , \nonumber\\ \mathscr{b}_\mu(x ) : = & ig_{{}_{\rm ym}}^{-1 } \sum_{j_1,\dots , j_6}\eta_{j_1\cdots j_6}[\bm{n}_{j_1}(x),\cdots,[\bm{n}_{j_5}(x ) , \partial_\mu \bm{n}_{j_6}(x ) ] \cdots ] \label{decom_cb } .\end{aligned}\ ] ] for the sake of convenience , we define the field @xmath236 for the highest weight @xmath237 by @xmath238 here @xmath233 commutes with @xmath236 : @xmath239 = 0 \label{cm } , \end{aligned}\ ] ] while @xmath234 is orghogonal to @xmath231 : @xmath240 the first term in the right - hand side of @xmath233 corresponds to the element of the cartan subalgebra @xmath7 and the second term to the remaining part @xmath241 which vanishes when the maximal stability group coincides with the maximal torus group @xmath242 ( this is the case for the maximal option of @xmath1 ) .
notice that @xmath234 is the extension of the @xmath1 cho connection to @xmath0 .
an appropriate set of the above fields will be used in the reformulation of the @xmath0 yang - mills theory .
we suppose that the dominant mode for quark confinement is the restricted field @xmath224 extracted from the original @xmath0 yang - mills field @xmath223 through the decomposition given in the above .
in fact , this observation is exemplified for the @xmath0 wilson loop operator by using the non - abelian stokes theorem in the same manner as in @xmath1 , as given in the next section .
in this section , we derive the non - abelian stokes theorem for the wilson loop operator in an arbitrary representation of @xmath0 gauge group using the color direction fields @xmath243 . before proceeding to the case of the gauge group @xmath0
, we discuss the general case .
it is known @xcite that the wilson loop operator defined for any lie algebra valued yang - mills field @xmath3 and the irreducible ( unitary ) representation @xmath50 is cast into the following ( path - integral ) representation in ( [ wl ] ) as @xmath244)\bigr)\ } , \notag\\ & r(\bm m(x ) ) = \lambda_j r(g(x))r(h_j)r(g(x))^\dag = \lambda_j r({\rm ad}_{g(x)}(h_j ) ) , \notag\\ & \omega_\mu(x ) = ig^{-1}_{{}_{\rm ym}}r(g(x))\partial_\mu r(g(x))^\dag . \notag\end{aligned}\ ] ] to simplify the notation , we omit the symbol @xmath74 throughout this section , appendix [ sec : dnast ] and [ sec : c ] . ] : @xmath245 = \int[d\mu(g)]_\sigma \exp \left [ -ig_{{}_{\rm ym } } \int_{\sigma:\partial\sigma = c } f^g\right ] , \notag\\ & f^g = \frac12 f^g_{\mu\nu } dx^\mu\wedge dx^\nu \notag\\ & f^g_{\mu\nu}(x ) = \kappa\ { \partial_\mu { \rm tr}(\bm{m}(x)\mathscr{a}_\nu(x ) ) - \partial_\nu { \rm tr}(\bm{m}(x)\mathscr{a}_\mu(x ) ) + ig_{{}_{\rm ym } } { \rm tr}(\bm{m}(x ) [ \omega_\mu(x),\omega_\nu(x)])\ } , \notag\\ & \omega_\mu(x ) : = ig_{{}_{\rm ym}}^{-1 } g(x)\partial_\mu g^\dag(x ) , \quad g(x)\in g , \label{wl}\end{aligned}\ ] ] where @xmath246_\sigma$ ] is the product measure of the haar measure on the gauge group @xmath5 over @xmath247 and @xmath66 in @xmath248 is the highest weight vector of the representation @xmath50 . here
the gauge - invariant field strength @xmath249 is equal to the non - abelian field strength @xmath250:=\partial_\mu \mathscr{v}_\nu- \partial_\nu \mathscr{v}_\mu -ig_{{}_{\rm ym } } [ \mathscr{v}_\mu , \mathscr{v}_\nu]$ ] of the restricted field @xmath251 ( in the decomposition @xmath8 ) projected to the color field @xmath252 : @xmath253\ } = \lambda_j f_{\mu\nu}^{(j ) } , \quad f_{\mu\nu}^{(j ) } = { \rm tr}\ { \bm{n}_j \mathscr{f}_{\mu\nu } [ \mathscr{v}]\ } .
\label{v } \ ] ] therefore , the restricted field @xmath251 is regarded as the dominant mode for quark confinement , since the remaining field @xmath254 does not contribute to the wilson loop operator .
the derivation of this fact is given in appendix [ sec : c ] .
let @xmath26 be an arbitrary element of the lie algebra @xmath255 .
suppose that @xmath26 is decomposed as @xmath256\cdots ] , \notag\\ \label{40 } & \mathscr{f}_{g/\tilde{h } } = \sum_{j_1,\dots , j_n } \eta_{j_1\cdots j_n}[h_{j_1},\cdots,[h_{j_n},\mathscr{f}]\cdots],\end{aligned}\ ] ] where @xmath53 is the rank of the gauge group .
at least , this relation for the decomposition has already been proved for @xmath0 in the previous section , and the method is applicable to any semi - simple compact lie group . in order to complete the non - abelian stokes theorem
, we can follow the same procedures as those for @xmath1 given in @xcite , if @xmath257 does not have the part belonging to the @xmath258 : @xmath259 this enables us to rewrite @xmath260 $ ] in terms of the color fields @xmath261 , which is indeed shown in appendix [ sec : dnast ] . the relevant relation ( [ 42 ] )
is indeed verified as follows . by applying ( [ 40 ] ) to @xmath262 , we obtain the decomposition : @xmath263\cdots ] .
\label{43}\end{aligned}\ ] ] the first term on the right - hand side vanishes , since @xmath264 ) \notag\\ & = 0,\end{aligned}\ ] ] where we have used @xmath265 in the second equality , the relation @xmath266 following from @xmath267 in the third equality and the cyclicity of the trace in the fourth equality .
in addition , by taking account of @xmath268 , \end{aligned}\ ] ] the second term is rewritten as @xmath269\cdots ] & = \sum_{i_1,\dots , i_n } \xi_{i_1\dots i_n}[h_{i_1},\cdots,[h_{i_n},[\lambda\cdot h , \partial_\mu g^\dag g]]\cdots ] \notag\\ & = \sum_{i_1,\dots , i_n } \xi_{i_1\dots i_n}[\lambda\cdot h,[h_{i_1},\cdots,[h_{i_n } , \partial_\mu g^\dag g]\cdots ] ] \notag\\ & = [ \lambda\cdot h , ( \partial_\mu g^\dag g)_{\tilde{h } } ] , \label{46}\end{aligned}\ ] ] where we have used the commuting property ( [ 18a ] ) in the second equality and ( [ 40 ] ) for @xmath270 in the third equality . by substituting the cartan decomposition of @xmath271 in @xmath146 given by @xmath272 into ( [ 46 ] ) , we find that the second term also vanishes , @xmath273
\notag\\ & = \sum_{\alpha\in \mathcal{r}_+ : e_{\pm\alpha}\in \tilde{h}}\lambda\cdot\alpha((\partial_\mu g^\dag g)_\alpha^*e_\alpha - ( \partial_\mu g^\dag g)_\alpha e_{-\alpha } ) \notag\\ & = 0,\end{aligned}\ ] ] where we have used @xmath274 for @xmath43 satisfying @xmath275 . thus we have confirmed ( [ 42 ] ) .
thus we obtain the final form of wilson loop operator as of ( [ nast - gen ] ) using the inner product instead of using the trace as @xmath276\cdots ] ] ) , \notag\end{aligned}\ ] ] so that the wilson loop depends on the representation only through the highest weight vector @xmath66 . ]
@xmath245 = \int[d\mu(g)]_\sigma \exp \left [ -ig_{{}_{\rm ym } } \int_{\sigma:\partial\sigma = c } f^g\right ] , \notag\\ & f^g = \frac12 f^g_{\mu\nu } dx^\mu\wedge dx^\nu , \notag\\ & f^g_{\mu\nu}(x ) = \kappa\ { \partial_\mu { \rm tr}(\bm{m}(x)\mathscr{a}_\nu(x ) ) - \partial_\nu { \rm tr}(\bm{m}(x)\mathscr{a}_\mu(x ) ) \notag\\ & \qquad \qquad+ ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n } { \rm tr}(\bm{m}(x)[\partial_\mu\bm{n}_{i_1}(x),[\bm{n}_{i_2},\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\nu\bm{n}_{i_n}(x)]\cdots]])\}. \label{nast - gen}\end{aligned}\ ] ] the detail of the derivation of ( [ nast - gen ] ) is given in appendix [ sec : dnast ] , which is almost the same as that given in @xcite for @xmath1 , once ( [ 42 ] ) is established . in each case of representations , we can write the new form for the wilson loop operator using the decomposition formula based on the above general consideration : 1 . for @xmath277
, the wilson loop operator is written as ( [ nast - gen ] ) where @xmath278 and @xmath279 where @xmath195 is defined in ( [ zeta ] ) .
2 . for @xmath280 , the wilson loop operator is written as ( [ nast - gen ] ) where @xmath278 and @xmath281 where @xmath203 is defined in ( [ zeta ] ) .
3 . for @xmath282
, the wilson loop operator is written as ( [ nast - gen ] ) where @xmath278 and @xmath283 where @xmath209 is defined in ( [ zeta ] ) .
+ by using @xmath284 , another form is obtained as @xmath285\cdots]])\ } \notag\\ & f^{(2)}_{\mu\nu}(x ) = \kappa\ { \partial_\mu { \rm tr}(\bm{n}_2(x)\mathscr{a}_\nu(x ) ) - \partial_\nu { \rm tr}(\bm{n}_2(x)\mathscr{a}_\mu(x ) ) \notag\\ & \qquad \qquad+ ig_{{}_{\rm ym}}^{-1}\sum_{j_1,\dots , j_6 } \zeta_{j_1\cdots j_6 } { \rm tr}(\bm{n}_2(x)[\partial_\mu\bm{n}_{j_1}(x),[\bm{n}_{j_2},\cdots,[\bm{n}_{j_{5}}(x),\partial_\nu\bm{n}_{j_6}(x)]\cdots]])\}.\end{aligned}\ ] ]
we can define magnetic - monopole current @xmath42 as the co - differential of the hodge dual of @xmath286 : @xmath287 in the @xmath288-dimensional spacetime , @xmath42 is expressed by a differential form , @xmath289-form . for @xmath290 , especially , the magnetic monopole current reads @xmath291 then , the magnetic charge @xmath292 is defined by @xmath293 we examine the quantization condition for the magnetic charge .
the magnetic charge can have nonzero value because the map defined by @xmath294 has the nontrivial homotopy group : @xmath295 because the value of the magnetic charge depends only on the topological character of @xmath296 , we can use specific group elements @xmath297 to obtain the quantization condition for the magnetic charge .
now , we consider a case in which @xmath213 belongs to @xmath31 . in this case , @xmath249 reduces to @xmath298 + \bm{n}_1[\partial_\mu\bm{n}_2,\partial_\nu\bm{n}_2])\ } , \notag\\ f^{(2)}_{\mu\nu } = & \kappa\{\partial_\mu{\rm tr}(\bm{n}_2\mathscr{a}_\nu ) - 2\partial_\nu{\rm tr}(\bm{n}_2\mathscr{a}_\mu ) - ig_{{}_{\rm ym}}^{-1 } { \rm tr}(\bm{n}_2[\partial_\mu\bm{n}_1,\partial_\nu\bm{n}_1 ] + \bm{n}_2[\partial_\mu\bm{n}_2,\partial_\nu\bm{n}_2])\ } \nonumber\\ = & \kappa \left\{\partial_\mu{\rm tr}(\bm{n}_2\mathscr{a}_\nu ) - 2\partial_\nu{\rm tr}(\bm{n}_2\mathscr{a}_\mu ) - \frac{4}{3 } ig_{{}_{\rm ym}}^{-1 } { \rm tr } ( \bm{n}_2[\partial_\mu\bm{n}_2,\partial_\nu\bm{n}_2 ] ) \right\}.\end{aligned}\ ] ] here notice that two field strengths @xmath299 and @xmath300 appear in the non - abelian stokes theorem for @xmath31 .
it is shown @xcite that the two kinds of the gauge - invariant charges @xmath301 and @xmath302 obey the different quantization conditions : @xmath303 thus , we obtain the quantization condition for the magnetic charge in @xmath0 : @xmath304 where we have defined @xmath305 , which can take an arbitrary integer . the observation based on the homotopy group ( [ 77 ] )
that there need to be two integers in @xmath292 .
there exist already two integers in @xmath292 .
therefore , it is enough to consider a case @xmath306 for deriving the quantization condition for the magnetic charge in @xmath0 .
for the exceptional group @xmath0 , we have first shown that there exist three cases of the maximal stability subgroup .
then , we have derived the gauge - covariant decomposition formula which is written using the multiple commutators with the color direction fields , in accord with each stability group .
moreover , we have obtained the non - abelian stokes theorem for the wilson loop operator that is written in terms of the relevant color direction fields .
these results indicate that there exist three options for the reformulation of the @xmath0 yang - mills theory . in any option , we need the two kinds of color fields , since the two cartan generators are inevitably required in the decomposition formula , in marked contrast to the minimal option of @xmath1 group .
nevertheless , each option would be utilized for describing confinement of quarks in the relevant representation of @xmath0 .
this is because the the non - abelian stokes theorem for the wilson loop operator is attributed to the decomposition formula available to a given representation .
this would be confirmed more explicitly when the reformulation is ready to be checked .
the method we have used in this paper for obtaining the decomposition formula would be so general that the decomposition formula is written for any semi - simple lie group using the multiple commutators with the cartan generators .
in addition , once the decomposition formula given in the above is obtained , we can immediately obtain the expression of the wilson loop operator written in terms of the color direction fields , because we derived the non - abelian stokes theorem in a general way .
this observation suggests that the reformulation of the yang - mills theory with an arbitrary semi - simple gauge group would be possible .
this work is supported by grants - in - aid for scientific research ( c ) no.24540252 and ( c ) no.15k05042 from the japan society for the promotion of science ( jsps ) .
in section [ sec : decom ] , we have seen that the sextuple commutator is used to obtain the decomposition formula for @xmath0 . in this appendix
we show that such a formula for @xmath0 can not be obtained by taking the commutator less than six times . taking the commutator odd number of times , we obtain @xmath307\cdots ] = \sum_{\alpha\in\mathcal{r}_+ } ( \alpha_{j_1}\cdots\alpha_{j_{2n+1 } } \mathscr{f}_\alpha^ * e_\alpha - \alpha_{j_1}\cdots\alpha_{j_{2n+1 } } \mathscr{f}_\alpha e_{-\alpha } ) .\end{aligned}\ ] ] in this expression the sign of the term @xmath80 is different from the sign of the term @xmath77 and therefore this is not appropriate .
thus we see that we just need to consider the cases of double and quadruple commutators .
first , we consider the case of double commutators .
we can decompose an arbitrary element @xmath26 of the lie algebra if and only if there are real numbers @xmath308 , @xmath309 and @xmath310 that satisfy @xmath311 + k_2[h_1,[h_2,\mathscr{f } ] ] + k_3[h_2,[h_2,\mathscr{f } ] ] = \sum_{\alpha\in\mathcal{r}_+:e_{\pm\alpha}\notin\tilde{h}}(\mathscr{f}_\alpha^ * e_\alpha + \mathscr{f}_\alpha e_{-\alpha}).\end{aligned}\ ] ] using the cartan decomposition of @xmath26 , we find that the left hand side of ( [ a1 ] ) is equal to @xmath312 where we put @xmath313
. hence ( [ a1 ] ) is equivalent to @xmath314 in the case of @xmath315 , the three equations ( [ a4 ] ) for @xmath128 , @xmath132 and @xmath138 can be written in a matrix form as @xmath316 the solution of this equation is @xmath317 , @xmath318 .
this solution is consistent with the decomposition formula for @xmath31 .
but , these values of @xmath308 , @xmath309 and @xmath310 do not satisfy the equation ( [ a4 ] ) for @xmath319 , @xmath320 and @xmath321 . for example , the equation ( [ a4 ] ) for @xmath320 is given by @xmath322 which is however not satisfied by @xmath318 .
therefore , there is no solution for all of ( [ a4 ] ) . in the case of @xmath323 and of @xmath206 , the equation ( [ a4 ] ) for @xmath128 , @xmath132 and @xmath138
reads @xmath324 the solution of this equation is @xmath325 , @xmath326 .
this solution is also consistent with the decomposition formula for @xmath31 .
but this does not satisfy the equation ( [ a4 ] ) for @xmath319 , @xmath320 and @xmath321 .
therefore , also in this case , there are no solutions for all of ( [ a4 ] ) .
thus we confirm that there are no decomposition formulae using double commutators for all representations .
next , we consider the case of quadruple commutators .
there exits the decomposition formulae if and only if there are real numbers @xmath308 , @xmath309 , @xmath310 , @xmath327 and @xmath328 that satisfy @xmath329 ] ] + k_2[h_1,[h_1,[h_1,[h_2,\mathscr{f } ] ] ] ] + k_3[h_1,[h_1,[h_2,[h_2,\mathscr{f } ] ] ] ] + k_4[h_1,[h_2,[h_2,[h_2,\mathscr{f } ] ] ] ] \notag\\ & + k_5[h_2,[h_2,[h_2,[h_2,\mathscr{f } ] ] ] ] = \sum_{\alpha\in\mathcal{r}_+:e_{\pm\alpha}\notin\tilde{h}}(\mathscr{f}_\alpha^ * e_\alpha + \mathscr{f}_\alpha e_{-\alpha}).\end{aligned}\ ] ] this is equivalent to @xmath330 the equivalent matrix form is given as @xmath331 calculating the rank of the matrix , we see that these equations do not have the solutions .
thus we have confirmed that quadruple commutators are not enough to obtain the desired decomposition formula .
from the fact ( [ 42 ] ) that the @xmath19 part of @xmath332 is vanishing , we have @xmath333\cdots].\end{aligned}\ ] ] multiplying both sides of this equation by @xmath297 from the left and by @xmath334 from the right , we obtain @xmath335\cdots ] \notag\\ & = \sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n}[\bm{n}_{i_1}(x),\cdots,[\bm{m}(x),\partial_\mu\bm{n}_{i_n}(x)]\cdots ] \notag\\ & = \sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n}[\bm{m},[\bm{n}_{i_1}(x),\cdots,[\bm{n}_{i_{n-1}},\partial_\mu\bm{n}_{i_n}(x)]\cdots ] ] \notag\\ & = ig_{{}_{\rm ym } } [ \mathscr{b}_\mu(x ) , \bm{m}(x ) ] , \label{49}\end{aligned}\ ] ] where we have used @xmath336 = [ \bm{m},\partial_\mu\bm{n}_i]$ ] following from @xmath337 = 0 $ ] in the second equality , the commuting property ( [ 18a ] ) in the third equality , and we have introduced @xmath338\cdots ] .
\end{aligned}\ ] ] in last equality . on the other hand ,
we find @xmath339 .
\label{51}\end{aligned}\ ] ] combining ( [ 49 ] ) and ( [ 51 ] ) , we obtain @xmath340 = [ \mathscr{b}_\mu(x),\bm{m}(x ) ] .\label{b5}\end{aligned}\ ] ] using this relation , we can rewrite the third term in @xmath341 as @xmath342 ) & = ig_{{}_{\rm ym } } { \rm tr}([\bm{m}(x ) , \omega_\mu(x)]\omega_\nu(x ) ) \notag\\ & = ig_{{}_{\rm ym } } { \rm tr}([\bm{m}(x ) , \mathscr{b}_\mu(x)]\omega_\nu(x ) ) \notag\\ & = ig_{{}_{\rm ym } } { \rm tr } ( [ \omega_\nu(x),\bm{m}(x)]\mathscr{b}_\mu(x ) ) \notag\\ & = { \rm tr}(\partial_\nu\bm{m}(x)\mathscr{b}_\mu(x ) ) \notag\\ & = ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n } { \rm tr}(\partial_\nu\bm{m}(x)[\bm{n}_{i_1}(x),\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\mu\bm{n}_{i_n}(x)]\cdots ] ) \notag\\ & = ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n } { \rm tr}([\partial_\nu\bm{m}(x),\bm{n}_{i_1}(x)][\bm{n}_{i_2},\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\mu\bm{n}_{i_n}(x)]\cdots ] ) \notag\\ & = -ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n } { \rm tr}([\bm{m}(x),\partial_\nu\bm{n}_{i_1}(x)][\bm{n}_{i_2},\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\mu\bm{n}_{i_n}(x)]\cdots ] ) \notag\\ & = -ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots
i_n } { \rm tr}(\bm{m}(x)[\partial_\nu\bm{n}_{i_1}(x),[\bm{n}_{i_2},\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\mu\bm{n}_{i_n}(x)]\cdots]])\notag\\ & = ig_{{}_{\rm ym}}^{-1}\sum_{i_1,\dots , i_n } \eta_{i_1\cdots i_n } { \rm tr}(\bm{m}(x)[\partial_\mu\bm{n}_{i_1}(x),[\bm{n}_{i_2},\cdots,[\bm{n}_{i_{n-1}}(x),\partial_\nu\bm{n}_{i_n}(x)]\cdots]]),\end{aligned}\ ] ] where we have used the cyclicity of the trace in first , third , sixth and eighth equality , ( [ 51 ] ) in fourth equality , @xmath343 = -[\bm{m},\partial_\nu\bm{n}_{i_1}]$ ] following from @xmath344=0 $ ] in seventh equality and the fact that the first expression is anti - symmetric in @xmath345 and @xmath124 in the last equality .
this completes the proof of the non - abelian stokes theorem ( [ nast - gen ] ) .
in this appendix , we show that the remaining field @xmath347 does not contribute to the wilson loop operator by deriving the equality ( @xmath348 ) . using the decomposition ( [ decom_cb ] )
, we obtain @xmath349 ) & = { \rm
tr}(\bm m(\partial_\mu\mathscr c_\nu -\partial_\nu\mathscr c_\mu -ig_{{}_{ym}}[\mathscr c_\mu , \mathscr c_\nu ] -ig_{{}_{ym}}[\mathscr c_\mu , \mathscr b_\nu ] -ig_{{}_{ym}}[\mathscr b_\mu , \mathscr c_\nu ] \notag\\ & \quad+ \partial_\mu\mathscr b_\nu -\partial_\nu\mathscr b_\mu -ig_{{}_{ym}}[\mathscr b_\mu , \mathscr b_\nu ] ) .\end{aligned}\ ] ] from the fact ( [ cm ] ) we see that the third , fourth and fifth terms vanish .
thus we obtain @xmath350 ) = { \rm tr}(\bm m(\partial_\mu\mathscr c_\nu -\partial_\nu\mathscr c_\mu + \partial_\mu\mathscr b_\nu - \partial_\nu\mathscr b_\mu -ig_{{}_{\rm ym}}[\mathscr b_\mu,\mathscr b_\nu])).\label{c1}\end{aligned}\ ] ] the first term of ( [ c1 ] ) reads @xmath351 where we have used @xmath352 and @xmath353 .
the third term of ( [ c1 ] ) reads @xmath354\mathscr b_\nu ) \notag\\ & = ig_{{}_{\rm ym}}{\rm tr}(\bm m[\mathscr b_\mu,\mathscr b_\nu ] ) , \label{c3}\end{aligned}\ ] ] where we have used @xmath355 in the first equality , ( [ 49 ] ) in the second equality and the cyclicity of the trace in the last equality .
thus we obtain @xmath356 ) \notag\\ & = \partial_\mu{\rm tr}(\bm m\mathscr a_\nu ) - \partial_\nu{\rm
tr}(\bm m\mathscr a_\mu ) + ig_{{}_{\rm ym } } { \rm tr}(\bm m[\omega_\mu , \omega_\nu])\notag\\ & = \frac1\kappa f^g_{\mu\nu } , \end{aligned}\ ] ] where we have used ( [ c2 ] ) and ( [ c3 ] ) in the first equality and ( [ b5 ] ) in the second equality
. l. faddeev and a.j .
niemi , phys .
b*449 * , 214218 ( 1999 ) .
[ hep - th/9812090 ] + l. faddeev and a.j .
niemi , phys .
b*464 * , 9093 ( 1999 ) .
[ hep - th/9907180 ] + t.a .
bolokhov and l.d .
faddeev , theoretical and mathematical physics , * 139 * , 679692 ( 2004 ) .
y. nambu , phys .
d*10 * , 42624268 ( 1974 ) .
+ g. t hooft , in : high energy physics , edited by a. zichichi ( editorice compositori , bologna , 1975 ) .
+ s. mandelstam , phys .
report*23 * , 245249 ( 1976 ) .
+ a.m. polyakov , phys .
b*59 * , 8284 ( 1975 ) .
b*120 * , 429458 ( 1977 ) .
l. del debbio , m. faber , j. greensite , and s. olejnik , phys .
rev . d*55 * , 22982306 ( 1997 ) hep - lat/9610005 , + m. faber , j. greensite , and s. olejnik , phys .
rev . d*57 * , 26032609 ( 1998 ) .
[ hep - lat/9710039 ] + l. del debbio , m. faber , j. giedt , j. greensite , and s. olejnik , phys .
rev . d*58 * , 094501 ( 1998 ) .
[ hep - lat/9801027 ]
|
we give a gauge - covariant decomposition of the yang - mills field with an exceptional gauge group @xmath0 , which extends the field decomposition invented by cho , duan - ge , and faddeev - niemi for the @xmath1 yang - mills field . as an application of the decomposition
, we derive a new expression of the non - abelian stokes theorem for the wilson loop operator in an arbitrary representation of @xmath0 .
the resulting new form is used to define gauge - invariant magnetic monopoles in the @xmath0 yang - mills theory .
moreover , we obtain the quantization condition to be satisfied by the resulting magnetic charge . the method given in this paper
is general enough to be applicable to any semi - simple lie group other than @xmath1 and @xmath0 .
|
the field of micro - machining is forcing a profound redefinition of the nature and attributes of electronic devices .
this technology allows fabrication of a variety of on - chip fully integrated micromechanical sensors and actuators with a rapidly growing range of applications . in many cases ,
it is highly desirable to shrink the size of mechanical elements down to the nano - scale @xcite .
this allows enhancing the speed of operation by increasing the frequencies of mechanical resonances and improving their sensitivity as sensors .
furthermore , as devices become smaller , their power consumption decreases and the cost of mass fabrication can be significantly lowered .
some key applications of microelectromechanical systems ( mems ) technology include magnetic resonance force microscopy ( mrfm ) @xcite and mass - sensing @xcite .
further miniaturization is also motivated by the quest for mesoscopic quantum effects in mechanical systems .
nonlinear effects are of great importance for micromechanical devices .
the relatively small applied forces needed for driving a micromechanical oscillator into the nonlinear regime are usually easily accessible @xcite .
thus , a variety of useful applications such as frequency synchronization @xcite , frequency mixing and conversion @xcite , parametric and intermodulation amplification @xcite , mechanical noise squeezing @xcite , stochastic resonance @xcite , and enhanced sensitivity mass detection can be implemented by applying modest driving forces .
furthermore , monitoring the displacement of a micromechanical resonator oscillating in the linear regime may be difficult when a displacement detector with high sensitivity is not available .
thus , in many cases the nonlinear regime is the only useful regime of operation .
another key property of systems based on mechanical oscillators is the rate of damping .
for example , in many cases the sensitivity of nems sensors is limited by thermal fluctuation , which is related to damping via the fluctuation dissipation theorem .
in general , micromechanical systems suffer from low quality factors @xmath0 relative to their macroscopic counterparts @xcite .
however , very little is currently known about the underlying physical mechanisms contributing to damping in these devices .
a variety of different physical mechanisms can contribute to damping , including bulk and surface defects @xcite , thermoelastic damping , nonlinear coupling to other modes , phonon - electron coupling , clamping loss @xcite , interaction with two level systems @xcite , etc .
identifying experimentally the contributing mechanisms in a given system can be highly challenging , as the dependence on a variety of parameters has to be examined systematically .
the archetypical model used to describe nonlinear micro- and nanomechanical oscillators is the duffing oscillator @xcite .
this model has been studied in great depth , and special emphasis has been given to the dynamics of the system near the bifurcation points . in order to describe dissipation processes , a linear damping model is usually employed , either as a phenomenological ansatz , or in the form of linear coupling to thermal bath , which represents the environment . however , nonlinear damping is known to be significant at least in some cases . for example , the effect of nonlinear damping for the case of strictly dissipative force , being proportional to the velocity to the nth power , on the response and bifurcations of driven duffing and other types of nonlinear oscillators @xcite has been studied extensively .
also , nonlinear damping plays an important role in parametrically excited mechanical resonators where without it , solutions will grow without bound . in spite of the fact that a massive body of literature exists which discusses the nonlinear elastic effects in micro- and nanomechanical oscillators as well as the consequences of nonlinear damping
, the quantitative experimental data on systems with nonlinear damping , especially those nearing bifurcation points , remains scarce .
furthermore , such systems impose special requirements on the experiment parameters and procedures , mainly due to the very slow response times near the bifurcation points .
straightforward evaluation of these requirements by simple measurements can facilitate accurate data acquisition and interpretation . in the present paper we study damping in a micromechanical oscillator operating in the nonlinear regime excited by an external periodic force at frequencies close to the mechanical fundamental mode .
we consider a duffing oscillator nonlinearly coupled to a thermal bath .
this coupling results in a nonlinear damping force proportional to the velocity multiplied by the displacement squared . as will be shown below
, this approach is equivalent to the case where the damping nonlinearity is proportional to the velocity cubed . in conjunction with a linear dissipation term
, it has also been shown to describe an effective quadratic drag term @xcite .
we find that nonlinear damping in our micromechanical oscillators is non - negligible , and has a significant impact on the oscillators response .
furthermore , we develop a theoretical one - dimensional model of the oscillator s behavior near the bifurcation point .
most of the parameters that govern this behavior can be estimated straightforwardly from frequency response measurements alone , not requiring exact measurement of oscillation amplitudes .
measuring these parameters under varying conditions provides important insights into the underlying physical mechanisms .
we use our results to estimate different dynamic parameters of an experimentally measured micromechanical beam response , and show how these estimations can be used to increase the accuracy of experimental measurements and to estimate measurement errors .
the main source of error is found to be the slowing down behavior near the bifurcation point , also known as the saddle node `` ghost '' @xcite .
we also investigate the possibility of thermal escape of the system from a stable node close to the bifurcation point and find that the probability of this event in our experiments is negligible . finally , we propose and analyze a continuum mechanics model of our micromechanical oscillator as a planar , weakly nonlinear strongly pretensioned , viscoelastic beam - string .
the analysis of this model illustrates a possible cause for non negligible nonlinear damping as observed in the experiment .
for the experiments we employ micromechanical oscillators in the form of doubly clamped beams made of pdau ( see fig .
[ fig : exp_setup ] ) .
the device is fabricated on a rectangular silicon - nitride membrane ( side length 100 - 200 ) by the means of electron beam lithography followed by thermal metal evaporation .
the membrane is then removed by electron cyclotron resonance ( ecr ) plasma etching , leaving the doubly clamped beam freely suspended .
the bulk micro - machining process used for sample fabrication is similar to the one described in .
the dimensions of the beams are : length 100 - 200 , width 0.25 - 1and thickness 0.2 , and the gap separating the beam and the electrode is 5 - 8 .
measurements of all mechanical properties are done _ in - situ _ by a scanning electron microscope ( sem ) ( working pressure @xmath1 ) , where the imaging system of the microscope is employed for displacement detection .
some of the samples were also measured using an optical displacement detection system described elsewhere @xcite . driving force
is applied to the beam by applying a voltage to the nearby electrode . with a relatively modest driving force
, the system is driven into the region of nonlinear oscillations .
[ htb ] we use a network analyzer for frequency domain measurements , as shown in fig .
[ fig : exp_setup_na ] . for time domain measurements of the slow varying envelope
we employ a lock - in amplifier , connected as show in fig .
[ fig : exp_setup_lockin ] .
the mechanical oscillator is excited by a monochromatic wave , whose amplitude is modulated by a square wave with low frequency ( 20 - 50 ) .
this results in bursting excitation , which allows measurement of ring - down behavior in time domain .
the lock - in amplifier is locked to the excitation frequency , and measures the amplitude of the slow envelope of the oscillator s response .
the lock - in amplifier time constant should be much smaller than the ring down time , which is governed by dissipation in the micromechanical system .
typically , in our experiments , the time constant is 100and the characteristic ring down time is 10 .
[ htb ] .,width=326 ] [ htb ] the displacement detection scheme described above is not exactly linear , because the amount of the detected secondary electrons or reflected light is not strictly proportional to the mechanical oscillator amplitude , but merely a monotonic function of the latter .
nonuniform distribution of primary electrons or light power in the spot increases this nonlinearity even further .
thus , some distortion in the measured response amplitude is introduced .
we excite the system close to its fundamental mode .
ignoring all higher modes allows us to describe the dynamics using a single degree of freedom @xmath2 .
in the main part of this study , no assumptions are made about the source of linear and nonlinear dissipation .
the energy dissipation is modeled phenomenologically by coupling the micromechanical oscillator to a thermal bath consisting of harmonic oscillators .
physically , several processes may be responsible for mechanical damping , including thermoelastic effects , friction at grain boundaries @xcite , bulk and surface impurities , electrical losses , clamping loss , etc .
we also regard the linear and nonlinear damping constants as independent of one another , although they probably result from same physical processes . in sec .
[ sec : nonlin_damping_source ] we consider one possible model connecting the linear and nonlinear dissipation coefficients , and compare its predictions to experimental data .
the hamiltonian of the system , which includes the mechanical beam and thermal bath modes coupled to it , is @xmath3 where @xmath4 describe the micromechanical beam , the thermal bath , and the interaction between them , respectively .
here , @xmath5 is the effective mass of the fundamental mode of the micromechanical beam , and @xmath6 and @xmath2 are the effective momentum and displacement of the beam .
also , @xmath7 is the elastic potential , and @xmath8 is the capacitive energy , where @xmath9 is the displacement dependent capacitance , @xmath10 is the gap between the electrode and the beam , and @xmath11 is the time dependent voltage applied between the electrode and the micromechanical beam .
the sum @xmath12 denotes summing over all relevant thermal bath modes , while @xmath13 is the frequency of one of the modes in the thermal bath with effective momentum @xmath14 and displacement @xmath15 , and @xmath16 is the effective mass of the same mode . finally , @xmath17 is a function describing the interaction strength of each thermal bath mode with the fundamental mode of the micromechanical beam .
the equations of motion resulting from are @xmath18 the formal solution of can be written as @xmath19 or , integrating by parts , @xmath20 where @xmath21 and @xmath22 are the initial conditions of the thermal mode displacement and velocity , respectively ; and @xmath23 denotes the coupling strength function @xmath17 evaluated at time @xmath24 . substituting into , one gets @xmath25 where @xmath26 is the noise , @xmath27 is the renormalized potential , and @xmath28 is the memory kernel . also , the initial slip term given by @xmath29 has been dropped @xcite . finally , the noise autocorrelation for an initial thermal ensemble is @xmath30 where @xmath31 is the effective temperature of the bath , and @xmath32 is the boltzmann s constant .
the last result is a particular form of the fluctuation - dissipation theorem .
we employ a nonlinear , quartic potential @xmath33 in order to describe the elastic properties of the micromechanical beam oscillator .
assuming @xmath17 to be polynomial in @xmath2 , it can be deduced from that only linear and quadratic terms in @xmath17 should be taken into account , i.e. , @xmath34 the memory kernel in this case is @xmath35 making the usual markovian ( short - time noise autocorrelation ) approximation , i.e. , @xmath36 , one obtains @xmath37 and the equation of motion becomes @xmath38 where @xmath39 is the linear damping constant , @xmath40 and @xmath41 are the nonlinear damping constants , @xmath42 is the linear spring constant and @xmath43 is the nonlinear spring constant . some clarifications regarding are in order .
the quadratic dissipation term @xmath44 has been dropped from the equation because it has no impact on the first order multiple scales analysis , which will be applied below .
an additional dissipation term proportional to the cubed velocity , @xmath45 , has been added artificially .
such term , although not easily derived using the analysis sketched above , may be required to describe some macroscopic friction mechanisms , such as losses associated with nonlinear electrical circuits .
it will be shown below that the impact of this term on the behavior of the system is very similar to the impact of @xmath46 .
the applied voltage is composed of large constant ( dc ) and small monochromatic components , namely , @xmath47 .
the one dimensional equation of motion can be rewritten as @xmath48 where @xmath49 , @xmath50 , @xmath51 , @xmath52 , and @xmath53 . in order to investigate the dynamics described by the equation of motion analytically
, we use the fact that nonlinearities of the micromechanical oscillator and the general energy dissipation rate are usually small ( as shown in sec .
[ sec : results ] , the linear quality factor in our systems has a typical value of several thousands ) . in the spirit of the standard multiple scales method @xcite , we introduce a dimensionless small parameter @xmath54 in , and regard the linear damping coefficient @xmath55 , the nonlinear damping coefficients @xmath56 and @xmath57 , the nonlinear spring constant @xmath58 , and the excitation amplitude @xmath59 as small .
it is also assumed that the maximal amplitude of mechanical vibrations is small compared to the gap between the electrode and the mechanical beam @xmath10 , i.e. , @xmath60 .
also , the frequency of excitation @xmath61 is tuned close to the fundamental mode of mechanical vibrations , namely , @xmath62 , where @xmath63 is a small detuning parameter .
retaining terms up to first order in @xmath54 in gives @xmath64\\ = f+2\epsilon \tilde f_0\cos \omega t , \label{eq : motion_with_epsilon}\end{gathered}\ ] ] where @xmath65 , and @xmath66 .
we have dropped the noise from the equation of motion , and will reintroduce its averaged counterpart later in the evolution equation . following @xcite , we introduce two time scales @xmath67 and @xmath68 , and assume the following form for the solution : @xmath69 it follows to the first order in @xmath54 that @xmath70 and can be separated according to different orders of @xmath54 , giving @xmath71 and @xmath72 the solution of is @xmath73 where @xmath74 is a complex amplitude and @xmath75 denotes complex conjugate .
the `` slow varying '' amplitude @xmath74 varies on a time scale of order @xmath76 or slower .
the secular equation @xcite , which follows from substitution of into , is @xmath77 where @xmath78 and @xmath79 where @xmath80 represents a constant shift in linear resonance frequency due to the constant electrostatic force @xmath81 .
equation is also known as evolution equation .
note that we have returned to the full physical quantities , i.e , dropped the tildes , for convenience .
also , one must always bear in mind that the accuracy of the evolution equation is limited to the assumptions considered at the beginning of this section . as was mentioned earlier
, both nonlinear dissipation terms give rise to identical terms in the evolution equation .
therefore , the behavior of these two dissipation cases is similar near the fundamental resonance frequency @xmath82 .
also , note that linear dissipation coefficient @xmath83 is not constant , but is rather quadratically dependent on the constant electrostatic force @xmath81 due to the nonlinear dissipation term @xmath84 .
the secular equation can be written as @xmath85 where dot denotes differentiation with respect to ( slow ) time , @xmath86 and @xmath87 is the averaged noise process with the following characteristics : @xmath88 the steady state amplitude can be found by setting @xmath89 , @xmath90 and taking a square of the evolution equation , resulting in @xmath91 this cubic equation of @xmath92 can have either one , two , or three different real roots , depending on the values of the detuning parameter @xmath93 and the excitation amplitude @xmath94 .
when @xmath95 is sufficiently small , i.e. , @xmath96 , the solutions of behave very much like the ordinary duffing equation solutions , to which reduces if @xmath97 and @xmath98 ( see fig . [ fig : duffing ] ) . .
in case @xmath99 ( where @xmath100 is some critical excitation force , dependent on the system parameters , see text ) , only one real solution exists and no bistability is possible . in case @xmath101
, the system is on the edge of bistability and one point exists , where @xmath92 vs. @xmath61 has an infinite slope . in case @xmath102 ,
the system is in bistable regime having three real solutions over some range of frequencies .
two of these solutions are stable .
the dashed line denotes the unstable solution.,width=326 ] the solution of can be also presented in polar form @xcite @xmath103 where @xmath104 and @xmath105 are real , and @xmath104 is assumed to be positive . separating the real and imaginary parts of , one obtains ( omitting the noise ) @xmath106 [ eq : evolution_polar ]
steady state solutions are defined by @xmath107 , @xmath108 , which results in .
the maximal amplitude @xmath109 can be found from by requiring @xmath110 where @xmath111 is the corresponding excitation frequency detuning .
this results in @xmath112 interestingly enough , the phase @xmath105 of the maximal response is always equal @xmath113 , i.e. , the maximal response is exactly out of phase with the excitation regardless the magnitude of the excitation , a feature well known for the linear case .
this general feature can be explained as follows . for an arbitrary response amplitude @xmath104 , there exist either two or no steady state @xmath105 solutions of .
if two solutions @xmath114 and @xmath115 exist , they must obey @xmath116 , as seen from .
it follows from that these two solutions correspond to two different values of @xmath93 .
however , at the point of maximum response the two solutions coincide , resulting in @xmath117 , i.e. , @xmath118 the system s behavior qualitatively changes when parameters such as the excitation amplitude and the frequency detuning are varied , as seen in fig .
[ fig : duffing ] . the parameter values at which these qualitative changes occur are called bifurcation ( jump ) points @xcite .
a jump in amplitude is characterized by the following condition : @xmath119 or , alternatively , @xmath120 applying this condition to yields @xmath121 where @xmath122 and @xmath123 denote the frequency detuning and the amplitude at the jump point , respectively .
when the system is on the edge of bistability , the two jump points coincide and has a single real solution at the point of critical frequency @xmath124 and critical amplitude @xmath125 .
the driving force at the critical point is denoted in fig .
[ fig : duffing ] as @xmath100 .
this point is defined by two conditions @xmath126 by applying these conditions one finds @xmath127 where @xmath125 is the corresponding critical amplitude . substituting the last result back into
, one finds @xmath128 where the upper sign should be used if @xmath129 , and the lower sign otherwise .
in general , @xmath130 is always positive , but @xmath131 can be either positive or negative . therefore , @xmath132 and @xmath6 are negative if @xmath133 ( soft spring ) , and positive if @xmath129 ( hard spring ) .
it follows from that the condition for the critical point to exist is @xmath134 without loss of generality , we will focus on the case of `` hard '' spring , i.e. , @xmath129 , @xmath135 , @xmath136 , as this is the case encountered in our experiments .
when the system approaches the bifurcation points , it exhibits some interesting features not existent elsewhere in the parametric phase space . in order to investigate the system s behavior in the vicinity of the jump points ,
it is useful to rewrite the slow envelope evolution equation as a two dimensional flow @xmath137 [ eq:2d_flow ] where we have defined @xmath138 ( i.e. , @xmath139 ) , and @xmath140 [ eq:2d_flow_fg ] the real - valued noise processes @xmath141 and @xmath142 have the following statistical properties : @xmath143 [ eq : ncns_properties ] at the fixed points , the following holds : @xmath144 . a typical phase space flow of the oscillator in bistable regime is shown in fig . [ fig : bistable_flow ] .
[ htb ] and @xmath145 are stable spiral nodes , whereas @xmath146 is the saddle - node , from which two manifolds emerge @xcite .
the green dotted line is the stable manifold ( `` separatrix '' ) , which separates different basins of attraction , belonging to different stable nodes @xmath147 and @xmath145 .
the magenta thick line is the unstable manifold .
two typical phase plane trajectories are shown by arrowed thin blue lines.,width=326 ] for small displacements near an arbitrary fixed point @xmath148 , namely , @xmath149 and @xmath150 , where @xmath151 and @xmath152 , the above nonlinear flow map can be approximated by its linearized counterpart @xmath153 where @xmath154 and the excitation frequency detuning @xmath93 as well as the external excitation amplitude @xmath94 are considered constant .
the subscripts in the matrix elements denote partial derivatives evaluated at @xmath155 , for example , @xmath156 the matrix @xmath157 is , therefore , the jacobian matrix of the system evaluated at the point @xmath155 .
it is straightforward to show that @xmath158 [ eq : m_elements ] two important relations follow immediately , @xmath159 [ eq : fxfygxgy ] the linearized system retains the general qualitative structure of the flow near the fixed points @xcite , in particular both eigenvalues of the matrix @xmath157 are negative at the stable nodes , denoted as @xmath147 and @xmath145 in fig .
[ fig : bistable_flow ] . at the saddle node @xmath146 , which is not stable , one eigenvalue of @xmath157 is positive , whereas the other is negative . the discussed duffing like systems exhibit saddle point bifurcations . at the bifurcation ,
one of the stable nodes and the saddle node coincide , resulting in a zero eigenvalue in @xmath157 .
the bifurcation ( `` jump '' ) point condition is , therefore , @xmath160 , which gives the same result as in ( [ eq : jump_point ] ) .
the case of well separated stable node and saddle node is shown in fig .
[ fig : far_saddle_node ] , and the case of almost coinciding stable and not stable fixed points is shown in fig .
[ fig : close_saddle_node ] , where the oscillator is on the verge of bifurcation .
[ htb ] ) and the stable node ( @xmath145 ) are well separated .
the green dotted line is the stable manifold ( `` separatrix '' ) and the magenta thick line is the unstable manifold .
a typical phase plane trajectory is shown by the arrowed thin blue line .
the absolute value of slow envelope s rate of change @xmath161 is represented by the background color .
the darkest parts correspond to the slowest motion in the phase space . at both fixed points @xmath146 and
@xmath145 the value of @xmath161 is zero.,width=326 ] [ htb ] ) and the stable node ( @xmath145 ) are close one to another .
the green dotted line is the stable manifold ( `` separatrix '' ) , and the magenta thick line is the unstable manifold .
phase plane trajectories are shown by the thin blue lines .
the absolute value of @xmath161 is represented by the background color .
the darkest parts correspond to the slowest motion in the phase space . at both fixed points @xmath146 and @xmath145 ,
the value of @xmath161 is zero .
note that the motion in the phase space becomes essentially one - dimensional and slows down significantly in the vicinity of the stable node @xmath145.,width=326 ] we note that in general @xmath162 and the slow eigenvalue near the bifurcation point can be estimated as @xmath163 where @xmath164 is a small frequency detuning from @xmath122 , i.e. @xmath165 . if the system in bistable regime is close to bifurcation then @xmath166 and the evolution of the system almost comes to stagnation , phenomena often referred to as critical slowing down . the motion in the vicinity of the stable node becomes slow and essentially one - dimensional along the unstable manifold .
we now turn to show this analytically . at the bifurcation
points the matrix @xmath157 is singular , i.e. , @xmath167 .
consequently , the raws of the matrix are linearly dependent , i.e. , @xmath168 where @xmath169 is some real constant .
using the last result , we may rewrite at the bifurcation point as @xmath170 where we have used the fact that at any fixed point ( stable or saddle - node ) @xmath144
. however , according to , at any fixed point @xmath171 . therefore , @xmath172 at the bifurcation point , and the matrix @xmath157 can be written as @xmath173 where @xmath174 [ eq : lambda_f_and_k ] it also follows from that @xmath175 due to the singularity of matrix @xmath157 at the bifurcation point , a second order taylor expansion must be used .
the flow map can be approximated near the bifurcation point by [ eq:2d_flow_orig_coor ] @xmath176 @xmath177 where all the derivatives denoted by subscripts are evaluated at the jump point @xmath178 , and @xmath179 the above system of differential equations can be simplified by using the following rotation transformation , shown in fig .
[ fig : dxdy2eta_xi ] , @xmath180 where @xmath181 . in these new coordinates ,
the system becomes @xmath182 [ eq : xi_eta_eqs ] where @xmath183 and @xmath184 is the differentiation operator @xmath185 the noise processes @xmath186 and @xmath187 have the same statistical properties as @xmath188 and @xmath189 .
[ htb ] is the saddle point , and @xmath145 is a stable node .
the effective flow between these two points , marked by arrows , is almost parallel to the rotated coordinate @xmath190 , while the rotated coordinate @xmath191 remains essentially constant , @xmath192 .
@xmath193 is the angle of coordinate rotation .
the velocity of the flow is largest at the point @xmath157 , between the saddle point and the stable node.,width=326 ] the time evolution of the system described by the differential equations has two distinct time scales .
motion along the coordinate @xmath191 is `` fast '' , and settling time is of order @xmath194 . the time development along the coordinate @xmath190 , however , is much slower , as will be shown below . on a time scale much longer than @xmath194 , the coordinate @xmath191 can be regarded as not explicitly dependent on time .
the momentary value of @xmath191 can be approximated as @xmath195 where we have neglected all terms proportional to @xmath196 and @xmath197 .
the motion along the coordinate @xmath190 is governed by a slow evolution equation , combining which with results in @xmath198 where @xmath199 @xmath200 .
\label{eq : b } \end{gathered}\ ] ] [ eq : eta_0_and_b ] note that the noise process @xmath186 does not play a significant role in the dynamics of the system , because the system is strongly confined in @xmath191 direction .
such noise squeezing is a general feature of systems nearing saddle - point bifurcation .
two qualitatively different cases of should be recognized .
the first case is of a system in a bistable regime with a stable ( quasi stable , as we will see below ) and non stable ( saddle ) fixed points close enough to a bifurcation point . in this case , the one dimensional motion is equivalent to a motion of a massless particle in a confining cubic `` potential '' @xmath201 as shown in fig .
[ fig : one_dim_potential ] .
[ htb ] .,width=326 ] figure [ fig : one_dim_simulation_fdomain ] depicts the location of the fixed points and the bifurcation point on a frequency response curve in this case .
figure [ fig : one_dim_simulation ] shows a comparison between the exact simulation of the system s motion near the bifurcation point and the analytical result . [ htb ] and @xmath145 , and a saddle node @xmath146 in a bistable regime close to a bifurcation point .
@xmath164 , which is negative in this case , is the frequency difference between the excitation frequency and the jump point frequency @xmath202 .
the scales of the axes are arbitrary.,width=326 ] [ htb ] for different values of detuning @xmath164 .
@xmath122 is the jump point ( bifurcation ) detuning .
@xmath203 in all cases , @xmath204 and @xmath205 are the saddle node positions for different values of @xmath164 . similarly , @xmath206 and @xmath207 are the stable node positions for different values of @xmath164 .
exact values of @xmath208 are shown by solid lines .
the dashed lines are the results of analytical approximation .
the scales of the axes are arbitrary.,width=326 ] the quasi one - dimensional system described above is obviously not stable @xcite .
the rate of escape from the vicinity of the quasi stable fixed point is @xcite @xmath209 where @xmath210 characteristic time of thermal escape @xmath211 can be shown to be @xmath212 where @xmath213 [ eq : tau_therm ] this is a mean time in which the system escapes from the stable node near bifurcation point to the other stable solution of due to thermal noise @xmath187 , and the @xmath214 power law is correct as long as @xmath215 @xcite .
the second case describes a system which has undergone saddle bifurcation , i.e. , an annihilation of the stable and non stable points has occurred .
the phase plane motion close to the bifurcation point is still one dimensional , however , @xmath216 changes its sign .
therefore , the motion is not confined any more , but is still very slow in the vicinity of the bifurcation point , because @xmath217 , as follows from eq . .
the system starts converging to the single remaining stable fixed point , but is significantly slowed down , and lingers in the vicinity of the bifurcation point due to the saddle node `` ghost '' . as the system spends most of its time of convergence near the saddle node `` ghost '' , this slow time of convergence @xmath218 can be roughly estimated as @xcite @xmath219 note that @xmath220 , due to .
the analytical results presented above allow us to use data acquired in relatively simple experiments in order to estimate several important dynamic parameters of the micromechanical beam .
we note that data acquisition using e - beam or optical beam interaction with vibrating elastic element does not readily enable extraction of displacement values .
in contrast , the frequencies of important dynamical features , including maximum and jump points , can be measured with high accuracy using standard laboratory equipment , such as network analyzers and lockin amplifiers .
therefore , it is desirable to be able to extract as much data as possible from the frequency measurements . if the system can be brought to the verge of bistable regime , i.e. , @xmath221 , the nonlinear damping parameter @xmath6 can be readily determined using eq . .
the same coefficient can also be extracted from the measurements of the oscillator s frequency response in the bistable regime . in general , the sum of the three solutions for @xmath92 at any given frequency can be found from eq . .
this is employed for the jump point at @xmath222 seen in fig .
[ fig : duffing ] . using eq . to calibrate the measured response at this jump point one
has @xmath223 or @xmath224 where @xmath225 and @xmath226 are defined in fig .
[ fig : duffing ] . due to the frequency proximity between the maximum point and the jump point at @xmath227 ,
the inaccuracy of such a calibration is small .
moreover , as long as excitation amplitude is high enough , @xmath226 is much smaller than @xmath225 and even considerable inaccuracy in @xmath228 estimation will not have any significant impact .
this equation can be used to estimate @xmath6 for different excitation amplitudes at which the micromechanical oscillator exhibits bistable behavior , i.e. , @xmath229 .
it is especially useful if the system is strongly nonlinear and can not be measured near its critical point due to high noise floor or low sensitivity of the displacement detectors used .
another method for estimating the value of @xmath6 requires measurement of free ring down transient of the micromechanical oscillator and can be employed also at low excitations , when the system does not exhibit bistable behavior , i.e. , @xmath230 .
the polar form of the evolution equation is especially well suited for the analysis of the system s behavior in time domain .
starting from eq . and
applying the free ring down condition @xmath231 , one finds @xmath232 where @xmath233 is the amplitude at @xmath234 . in particular
, consider a case in which the system is excited at its maximal response frequency detuning @xmath111 , i.e. , @xmath235 . then , after turning the excitation off , the amplitude during the free ring down process described by eq . can be written as @xmath236 the ring down amplitude measured in time domain can be fitted to the last result .
in addition to nonlinear damping parameter @xmath6 , most parameters defined above can be easily estimated from frequency measurements near the jump point shown in fig .
[ fig : duffing ] if the following conditions are satisfied .
the first condition is @xmath237 which can be satisfied by exciting the micromechanical beam oscillator in the bistable regime strongly enough , i.e. , @xmath238 .
the immediate consequence of the first condition is @xmath239 i.e. , @xmath240 , as described above . using eq .
, it follows from eq .
that @xmath241 . from the last result and from eqs . , , and , the following approximations follow immediately : @xmath242 [ eq : approx_k_lf_yjxj ] as shown in sec .
[ sec : slow_env_approx ] , eq .
, at the maximum response point @xmath243 , the following holds : @xmath244 .
therefore , in view of our assumptions described above , we may write @xmath245 consequently , @xmath246 and @xmath247 , \end{gathered}\ ] ] which follows from eq . .
the time @xmath218 , which describes the slowing down near the saddle - node `` ghost '' described above eq .
, can be expressed as @xmath248 where @xmath249}}. \label{eq : approx_y}\ ] ] finally , we turn to estimate the value of the thermal escape time @xmath211 given by eq . . using the same assumptions as above ,
we find @xmath250 [ eq : tau_therm_approx ] unlike in the previous approximations , one has to know at least the order of magnitude of the response amplitude in the vicinity of the jump point ( in addition to effective noise temperature @xmath31 and effective mass @xmath5 ) in order to approximate @xmath211 appropriately .
the same is also true for estimation attempt of the physical nonlinear constants @xmath251 [ eq : alpha3_gamma3_approx ] for more accurate estimation , one of several existing kinds of fitting procedures can be utilized .
however , the order of magnitude estimations often fully satisfy the practical requirements .
the above discussion of parameters evaluation using experimental data , especially in frequency domain , emphasizes the importance of accurate frequency measurements .
however , the slowing down of the oscillator s response near the bifurcation points poses strict limitations on the rates of excitation frequency or amplitude sweeps used in such measurements @xcite .
this is to say that special care must be taken by the experimentalist choosing a correct sweep rate for the measurement in order to obtain the smallest error possible .
fortunately , this error can easily be estimated based on our previous analysis .
let @xmath252 represent the frequency sweep rate in the frequency response measurement .
for example , using network analyzer in part of our experiments , we define @xmath253 in order to estimate the inaccuracy , @xmath254 , in the measured value of the bifurcation point detuning , @xmath122 , which results from nonzero frequency sweep rate , the following expression may be used : @xmath255 whose solution is @xmath256 note that this error is a systematic one - the measured jump point will always be shifted in the direction of the frequency sweep .
obviously , the first step towards accurate measuring of @xmath122 is to ensure that the established value of the bifurcation point detuning does not change when the sweep rate is further reduced .
another possible source of uncertainty in frequency measurements near the bifurcation point is the thermal escape process .
the error introduced by this process tends to shift the measured jump point detuning in the direction opposite to the direction of the frequency sweep .
moreover , unlike the error arising from slowing down process , this inaccuracy can not be totally eliminated by reducing the sweep rate .
however , as will be shown in sec .
[ sec : results_parameter_extraction ] , in our case this error is negligible .
a typical measured response of the fundamental mode of a 200long beam occurring at the resonance frequency of 123.2measured with @xmath257 and varying excitation amplitude is seen in fig .
[ fig : raw_data ] .
the linear regime is shown in the frequency response diagram and damping backbone curve depicted in figs .
[ fig : experiment_linear_na ] and [ fig : quality_factor_backbone ] respectively , for a 125long beam with fundamental mode resonant frequency 885.53and @xmath257 .
we derive the value of @xmath258 from the linear response at low excitation amplitude and find @xmath259 for 200beam and @xmath260 for 125beam . and with varying peak - to - peak excitation amplitude @xmath261 of a 200long beam with fundamental mode occurring at 123.2 .
the excitation amplitude is shown on the graphs .
the oscillator exhibits bistable behavior at all excitation amplitudes except for the lowest one .
the vertical axis is in arbitrary units.,width=326 ] .
the linear regime is defined as a regime in which the frequency response function is symmetric around the resonance frequency . in the main panel ,
the measured responses with three different excitation amplitudes are shown .
blue circles correspond to @xmath262 , green rectangles correspond to @xmath263 , and red triangles correspond to @xmath264 .
solid black lines show the fitted lorentzian shapes .
vertical scale is in arbitrary units.,width=326 ] vs. inverse quality factor @xmath265 in a linear regime .
large black diamonds correspond to the frequency responses depicted in fig .
[ fig : experiment_linear_na ] .
the measured averaged quality factor is @xmath266 .
other experimental parameters are similar to those described in fig .
[ fig : experiment_linear_na ] caption .
vertical scale is in arbitrary units.,width=326 ] as shown in sec .
[ sec : parameter_extraction ] , the value of @xmath6 can be estimated for different excitation amplitudes using eqs . and .
typical results of applying these methods to experimental data from a micromechanical beam oscillator can be seen in fig .
[ fig : experiment_p_extract ] . using these procedures
we find @xmath267 for the 200long beam and @xmath268 for the 125long beam .
we also estimate @xmath6 from the critical point detuning using eq . , and obtain similar values . vs. excitation amplitude .
the excitation amplitude on the horizontal axis is normalized by the respective critical excitation amplitude @xmath269 .
* a. * 200long beam with fundamental mode occurring at 123.20and @xmath259 .
the values of @xmath6 extracted from frequency domain jump point measurements ( see eq . ) are represented by blue circles .
red dashed line represents the value of @xmath270 evaluated using the critical point frequency detuning @xmath124 ( see eq . ) .
the critical excitation voltage is 50 , and @xmath257 . *
b. * 120long beam with fundamental mode occurring at 885.53and @xmath260 .
the values of @xmath6 extracted from time domain ring down measurements according to eq . and frequency domain jump point measurements ( see eq . ) are represented by green squares and blue circles respectively .
red dashed line represents the value of @xmath271 evaluated using the critical point frequency detuning @xmath124 that is given by eq . .
the critical excitation voltage is 105 , and @xmath272.,width=326 ] in order to illustrate the procedures derived in sec .
[ sec : parameter_extraction ] , we evaluate the main parameters of slow envelope dynamics of the 125long beam in a particular case in which @xmath272 and the excitation voltage amplitude is 140 .
the quality factor of the beam , as measured in the linear regime , is @xmath260 . the results that can be derived from frequency measurements only , i.e. , the results corresponding to eqs . , and ,
are summarized in table [ tab : eval_freq_meas_parameters ] .
.parameters of the slow envelope dynamics of a 125long beam . applied dc voltage is 15and excitation voltage amplitude is 140 .
the critical excitation voltage is 105 .
quality factor is @xmath260 . [
cols="<,^,^ " , ] we estimate below the thermal escape time for @xmath273 ( see eq . ) .
however , the value of the exponent , @xmath274 at @xmath275 , makes the thermal escape time at this detuning value extremely large .
therefore , in our experiments , the thermal escape process does not contribute significantly to the total inaccuracy in frequency measurements near the bifurcation point , at least for effective noise temperatures lower than @xmath276 , at which the assumption @xmath215 is no longer valid . finally , it is interesting to compare the nonlinear dissipation term @xmath277 and the linear dissipation term @xmath83 in the evolution equation .
it follows from the above assumptions and the values in table [ tab : eval_special_parameters ] that for our chosen example @xmath278 in order to verify the correctness of our approximated solution achieved by multiple scales method , we compare the results of direct integration of the full motion equation with the steady state solution of the evolution equation .
we use the results from tables [ tab : eval_freq_meas_parameters ] and [ tab : eval_special_parameters ] for @xmath279 , and @xmath130 .
we also estimate the effective mass @xmath5 to be @xmath280 , the effective capacitance to be of order of @xmath281 , the dc voltage @xmath282 , the ac voltage @xmath283 , and take the distance @xmath10 to be the actual distance between the electrode and the mechanical beam , i.e. , @xmath284 .
the resulting excitation force amplitude is @xmath285 , the constant force is @xmath286 ( see eq . ) , and the constant resonance frequency shift is @xmath287 ( see eq . ) . in fig . [ fig : model_validity ] , the exact numerical integration of eq .
is compared with the solution of the approximated frequency response equation .
a very good correspondence between the two solutions is achieved , which validates the approximations applied in sec .
[ sec : slow_env_approx ] .
it follows from our experimental results that the nonlinear damping constant @xmath6 can be estimated with a high degree of confidence by measuring the micromechanical oscillator bistable response in the frequency domain .
the values of @xmath6 that we find , @xmath288 , obviously are not negligible . referring to eqs . and
, we see that the considered micromechanical oscillators exhibit a damping nonlinearity that has a measurable impact on both the amplitude and frequency offset of the critical point , as well as on jump points in the bistable region . on the other hand ,
these values are significantly smaller then the critical value @xmath289 , which would prevent the system from exhibiting bistable behavior .
two methods of estimating the value of @xmath6 from frequency domain measurements were used .
the first is based on a single measurement of the critical point and provides a simple means for estimating the value of @xmath6 by experimentally measuring the linear quality factor @xmath0 at low excitation amplitude and the critical frequency shift @xmath124 only ( see eq . ) .
the second can be used for any excitation amplitude that drives the system into bistable regime , but requires a comparison of different response amplitudes ( see eq . ) .
both these methods yield similar results , however , the second one , although being less accurate , allows the experimentalist to estimate when the limit of hard excitation @xcite is approached and the first order multiple scales analysis used in this study becomes inadequate . in this limit of strong excitation ,
the extracted values of @xmath6 start to diverge significantly from the results obtained at low excitation amplitudes .
our results , especially fig .
[ fig : experiment_p_extract ] , and the analysis of the validity of our approximations , which was carried out in sec .
[ sec : model_validity ] , suggest that the analysis method employed by us is adequate for a wide range of excitation amplitudes .
the third method described above allows one to estimate the value of @xmath6 from time domain measurements of the free ring down of the micromechanical beam oscillator based on eq . .
although fitting results of time domain measurements to a theoretical curve introduces large inaccuracy , this method is invaluable in cases where the bistable regime can not be achieved , e.g. , due to prohibitively large amplitudes involved and the risk of pull - in . by using the approximations developed in sec .
[ sec : parameter_extraction ] , we were able to estimate different parameters describing the slow envelope dynamics of our oscillators , summarized in table [ tab : eval_freq_meas_parameters ] .
the most important and , as far as we know , novel result is the direct estimation of the slowing down time @xmath218 that is given by eq . , which governs the system s dynamics in the vicinity of bifurcation point
. in turn ,
this result is used to quantitatively evaluate the error introduced to the frequency measurements by the slowing down process , @xmath254 that is given by eq . , which in the example studied is @xmath290 .
it can be seen that even slow sweeping rate ( as compared to quasistatic rate in the linear case , which is of order of one resonant width per ring down time ) can introduce a significant inaccuracy in the measured response of a micromechanical beam oscillator near bifurcation points . in our case , the inaccuracy in @xmath122 is about 3% , but the inaccuracy in @xmath291 is probably much larger .
the nonlinear damping constant @xmath6 plays an important role in all the dynamical parameters . in the value of @xmath292 that isgiven by eq .
in our example , @xmath6-dependent term constitutes about 30% of the value .
the same is true for other parameters as well .
also , we make order of magnitude estimations of thermal escape time @xmath293 ( see eq . ) , @xmath131 , and @xmath130 ( see eq . ) , which are summarized in table [ tab : eval_special_parameters ] .
these approximations can be used in order to construct an accurate model of the effective one - dimensional movement of the system in the vicinity of a bifurcation point , especially if accurate enough estimations of the oscillator s amplitude and effective mass can be made . in our case , only the order of magnitude of the parameters can be estimated .
however , we were able to estimate the thermal escape time , and found the thermal escape process to be a non negligible source of inaccuracy in the frequency measurements only at very high effective noise temperatures of order @xmath276 .
this result can be compared to a result from our previous work @xcite . in that work
, a micromechanical beam oscillator similar to the ones used here was excited at a frequency between the bifurcation points .
the intensity of voltage noise needed to cause transitions between these stable states was found to be @xmath294 , with noise bandwidth of 10 .
the resulting voltage noise density is @xmath295 , which corresponds to an effective noise temperature @xmath296 . in the case of thermal escape
described here , the two stable states are highly asymmetrical . the effective noise temperature of @xmath276 , which invalidates the estimations of very slow thermal escape rate in sec .
[ sec : results_parameter_extraction ] , corresponds to voltage noise density of @xmath297 , giving the total voltage noise intensity of 1.6 .
finally , we can also estimate the relative contribution of the nonlinear damping term @xmath277 in the evolution equation , and find it to be a non negligible one - tenth of the linear term @xmath83 ( see eq . ) at the estimated amplitude of @xmath298 .
the nature of nonlinear damping is not discussed in this work . however , nonlinear damping can be , in part , closely related to material behavior with a linear dissipation law that operates within a geometrically nonlinear regime . here
, we investigate one possible mechanism , originating from a voigt - kelvin type of dissipation model which describes internal viscoelastic damping in the form of a parallel spring and dashpot . before we proceed to build the model ,
one technical remark is in order .
the notations in this section follow the standard ones used in continuum mechanics , and some parameters used above are redefined below .
however , the end results are brought back to the form of .
following leamy and gottlieb , we consider a planar weakly nonlinear pretensioned , viscoelastic string augmented by linear euler - bernoulli bending , which incorporates a voigt - kelvin constitutive relationship where the stress is a linear function of the strain and strain rate @xcite : @xmath299 where @xmath300 is the stress , @xmath301 is the strain , @xmath302 is the material young modulus , @xmath303 is a viscoelastic damping parameter , and subscripts denote differentiation with respect to the corresponding variable .
the equations of motion of the beam - string are @xmath304_{\tilde s}=0,\end{gathered}\ ] ] @xmath305_{\tilde s}=q_{\tilde w},\end{gathered}\ ] ] [ eq : string_eq_of_motion_orig ] where @xmath306 is the pretension , @xmath307 is the material density , @xmath308 is the material coordinate along the beam , @xmath104 and @xmath309 are the elastic element cross - sectional area and moment of inertia , respectively .
also , @xmath310 and @xmath311 are the respective longitudinal and transverse components of an elastic field .
the generalized transverse force component @xmath312 is due to external electrodynamic actuation .
note that for a parallel plate approximation , @xmath313 ^ 2}{(d-\tilde w)^2},\ ] ] where @xmath314 , @xmath315 , @xmath10 and @xmath61 are as those defined in eq .
, and @xmath316 is a proportionality coefficient dependent on the exact geometry of the mechanical oscillator .
we rescale the elastic field components @xmath317 and @xmath318 , and the material coordinate @xmath319 by the beam length @xmath320 , and time by the pretension @xmath321 to yield a coupled set of dimensionless partial differential equations for the beam - string : @xmath322_s=0 , \label { eq : string_eq_of_motion_nondim_longitudinal}\ ] ] @xmath323_s = q_w , \label{eq : string_eq_of_motion_nondim}\end{gathered}\ ] ] where @xmath324 , @xmath325 , @xmath326 and @xmath327 other dimensionless parameters include the effects of weak bending @xmath328 , a strong nonlinear pretension @xmath329 , a small slenderness ratio @xmath330 ( because @xmath331 , where @xmath332 is the beam - string radius of gyration @xcite ) , and finite viscoelastic damping @xmath164 : @xmath333 note that @xmath334 defines the ratio between the longitudinal and transverse wave speeds @xcite .
the rescaled parallel plate approximation is thus : @xmath335 ^ 2}{(\gamma - w)^2},\ ] ] where @xmath336 we note that as the first longitudinal natural frequency is much higher than the first transverse natural frequency ( @xmath337 ) , the longitudinal inertia and damping terms in eq .
can be neglected to yield a simple spatial relationship between the transverse and longitudinal derivatives .
incorporating fixed boundary conditions ( @xmath338 enables integration of the resulting relationship to yield : @xmath339 where @xmath340 thus , the resulting weakly nonlinear beam - string initial boundary value problem consists of an integro - differential equation for the transverse mode : @xmath341 where @xmath342 in order to facilitate comparison of the continuum model with the lumped mass model in eq . , we consider a localized electrodynamic force @xmath343 .
we reduce the integro - differential field equation in and its fixed boundary conditions to a modal dynamical system via an assumed single mode galerkin assumption , @xmath344 , using a harmonic string mode @xmath345 : @xmath346+i_4 \left(\alpha q+\mu\delta q_\tau\right)\\ = i_5 \eta\frac{\left[1+\epsilon\cos(\omega\tau)\right]^2}{\left(\gamma - i_5 q\right)^2},\end{gathered}\ ] ] where @xmath347 and the integral coefficients are : @xmath348 it is convenient to rescale the maximal response @xmath349 , where @xmath350 , by the dimensionless gap @xmath351 , and to rescale time by the unperturbed @xmath352 natural frequency @xmath353 , where @xmath354 .
the resulting dynamical system is : @xmath355 ^ 2}{(1-z)^2 } , \label{eq : string_eq_of_motion_one_dim}\ ] ] where @xmath356 note that the ratio between nonlinear and linear damping in eq .
consists of only the beam - string geometric properties @xcite .
for example , a typical ratio is @xmath357 for a beam - string with a prismatic cross - section , where @xmath358 is the dimension of the beam - string in the transverse direction @xmath359 , and @xmath284 is the resonator gap .
the last equation can be compared , after rescaling , to the dimensional equation , which we rewrite here for convenience after some rearrangement and simplification ( e.g. @xmath360 ) : @xmath361 ^ 2}{\left(1-\frac{x}{d } \right)^2}. \label{eq : full_eq_of_motion_for_comparison}\end{gathered}\ ] ] the comparison of eq . with eq .
results in : @xmath362 [ eq : param_estimations_continuous_model ] the last results can be used to estimate the lower bound of nonlinear damping due to nonlinear pretension of a viscoelastic string . using eqs .
, , , and @xmath363 for prismatic cross - section , one has @xmath364 where @xmath365 denotes the dimension of the beam - string in the transverse direction @xmath359 .
it is possible to estimate the order of magnitude of @xmath366 in eq . for metals using the fact that the young modulus of bulk metals @xmath367 .
also , the largest value of @xmath368 that is still compatible with elastic behavior can be approximated by half the ultimate tensile strength , which is about @xmath369 for most metals . for our beam - strings discussed above , @xmath370 . using these values results in @xmath371 .
for longer and wider beams ( @xmath372 ) fabricated and measured using the same methods @xcite , the lower bound on nonlinear damping coefficient given by eq .
is @xmath373 , while the range of values extracted from the experiment is @xmath374 @xcite .
although the elastic properties of a specific metal or alloy used in micro machined devices might differ significantly from the bulk values , they are still likely to fall inside the ranges defined above .
therefore , a linear viscoelastic process with a pure voigt - kelvin dissipation model can serve as a possible lower bound but can not account for the main part of nonlinear dissipation rate found in our experiments .
unfortunately , theory describing the processes underlying nonlinear damping in micromechanical beam is virtually non - existent at this moment , and no clear tendencies in the value of @xmath6 were observed during the experiments .
therefore , the exact behavior of nonlinear damping term during beam scaling and its dependence on the linear @xmath0 of the structure remains elusive .
further experiments with wider range of micromechanical beams are needed to establish this behavior and to pinpoint the most significant mechanisms of dissipation .
in this study , the nonlinear dynamical behavior of an electrically excited micromechanical doubly clamped beam oscillator was investigated in vacuum .
the micromechanical beam was modeled as a duffing - like single degree of freedom oscillator , nonlinearly coupled to a thermal bath . using the method of multiple scales
, we were able to construct a detailed model of slow envelope behavior of the system , including effective noise terms .
it follows from the model that nonlinear damping plays an important role in the dynamics of the micromechanical beam oscillator .
several methods for experimental evaluation of the nonlinear damping contribution were proposed , applicable at different experimental situations .
these methods were compared experimentally and shown to provide similar results .
the experimental values of the nonlinear damping constant are non negligible for all the beams measured .
also , the slow envelope model was used to describe the behavior of the system close to bifurcation points in the presence of nonlinear damping . in the vicinity of these points , the dynamics of the system
is significantly slowed down , and the phase plane motion becomes essentially one - dimensional .
we have defined several parameters that govern the dynamics of the micromechanical beam oscillator in these conditions , and have provided simple approximations that can be used to estimate these parameters from experimental data .
the approximations developed in this study can be utilized by the experimentalist in order to estimate the inaccuracy of frequency response measurements of duffing - like oscillators in the vicinity of bifurcation points . applying these results to our samples
, we have found that thermal escape process near the bifurcation point causes measurement inaccuracy that is negligible .
in contrast , the slowing down phenomenon , which is a characteristic of saddle - node bifurcation , can contribute a significant error to the measured frequency response .
this error is non negligible even at relatively slow frequency sweeping rates .
similar methods can be utilized for other parameter sweeping measurements , such as excitation amplitude sweeping . as part of an effort to explain the origins of the nonlinear damping , we have formulated and analyzed a model of a planar , weakly nonlinear pretensioned , viscoelastic string augmented by linear euler - bernoulli bending , which incorporates a voigt - kelvin constitutive relationship .
this model exemplifies one of the possible causes of non negligible nonlinear damping observed in the experiment .
based on this model , we have determined a simple relation connecting the maximal expected value of the nonlinear damping parameter , the bulk young modulus of the material , and its yielding stress . however , while this model can serve as a lower bound , it can not account for the full magnitude of the nonlinear damping measured in the experiment .
additional experimental and theoretical work is required to enhance our understanding of the phenomenon of nonlinear damping in microelectromechanical systems . in this work
we have demonstrated conclusively that nonlinear damping in micromechanical doubly - clamped beam oscillator may play an important role .
the methods presented in this paper may allow a systematic study of nonlinear damping in micro- and nanomechanical oscillators , which may help revealing the underlying physical mechanisms .
we would like to thank r. lifshitz for many fruitful discussions .
this work was partially supported by intel corporation , the israeli ministry of science , the israel science foundation , the german israel foundation , and the russell berry foundation .
dykman and m.a .
theory of nonlinear oscillator interacting with a medium . in i.
m. khalatnikov , editor , _
soviet scientific reviews , section a , physics reviews _ , volume 5 , pages 265441 .
harwood academic , 1984 .
r lifshitz and m.c . cross .
nonlinear dynamics of nanomechanical and micromechanical resonators . in heinz
georg schuster , editor , _ reviews of nonlinear dynamics and complexity _ , volume 1 , pages 148 .
wiley - vch , 2008 .
p. hnggi .
generalized langevin equations : a useful tool for the perplexed modeller of nonequilibrium fluctuations ? in _ stochastic dynamics _
, volume 484 of _ lecture notes in physics _ , pages 1522 .
springer - verlag , 1997 .
|
nonlinear elastic effects play an important role in the dynamics of microelectromechanical systems ( mems ) .
a duffing oscillator is widely used as an archetypical model of mechanical resonators with nonlinear elastic behavior .
in contrast , nonlinear dissipation effects in micromechanical oscillators are often overlooked . in this work ,
we consider a doubly clamped micromechanical beam oscillator , which exhibits nonlinearity in both elastic and dissipative properties .
the dynamics of the oscillator is measured in both frequency and time domains and compared to theoretical predictions based on a duffing - like model with nonlinear dissipation .
we especially focus on the behavior of the system near bifurcation points .
the results show that nonlinear dissipation can have a significant impact on the dynamics of micromechanical systems . to account for the results ,
we have developed a continuous model of a geometrically nonlinear beam - string with a linear voigt - kelvin viscoelastic constitutive law , which shows a relation between linear and nonlinear damping .
however , the experimental results suggest that this model alone can not fully account for all the experimentally observed nonlinear dissipation , and that additional nonlinear dissipative processes exist in our devices .
|
the recent explosion@xcite of research on the electronic properties of single layer and stacked multilayer graphene sheets has been driven by advances in material preparation methods@xcite , by the unusual@xcite electronic properties of these materials including unusual quantum hall effects@xcite , and by hopes that these elegantly tunable systems might be useful electronic materials . in this paper
, we demonstrate an unanticipated low - energy property of graphene multilayers , which follows from an interplay between interlayer tunneling and the chiral properties of low - energy quasiparticles in an isolated graphene sheet .
our conclusions apply in the strongest form to models with only nearest - neighbor interlayer tunneling , but are valid over a broad field range as we explain below .
we find that the low - energy band structure of any graphene multilayer consists of a set of independent pseudospin doublets . within each doublet ,
the bands are described by a pseudospin hamiltonian of the form @xmath4 , \label{eq : chiralband}\ ] ] where @xmath5 is a pauli matrix acting on the doublet pseudospin , @xmath6 is an envelope function momentum measured from either the @xmath7 or @xmath8 corner of the honeycomb lattice s brillouin - zone@xcite , @xmath9 , and @xmath10 is the orientation of @xmath6 .
the @xmath11 sign in eq .
( [ eq : chiralband ] ) assumes the opposite signs in graphene s @xmath7 and @xmath8 valleys . following the earlier work on graphene bilayers@xcite
, we refer to @xmath12 as the chirality index of a doublet . in the presence of a perpendicular magnetic field @xmath13
, @xmath14 yields the @xmath12 landau levels at @xmath15 and @xmath16 levels with @xmath17 . taking the twofold spin and valley degeneracies into account
, the number of independent zero - energy band eigenstates at the dirac point ( @xmath18 ) is therefore @xmath19 , where @xmath20 is the number of pseudospin doublets .
we find that , although @xmath21 depends on the details of the stacking sequence , @xmath22 in an @xmath0-layer stack .
it follows from eq .
( [ eq : sumrule ] ) that the hall conductivity of an @xmath0-layer stack has strong integer quantum hall effects with plateau conductivities , @xmath23 where @xmath3 is a non - negative integer . and @xmath24 .
given a starting graphene sheet , the honeycomb lattice for the next layer is positioned by displacing either @xmath25 or @xmath24 sublattice carbon atoms along a honeycomb edge .
this results in three distinct two - dimensional sheets , labeled by a , b , and c. representative @xmath25 and @xmath24 sublattice positions in a , b , and c layers are identified in this illustration .
it is also possible to transform between layer types by rotating by 60@xmath26 about one of the carbon atoms .
( b ) each added layer cycles around this stacking triangle in either the right - handed or the left - handed sense .
reversals of the sense of this rotation tend to increase the number of low - energy pseudospin doublets @xmath20 . in graphite ,
the bernal ( ab ) stacking corresponds to a reversal at every step and orthorhombic ( abc ) stacking corresponds to no reversals . ]
when one graphene layer is placed on another , it is energetically favorable@xcite for the atoms of either @xmath25 or @xmath24 sublattices to be displaced along the honeycomb edges , as illustrated in fig .
[ fig : stack ] .
this stacking rule implies the three distinct but equivalent projections ( labeled a , b , and c ) of the three - dimensional structure s honeycomb - lattice layers onto the @xmath27-@xmath28 plane and @xmath29 distinct @xmath0-layer stack sequences .
when a b layer is placed on an a layer , a c layer on a b layer , or an a layer on a c layer , the @xmath25 sites of the upper layer are above the @xmath24 sites of the lower layer and therefore linked by the nearest interlayer neighbor @xmath30-orbital hopping amplitude @xmath31 . for the corresponding anticyclic stacking choices ( a on b , b on c , or c on a ) , it is the @xmath24 sites of the upper layer and the @xmath25 sites of the lower layer that are linked .
layer stacks . the low - energy band and the landau level structure of a graphene stack
are readily read off these diagrams as explained in the text .
shaded ovals link @xmath25 and @xmath24 nearest interlayer neighbors . ]
all distinct @xmath32 , @xmath33 , and @xmath34 layer stacks are illustrated in fig .
[ fig : min_diagrams ] , in which we have arbitrarily labeled the first two layers starting from the bottom as a and b. the low - energy band and the landau level structure can be read off these diagrams by partitioning a stack using the following rules , which are justified in the following section .
( i ) identify the longest nonoverlapping segments within which there are no reversals of stacking sense .
when there is ambiguity in the selection of nonoverlapping segments , choose the partitioning which incorporates the largest number of layers .
each segment ( including for interior segments the end layers at which reversals take place ) defines a @xmath12-layer partition of the stack and may be associated with a chirality @xmath12 doublet .
( ii ) iteratively partition of the remaining segments of the stack into smaller @xmath12 elements , excluding layers contained within previously identified partitions , until all layers are exhausted .
chirality decompositions which follow from these rules are summarized in table [ tab : decomposition ] .
note that this procedure can result in @xmath35 doublets associated with separated single layers which remain at the last step in the partitioning process . in applying these rules ,
the simplest case is cyclic abc stacking for which there are no stacking sense reversals and therefore a single @xmath36 partition . in the opposite limit , ab stacking , the stacking sense is reversed in every layer and the rules imply @xmath37 partitions with @xmath38 for even @xmath0 , and when @xmath0 is odd a remaining @xmath35 partition . between these two limits , a rich variety of qualitatively distinct low - energy behaviors
for example , in the abcb stacked tetralayer , abc is identified as a @xmath39 doublet and the remaining b layer gives a @xmath35 doublet .
the low - energy band structure and the landau level structure of this stack , as illustrated in fig .
[ fig : stack_abcb ] , have two sets of low - energy bands with @xmath40 , landau levels with @xmath41 , and four zero - energy landau levels per spin and valley .
all these properties are predicted by the partitioning rules .
we have explicitly checked that the rules correctly reproduce the low - energy electronic structure for all stacking sequences up to @xmath42 . because each layer is a member of one and only one partition ,
the partitioning rules imply the chirality sum rule in eq .
( [ eq : sumrule ] ) . .chirality decomposition for @xmath43 layer stacks .
[ cols="<,<,<,<",options="header " , ] ev and nearest interlayer neighbor hopping @xmath44.,title="fig : " ] ev and nearest interlayer neighbor hopping @xmath44.,title="fig : " ]
we start from the well - known @xmath35 massless dirac equation@xcite @xmath45 model for isolated sheets , @xmath46 where @xmath47 and @xmath48 is the quasiparticle velocity . in the presence of an external magnetic field , @xmath49 and @xmath30
are proportional to the landau level raising and lowering operators , so that eq .
( [ eq : hamiltonian_md ] ) implies the presence of one macroscopically degenerate landau level at the dirac point for each spin and valley , and therefore , to the @xmath50 quantum hall effect@xcite of eq .
( [ eq : iqhe ] ) .
an @xmath0-layer stack has a two - dimensional band structure with @xmath51 atoms per unit cell .
the hamiltonian can be written as @xmath52 where @xmath53 accounts for interlayer tunneling and @xmath54 for intralayer tunneling .
@xmath54 is the direct product of massless dirac model hamiltonians @xmath55 for the sublattice pseudospin degrees of freedom of each layer .
we construct a low - energy hamiltonian by first identifying the zero - energy eigenstates of @xmath56 and then treating @xmath54 as a perturbation . referring to fig .
[ fig : min_diagrams ] , we see that @xmath53 is the direct product of a set of finite - length one - dimensional ( 1d ) tight - binding chains and a null matrix with dimension equal to the number of isolated sites .
an elementary calculation gives the following eigenvalues and eigenvectors for a chain of length @xmath57 : @xmath58 where @xmath59 is the chain eigenvalue index and @xmath60 .
note that odd @xmath57 chains have a zero - energy eigenstate with an eigenvector that has nonzero amplitudes , constant in magnitude and alternating in sign , on the sublattice of the chain ends .
the set of zero - energy eigenstates of @xmath53 consists of the states localized on isolated sites and the single zero - energy eigenstates of each odd - length chain .
the low - energy effective hamiltonian is evaluated by applying leading order degenerate state perturbation theory to the zero - energy subspace .
the matrix element of the effective hamiltonian between degenerate zero - energy states @xmath61 and @xmath62 is given by@xcite @xmath63^{n-1}\left|\psi_{r'}\right>,\ ] ] where @xmath3 is the smallest positive integer for which the matrix element is nonzero , and @xmath64 , @xmath65 is a projection operator onto the zero - energy subspace . to understand the structure of this hamiltonian , it is helpful to start with some simple examples . for abc stacked @xmath0-layer graphene , the zero - energy states are the two isolated site states in top and bottom layers , @xmath66 and @xmath67 .
the high - energy hilbert space is formed from a set of two - site chains .
because @xmath68 is diagonal in layer index and @xmath56 ( and hence @xmath69 ) can change the layer index by one unit , the lowest order at which @xmath66 and @xmath67 are coupled is @xmath70 .
it follows that @xmath71 thus , the effective hamiltonian of @xmath0-layer graphene with abc stacking has a single @xmath36 doublet@xcite . for ab
stacked @xmath0-layer graphene , the high - energy hilbert space consists of a single @xmath0-site 1d chain , excluding its zero- energy eigenvalue when @xmath0 is odd .
there is an isolated site in each layer which is connected to both its neighbors at order @xmath72 forming an isolated site chain .
when @xmath0 is even , this chain is diagonalized by @xmath37 , @xmath38 doublets formed between @xmath25-sublattice and @xmath24-sublattice chain states@xcite .
when @xmath0 is odd , the zero - energy chain state is mapped to an equal - magnitude oscillating - sign linear combination of isolated site states by intralayer tunneling at order @xmath73 , yielding a @xmath35 doublet .
the @xmath74 , @xmath38 doublets are then formed between @xmath25-sublattice and @xmath24-sublattice isolated site chain states in the orthogonal portion of the isolated state subspace . a more complex and more typical example is realized by placing a single reversed layer on top of abc stacked n - layer graphene with @xmath75 . in this case , we obtain @xmath76 using a @xmath77 basis .
here , @xmath78 and @xmath79 is the three - site chain zero - energy state , @xmath80 . the first @xmath81 block in eq .
( [ eq : hamiltonian_abcb_eff ] ) gives a @xmath35 doublet with a reduced velocity .
the @xmath36 doublet in this instance includes both the @xmath82 subspace contribution and an equal contribution due to perturbative coupling to the @xmath83 subspace .
the final hamiltonian is reduced to @xmath84 where @xmath85 the relationship between the electronic structure of a general stack and the partitioning procedure explained above can be understood as follows .
first , note that a partition with chirality @xmath12 has isolated sites in its terminal layers that are coupled at order @xmath12 in perturbation theory .
in the case of @xmath35 partition , the chain opposite to the single isolated site always has an odd length and provides the zero - energy partner ; isolated site to chain coupling therefore always occurs at first order .
next , consider the perturbation theory , truncating at successively higher orders .
when truncated at first order , the @xmath35 partitions are isolated by higher @xmath12 blocks within which the hamiltonian vanishes .
each @xmath35 partition therefore yields a separate massless dirac equation with velocities@xcite that can be smaller than the graphene sheet dirac velocity .
when the perturbation theory is truncated at second order , the hamiltonian becomes nonzero within the @xmath38 partitions .
the eigenenergies within the @xmath35 partitions are parametrically larger , and the hamiltonian within the @xmath86 partitions is still zero . to leading order therefore , the @xmath38 partitions are separated , and their isolated states are coupled at the second order in perturbation theory so that each provides a @xmath38 doublet such as that of an isolated bilayer
. if two or more @xmath38 partitions are adjacent , then their hamiltonians do not separate . in this case
, there is a chain of second order couplings between isolated states , such as those of an even - length ab stack , but the end result is still @xmath38 doublet for each @xmath38 partition .
the identification between partitions and chiral doublets can be established by continuing this consideration up to the highest values of @xmath12 which occur for a particular stack .
then , the effective hamiltonian of any @xmath0-layer graphene is as follows : @xmath87 with the sum rule in eq .
( [ eq : sumrule ] ) .
note that @xmath20 is half the sum of the number of isolated sites and the number of odd - length chains .
valley as a function of @xmath88 for an ab stacked bilayer for ( a ) @xmath89 t and ( b ) @xmath90 t , and as a function of @xmath91 for an aba stacked trilayer for ( c ) @xmath90 t and ( d ) @xmath92 t. here @xmath93 ev , @xmath44 , and @xmath94 , with @xmath95 , were used.,title="fig : " ] valley as a function of @xmath88 for an ab stacked bilayer for ( a ) @xmath89 t and ( b ) @xmath90 t , and as a function of @xmath91 for an aba stacked trilayer for ( c ) @xmath90 t and ( d ) @xmath92 t. here @xmath93 ev , @xmath44 , and @xmath94 , with @xmath95 , were used.,title="fig : " ] the minimal model we have used to derive these results is approximately valid in the broad intermediate magnetic field @xmath13 range between @xmath96 and @xmath97 t , over which the intralayer hopping energy in field ( @xmath98 where @xmath99^{1/2}$ ] is the magnetic length ) is larger than the distant neighbor interlayer hopping amplitudes that we have neglected ( @xmath100 mev ) , but still smaller than @xmath31 .
for example , if we consider @xmath101 hopping process in aba stacked trilayer in fig .
[ fig : min_diagrams ] , the valid range of magnetic field for the minimal model is given by @xmath102 when @xmath91 does not play an important role ( in @xmath103 stacks , for example ) , the lower limit of the validity range is parametrically smaller .
the minimum field in bilayers has been estimated to be @xmath104 t@xcite , by comparing intralayer hopping with the @xmath105 ev interlayer hopping amplitude , @xmath106 where @xmath107 and @xmath108 is a lattice constant of graphene .
figure [ fig:2nn ] shows the landau level spectrum at the @xmath7 valley as a function of @xmath88 for an ab stacked bilayer and as a function of @xmath91 for an aba stacked trilayer . in the case of the bilayer , the dependence of the landau levels on @xmath88 is weak for @xmath13 larger than 1 t , whereas in the case of the trilayer , the landau level spectrum still strongly depends on @xmath91 for @xmath90 t , but the dependence becomes weak for @xmath13 above 10 t , confirming the above argument . in fig .
[ fig : iqhe4 ] , we plot the noninteracting hall conductivity as a function of the fermi energy for the four distinct four - layer stacks .
when electron - electron interactions are included at an electrostatic mean - field ( hartree ) level , charge redistribution changes the positions of plateaus which are away from the dirac point , but it does not alter the conductivity jumps at the dirac point . if we allow the remote hopping , a small @xmath109 plateau appears at the dirac point for the abab staked tetralayer due to the energy gap opened by @xmath91 .
this small gap is likely to be important only in very weak disorder samples , in which case electron - electron interactions beyond the hartree level are also likely to be important@xcite .
the property that the hall conductivity jumps by four units on crossing the dirac point for arbitrarily stacked tetralayer graphene is the most obvious experimental manifestation of the chirality sum rule discussed in this paper . in practice charged multilayers ( @xmath110 ) would normally be prepared by placing the system on one side of an electrode and gating . even though gating will redistribute charge and shift energies differently in different layers
, the landau level bunching we discussed should still be clearly reflected in quantum hall effect measurements . in numerical calculations
, we have found that for many stacks , the @xmath0-fold degeneracy at @xmath1 persists even outside the field range over which the minimal model is accurate . because of the topological character of the quantum hall effect , we expect the plateaus at @xmath111 to be exceptionally robust in most stacks , and that plateaus at the smaller hall conductivity samples will require exceptionally high quality samples .
we do not anticipate strong quantum hall effects with @xmath112 unless gaps between the landau levels are enhanced by interactions@xcite in very high quality@xcite samples .
finally , we note that chiral two - dimensional electron system tend toward momentum - space vortex states in which charge is spontaneously shifted between layers@xcite and that these instabilities are stronger in systems with larger @xmath12 .
the present work identifies abc stacked multilayer graphene as the most likely candidate for these exotic states .
999 for recent technical and popular reviews see a. k. geim and k. s. novoselov , nature mat . *
6 * , 183 ( 2007 ) ; a. k. geim and a. h. macdonald , phys .
today * 60 * ( 8) , 35 ( 2007 ) . k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , y. zhang , s.v .
dubonos , i. v. grigorieva , and a. a. firsov , science * 306 * , 666 ( 2004 ) . c. berger , z. song , t. li , x. li , a. y. ogbazghi , r. feng , z. dai , a. n. marchenkov , e. h. conrad , p. n. first , and w. a. de heer , phys .
b * 108 * , 19912 ( 2004 ) .
t. ohta , a. bostwick , t. seyller , k. horn , and e. rotenberg , science * 313 * , 951 ( 2006 ) .
a. rycerz , j. tworzydl , and c. w. j. beenakker , nat .
phys . 3 , * 172 *
( 2007 ) . v. v. cheianov , v. falko , and b. l. altshuler , science * 315 * , 5816 ( 2007 ) . k. s. novoselov , a. k. geim , s.v .
morozov , d. jiang , m. i. katsnelson , i. v. grigorieva , s. v. dubonos , and a. a. firsov , nature * 438 * , 197 ( 2005 ) . y. zhang , y. w. tan , h. l. stormer , and p. kim , nature * 438 * , 201 ( 2005 ) .
e. mccann and v.i .
falko , , 086805 ( 2006 ) .
j. c. charlier , j. p. michenaud , and x. gonze , , 4531 ( 1992 ) .
j. j. sakurai , _ modern quantum mechanics _ ( addison wesley , reading , 1994 ) .
the pure ab and abc cases have been analyzed in recent related papers : f. guinea , a. h. castro neto , and n. m. r. peres , , 245426 ( 2006 ) ; m. koshino and t. ando , , 085425 ( 2007 ) ; j. l. maes , f. guinea , and m. a. h. vozmediano , , 155424 ( 2007 ) ; m. nakamura and l. hirasawa , , 045429 ( 2008 ) .
the velocity of the @xmath35 doublets is determined by the strength of the coupling between the odd - length chain zero - energy state and isolated states on the sublattice opposite to the chain ends . for a chain of length @xmath113 , the chain s zero - energy state
has nonzero amplitude on the @xmath0 odd - index sites .
the velocity is reduced from the single sheet velocity by a factor of @xmath114 , where @xmath57 is the number of isolated sites opposite to the @xmath0 odd - index sites . in a similar manner ,
higher @xmath12 doublet hamiltonians are sometimes altered by a multiplicative factor by perturbative coupling to smaller @xmath12 doublets as in the single reversed layer example .
d. a. abanin , p. a. lee , and l.s .
levitov , , 176803 ( 2006 ) ; j. alicea and m. p. a. fisher , , 075422 ( 2006 ) ; m. o. goerbig , r. moessner , and b. doucot , , 161407(r ) ( 2006 ) ; k. yang , s. das sarma , and a. h. macdonald , , 075423 ( 2006 ) ; y. barlas , r. cote , k. nomura , and a. h. macdonald , arxiv:0803.0044 ( unpublished ) . k. nomura and a. h. macdonald , , 256602 ( 2006 ) .
hongki min , g. borghi , m. polini , and a. h. macdonald , , 041407(r ) ( 2008 ) .
|
we show that the low - energy electronic structure of arbitrarily stacked graphene multilayers with nearest - neighbor interlayer tunneling consists of chiral pseudospin doublets .
although the number of doublets in an @xmath0-layer system depends on the stacking sequence , the pseudospin chirality sum is always @xmath0 .
@xmath0-layer stacks have @xmath0 distinct landau levels at @xmath1 for each spin and valley , and quantized hall conductivity @xmath2 where @xmath3 is a non - negative integer .
|
a localized excess of emission in the microwave region was first detected in the @xmath0/dmr data and was initially attributed to free free emission ( kogut et al .
1996a , 1996b ) .
since then this anomalous emission has been detected by a number of authors ( de oliveira - costa et al .
2002 , 2004 ; banday et al . 2003 ; finkbeiner et al . 2004 ; watson et al . 2005 ; fernndez - cerezo et al .
2006 ) , and has been nicknamed ` @xmath1 @xmath2 ' . although initially ascribed to thermal bremsstrahlung in view of its strong correlation with thermal dust , low h@xmath3 surface brightness measurements ( leitch et al .
1997 ) implied gas temperatures in excess of @xmath4k , which were rejected on energetic grounds by draine & lazarian ( 1998a ) .
its physical mechanism has yet to be constrained ; the most popular interpretation is that of rapidly rotating dust grains , or @xmath5 @xmath6 ( draine & lazarian 1998a , 1998b ) .
other mechanisms which have been proposed include magnetic dust emission ( draine & lazarian 1999 ) , flat spectrum synchrotron ( bennett et al .
2003b ) , and bremsstrahlung from very hot electrons ( leitch et al . 1997 ) .
spinning dust causes an excess of emission in the 10 50ghz region of the spectrum , where the combination of synchrotron , bremsstrahlung and thermal dust emission is a minimum , and is problematic especially for cmb experiments which utilise this region to minimise foreground contamination and avoid atmospheric emission .
consequently the presence of a poorly constrained foreground such as this anomalous dust emission is a potentially serious problem for cmb observers and needs to be better understood in order to be correctly removed .
this has led to several directed observations ( finkbeiner et al .
2004 ; watson et al . 2005
; casassus et al .
2004 , 2006 ; dickinson et al .
2006 ) being made towards targets suggested by the theoretical predictions of draine & lazarian ( 1998b ; hereinafter dl98b ) . whilst these have been mainly directed at hii regions and dark clouds
, dl98b also suggest that it may be possible to detect @xmath7ghz emission from spinning dust in photodissociation regions . here
we present observations at 33ghz towards two hii regions and one supernova remnant ( snr ) for which infrared observations imply the presence of significant photodissociation regions ( pdrs ) .
the vsa is a 14 element interferometer sited at the teide observatory , in tenerife , at an altitude of 2400 m .
the vsa operates in a single 1.5ghz wide channel at a central frequency of 33ghz ( scott et al .
the 14 antennas use hemt amplifiers with typical system temperature @xmath835k . in its new super - extended configuration
the vsa uses mirrors of diameter 65 cm .
the horn - reflector antennas are mounted on a tilt table hinged east west and each antenna individually tracks the observed field by rotating its horn axis perpendicularly to the table hinge , wavefront coherence being maintained with an electronic path compensator system ( watson et al .
2003 ) .
the individual tracking of the vsa antennas allows for the filtering of contaminating signals .
these may be celestial sources , such as the sun and moon , or ground - spill and other environment based spurious signals .
the vsa also uses a ground - shield to minimise ground - spill .
the consequences of this design are two - fold .
first , the vsa is able to observe continuously and can filter out emission from the sun and moon when they are as close as @xmath9 .
second , the vsa is unaffected by ground - spill contamination for fields within 35@xmath10 of the zenith and so is able to make direct images of the sky , rather than employing the lead
trail approach of many other interferometers operating in the microwave band .
observations were carried out as single pointings with a primary beam of 72arcmin fwhm .
the data were calibrated using taua and casa in accordance with vsa reduction procedures .
this calibration , along with appropriate flagging and filtering of the data , was performed using the special purpose package reduce developed specifically for reduction of vsa observations .
details of the vsa reduction and calibration procedure can be found in dickinson et al ( 2004 ) and references therein .
many surveys at lower frequency do not have similar resolution to the super - extended vsa , making comparison difficult .
instead we extrapolate from the effelsberg 100 m telescope at 2.7ghz ( f@xmath11rst et al .
these data are single dish maps with a circular beam of 4.3arcmin fwhm . for a robust comparison
the 2.7ghz effelsberg maps were multiplied by the primary beam of the vsa and visibility sampling was performed in the @xmath12plane with a correction for the effelsberg beam .
the hii region lph96 ( ra = 06@xmath13 36@xmath14 40@xmath15 , @xmath16 = + 10@xmath10
46@xmath17 28@xmath18 , j2000 ) has been observed with the green bank 43 m telescope between 5 and 10ghz ( finkbeiner et al . , 2002 ) and was shown to have a rising spectrum consistent with that expected from spinning dust .
however , a pointed observation made with the cbi telescope ( dickinson et al . , 2006 ) at 31 ghz shows emission consistent with an approximately flat spectrum source with only little possibility of spinning dust . at 33ghz the peak flux density of lph96 is [email protected] beam@xmath20 .
the synthesized beam of the vsa towards lph96 is [email protected]@xmath22 and the source is slightly resolved with structure extending towards the north and southwest .
the 2.7ghz data after sampling gives a peak flux density of [email protected] jy beam@xmath20 . comparing this with the vsa flux density we find a spectral index @xmath3 = [email protected] , where the errors are calculated from the thermal noise outside the beam on the 2.7 and 33 ghz maps .
the spectral index , @xmath3 , is here defined so that flux density scales as @xmath23 .
this calculation fails to take into account systematic errors ; taking errors of 5 percent on the flux density scales at each frequency gives @xmath3 = [email protected] .
the vsa measurement shows no indication of the excess emission which would be expected at 33ghz from the warm neutral medium ( wnm ) spinning dust model of dl98b consistent with the green bank data .
the result is more consistent with that found by the cbi telescope ( dickinson et al .
2006 ) who found emission at 31ghz consistent only with @xmath24 .
dickinson et al . also note that the galactic plane survey of langston et al .
( 2000 ) failed to detect lph96 at 14.35 ghz with a detection limit of 2.5jy .
the supernova remnant ( snr ) 3c396 ( @[email protected] ) , is a shell - like remnant at radio frequencies , with a mean angular diameter of @xmath27 ( patnaik et al.1990 ) .
its spectral behaviour has been extensively studied in the radio , most notably by patnaik , and has a non - thermal radio spectrum with @xmath28 between @xmath29 mhz and @xmath30 ghz .
below 30 mhz catalogued flux densities for 3c396 are contaminated by the near by steep - spectrum pulsar , psr [email protected] ( manchester & taylor 1981 ) .
a pulsar wind nebula near the centre of this snr has been detected in x - rays ( see olbert et al . 2003 ) , but the remnant has not been detected optically .
only a lower limit of 7.7 kpc for its distance is available from its hi absorption observation ( caswell et al.1975 ) .
patnaik et al .
conclude that the neighbouring hii region nrao 591 is likely to be at a distance of @xmath32 kpc .
3c396 was observed in august 2006 in a short observation of 1.8 hours .
mapping and clean - based deconvolution were performed using the aips package .
the observation was not limited by thermal noise but rather by the dynamic range of the telescope and has rms noise of 34.0mjy .
the vsa map is shown in fig .
[ fig:3c396vsa ] ; it shows both 3c396 and also the hii region nrao 591 to the north west of the remnant .
the vsa beam towards 3c396 is @xmath33arcmin@xmath22 .
interesting features include a protrusion to the north and a faint detection of the `` blow - out '' tail to the north east , both of which are also present at longer radio wavelengths .
fits were obtained using the aips task jmfit by drawing a bounding box around both sources and fitting for two gaussians and a base level . the peak flux density for 3c396
was found to be [email protected] beam@xmath20 and the integrated flux density [email protected] at 33ghz , with the peak at 19@xmath1301@xmath1443@xmath34 + 05@xmath1022@xmath3522@xmath36 ( b1950 ) . for the secondary source nrao 591
we find a peak flux density of [email protected] beam@xmath20 and an integrated flux density of [email protected] centered at 19@xmath1300@xmath1447@xmath37 + 05@xmath1031@xmath3506@xmath38 ( b1950 ) .
the errors here include contributions from the rms noise on the observation , the statistical error from the gaussian fits , and a conservative 5% error from the flux calibration which dominates the overall value .
a complete discussion of the vsa flux calibration may be found in dickinson et al .
( 2004 ) . 2 , @xmath261 , 1 , 2 , 4 , 8 , 16 , 32 and 64 @xmath39.[fig:3c396vsa ] ]
again we use observations from the effelsberg 100 m telescope to examine the emission at lower radio frequencies . using effelsberg data at 1.4 ( reich et al .
1997 ) and 2.7ghz we convolve the data at 2.7ghz to match the 1.4ghz resolution of 9.4arcmin and find flux densities of 14.9 @xmath19 1.3 jy and 11.4 @xmath19 1.3 jy at 1.4 and 2.7ghz , respectively .
these values agree with those of reich et al .
( 1990 ) and give a spectral index of @xmath3 = 0.46 , which is consistent with the mean of 0.45 found for shell - type snrs ( green , 2004 ; 2006 ) .
after sampling the effelsberg observation at 2.7ghz to match the _
uv_coverage of the super - extended vsa towards 3c396 we find a flux density of [email protected] .
our best fit index implies a flux density of 3.4jy at 33ghz extrapolated from the effelsberg data .
we investigate the spectrum of 3c396 using our own flux density and other published values ; all errors are quoted to 1@xmath39 .
taken at face value the vsa observation of 3c396 would imply an index @xmath40 = [email protected] , broadly consistent with a region of thermal emission .
alternatively , if we assume the index @xmath41 of 0.46 determined from the effelsberg data , then the vsa flux density would imply an excess of emission seen at microwave frequencies towards this source . a large number of observations between 400mhz and 5ghz exist and to confirm this spectral index we compile a spectrum using flux densities taken from patnaik et al .
( 1990 ) who made corrections to the original measurements to bring them onto the flux density scale of baars et al .
the integrated spectrum of 3c396 is shown in fig .
[ fig:3c396spec ] .
data from patnaik et al . is shown as crosses , the data from effelsberg as filled squares and the vsa data as an unfilled diamond . performing a weighted least squares fit to the patnaik and effelsberg data we find an index of @xmath42 .
the vsa measurement at 33ghz is inconsistent with this spectrum , shown in fig .
[ fig:3c396spec ] as a solid line . including the vsa data at 33ghz we find a spectral index of @xmath43 , shown as a dashed line . however , @xmath44 values for the two fitted spectra show that the spectral index of 0.42 ( @xmath45 , 31 d.o.f . ,
p(1.07 ) = 0.366 ) is a better fit to the data .
indeed a spectral index of 0.32 ( @xmath46 , 32 d.o.f . ,
p(1.95 ) = 0.001 ) would be unusually flat for a supernova remnant with less than 7% of shell like or possible shell like supernova remnants having a spectral index of less than or equal to 0.32 ( green 2006 ) .
however , we hesitate to over - interpret this statistic since the catalogued spectral indices are by no means uniform in quality . since previous observations ( finkbeiner et al .
2004 ) have suggested the presence of spinning dust in hii regions we have also compiled a spectrum for nrao591 .
we use flux densities compiled by patnaik et al .
( 1990 ) which are plotted in fig .
[ fig : nraospec ] and find a spectrum compatible with that of optically - thin free free emission .
m data overlaid with the vsa observation at 33ghz .
contours are @xmath262 , @xmath261 , 1 , 2 , 4 , 8 , 16 , 32 and 64@xmath39.[fig:100um ] ] for the source 3c396 we constrain the possible excess emission relative to the non - thermal contribution at 33ghz by subtracting a non - thermal model extrapolated from lower frequency . for the flux densities between 400mhz and 10ghz , shown in fig .
[ fig:3c396spec ] , a best fit spectral index of @xmath47 was found .
this index implies a flux density of @xmath480.11jy at 33ghz , where the error is statistical only .
this leaves @xmath490.35jy that may be due to an anomalous component .
if we include a 5 percent error on the flux calibration these errors are increased to 0.24 and 0.41jy , respectively .
the excess accounts for 37% of the total 33ghz flux density .
radio recombination line observations ( anantharamaiah , 1985 ) put an upper limit on the emission measure of 3c396 of 280@xmath50pc and fits an electron temperature of 5000k for the gas .
this implies an upper limit on the free free emission from the snr of 0.061jybeam@xmath20 at 33ghz .
the beam sizes used for this measurement are poorly matched to that of the vsa and may cause the values to be over - estimated . taking this into account
it can be seen that the free free contribution is small compared to the non - thermal and can only account for @xmath512% of the excess emission .
in addition , models including both free free and non - thermal contributions provide poor fits to the patnaik et al . data .
the possibility of the vsa seeing a secondary radio source in projection towards 3c396 is small .
radio sources with flux densities @xmath52mjy are seen with a frequency of 0.2/deg@xmath22 .
the possibility of a rising spectrum source at @xmath53 5ghz reduces this number by a factor of 10 ( waldram et al . , 2003 ; cleary et al . ,
2005 ) .
olbert et al .
( 2003 ) report the presence of a small pulsar wind nebula ( pwn ) within the 3c396 snr , although the authors note the lack of any corresponding radio feature in high resolution 20 cm vla images of the remnant ( dyer & reynolds 1999 ) . for this pwn ( or plerion - like component ) , to account for the excess flux density seen at 33ghz it would be very obvious at 1.4ghz , where it would have to contribute approximately 1/6 of the total flux density , assuming a flat spectral index
olbert et al .
however suggest that its contribution is @xmath541/25 of the total radio flux density at 1.4ghz . the infra - red emission at 100@xmath55 m towards 3c396 is shown in fig .
[ fig:100um ] . in this region of the galactic plane it is not certain whether the emission is associated with the snr remnant , or is merely a projection . reach et al .
( 2006 ) suggest that higher resolution spitzer data shows emission at 3.6 to 8@xmath55 m which may be associated with the snr .
they find irac colours for these regions consistent with both pdrs and hii regions .
if the excess emission seen at 33ghz from 3c396 is associated with these regions then it is possible it may arise from the dipole emission of draine & lazarian ( 1998a;1998b ) .
however , the discrepancy in resolution between the spitzer and vsa telescopes precludes a more detailed spatial analysis .
this possibility is illustrated in figure [ fig:3c396spec2 ] where the data is shown with the wnm spinning dust model of dl98b .
for clarity we have binned the data of patnaik et al . at similar frequencies , and have excluded one point at 10.7ghz .
in addition , we note that the wnm model of dl98b , where the column density towards 3c396 is determined from the full - sky magnitude map of schlegel , finkbeiner & davis ( 1998 ) , predicts a peak flux density at 33ghz of 3.5jybeam@xmath20 at the resolution of the vsa due to spinning dust .
this gives an integrated flux density of approximately 7jy , using the ratio of peak to integrated flux found with the vsa .
however , it is likely that this method overestimates the integrated flux density since the morphology of the thermal dust emission appears more compact than the emission seen with the vsa .
it is however within a factor of 3 of the excess emission we see at 33ghz . in conclusion ,
we have assessed the possibility of spinning dust emission at 33ghz towards the snr 3c396 .
apart from cas a and tau a , few snr have been studied in the microwave region .
consequently , in order to confirm this possibility further measurements are required in the range 1020ghz .
we thank the staff of the mullard radio astronomy observatory , the teide observatory and the jodrell bank observatory for their invaluable assistance in the commissioning and operation of the vsa .
the vsa is supported by pparc and the iac . as acknowledges the support of a pparc studentship .
part of the research described in this paper was carried out at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration .
d. a. , 2006 , ` a catalogue of galactic supernova remnants ( 2006 april version ) ' , astrophysics group , cavendish laboratory , cambridge , united kingdom ( available at http://www.mrao.cam.ac.uk/surveys/snrs/ ) .
|
we present results from observations made at 33 ghz with the very small array ( vsa ) telescope towards potential candidates in the galactic plane for spinning dust emission . in the cases of the diffuse hii regions lph96 and nrao591 we find no evidence for anomalous emission and , in combination with effelsberg data at 1.4 and 2.7 ghz , confirm that their spectra are consistent with optically thin free free emission . in the case of the infra - red bright snr 3c396 we find emission inconsistent with a purely non - thermal spectrum and discuss the possibility of this excess arising from either a spinning dust component or a shallow spectrum pwn , although we conclude that the second case is unlikely given the strong constraints available from lower frequency radio images .
radiation mechanisms : general radio continuum : ism dust , extinction ism : individual : 3c396
|
the non - abelian gauge theory is renormalizable in the sense that one can send the momentum cutoff infinity while keeping the strength of the interaction finite . in such a theory with an infinite cutoff , any observable
must be renormalized one and hence in gauge theory how to construct renormalized quantities is a fundamental question .
the gradient or wilson flow @xcite provides a surprisingly versatile method to define renormalized quantities in gauge theory , without explicitly referring to the perturbative renormalization .
see also ref .
since renormalized quantities have the meaning being independent of regularization , this method is especially useful in the context of lattice regularization with which one clearly wants to define renormalized quantities without referring to perturbation theory as much as possible . that the bare gauge field and its composite operators evolved by the gradient flow possess finite correlation functions without any wave function renormalization
was proven in ref .
@xcite to all orders of perturbation theory .
the gradient flow has been generalized to the flow of the fermion field @xcite and it is known that wave function renormalization of the flowed fermion elementary field makes any correlation functions and composite operators finite .
backed by these theoretical understandings , the gradient flow has been applied to many problems in lattice gauge theory ; refs .
@xcite and @xcite are recent reviews .
more recently , as the reference list @xcite shows , the number of related works is rapidly growing ( see also proceedings of the recent lattice conferences ) . in most of these works ,
the gradient flow is used to the scale setting , definition of a non - perturbative running coupling , and to define the topological charge .
another interesting branch of application is to define a wide class of renormalized composite operators using the flow , such as the energy
momentum tensor @xcite and the axial - vector current @xcite . in refs .
@xcite , applications of the gradient flow for the construction of the energy
momentum tensor are studied from a somewhat different perspective from refs .
see also refs .
@xcite for related works . in the present paper ,
we present an alternative perturbative proof of the renormalizability of the system defined by the gradient flow and the fermion flow . although the proof of the renormalizability has already been given in refs .
@xcite , at some stages of the proof , the explanations in refs .
@xcite seem rather concise . in the present paper
, we tried to improve the presentations in the following points .
as refs .
@xcite , we use a @xmath0-dimensional field theoretical representation of the system defined by the flow .
however , to show the renormalizability , we employ the ward
takahashi ( wt ) relation associated with the becchi
stora ( brs ) symmetry @xcite expressed in terms of the effective action ( the generating functional of the 1pi correlation functions ) , the zinn - justin equation @xcite ( see ref @xcite for example for the exposition ) .
this type of proof , that was adopted also in ref .
@xcite to prove the renormalizability of the gradient flow in the 2d @xmath1 non - linear sigma model , is easy to follow step by step , although tends to become somewhat lengthy .
also , in the present paper , as in ref .
@xcite , we provide a precise definition of the flow - time derivative by the forward difference prescription .
we show that this prescription is consistent with the above brs symmetry on which the derivation of the wt relation relies .
this precise definition resolves a certain subtle issue concerning the so - called `` flow - line loop '' ; the vanishing - ness of the flow - line loop is crucially important for the renormalizability proof .
the prescription also clearly illustrates the origin of the `` boundary counterterm '' .
we believe that our proof is quite accessible because of these aspects .
this paper is organized as follows . in sect .
[ sec:2 ] , we recapitulate the basic definition of the gradient flow and the fermion flow in vector - like gauge theories .
we then elucidate the perturbative expansion of the flowed system . in sect .
[ sec:3 ] , we introduce a @xmath0-dimensional local field theory @xcite whose feynman rules reproduce the perturbative expansion of the flowed system in sect .
[ sec:2.2 ] .
we lay some emphasis on the usefulness of discretized flow time ; this resolves a certain subtle issue concerning a `` flow - line loop '' , that is crucial for the perturbative equivalence of the @xmath0-dimensional field theory and the flowed system . section [ sec:4 ] is the main part of the present paper , the proof of the renormalizability .
first , we study the case of the pure yang mills theory . in sect .
[ sec:4.1 ] , we derive the wt relation in the @xmath0-dimensional field theory . after stating our way of multiplicative renormalization ( eqs . ) and expressing the wt relation in terms of renormalized quantities ( sect . [ sec:4.2 ] )
, we argue the most general form of the divergent part in the effective action ( sect .
[ sec:4.3 ] ) .
finally , by examining the constraint imposed by the wt relation on the divergent part , we find the validity of our renormalization
[ sec:4.4 ] ) to all orders of perturbation theory . in sect .
[ sec:4.5 ] , the proof is generalized to include the fermion fields .
the renormalization factors are defined by eqs .
. here , we find that the flowed fermion fields require wave function renormalization @xcite , in contrast to the flowed gauge field . in sect .
[ sec:4.6 ] , we argue the finiteness of composite operators of the flowed fields . section [ sec:5 ] is devoted for the conclusion .
here is our notation : the spacetime dimension is set to @xmath2 because in what follows we implicitly assume dimensional regularization that preserves the gauge symmetry in perturbation theory .
the signature of the metric is euclidean and all dirac matrices are hermitian .
the summation over repeated lorentz and gauge indices is understood .
the generators of the gauge group are normalized as @xmath3 we use the following abbreviation for the momentum integral : @xmath4
we consider the gradient flow and the fermion flow in the vector - like gauge theory defined by @xmath6 + \int d^dx\,\bar{\psi}(x)({{\ooalign{\hfil/\hfil\crcr$d$}}}+m_0)\psi(x ) , \label{eq:(2.1)}\ ] ] where @xmath7 and @xmath8 are the bare gauge coupling and the bare mass parameter , respectively .
the field strength is given by @xmath9 \label{eq:(2.2)}\ ] ] from the gauge potential @xmath10 and the covariant derivative on the fermion field @xmath11 in @xmath12 is defined by @xmath13 to define perturbation theory , we introduce the gauge fixing term and the faddeev
popov ghost term by @xmath14\right\ } , \label{eq:(2.4)}\ ] ] where @xmath15 is the nilpotent brs transformation on the gauge field , the ghost field ( @xmath16 ) , the anti - ghost field ( @xmath17 ) , and the nakanishi lautrup field ( @xmath18 ) : @xmath19 from these constructions , the above @xmath5-dimensional field theory is manifestly brs invariant : @xmath20 the yang
mills gradient flow is an evolution of a gauge field configuration @xmath21 along a fictitious time ( the flow time ) @xmath22 , according to the flow equation @xcite @xmath23 where @xmath24 is the field strength of the flowed field , @xmath25 , \label{eq:(2.10)}\ ] ] and @xmath26 is the covariant derivative on the gauge field , @xmath27 . \label{eq:(2.11)}\ ] ] the first term on the right - hand side of the flow equation is the `` gradient '' in the functional space , @xmath28 , where @xmath29 is the yang
mills action integral for the flowed field .
the flow equation has a form of the diffusion equation and the evolution for @xmath30 effectively damps high - frequency modes in the configuration .
the second term on the right - hand side of the flow equation is a `` gauge fixing term '' and it makes the integrand of the feynman integral well behaved .
it can be shown @xcite that , however , any gauge - invariant quantity which does not contain the flow - time derivative @xmath31 is independent of the `` gauge fixing parameter '' @xmath32 .
an evolution of fermion fields being similar to the above can be introduced .
a possible choice is @xcite @xmath33 \chi(t , x ) , & \chi(t=0,x)&=\psi(x ) , \label{eq:(2.12)}\\ & \partial_t\bar{\chi}(t , x ) = \bar{\chi}(t , x ) \left[\overleftarrow{\delta } + \alpha_0\partial_\mu b_\mu(t , x)\right ] , & \bar{\chi}(t=0,x)&=\bar{\psi}(x ) , \label{eq:(2.13)}\end{aligned}\ ] ] where @xmath34 these evolutions of the fermion fields are referred to as the fermion flow in what follows . the flow equation for the gauge field , eq .
, can be formally solved as @xcite @xmath35 , \label{eq:(2.16)}\ ] ] where @xmath36 \label{eq:(2.17)}\ ] ] is termed the _ heat kernel _ in what follows , and @xmath37 -[b_\nu,\partial_\mu b_\nu ] + ( \alpha_0 - 1)[b_\mu,\partial_\nu b_\nu ] + [ b_\nu,[b_\nu , b_\mu ] ] \label{eq:(2.18)}\ ] ] arises from the non - linear terms in the flow equation .
then by iteratively solving eq .
, we have a perturbative expansion for the flowed field @xmath38 in terms of the initial value @xmath39 .
note the retarded nature of the solution ; the heat kernels connect a sequence of points , @xmath40 , @xmath41 , @xmath42 , , in the temporal order @xmath43 .
note also that the gaussian damping factor in the heat kernel @xmath44 behaves as @xmath45 or @xmath46 , where @xmath47 and @xmath48 are flow times at end points of the heat kernel , so the damping factors become inactive when the two flow times @xmath47 and @xmath48 coincide . for the fermion flow and
, we have @xmath49 , \label{eq:(2.19)}\\ \bar{\chi}(t , x)&=\int d^dy\ , \left [ \bar{\psi}(y)k_t(x - y ) + \int_0^tds\,\bar{\chi}(s , y)\overleftarrow{\delta}'k_{t - s}(x - y ) \right ] , \label{eq:(2.20)}\end{aligned}\ ] ] where @xmath50 is also referred to as the _ heat kernel _ , and @xmath51 by iteratively solving eqs . and , we again have a perturbative expansion of the fermion flow .
again note the retarded nature of the solutions and the fact that the gaussian factor behaves as @xmath45 .
the initial values for the above flows , @xmath21 , @xmath11 , and @xmath52 , are quantum fields being subject to the functional integral .
correlation functions of the flowed fields are thus obtained by expressing the flowed fields in terms of the initial values following the above procedures and then carrying out the functional integral over the latter .
for example , in the tree - level approximation , we have @xmath53 \label{eq:(2.24)}\ ] ] and , similarly for the fermion field , @xmath54 in the tree - level approximation .
eqs . and
are referred to as _ propagators _ in what follows . in a diagrammatic representation of the above perturbative expansion of the flowed system ( see refs .
@xcite for details ) , we thus have two types of `` lines '' , one represents the heat kernels and , and another is for the propagators and .
the former line is called the _ flow line _ ( or heat kernel ) , while the latter is simply called the propagator .
note that the propagators carry the gaussian damping factor of the form @xmath55 or @xmath56 , where @xmath47 and @xmath48 are flow times at end points of the propagator , so the damping factor is always active even when the two flow times @xmath47 and @xmath48 coincide as far as @xmath30 or @xmath57 ; this is a crucial difference from the heat kernel or the flow line .
also we have two types of vertices , one arises from the non - linear terms in the flow equations , eqs . , , and , and another represents the conventional interaction vertices in eq . .
an important feature of this diagrammatic representation of the flowed system is that there is no closed loop totally consisting of the flow lines ( or heat kernels ) .
this property , which turns out to be crucial for the renormalizability of the flowed system , follows from the fact that to form a loop in a diagram , we have to wick contract the initial values and then the contraction results in one of the propagators and , but not heat kernels . the statement of the renormalizability of the flowed system , first proved in refs .
@xcite to all orders of perturbation theory , is that any correlation functions and composite operators of flowed fields are made finite by the conventional renormalization in the gauge theory and wave function renormalization of the flowed fermion fields . in particular ,
no wave function renormalization of the flowed gauge field is required .
the purpose of the present paper is to give an another proof of this statement .
the total action of a @xmath0-dimensional local field theory we will consider is @xmath58 where @xmath5-dimensional actions , @xmath29 , @xmath59 , and @xmath60 , are already introduced in sect . [ sec:2.1 ] .
the @xmath0-dimensional actions are defined by @xcite @xmath61 \right\ } \notag\\ & \qquad { } + \int_0^\infty dt\,\int d^dx\ , \bigl\{\bar{\lambda}(t , x ) \left[\partial_t-\delta+\alpha_0\partial_\mu b_\mu(t , x)\right]\chi(t , x ) \notag\\ & \qquad\qquad\qquad\qquad\qquad{}+ \bar{\chi}(t , x ) \left[\overleftarrow{\partial}_t -\overleftarrow{\delta}-\alpha_0\partial_\mu b_\mu(t , x)\right ] \lambda(t , x ) \bigr\ } \label{eq:(3.2)}\end{aligned}\ ] ] and @xmath62\right\}. \label{eq:(3.3)}\ ] ] in eqs . and
, @xmath0-dimensional fields , @xmath38 , @xmath63 , and @xmath64 , are subject to the boundary conditions , @xmath65 corresponding to the initial conditions of the flow equations in eqs . , , and .
the lagrange multiplier fields , @xmath66 , @xmath67 , and @xmath68 , on the other hand , obey free boundary conditions , that is , @xmath69 , @xmath70 , and @xmath71 are integrated as independent variables in the functional integral . in eq .
, @xmath72 is a @xmath0-dimensional ghost field which obeys the boundary condition , @xmath73 and @xmath74 is a ghost analogue of the lagrange multiplier field ; it also obeys the free boundary condition at @xmath75 .
in addition to eqs .
, we introduce the brs transformation on the above @xmath0-dimensional fields by @xmath76 , & & \label{eq:(3.7 ) } \\
\delta\chi(t , x)&=-d(t , x)\chi(t , x ) , & \delta\bar{\chi}(t , x)&=-\bar{\chi}(t , x)d(t , x ) , \label{eq:(3.8 ) } \\
\delta\lambda(t , x)&=-d(t , x)\lambda(t , x ) , & \delta\bar{\lambda}(t , x)&=-\bar{\lambda}(t , x)d(t , x ) .
\label{eq:(3.9)}\end{aligned}\ ] ] these are simply a standard form of the brs transformation in which the parameter of the gauge transformation is replaced by the @xmath0-dimensional ghost field @xmath72 ; here @xmath77 is regarded as a field in the adjoint representation and @xmath67 ( @xmath68 ) is regarded as a field in the same representation as @xmath63 ( @xmath64 ) . by this construction ,
the nilpotency @xmath78 on these fields immediately follows as for the conventional @xmath5-dimensional brs transformation . on the other hand , for the `` ghost lagrange multiplier '' @xmath79
, we define the brs transformation as @xmath80 by a somewhat troublesome calculation , one can confirm that this transformation on @xmath79 is nilpotent @xmath78 without using any equation of motion .
one may wonder how this intricate structure of nilpotent @xmath81 arises . in appendix [ app :
a ] , we show that this structure of @xmath81 can be naturally understood if one considers an enlarged field space in which the covariance under a @xmath0-dimensional gauge transformation is manifest .
one can also confirm by a direct calculation that the @xmath0-dimensional system is brs invariant : @xmath82 this can be seen by writing @xmath83
\notag\\ & \qquad\qquad { } -2\int_0^\infty dt\int d^dx\ , \operatorname{tr}\bar{d}(t , x)e(t , x ) , \label{eq:(3.12)}\end{aligned}\ ] ] where @xmath84\chi(t , x ) , \label{eq:(3.15 ) } \\ \bar{f}(t , x)&\equiv \bar{\chi}(t , x ) \left[\overleftarrow{\partial}_t -\overleftarrow{\delta}-\alpha_0\partial_\mu b_\mu(t , x)\right ] , \label{eq:(3.16)}\end{aligned}\ ] ] and then using the relations @xcite @xmath85+d_\mu e(t , x ) , \label{eq:(3.17 ) } \\
\delta e(t , x)&=-\left\{e(t , x),d(t , x)\right\ } , \label{eq:(3.18 ) } \\
\delta f(t , x)&=-d(t , x)f(t , x)-e(t , x)\chi(t , x ) , \label{eq:(3.19 ) } \\
\delta\bar{f}(t , x)&=-\bar{f}(t , x)d(t , x)-\bar{\chi}(t , x)e(t , x ) .
\label{eq:(3.20)}\end{aligned}\ ] ] again , this ( apparently miraculous ) brs invariance of the action can be naturally understood in the enlarged field space ( see appendix [ app : a ] ) . in this subsection , we show that perturbation theory in the above @xmath0-dimensional field theory precisely reproduces the perturbative expansion of the gradient flow we have considered in sect .
[ sec:2.3 ] .
this fact then allows us to use the @xmath0-dimensional field theory to prove the perturbative renormalizability of the flowed system . for this
, it is quite helpful to specify the precise meaning of the flow - time derivative @xmath31 . in what follows
, we understand that the flow time is discretized in the step @xmath86 .
the integral is thus given by the sum , @xmath87 for the flow - time derivative , we adopt the _ forward difference prescription _ , @xmath88 \label{eq:(3.22)}\ ] ] for any field @xmath89 .
in particular , this prescription implies that for @xmath75 , @xmath90 , & & \label{eq:(3.23 ) } \\
\partial_t\chi(t=0,x ) & \equiv\frac{1}{\epsilon}\left[\chi(t=\epsilon , x)-\psi(x)\right ] , & \bar{\chi}(t=0,x)\overleftarrow{\partial}_t & \equiv\frac{1}{\epsilon } \left[\bar{\chi}(t=\epsilon , x)-\bar{\psi}(x)\right ] , \label{eq:(3.24 ) } \\ \partial_t d(t=0,x ) & \equiv\frac{1}{\epsilon}\left[d(t=\epsilon , x)-c(x)\right ] , & & \label{eq:(3.25)}\end{aligned}\ ] ] because of the boundary conditions and .
the ordering of removing regularizations is also quite important : after applying the feynman rules to a given set of diagrams , we understand that we first take the continuous flow - time limit @xmath91 and then , after the renormalization , remove the dimensional regularization , @xmath92 ( or other @xmath93-dimensional regularization such as the lattice ) . under the above prescription , to obtain the tree - level propagators , we introduce the heat kernels with the discretized flow time @xcite by @xmath94 , \label{eq:(3.26 ) } \\
k_t^\epsilon(x;\alpha_0 ) & \equiv\int_p e^{ipx } ( 1-\epsilon\alpha_0p^2)^{t/\epsilon}. \label{eq:(3.27)}\end{aligned}\ ] ] these quantities satisfy , by construction , @xmath95 & = \partial_\rho\partial_\rho k_t^\epsilon(x)_{\mu\nu } + ( \alpha_0 - 1)\partial_\mu\partial_\rho k_t^\epsilon(x)_{\rho\nu } , \label{eq:(3.28 ) } \\
k_0^\epsilon(x)_{\mu\nu}&=\delta_{\mu\nu}\delta(x ) , \label{eq:(3.29)}\end{aligned}\ ] ] and @xmath96 & = \alpha_0\partial_\mu\partial_\mu k_t^\epsilon(x;\alpha_0 ) , \label{eq:(3.30 ) } \\ k_0^\epsilon(x;\alpha_0)&=\delta(x ) .
\label{eq:(3.31)}\end{aligned}\ ] ] first , we consider the propagators containing the flowed gauge field @xmath38 .
we change the variable from @xmath97 to @xmath98 as @xcite @xmath99 then , because of eq . , the boundary condition becomes simply @xmath100 and , because of eq . , the kinetic term of @xmath98 is given by @xmath101 -\partial_\nu\partial_\nu b_\mu^a(t , x ) + ( 1-\alpha_0)\partial_\mu\partial_\nu b_\nu^a(t , x)\biggr\}+\dotsb .
\label{eq:(3.34)}\end{aligned}\ ] ] the corresponding schwinger dyson equation in the tree - level yields , @xmath102 l_\nu^b(s , y ) + \left [ -\delta_{\mu\rho}\partial_\sigma\partial_\sigma + ( 1-\alpha_0)\partial_\mu\partial_\rho \right]b_\rho^a(t , x)l_\nu^b(s , y)\right\rangle_0 \notag\\ & = \frac{1}{\epsilon}\delta^{ab}\delta_{\mu\nu}\delta_{t , s}\delta(x - y ) . \label{eq:(3.35)}\end{aligned}\ ] ] this equation can then be solved step by step in the discretized @xmath47 , starting from @xmath75 at which @xmath103 because of the boundary condition .
in particular , we find @xmath104 and generally , @xmath105 where @xmath106 is a `` regularized step function '' , @xmath107 we then note that , from the structure of the action , the @xmath108 and @xmath109 propagators in the tree - level vanish .
thus , by using eq . , we find @xmath110 . \label{eq:(3.42)}\end{aligned}\ ] ] these results show that , in the @xmath91 limit , the former @xmath111 propagator becomes the heat kernel ( or flow line ) in the perturbative expansion of the gradient flow , eq . , and the latter @xmath112 propagator reproduces eq . . in a similar way
, we find for other flowed fields , @xmath113 which , for @xmath91 , reproduce heat kernels and propagators in sect .
[ sec:2.3 ] . from the structure of @xmath114 in eq . , it is obvious that the interaction vertices derived from eq .
are identical to the non - linear terms in the flow equations , @xmath115 in eq . and @xmath116 and @xmath117 in eqs . and .
thus , we see that the feynman rules of the above @xmath0-dimensional field theory basically coincide with the perturbative expansion of the flowed system in sect .
[ sec:2.3 ] .
a possible subtlety arises , however , for a feynman loop consisting only of the @xmath111 propagators ( and similarly for a loop consisting only of the @xmath118 and @xmath119 propagators ) , because although we can write such a feynman diagram in the @xmath0-dimensional field theory , we have no such loop diagram of the flow lines ( heat kernels ) in the perturbative expansion of the flowed system , as we have noted at the end of sect . [ sec:2.3 ] . in reality
, this possible discrepancy does not arise according to the above forward difference prescription , because we have for example @xmath120 for @xmath121 .
this implies that such a loop consisting only of the @xmath111 propagators ( and a similar loop of the @xmath118 or @xmath119 propagators ) identically vanishes ; this is the role of the forward difference prescription .
a similar remark applies to a loop consisting of the @xmath122 propagators .
the interaction vertex in @xmath123 in eq .
does not contribute to correlation function without containing @xmath124 and @xmath125 , because any loop consisting of the @xmath122 propagator vanishes because of the property @xmath126 for @xmath121 .
note that we are interested only in correlation functions without containing @xmath124 and @xmath125 ; these fields are auxiliary tools simply to derive the wt relation associated with the brs symmetry .
this completes our argument for the perturbative equivalence of the @xmath0-dimensional field theory and the system defined by the flow equations .
we can now study the structure of the renormalization in the flowed system , by employing the @xmath0-dimensional local field theory .
in this section , we give a proof of the renormalizability of the flowed system on the basis of the @xmath0-dimensional local field theory introduced in the previous section .
to avoid our basic reasoning from being made obscure by too complicated mathematical expressions , we first extensively study the case of the pure yang mills theory .
then , in sect
. [ sec:4.5 ] , we explain how the proof is modified by the inclusion of fermion fields . in this subsection , we derive the wt relation associated with the brs symmetry ( the zinn - justin equation ) of the @xmath0-dimensional field theory . the argument is standard , except a caution associated with non - trivial relations among field variables , eqs . and .
first , we introduce the external sources for each field by @xmath127 \notag\\ & \qquad { } + 2\fint_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ j_\mu^b(t , x)b_\mu(t , x)+j^d(t , x)d(t , x)\right ] \notag\\ & \qquad\qquad { } + 2\int_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ j_\mu^l(t , x)l_\mu(t , x)+j^{\bar{d}}(t , x)\bar{d}(t , x)\right ] .
\label{eq:(4.1)}\end{aligned}\ ] ] in writing this , we have noted that the field @xmath128 is not an independent dynamical variable but rather @xmath129 .
thus , we should not include the external source @xmath130 for @xmath38 at @xmath75 , @xmath131 , because @xmath131 can not be distinguished from @xmath132 . in order to take this point into account
, we have introduced a flow - time integration with the @xmath75 time - slice being excluded : @xmath133 this remark is applied also to the external source @xmath134 for @xmath72 , because @xmath135 is not an independent variable .
we next introduce another source term for the brs transformations of each field : @xmath136 \notag \\ & \qquad { } + 2\fint_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ k_\mu^b(t , x)d_\mu d(t , x)-k^d(t , x)d(t , x)^2\right ] \notag \\ & \qquad\qquad { } + 2\int_0^\infty dt\,\int d^dx\ , \operatorname{tr}\bigl\ { k_\mu^l(t , x)[l_\mu(t , x),d(t , x ) ] \notag \\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad { } + k^{\bar{d}}(t , x)\left[d_\mu l_\mu(t , x ) -\{d(t , x),\bar{d}(t , x)\}\right]\bigr\}. \label{eq:(4.3)}\end{aligned}\ ] ] the symbol @xmath137 is used for the same reason as for eq . .
note that because of the nilpotency of the brs transformation @xmath78 , @xmath138 . then , as usual , we consider the brs transformation @xmath15 of integration variables in the functional integral @xmath139}\equiv\int d\mu\,e^{-s_{\text{tot}}-s_j - s_k } , \label{eq:(4.4)}\ ] ] where @xmath140 is the functional integration measure .
then , since @xmath141 and @xmath142 are invariant under @xmath15 while @xmath143 is not , we have the following identity among correlation functions : @xmath144 \right\rangle \notag \\ & \qquad { } + \left\langle-2\fint_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ j_\mu^b(t , x)d_\mu d(t , x)+j^d(t , x)d(t , x)^2\right]\right\rangle \notag \\ & \qquad\qquad { } + \biggl\langle-2\int_0^\infty dt\,\int d^dx\ , \operatorname{tr}\bigl\ { j_\mu^l(t , x)[l_\mu(t , x),d(t , x ) ] \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad { } -j^{\bar{d}}\left[d_\mu l_\mu(t , x ) -\{d(t , x),\bar{d}(t , x)\}\right]\bigr\ } \biggr\rangle=0 . \label{eq:(4.5)}\end{aligned}\ ] ] in reality , this identity is broken by @xmath145 terms under the discrete flow - time prescription that we are assuming at present . to explicitly see those @xmath145 terms safely vanish for the @xmath91 limit
requires a little work and appendix [ app : b ] is devoted to this task . in what follows ,
we neglect those @xmath145 terms by assuming that the limit @xmath91 is taken in an appropriate stage .- dimensional field theory @xcite in which the brs symmetry is manifest and then study the perturbative renormalizability by using this regularization . ] we then introduce the effective action @xmath146 , the generating functional of one - particle irreducible ( 1pi ) diagrams in the presence of @xmath147 external sources , by the legendre transformation from external sources to the expectation values by , @xmath148 \notag\\ & \equiv w[j , k ] \notag\\ & \qquad { } + 2\int d^dx\ , \operatorname{tr}\left [ j_\mu^a(x)a_\mu(x)+j^c(x)c(x)+j^{\bar{c}}(x)\bar{c}(x)+j^b(x)b(x)\right ] \notag\\ & \qquad\qquad { } + 2\fint_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ j_\mu^b(t , x)b_\mu(t , x)+j^d(t , x)d(t , x)\right ] \notag\\ & \qquad\qquad\qquad { } + 2\int_0^\infty dt\,\int d^dx\ , \operatorname{tr}\left [ j_\mu^l(t , x)l_\mu(t , x)+j^{\bar{d}}(t , x)\bar{d}(t , x)\right ] , \label{eq:(4.6)}\end{aligned}\ ] ] where field variables in this expression , such as @xmath21 , stand for expectation values rather than dynamical variables .
note that the @xmath147 external sources are not legendre transformed ; the dependence on @xmath147 is thus taken over from @xmath149 to @xmath146 . then , noting the relation such as @xmath150 , the wt relation is expressed in terms of the effective action @xmath146 as @xmath151 \notag\\ & \qquad { } + \fint_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\mathit{\gamma}}{\delta b_\mu^a(t , x ) } \frac{\delta\mathit{\gamma}}{\delta k_\mu^{ba}(t , x ) } + \frac{\delta\mathit{\gamma}}{\delta d^a(t , x ) } \frac{\delta\mathit{\gamma}}{\delta k^{da}(t , x ) } \right ] \notag\\ & \qquad\qquad { } + \int_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\mathit{\gamma}}{\delta l_\mu^a(t , x ) } \frac{\delta\mathit{\gamma}}{\delta k_\mu^{la}(t , x ) } + \frac{\delta\mathit{\gamma}}{\delta\bar{d}^a(t , x ) } \frac{\delta\mathit{\gamma}}{\delta k^{\bar{d}a}(t , x ) } \right ] = 0 .
\label{eq:(4.7)}\end{aligned}\ ] ] next , we note that the equations of motion ( the schwinger dyson equations ) , @xmath152 \right\rangle -j^{ba}(x)&=0 , \label{eq:(4.8 ) } \\
\left\langle -\frac{1}{g_0 ^ 2}\partial_\mu d_\mu c^a(x)\right\rangle + j^{\bar{c}a}(x)&=0 , \label{eq:(4.9)}\end{aligned}\ ] ] are expressed in terms of the effective action as @xmath153 , \label{eq:(4.10 ) } \\ & \frac{\delta\mathit{\gamma}}{\delta\bar{c}^a(x ) } -\frac{1}{g_0 ^ 2}\partial_\mu\frac{\delta\mathit{\gamma}}{\delta k_\mu^{aa}(x ) } = 0 . \label{eq:(4.11)}\end{aligned}\ ] ] these relations show that , by defining @xmath154 and @xmath155 \notag\\ & \equiv\mathit{\gamma } + \frac{2}{g_0 ^ 2}\int d^dx\ , \operatorname{tr}\left\ { b(x)\left[\partial_\mu a_\mu(x)-\frac{1}{2\lambda_0}b(x ) \right]\right\ } , \label{eq:(4.13)}\end{aligned}\ ] ] the dependence on @xmath156 and @xmath157 are completely eliminated ; @xmath158 does not depend on neither @xmath156 nor @xmath157 .
the reduced effective action @xmath158 then obeys @xmath159 \notag\\ & \qquad { } + \fint_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\tilde{\mathit{\gamma}}}{\delta b_\mu^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k_\mu^{ba}(t , x ) } + \frac{\delta\tilde{\mathit{\gamma}}}{\delta d^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k^{da}(t , x ) } \right ] \notag\\ & \qquad\qquad { } + \int_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\tilde{\mathit{\gamma}}}{\delta l_\mu^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k_\mu^{la}(t , x ) } + \frac{\delta\tilde{\mathit{\gamma}}}{\delta\bar{d}^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k^{\bar{d}a}(t , x ) } \right ] = 0 .
\label{eq:(4.14)}\end{aligned}\ ] ] this is the wt relation following from the brs invariance of the @xmath0-dimensional field theory . now
, our statement of renormalizability is that the generating functional of the 1pi correlation functions @xmath146 can be made finite in terms of renormalized quantities , by choosing three renormalization constants , @xmath160 , and @xmath161 ; see sect .
[ sec:4.5 ] . ]
@xmath162 , @xmath163 , and @xmath164 in @xmath165 and @xmath166 appropriately order by order in the loop expansion of perturbation theory .
note that the above statement in particular claims that the flowed gauge field @xmath38 _ does not require any wave function renormalization_. is not renormalized . ] from eq . , it suffices to show the finiteness of the reduced effective action @xmath158 , because @xmath167 = -\frac{2}{\mu^{2\varepsilon}g^2 } \operatorname{tr}\left\ { b_r\left[\partial_\mu ( a_r)_\mu-\frac{1}{2\lambda}b_r \right]\right\ } , \label{eq:(4.20)}\ ] ] is finite with the above definitions . for @xmath158 , defining @xmath168 the wt relation in terms of the renormalized quantities yields , @xmath169 \notag\\ & \qquad { } + \fint_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\tilde{\mathit{\gamma}}}{\delta b_\mu^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k_\mu^{ba}(t , x ) } + \frac{\delta\tilde{\mathit{\gamma}}}{\delta d^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k^{da}(t , x ) } \right ] \notag\\ & \qquad\qquad { } + \int_0^\infty dt\,\int d^dx\ , \left [ \frac{\delta\tilde{\mathit{\gamma}}}{\delta l_\mu^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k_\mu^{la}(t , x ) } + \frac{\delta\tilde{\mathit{\gamma}}}{\delta\bar{d}^a(t , x ) } \frac{\delta\tilde{\mathit{\gamma}}}{\delta k^{\bar{d}a}(t , x ) } \right ] = 0 . \label{eq:(4.22)}\end{aligned}\ ] ] let us now consider the loop expansion of @xmath158 in the renormalized perturbation theory , @xmath170 where @xmath171 denotes the order of the loop expansion ; the tree - level effective action @xmath172 is given by the tree - level action in renormalized quantities : @xmath173 corresponding to the tree - level action , the counterterm in the renormalized perturbation theory is given by @xmath174 note that this counterterm contains in particular the `` boundary counterterm '' @xmath175 @xcite , @xmath176 \notag\\ & \qquad { } + 2\int d^dx\,\operatorname{tr}\left[\bar{d}(0,x)(\tilde{z}_3z^{1/2}z_3^{1/2}-1 ) c_r(x)\right ] , \label{eq:(4.26)}\end{aligned}\ ] ] because in eqs . and , the @xmath47-derivatives at @xmath75 are defined by eqs . and and @xmath21 and @xmath177 are renormalized as @xmath178 and @xmath179 , while @xmath180 and @xmath181 are not renormalized .
we also note that the tree - level action yields the tree - level propagators @xmath182 and ( in the @xmath91 limit ) @xmath183 . \label{eq:(4.28)}\ ] ] in the renormalized perturbation theory , these propagators are used . our proof for the renormalizability proceeds by induction
: suppose that , to the @xmath171-th order of the loop expansion , @xmath162 , @xmath163 , and @xmath164 can be chosen so that @xmath184 is finite
. then , considering the divergent part of the wt relation in the @xmath185-th loop order , we have @xmath186 where @xmath187 denotes the divergent part of @xmath188 and @xmath189 \notag\\ & \qquad { } -\epsilon\int d^dx\ , \biggl [ \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta l_\mu^a(0,x ) } \frac{\delta}{\delta k_\mu^{la}(0,x ) } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta k_\mu^{la}(0,x ) } \frac{\delta}{\delta l_\mu^a(0,x ) } \notag\\ & \qquad\qquad\qquad\qquad { } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta\bar{d}^a(0,x ) } \frac{\delta}{\delta k^{\bar{d}a}(0,x ) } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta k^{\bar{d}a}(0,x ) } \frac{\delta}{\delta\bar{d}^a(0,x ) } \biggr ] \notag\\ & \qquad\qquad\qquad\qquad\qquad { } + ( \text{derivatives w.r.t.\ field variables at $ t\neq0 $ } ) .
\label{eq:(4.30)}\end{aligned}\ ] ] in this expression , we have explicitly used the discrete time prescription .-dimensional functional derivative , such as @xmath190 , should be read as @xmath191 , when acting on a @xmath5-dimensional integral such as @xmath192 .
we also discard terms being higher order in @xmath86 in following calculations . ]
one can confirm that because of the brs invariance of the action @xmath193 , @xmath194 is nilpotent : @xmath195 eqs . and are basic relations for the proof .
next , we have to clarify the most general form of the divergent part @xmath187 .
first of all , we note that there is no `` bulk '' divergence that is given by a @xmath0-dimensional integral @xmath196 of a local polynomial of fields .
this follows from the fact that a feynman loop that consists only of the flow lines ( the heat kernels ) identically vanishes as we have observed .
thus the integrand of the loop integral for a loop diagram whose vertices reside in the `` bulk '' of the @xmath0-dimensional spacetime , @xmath30 , contains at least one propagator which provides the gaussian damping factor @xmath197 .
then the loop - momentum integral is absolutely convergent ; there is no `` bulk '' divergence . thus @xmath187 consists of the @xmath5-dimensional integral @xmath198 of a local polynomial of fields , the `` boundary '' divergence .
the mass dimension of the local polynomial is less or equal to @xmath93 for @xmath199 and the ghost number is @xmath200 .
let us next show that @xmath201,\qquad -2\int d^dx\,\operatorname{tr}\left[\partial_td(0,x)\bar{c}_r(x)\right ] , \label{eq:(4.32)}\ ] ] can not appear in @xmath187 , although the mass dimension and the ghost number perfectly match .
note that two combinations in eq .
contain , according to our prescription , @xmath180 and @xmath181 , respectively . that the terms in eq
. can not appear in the divergent part follows again from the fact that a feynman loop that consists only of the flow lines , the @xmath111 and @xmath122 propagators , identically vanishes .
because of this fact and from the structure of the interaction vertices containing @xmath97 or @xmath124 in @xmath202 , we see that any divergent term that contains @xmath97 or @xmath124 must involve @xmath203 , @xmath125 , or the external source @xmath147 .
the combinations in eq .
do not satisfy this criterion .
next , consider terms being proportional to the flow - time derivative at @xmath75 , @xmath204 or @xmath205 in the wt relation @xmath206 . in @xmath207 in eq .
, only the coefficient functions , @xmath208 contain the flow - time derivative at @xmath75 .
since @xmath207 has the structure @xmath209 taking into account the fact that @xmath187 contains neither @xmath204 nor @xmath205 as we have observed above , we infer that @xmath187 can not contain external fields , @xmath210 and @xmath211
. finally , the mass dimension for @xmath199 and the ghost number of @xmath212 ( @xmath213 ) are @xmath93 and @xmath214 ( @xmath93 and @xmath215 ) , respectively .
for this reason , these sources can not appear in @xmath187 . from these considerations
, we find that the most general form of the divergent part is given by @xmath216 where @xmath217 is a local polynomial of fields ; its mass dimension is at most @xmath93 for @xmath199 and the ghost number is @xmath200 .
our strategy to analyze the wt relation is to decompose it into a @xmath5-dimensional part that is well - understood in the renormalization of the yang
mills theory and an exotic part that originates from the @xmath0-dimensional fields .
thus , in eq . , we set @xmath218 , where @xmath219 note that @xmath220 contains at least one @xmath203 or one @xmath125
. we also decompose @xmath194 into @xmath221 , where @xmath222 is the `` brs operator '' in the original yang
mills theory @xmath223 \label{eq:(4.38)}\end{aligned}\ ] ] and @xmath224 is the remaining @xmath225 under the above decompositions , we see that the wt relation is decomposed as @xmath226 to conclude this , we have to note that @xmath220 is independent of @xmath227 and @xmath228 and thus @xmath229 and @xmath230 in @xmath224 , which might produce terms independent of @xmath203 and @xmath125 , do not contribute to the wt relation
. now , the general solution to eq . is well - known ; it is given by the sum of a gauge invariant part and the `` brs trivial part '' @xcite : @xmath231_{\mu\nu}(x)[f_r]_{\mu\nu}(x)\right ] \notag\\ & \qquad { } -2\tilde{\mathit{\gamma}}_d^{(0)}\ast \int d^dx\,\operatorname{tr}\left[y_1(\tilde{k}_r^a)_\mu(x)(a_r)_\mu(x ) + y_2k_r^c(x)c_r(x)\right ] , \label{eq:(4.42)}\end{aligned}\ ] ] where @xmath232 , @xmath233 , and @xmath234 are ( divergent ) constants , and @xmath235_{\mu\nu}(x)\equiv \partial_\mu(a_r)_\nu(x)-\partial_\nu(a_r)_\mu(x ) + [ ( a_r)_\mu,(a_r)_\nu](x ) .
\label{eq:(4.43)}\ ] ] substituting this in eq . and using eq . , we have @xmath236_{\mu\nu}(x)[f_r]_{\mu\nu}(x)\right ] \notag\\ & \qquad { } -2\tilde{\mathit{\gamma}}^{(0)}\ast \int d^dx\,\operatorname{tr}\left[y_1(\tilde{k}_r^a)_\mu(x)(a_r)_\mu(x ) + y_2k_r^c(x)c_r(x)\right ] \notag\\ & \qquad\qquad { } + \tilde{\tilde{\mathit{\gamma}}}_{d+1}^{(\ell+1)\text{div } } , \label{eq:(4.44)}\end{aligned}\ ] ] where @xmath237 .
\label{eq:(4.45)}\ ] ] this is the general form of @xmath187 , that is consistent with the wt relation .
@xmath238 itself is a boundary integral of a local polynomial of fields , that must contain at least one @xmath203 or one @xmath125 .
the general form of such term is @xmath239 . \label{eq:(4.46)}\ ] ] also , because of the nilpotency , the wt relation for eq . requires ( the gauge invariant part in the first line of eq .
is also invariant under @xmath194 ) @xmath240 -z_2\bar{d}(t , x)c_r(x)^2\bigl]=0 .
\label{eq:(4.47)}\end{aligned}\ ] ] this shows that @xmath241 and , by eq .
, @xmath242 .
thus , going back to eq . , we conclude that the most general form of @xmath187 is @xmath236_{\mu\nu}(x)[f_r]_{\mu\nu}(x)\right ] \notag\\ & \qquad { } -2\tilde{\mathit{\gamma}}^{(0)}\ast \int d^dx\,\operatorname{tr}\left[y_1(\tilde{k}_r^a)_\mu(x)(a_r)_\mu(x ) + y_2k_r^c(x)c_r(x)\right ] \notag\\ & = -\frac{1}{2\mu^{2\varepsilon}g^2 } \int d^dx\,\operatorname{tr}\bigl\ { ( x_1 - 2y_1 ) \left[\partial_\mu(a_r)_\nu-\partial_\nu(a_r)_\mu\right]^2 \notag\\ & \qquad\qquad\qquad\qquad\qquad { } + 2(x_1 - 3y_1 ) \left[\partial_\mu(a_r)_\nu(x)-\partial_\nu(a_r)_\mu(x)\right ] \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad { } \times[(a_r)_\mu(x),(a_r)_\nu(x ) ] \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad { } + ( x_1 - 4y_1 ) [ ( a_r)_\mu(x),(a_r)_\nu(x)]^2 \bigr\ } \notag\\ & \qquad { } -2\int d^dx\,\operatorname{tr}\bigl [ -(y_1+y_2)(\tilde{k}_r^a)_\mu(x)\partial_\mu c_r(x ) -y_2(\tilde{k}_r^a)_\mu(x)[(a_r)_\mu(x),c_r(x ) ] \notag\\ & \qquad\qquad\qquad\qquad\qquad { } + y_2k_r^c(x)c_r(x)^2\bigr ] \notag\\ & \qquad\qquad { } -2\int d^dx\,\operatorname{tr}\left [ y_1l_\mu(0,x)(a_r)_\mu(x ) -y_2\bar{d}(0,x)c_r(x)\right ] , \label{eq:(4.48)}\end{aligned}\ ] ] with three constants @xmath232 , @xmath233 , and @xmath234 .
then , it is straightforward to see that if one chooses the @xmath185-th loop order coefficients in @xmath243 , @xmath244 , and @xmath245 , as . ] @xmath246 the counterterm in eq .
precisely cancels the @xmath185-th order divergence .
in particular , the boundary counterterm cancels the last line of eq . .
this completes the mathematical induction : we have shown that an appropriate order by order choice of renormalization constants in eqs .
makes the generating functional of the 1pi diagrams thus all correlation functions of renormalized fields finite . in particular , the flowed gauge field does not require wave function renormalization . one can include the fermion fields into the above argument .
corresponding to the brs transformations in eqs . , , and , the action @xmath142 has additional terms , @xmath247 \notag\\ & \qquad\qquad { } + \fint_0^\infty dt\,\int d^dx\left [ \bar{k}^\chi(t , x)d(t , x)\chi(t , x)+\bar{\chi}(t , x)d(t , x)k^\chi(t , x ) \right ] \notag\\ & \qquad\qquad\qquad { } + \int_0^\infty dt\,\int d^dx\left [ \bar{k}^\lambda(t , x)d(t , x)\lambda(t , x)+\bar{\lambda}(t , x)d(t , x)k^\lambda(t , x ) \right ] .
\label{eq:(4.50)}\end{aligned}\ ] ] the renormalization constants are introduced as @xmath248 and @xmath249 and @xmath250 the derivative operator @xmath194 has additional terms , @xmath251 \notag\\ & \qquad\qquad { } -\epsilon\int d^dx\ , \biggl [ \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta\lambda_r(0,x ) } \frac{\delta}{\delta\bar{k}_r^\lambda(0,x ) } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta\bar{k}_r^\lambda(0,x ) } \frac{\delta}{\delta\lambda_r(0,x ) } \notag\\ & \qquad\qquad\qquad\qquad\qquad { } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta k_r^\lambda(0,x ) } \frac{\delta}{\delta\bar{\lambda}_r(0,x ) } + \frac{\delta\tilde{\mathit{\gamma}}^{(0)}}{\delta\bar{\lambda}_r(0,x ) } \frac{\delta}{\delta k_r^\lambda(0,x ) } \biggr ] \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad { } + ( \text{derivatives w.r.t.\ field variables at $ t\neq0 $ } ) .
\label{eq:(4.58)}\end{aligned}\ ] ] following the same line of argument as the pure yang
mills case , the representation of the possible divergent term , eq . , is modified to @xmath252\psi_r(x ) + w_2m_r\bar{\psi}_r(x)\psi_r(x)\right\ } \notag\\ & \qquad\qquad { } + \tilde{\mathit{\gamma}}^{(0)}\ast \int d^dx\,w_3 \left[\bar{k}_r^\psi(x)\psi_r(x)+\bar{\psi}_r(x)k_r^\psi(x)\right ] \notag\\ & \qquad\qquad\qquad { } + \int d^dx\,\xi_1 \left[\bar{\lambda}_r(0,x)\psi_r(x)+\bar{\psi}_r(x)\lambda_r(0,x)\right ] , \label{eq:(4.59)}\end{aligned}\ ] ] where @xmath253 , @xmath254 , @xmath255 , and @xmath256 are ( divergent ) constants .
this time , unlike the pure yang
mills case , the last term of eq .
is manifestly brs ( or gauge ) invariant and it is annihilated by @xmath194 .
thus we can not conclude @xmath257 from the wt relation @xmath206 . and @xmath258 .
] eq . then yields @xmath259 \notag\\ & \qquad\qquad\qquad\qquad\qquad\qquad\qquad { } + ( -w_3+\xi_1)\left [ \bar{\lambda}_r(0,x)\psi_r(x)+\bar{\psi}_r(x)\lambda_r(0,x ) \right ] \bigr\}. \label{eq:(4.60)}\end{aligned}\ ] ]
these divergences can be canceled by setting the @xmath185-th order renormalization constants as @xmath260 in particular , the last line of eq .
is canceled by the boundary counterterm arising from the flow time derivative in eq . .
note that there is no particular reason for @xmath261 ( i.e. , @xmath262 ) to hold and consequently the lagrange multiplier fields for the flowed fermion fields , @xmath68 and @xmath67 , _ must be renormalized_. this is a crucial difference from the gauge field for which @xmath77 is not renormalized .
finally , the renormalization of the flowed fermion fields @xmath63 and @xmath64 in eq .
being reciprocal to that of @xmath67 and @xmath68 is required , _ not _ to generate a `` bulk counterterm '' from @xmath114 .
as we emphasized , there is no divergence written as a @xmath0-dimensional integral of a local polynomial of fields .
thus we should not have any @xmath0-dimensional counterterm .
this requirement fixes the renormalization factors in eq . .
thus , unlike the flowed gauge field , the flowed fermion field @xmath63 and @xmath64 requires wave function renormalization . as eq
. indicates , the required renormalization constant @xmath161 is generally different from that for the original @xmath5-dimensional fermion field , @xmath263 ; an explicit perturbative calculation confirms that this is actually the case . from this observation , it is expected that this wave function renormalization of a flowed field persists also for generic matter fields .
finally , we mention on the finiteness of local products of flowed fields ( i.e. , composite operators ) . in the above
, we have shown that a correlation function of flowed fields @xmath264 is finite under the conventional renormalization in the original @xmath93-dimensional gauge theory and an appropriate choice of the wave function renormalization factor @xmath161 .
we can further see that this finiteness persists , even if some of spacetime points ( say , @xmath232 and @xmath265 in eq . )
coincide , as far as the corresponding flow times ( @xmath266 and @xmath267 in this example ) are strictly positive .
this implies that any local products of flowed elementary fields ( with the factor @xmath268 for flowed fermion fields ) is finite .
because of this remarkable property , the flow provides a versatile method to define renormalized composite operators .
this finiteness follows again from the fact that a feynman loop that consists only of the flow lines ( the heat kernels ) identically vanishes .
let us take the above situation in which two spacetime points @xmath232 and @xmath265 coincide . in the loop that contains the points @xmath269 and @xmath270 ,
there must exist at least one propagator .
the propagator carries the gaussian damping factor @xmath271 , where @xmath272 denotes the loop momentum and @xmath273 and @xmath274 are flow times at end points of the propagator .
in this situation , if @xmath275 or @xmath276 , then the loop integral for the loop absolutely converges because of the damping factor .
if @xmath277 , on the other hand , one of the propagators or the heat kernels contained in the loop gives rise to the factor that behaves like @xmath278 or @xmath279 . thus the loop integral again converges .
this shows that the coincidence @xmath280 does not produces new divergence as far as the corresponding flow times are strictly positive .
in the present paper , we have given another proof of the renormalizability of the system defined by the gradient flow and the fermion flow in vector - like gauge theories , the theorem first proven in ref .
@xcite : any correlation functions and composite operators of flowed fields are made finite by the conventional renormalization of gauge theory and wave function renormalization of the flowed fermion fields .
we believe that our proof is quite accessible if not elegant .
we hope that our proof will be helpful to understand the essential feature of the flow that underlies its various applications in ( lattice ) gauge theory .
the prototype of the present proof was presented by h. s. in a series of lectures at the yukawa institute for theoretical physics , kyoto university .
we would like to thank the participants , especially , sinya aoki , takahiro doi , kengo kikuchi , hiroshi kunitomo , and shigeki sugimoto , for valuable questions and comments which made our proof precise and concise .
the work of h. s. is supported in part by grant - in - aid for scientific research no .
in this appendix , we show that the brs transformation of the ghost lagrange multiplier @xmath125 , eq . , and the brs invariance of the @xmath0-dimensional action @xmath281 , eq . , can naturally be understood if one enlarges the field space so that the gauge covariance in a @xmath0-dimensional sense manifest .
the idea of this enlarged field space can be found in ref .
@xcite in the context of the stochastic quantization and also in ref .
@xcite in the context of the gradient flow . for this
, we introduce the gauge potential in the flow - time direction , @xmath282 , and set @xmath283 , \label{eq:(a1 ) } \\
d_t&\equiv\partial_t+b_t , \label{eq:(a2 ) } \\
\overleftarrow{d}_t&\equiv\overleftarrow{\partial}_t - b_t , \label{eq:(a3)}\end{aligned}\ ] ] and consider a @xmath0-dimensional action @xmath284
\right\ } \notag\\ & \qquad { } + \int_0^\infty dt\,\int d^dx\ , \left [ \bar{\lambda}(t , x)\left(d_t-\delta\right)\chi(t , x ) + \bar{\chi}(t , x ) \left(\overleftarrow{d}_t-\overleftarrow{\delta}\right)\lambda(t , x ) \right ] .
\label{eq:(a4)}\end{aligned}\ ] ] we regard the new gauge potential at @xmath75 , @xmath285 , as an independent integration variable . the above @xmath0-dimensional action @xmath286 is manifestly invariant under the gauge transformation in the @xmath0-dimensional sense
. correspondingly , @xmath286 is invariant under the following @xmath0-dimensional brs transformation : @xmath287 , & & \label{eq:(a7 ) } \\
\hat{\delta}\chi(t , x)&=-d(t , x)\chi(t , x ) , & \hat{\delta}\bar{\chi}(t , x)&=-\bar{\chi}(t , x)d(t , x ) , \label{eq:(a8 ) } \\
\hat{\delta}\lambda(t , x)&=-d(t , x)\lambda(t , x ) , & \hat{\delta}\bar{\lambda}(t , x)&=-\bar{\lambda}(t , x)d(t , x ) , \label{eq:(a9 ) } \\
\hat{\delta}d(t , x)&=-d(t , x)^2 , \label{eq:(a10)}\end{aligned}\ ] ] where @xmath72 is a @xmath0-dimensional ghost field . as usual , the above transformation is nilpotent by construction , @xmath288 .
next , we introduce a @xmath0-dimensional `` gauge fixing term '' @xmath289 and a corresponding `` ghost - anti - ghost term '' @xmath290 by @xmath291\right\}. \label{eq:(a11)}\ ] ] in this expression , @xmath79 is a @xmath0-dimensional analogue of the anti - ghost and its brs transformation is defined by @xmath292 where @xmath293 is a @xmath0-dimensional analogue of the nakanishi lautrup field .
clearly , the nilpotency @xmath288 is preserved with this definition .
thus , by construction , the above action whose explicit form is given by @xmath294\right\ } \notag\\ & \qquad{}-2\int_0^\infty dt\int d^dx\ , \operatorname{tr}\left\ { \bar{d}(t , x)\left[d_t d(t , x)-\alpha_0\partial_\mu d_\mu d(t , x)\right ] \right\ } , \label{eq:(a13)}\end{aligned}\ ] ] is manifestly brs invariant : @xmath295 so far , the nilpotency and the brs invariance of the action are manifest .
we will now consider a reduced field space in which @xmath296 and @xmath297 are eliminated by using the equations of motion ( we denote them eom in what follows ) : @xmath298 let us denote the brs transformation after the elimination of @xmath296 and @xmath297 as @xmath15 .
this is given by eqs .
in the text and @xmath299 interestingly , even after this elimination of @xmath296 and @xmath297 , the brs transformation remains nilpotent , @xmath78 .
the only non - trivial relation for this is @xmath300 \notag\\ & = 0 .
\label{eq:(a18)}\end{aligned}\ ] ] this relation may directly be confirmed ( cf .
ref @xcite ) .
instead , one can see this by noting the commutation relation , @xmath301 = -f^{abc}d^b(t , x)\frac{\delta}{\delta b_t^c(t , x ) } , \label{eq:(a19)}\ ] ] which immediately follows from the definition of @xmath302 . by applying this to @xmath303 and using @xmath304 ,
we have @xmath305 since the right - hand side vanishes under the eom , @xmath306 \right\}\right|_{\text{eom}}=0 , \label{eq:(a21)}\ ] ] where we have used @xmath307 .
finally , since @xmath308 , we have @xmath309=0 . \label{eq:(a22)}\ ] ] this is the relation .
if we integrate over @xmath310 , @xmath282 is eliminated from the system and we have the action in the reduced field space @xmath281 in eqs . and .
we can see that the reduced action @xmath281 is invariant under the reduced brs transformation @xmath15 .
the argument proceeds as follows : @xmath311
\notag\\ & = \biggl\{\biggl [ \int_0^\infty dt\int d^dx\ , \alpha_0\partial_\mu
d_\mu d^a(t , x)\,\frac{\delta}{\delta b_t^a(t , x ) } \notag\\ & \qquad\qquad { } + \hat{\delta } -\int_0^\infty dt\int d^dx\ , d_td^a(t , x)\,\frac{\delta}{\delta b_t^a(t , x ) } \biggr]\left(\hat{s}_{\text{fl}}+\hat{s}_{\text{gf}}+\hat{s}_{d\bar{d}}\right ) \biggr\}\biggr|_{\text{eom } } \notag\\ & = \left.\left [ \hat{\delta } \left(\hat{s}_{\text{fl}}+\hat{s}_{\text{gf}}+\hat{s}_{d\bar{d}}\right ) \right]\right|_{\text{eom } } \notag\\ & = 0 .
\label{eq:(a23)}\end{aligned}\ ] ] in the second equality , we have noted @xmath312\right|_{\text{eom } } -\left.\left[\hat{\delta}b_t(t , x)\right]\right|_{\text{eom } } & = \alpha_0\partial_\mu d_\mu d(t , x)-\left.d_td(t , x)\right|_{\text{eom } } , \label{eq:(a24 ) } \\
\delta\left.(\text{other fields})\right|_{\text{eom } } -\left.\left[\hat{\delta}(\text{other fields})\right]\right|_{\text{eom } } & = 0 , \label{eq:(a25)}\end{aligned}\ ] ] and , in the third equality , we have used the fact that @xmath313 under the eom .
the bottom line of the above argument is @xmath314 without using any equation of motion in the reduced field space ; this is the property that we employed in the main text .
throughout this appendix , the symbols @xmath31 and @xmath315 always stand for the difference operator and the sum , respectively , and do not refer to their continuum counterparts .
when the flow - time derivative is replaced by the difference operator , the leibniz rule does not hold and the brs transformations on quantities containing @xmath31 in eqs .
are modified by @xmath145 terms as even if the flow time is discretized . as a consequence , the brs invariance of the source term @xmath142 is preserved . ]
@xmath85+d_\mu e(t , x ) + \epsilon\left[\partial_t b_\mu(t , x),\partial_t d(t , x)\right ] , \label{eq:(b1 ) } \\
\delta e(t , x)&=-\left\{e(t , x),d(t , x)\right\ } -\epsilon\partial_t d(t , x)\partial_t d(t , x ) , \label{eq:(b2 ) } \\
\delta f(t , x)&=-d(t , x)f(t , x)-e(t , x)\chi(t , x ) -\epsilon\partial_t d(t , x)\partial_t\chi(t , x ) , \label{eq:(b3 ) } \\
\delta\bar{f}(t , x)&=-\bar{f}(t , x)d(t , x)-\bar{\chi}(t , x)e(t , x ) -\epsilon\partial_t\bar{\chi}(t , x)\partial_t d(t , x ) .
\label{eq:(b4)}\end{aligned}\ ] ] as a consequence , the brs invariance of the action , eq . , and
the resultant wt relation are broken by @xmath145 terms . to be definite , we have @xmath316 \right ] \notag\\ & \qquad { } + \int_0^\infty dt\int d^dx\,\left [ \bar{\lambda}(t , x)\epsilon\partial_t d(t , x)\partial_t\chi(t , x ) -\epsilon\partial_t\bar{\chi}(t , x)\partial_t d(t , x)\lambda(t , x)\right ] \notag\\ & \qquad\qquad { } -2\int_0^\infty dt\int d^dx\,\operatorname{tr}\left[\bar{d}(t , x ) \epsilon\partial_t d(t , x)\partial_t d(t , x)\right ] .
\label{eq:(b5)}\end{aligned}\ ] ] the wt relation is then modified by the expectation value of this combination . in what follows ,
we show that the above @xmath145 terms do not contribute in the @xmath91 limit and the wt relation assumes the present form ; this point requires careful examination because some @xmath317 quantities are hidden in our feynman rules .
non - linear sigma model @xcite , similar seemingly @xmath145 terms survive even in the @xmath91 limit and play an important role . ]
first of all , eq . is an @xmath145 quantity and thus its contribution basically vanishes as @xmath91 unless some element of @xmath317 emerges .
the unique place in our feynman rules where such an @xmath317 quantity can appear is the @xmath318 case of @xmath319 where the `` delta function '' has been introduced by @xmath320 note that @xmath321 and this factor can potentially cancel the factor @xmath86 .
let us consider the expectation value of the right - hand side of eq . under the functional integral .
for example , for the first line of eq . , we consider a flow line starting from the factor @xmath322 .
the two - point function is attached to this flow line .
if this flow line ends at a certain interaction vertex contained in the action @xmath323 , since those interaction vertices do not contain @xmath324 , the factor @xmath321 does not arise because of the flow - time integration at the vertex , @xmath325 ; the contribution remains @xmath145 .
the cancellation @xmath326 can occur only when the flow line goes back to eq . itself ,
i.e. , only when eq .
is self - contracted by eq . .
a similar argument applies to other lines of eq . .
however , the self - contractions are proportional to the factor @xmath327 or @xmath328 which identically vanishes . for a complex @xmath5 . ]
this shows that the wt relation restores the desired form safely in the @xmath91 limit .
m. lscher , commun .
phys . * 293 * , 899 ( 2010 ) doi:10.1007/s00220 - 009 - 0953 - 7 [ arxiv:0907.5491 [ hep - lat ] ] .
m. lscher , jhep * 1008 * , 071 ( 2010 ) erratum : [ jhep * 1403 * , 092 ( 2014 ) ] doi:10.1007/jhep08(2010)071 , 10.1007/jhep03(2014)092 [ arxiv:1006.4518 [ hep - lat ] ] .
r. narayanan and h. neuberger , jhep * 0603 * , 064 ( 2006 ) doi:10.1088/1126 - 6708/2006/03/064 [ hep - th/0601210 ] .
m. lscher and p. weisz , jhep * 1102 * , 051 ( 2011 ) doi:10.1007/jhep02(2011)051 [ arxiv:1101.0963 [ hep - th ] ] .
m. lscher , jhep * 1304 * , 123 ( 2013 ) doi:10.1007/jhep04(2013)123 [ arxiv:1302.5246 [ hep - lat ] ] .
m. lscher , pos lattice * 2013 * , 016 ( 2014 ) [ arxiv:1308.5598 [ hep - lat ] ] .
a. ramos , pos lattice * 2014 * , 017 ( 2015 ) [ arxiv:1506.00118 [ hep - lat ] ] .
m. c , c. consonni , g. p. engel and l. giusti , phys . rev .
d * 92 * , no .
7 , 074502 ( 2015 ) doi:10.1103/physrevd.92.074502 [ arxiv:1506.06052 [ hep - lat ] ] .
z. fodor , k. holland , j. kuti , s. mondal , d. nogradi and c. h. wong , jhep * 1509 * , 039 ( 2015 ) doi:10.1007/jhep09(2015)039 [ arxiv:1506.06599 [ hep - lat ] ] .
s. datta , s. gupta , a. lahiri , a. lytle and p. majumdar , phys .
d * 92 * , no .
9 , 094509 ( 2015 ) doi:10.1103/physrevd.92.094509 [ arxiv:1507.00821 [ hep - lat ] ] .
a. shindler , t. luu and j. de vries , phys .
d * 92 * , no .
9 , 094518 ( 2015 ) doi:10.1103/physrevd.92.094518 [ arxiv:1507.02343 [ hep - lat ] ]
. h. suzuki , ptep * 2015 * , no .
10 , 103b03 ( 2015 ) doi:10.1093/ptep / ptv139 [ arxiv:1507.02360 [ hep - lat ] ] .
r. bellwied , s. borsanyi , z. fodor , s. d. katz , a. psztor , c. ratti and k. k. szabo , phys .
d * 92 * , no . 11 , 114505 ( 2015 ) doi:10.1103/physrevd.92.114505 [ arxiv:1507.04627 [ hep - lat ] ] .
a. abdel - rehim _ et al .
_ [ etm collaboration ] , arxiv:1507.05068 [ hep - lat ] .
a. hasenfratz , y. liu and c. y. h. huang , arxiv:1507.08260 [ hep - lat ] .
a. francis , o. kaczmarek , m. laine , t. neuhaus and h. ohno , phys .
d * 92 * , no .
11 , 116003 ( 2015 ) doi:10.1103/physrevd.92.116003 [ arxiv:1508.04543 [ hep - lat ] ] .
a. kagimura , a. tomiya and r. yamamura , arxiv:1508.04986 [ hep - lat ] .
a. ramos and s. sint , eur .
j. c * 76 * , no . 1 , 15 ( 2016 )
doi:10.1140/epjc / s10052 - 015 - 3831 - 9 [ arxiv:1508.05552 [ hep - lat ] ] . v. g. bornyakov _ et al .
_ , arxiv:1508.05916 [ hep - lat ]
. s. borsanyi _ et al .
_ , phys .
b * 752 * , 175 ( 2016 ) doi:10.1016/j.physletb.2015.11.020 [ arxiv:1508.06917 [ hep - lat ] ] .
h. fukaya _ et al . _
[ jlqcd collaboration ] , phys .
d * 92 * , no .
11 , 111501 ( 2015 ) doi:10.1103/physrevd.92.111501 [ arxiv:1509.00944 [ hep - lat ] ] . c. alexandrou , a. athenodorou and k. jansen , phys .
d * 92 * , no .
12 , 125014 ( 2015 ) doi:10.1103/physrevd.92.125014 [ arxiv:1509.04259 [ hep - lat ] ] .
p. petreczky and h .-
schadler , phys .
d * 92 * , no .
9 , 094517 ( 2015 ) doi:10.1103/physrevd.92.094517 [ arxiv:1509.07874 [ hep - lat ] ] .
a. chowdhury , a. harindranath and j. maiti , jhep * 1602 * , 134 ( 2016 ) doi:10.1007/jhep02(2016)134 [ arxiv:1509.07959 [ hep - lat ] ] . m. g. endres , r. c. brower , w. detmold , k. orginos and a. v. pochinsky , phys .
d * 92 * , no .
11 , 114516 ( 2015 ) doi:10.1103/physrevd.92.114516 [ arxiv:1510.04675 [ hep - lat ] ] .
t. degrand , rev .
* 88 * , 015001 ( 2016 ) doi:10.1103/revmodphys.88.015001 [ arxiv:1510.05018 [ hep - ph ] ] .
d. lin , k. ogawa and a. ramos , jhep * 1512 * , 103 ( 2015 ) doi:10.1007/jhep12(2015)103 [ arxiv:1510.05755 [ hep - lat ] ] . c. alexandrou , a. athenodorou , m. constantinou , k. hadjiyiannakou , k. jansen , g. koutsou , k. ottnad and m. petschlies , phys .
d * 93 * , 074503 ( 2016 ) doi:10.1103/physrevd.93.074503 [ arxiv:1510.05823 [ hep - lat ] ] .
r. yamamura , prog .
( 2016 ) 073b10 doi:10.1093/ptep / ptw097 [ arxiv:1510.08208 [ hep - lat ] ] .
p. a. boyle _ et al .
_ , phys .
d * 93 * , no . 5 , 054502 ( 2016 ) doi:10.1103/physrevd.93.054502 [ arxiv:1511.01950 [ hep - lat ] ] .
d. m. grabowska and d. b. kaplan , phys .
* 116 * , no .
21 , 211602 ( 2016 ) doi:10.1103/physrevlett.116.211602 [ arxiv:1511.03649 [ hep - lat ] ] .
r. c. brower , a. hasenfratz , c. rebbi , e. weinberg and o. witzel , phys .
d * 93 * , no . 7 , 075028 ( 2016 ) doi:10.1103/physrevd.93.075028 [ arxiv:1512.02576 [ hep - ph ] ] .
j. rantaharju , phys .
d * 93 * , no .
9 , 094516 ( 2016 ) doi:10.1103/physrevd.93.094516 [ arxiv:1512.02793 [ hep - lat ] ] . s. datta , s. gupta and a. lytle , phys .
d * 94 * , no .
9 , 094502 ( 2016 ) doi:10.1103/physrevd.94.094502 [ arxiv:1512.04892 [ hep - lat ] ] .
d. banerjee , m. bgli , k. holland , f. niedermayer , m. pepe , u. wenger and u. j. wiese , jhep * 1603 * , 116 ( 2016 ) doi:10.1007/jhep03(2016)116 [ arxiv:1512.04984 [ hep - lat ] ] .
s. mages , b. c. tth , s. borsnyi , z. fodor , s. katz and k. k. szab , arxiv:1512.06804 [ hep - lat ] . j. giedt , int . j. mod .
a * 31 * , no .
10 , 1630011 ( 2016 ) doi:10.1142/s0217751x16300118 [ arxiv:1512.09330 [ hep - lat ] ] .
k. fujikawa , jhep * 1603 * , 021 ( 2016 ) doi:10.1007/jhep03(2016)021 [ arxiv:1601.01578 [ hep - lat ] ] .
t. appelquist _ et al .
_ , phys .
d * 93 * , no .
11 , 114514 ( 2016 ) doi:10.1103/physrevd.93.114514 [ arxiv:1601.04027 [ hep - lat ] ] . m. g. vera and s. schaefer , phys .
d * 93 * , 074502 ( 2016 ) doi:10.1103/physrevd.93.074502 [ arxiv:1601.07155 [ hep - lat ] ] .
p. boyle , a. jttner , m. k. marinkovi , f. sanfilippo , m. spraggs and j. t. tsang , jhep * 1604 * , 037 ( 2016 ) doi:10.1007/jhep04(2016)037 [ arxiv:1602.04118 [ hep - lat ] ] .
r. arthur , v. drach , m. hansen , a. hietanen , c. pica and f. sannino , phys .
d * 94 * , no .
9 , 094507 ( 2016 ) doi:10.1103/physrevd.94.094507 [ arxiv:1602.06559 [ hep - lat ] ] .
m. wandelt , f. knechtli and m. gnther , jhep * 1610 * , 061 ( 2016 ) doi:10.1007/jhep10(2016)061 [ arxiv:1603.05532 [ hep - lat ] ] .
t. w. chiu , arxiv:1603.08854 [ hep - lat ] .
h. suzuki , ptep * 2013 * , 083b03 ( 2013 ) erratum : [ ptep * 2015 * , 079201 ( 2015 ) ] doi:10.1093/ptep / ptt059 , 10.1093/ptep / ptv094 [ arxiv:1304.0533 [ hep - lat ] ] . h. makino and h. suzuki , ptep * 2014 * , 063b02 ( 2014 ) erratum : [ ptep * 2015 * , 079202 ( 2015 ) ] doi:10.1093/ptep / ptu070 , 10.1093/ptep / ptv095 [ arxiv:1403.4772 [ hep - lat ] ] .
m. asakawa _ et al .
_ [ flowqcd collaboration ] , phys . rev .
d * 90 * , no . 1
, 011501 ( 2014 ) erratum : [ phys .
d * 92 * , no . 5 , 059902 ( 2015 ) ] doi:10.1103/physrevd.90.011501 , 10.1103/physrevd.92.059902 [ arxiv:1312.7492 [ hep - lat ] ] . m. kitazawa , m. asakawa , t. hatsuda , t. iritani , e. itou and h. suzuki , pos lattice * 2014 * , 022 ( 2014 ) [ arxiv:1412.4508 [ hep - lat ] ] .
e. itou , h. suzuki , y. taniguchi and t. umeda , pos lattice * 2015 * , 303 ( 2016 ) [ arxiv:1511.03009 [ hep - lat ] ] .
m. kitazawa , m. asakawa , t. hatsuda , t. iritani , e. itou and h. suzuki , pos lattice * 2015 * , 162 ( 2016 ) [ arxiv:1511.05235 [ hep - lat ] ]
. m. kitazawa , `` qcd thermodynamics from gradient flow , '' talk delivered at the 13th international workshop on qcd in extreme conditions , http://conf.ccnu.edu.cn/~xqcd2015/ y. taniguchi , `` qcd energy momentum tensor at finite temperature using gradient flow , '' talk delivered at the rbrc workshop on lattice gauge theories 2016 , https://indico.bnl.gov/conferencedisplay.py?confid=1628 t. endo , k. hieda , d. miura and h. suzuki , ptep * 2015 * , no .
5 , 053b03 ( 2015 ) doi:10.1093/ptep / ptv058 [ arxiv:1502.01809 [ hep - lat ] ] . l. del debbio , a. patella and a. rago , jhep * 1311 * , 212 ( 2013 ) doi:10.1007/jhep11(2013)212 [ arxiv:1306.1173 [ hep - th ] ] .
a. patella , l. del debbio and a. rago , pos lattice * 2013 * , 324 ( 2014 ) .
f. capponi , a. rago , l. del debbio , s. ehret and r. pellegrini , pos lattice * 2015 * , 306 ( 2016 ) [ arxiv:1512.02851 [ hep - lat ] ] .
f. capponi , l. del debbio , a. patella and a. rago , pos lattice * 2015 * , 302 ( 2016 ) [ arxiv:1512.04374 [ hep - lat ] ] .
m. pepe and l. giusti , pos lattice * 2014 * , 322 ( 2014 ) [ arxiv:1410.8371 [ hep - lat ] ] .
l. giusti and m. pepe , phys .
d * 91 * , 114504 ( 2015 ) doi:10.1103/physrevd.91.114504 [ arxiv:1503.07042 [ hep - lat ] ] .
m. pepe and l. giusti , pos eps * -hep2015 * , 376 ( 2015 ) [ arxiv:1510.04512 [ hep - ph ] ] . l. giusti and m. pepe , pos lattice * 2015 * , 211 ( 2016 ) [ arxiv:1511.03710 [ hep - lat ] ] . h. makino and h. suzuki , ptep * 2015 * , no .
3 , 033b08 ( 2015 ) doi:10.1093/ptep / ptv028 [ arxiv:1410.7538 [ hep - lat ] ] . h. makino , f. sugino and h. suzuki , ptep * 2015 * , no .
4 , 043b07 ( 2015 ) doi:10.1093/ptep / ptv044 [ arxiv:1412.8218 [ hep - lat ] ] . s. aoki , k. kikuchi and t. onogi , jhep * 1504 * , 156 ( 2015 ) doi:10.1007/jhep04(2015)156 [ arxiv:1412.8249 [ hep - th ] ] .
h. suzuki , ptep * 2015 * , no .
4 , 043b04 ( 2015 ) doi:10.1093/ptep / ptv036 [ arxiv:1501.04371 [ hep - lat ] ] . c. monahan and k. orginos , phys . rev .
d * 91 * , no . 7 , 074513 ( 2015 ) doi:10.1103/physrevd.91.074513 [ arxiv:1501.05348 [ hep - lat ] ] .
s. aoki , k. kikuchi and t. onogi , ptep * 2015 * , no .
10 , 101b01 ( 2015 ) doi:10.1093/ptep / ptv131 [ arxiv:1505.00131 [ hep - th ] ] .
n. kamata and s. sasaki , pos lattice * 2015 * , 301 ( 2016 ) [ arxiv:1511.06076 [ hep - lat ] ] . c. becchi , a. rouet and r. stora , annals phys .
* 98 * , 287 ( 1976 ) .
doi:10.1016/0003 - 4916(76)90156 - 1 j. zinn - justin , `` renormalization of gauge theories , '' saclay-d.ph-t-74-88 lectures given at int .
summer inst .
for theoretical physics , jul 29aug 9 , 1974 , bonn , west germany .
b. w. lee , `` gauge theories , '' conf .
c * 7507281 * , 79 ( 1975 ) .
j. c. taylor , _ gauge theories of weak interactions _ ( cambridge university press , cambridge , uk , 1976 ) .
j. zinn - justin , nucl .
b * 275 * , 135 ( 1986 ) .
doi:10.1016/0550 - 3213(86)90592 - 4 j. zinn - justin and d. zwanziger , nucl . phys .
b * 295 * , 297 ( 1988 ) .
doi:10.1016/0550 - 3213(88)90358 - 6 s. d. joglekar and b. w. lee , annals phys .
* 97 * , 160 ( 1976 ) .
doi:10.1016/0003 - 4916(76)90225 - 6 n. nakazawa , nucl . phys .
b * 335 * , 546 ( 1990 ) .
doi:10.1016/0550 - 3213(90)90517-h
|
we give an alternative perturbative proof of the renormalizability of the system defined by the gradient flow and the fermion flow in vector - like gauge theories .
|
there is an increasing interest in high - current applications of ion linacs , such as the transformation of radioactive waste , the production of tritium , and fusion drivers .
high currents of the order of 100 ma restrict beam losses below 1 ppm .
thorough studies are necessary to understand mechanisms of intense - beam losses , in particular , beam instabilities and halo formation .
most of the theoretical efforts so far have concentrated on the kapchinsky - vladimirsky ( kv ) distribution of particles in transverse phase space @xcite .
the kv beam density is uniform so that space - charge forces inside the beam are linear .
it allows an analytical investigation and results are used to predict the behavior of real beams . on the other hand , it is recognized that the kv model , in which all particles have the same transverse energy , is not a realistic beam distribution , e.g. @xcite .
the present paper compares the kv beam with other , nonlinear particle - density distributions , which can serve as better models for real beams .
we study a continuous axisymmetric ion beam in a uniform focusing channel , with longitudinal velocity @xmath0 .
the hamiltonian of the transverse motion ( @xmath1 ) is @xmath2 where @xmath3 and @xmath4 are ion mass and charge , @xmath5 is the focusing strength of the channel , @xmath6 , @xmath7 is the distance from the @xmath8-axis in the transverse plane , and @xmath9 ( @xmath10 , @xmath11 ) is the dimensionless transverse velocity .
the electric potential @xmath12 must satisfy the poisson equation @xmath13 where @xmath14 is the distribution function in the transverse non - relativistic 4-d phase space .
the integral on the rhs is the particle density @xmath15 .
since the hamiltonian ( [ h ] ) is an integral of motion , any distribution function of the form @xmath16 is a stationary distribution .
we consider a specific set of stationary distributions for which the beam has a sharp edge ( for all ions @xmath17 ) , namely , @xmath18 the normalization constants @xmath19 are chosen to satisfy @xmath20 , where @xmath15 is the particle density , and @xmath21 is the beam current .
the set includes the kv distribution , @xmath22 , as a formal limit of @xmath23 , as well as the waterbag ( wb ) distribution , @xmath24 , where @xmath25 is the step - function . for a detailed discussion of these two specific examples
see @xcite .
we introduce the function @xmath26 , because the density can be expressed from ( [ fn ] ) as @xmath27 .
physically , this function gives the maximal transverse velocity for a given radius , @xmath28 , and defines the boundary in the phase space @xmath29 .
it allows us to rewrite eq .
( [ pois ] ) as @xmath30'/r - \lambda^2 g^n(r ) = - 2 k_0 ^ 2 \ , \label{gsc}\ ] ] with boundary conditions @xmath31 , and @xmath32 is finite . here the parameter @xmath33 $ ] , where @xmath34 is the beam perveance , and @xmath35 is a constant .
particular solutions to ( [ gsc ] ) are easy to find for @xmath36 ( kv ) and @xmath37 ( wb ) @xcite . for @xmath38
a numerical solution is required . to compare different transverse distributions on a common basis , we consider rms - equivalent beams which have the same perveance @xmath39 , rms radius , and rms emittance @xmath40 . to characterize the space - charge strength
, one introduces an equivalent ( or rms ) tune depression @xmath41 which reduces to the usual one for the kv beam . for numerical simulations we use dimensionless variables : @xmath42 , and @xmath43 , where @xmath44 . in normalized variables
the beam matched radius is @xmath45 , where @xmath46 and @xmath47 . for the kv case , @xmath48 ,
so that @xmath49 .
the `` hats '' are omitted below to simplify notation .
we use particle - in - cell simulations to study beam stability and halo formation in the presence of instabilities .
a leap - frog integration is applied to trace the time evolution for a given initial phase - space distribution .
the space - charge radial electric field of an axisymmetric beam can be found from gauss law by counting the numbers of particles in cells of a finite radial grid , which extends up to four times the beam matched radius .
the initial phase - space state is populated randomly but in accordance with ( [ fn ] ) for a chosen @xmath50 .
the matched distributions remain stable except for a minor dilution related to numerical errors . however , even the matched kv beam is unstable for @xmath51 , in agreement with existing theory @xcite and earlier simulations @xcite .
the beam breathing oscillations are excited by loading a mismatched initial distribution @xmath52 , @xmath53 , where @xmath54 correspond to the matched one , and the mismatch parameter @xmath55 . a typical range of the simulation parameters : time step @xmath56 , where @xmath57 is the period of breathing oscillations , total number of particles @xmath58 , where @xmath59 , and radial mesh size @xmath60 .
the code performs simulations of about 100 breathing oscillations per cpu hour for @xmath61 on sun ultrasparc 1/170 .
the beam behavior is studied as a function of tune depression @xmath62 and mismatch @xmath63 . due to a discrete filling of a mismatched beam distribution in simulations and , for @xmath64 , due to non - linear space - charge forces ,
higher modes are excited in addition to the breathing mode .
some of them can be unstable depending on values of @xmath62 and @xmath63 . a detailed numerical study of stability and halo formation for the kv beam and its comparison with the theory predictions @xcite
have already been reported in @xcite .
here we compare results for different transverse distributions . in figs .
1 - 2 the maximal radius of the whole ensemble of particles is plotted versus the number of breathing oscillations for the kv and wb beams , for the particular case of @xmath65 and @xmath66 ( @xmath61 , @xmath67 ) .
comparison of figs .
1 and 2 shows that for these parameters the wb beam remains stable much longer than the kv one , but eventually it also blows up and some particles form a halo far from the beam core .
results for the @xmath68 distribution are similar to those for wb .
some results depend on simulation parameters ; e.g. , it takes a smaller number of the breathing periods for a beam to blow up if @xmath69 is smaller ( i.e. , higher noise )
. however , the maximum radius , as well as the fraction of particles outside the core , are practically independent of @xmath69 .
the number of particles which go into the halo and produce jumps of @xmath70 seen in figs . 1 - 2 , might be rather small .
we define the halo intensity @xmath71 as the number of particles outside the boundary @xmath72 divided by @xmath69 .
such a definition is arbitrary , but convenient to compare beam halos over a wide range of tune depressions . while the beam behavior in figs .
1 and 2 seems qualitatively similar , the halos for these two cases are very different : @xmath73 for kv , and about 100 times less for the wb , with only a few particles in the halo ( less than 10 of 256k ) . that is the reason for oscillations of @xmath70 in fig . 2
: these few halo particles can initially all come back to the core simultaneously . a qualitative picture of the beam behavior for various values of the tune depression and mismatch is shown in fig . 3 , and is practically the same for all distributions studied
h corresponds to beam instability with halo formation , usually with a noticeable emittance growth , u means that the beam is unstable but a halo is not observed in our simulations , and s indicates beam stability . the most surprising feature of the diagram is the lack of any significant dependence on @xmath62 for mismatched beams ; on the contrary , the qualitative changes depend primarily on @xmath63 .
when @xmath63 changes from 0.6 to 0.8 , the ratio @xmath74 decreases from 1.72 to 1.031.07 for the kv beam , and from 1.41.5 to 1.001.01 for the wb and @xmath68 . the number of breathing periods after which the beam radius starts to grow noticeably and the halo forms , has some dependence on @xmath62 ; it is smaller for small @xmath62 .
we performed a systematic study of the kv , wb , and @xmath68 distributions for tunes @xmath62 from 0.1 to 0.9 and mismatches @xmath63 from 0.6 to 1.0 .
figure 4 shows the ratios of the halo radius to that of the matched beam for the kv and wb beams with three different mismatches , @xmath75 , 0.7 , and 0.8 .
results for @xmath68 beam are not shown ; they are slightly lower than those for the wb beam .
the kv halo has a larger radius , especially with small space charge ( large @xmath62 ) , but for space - charge dominated beams , at very small @xmath62 , the ratios converge for all distributions .
the analytical model for the kv halo formation @xcite predicts finite values of @xmath76 between 2 and 2.5 depending on @xmath62 and @xmath63 .
one can see from simulations that it works well also for wb and @xmath68 beams .
simulation results for halo intensity @xmath71 are shown in fig . 5 for kv and wb distributions .
again , results for @xmath68 are just slightly lower than for the wb beam , and not shown .
the intensity depends essentially on the mismatch , and decreases quickly as the mismatch decreases .
the wb halo is about 23 times less intense than the kv halo for small space charge and large mismatch ( 0.6 and 0.7 ) but , for space - charge dominated beams , the intensities are about the same . for @xmath66 , however , the wb halo is at least an order of magnitude less intense than the kv one ; it is not even included in fig . 5 .
an apparent decrease of @xmath71 as @xmath62 decreases is due to the definition used : the halo boundary radius @xmath77 increases as @xmath78 .
if a fixed boundary is used instead , the same for all tunes , the halo intensity would be larger for larger space charge .
one more interesting feature is how fast the halo develops . for the kv beam , the process is usually rather fast , and the halo saturates after a few hundred breathing periods . for the wb and @xmath68 distributions ,
it continues to grow rather slowly , and asymptotic values are usually reached after a few thousand breathing oscillations ; it takes especially long for @xmath79 .
data plotted in fig .
5 correspond to the asymptotic values , after n=5000 breathing periods for wb and after n=600 for kv ( except @xmath80 , where kv results are also for n=5000 ) .
these 5000 breathing oscillations correspond to 510 km of the length for a typical machine , much longer than existing proton linacs .
our simulations show the qualitative similarity of the beam behavior for all transverse distributions studied .
the kv beam can be considered as an extreme case compared to the wb and @xmath68 distributions which are closer to real beams .
the halo intensity is a few times higher and saturates faster for the kv distribution than for the other two .
an interesting new observation is that for axisymmetric beams under consideration the beam stability and halo formation depend primarily on the mismatch , not on the tune shift .
the halo was clearly observed only for large mismatches , at least 20% , and its radius is in agreement with the analytical model @xcite for halo formation .
gluckstern , phys .
letters * 73 * , 1247 ( 1994 ) .
i.m . kapchinsky and v.v .
vladimirsky , in _ proceed .
conf . on high energy accelerators _
( cern , geneva , 1959 ) , p. 274 .
h. okamoto and m. ikegami , phys . rev .
e * 55 * , 4694 ( 1997 ) .
m. reiser , _ theory and design of charged particle beams _
( wiley , new york , 1993 ) .
gluckstern , in _ proceed . of the linac conference _
( fermilab , 1970 ) , p. 811 . i. hofmann , l.j .
laslett , l. smith and i. haber , particle accelerators * 13 * , 145 ( 1983 ) .
gluckstern , w .- h .
cheng and h. ye , phys .
letters * 75 * , 2835 ( 1995 ) .
gluckstern , w .- h .
cheng , s.s .
kurennoy and h. ye , phys . rev .
e * 54 * , 6788 ( 1996 ) .
|
beam stability and halo formation in high - intensity axisymmetric 2d beams in a uniform focusing channel are analyzed using particle - in - cell simulations .
the tune depression - mismatch space is explored for the uniform ( kv ) distribution of the particle transverse - phase - space density , as well as for more realistic ones ( in particular , the water - bag distribution ) , to determine the stability limits and halo parameters .
the numerical results show an agreement with predictions of the analytical model for halo formation @xcite .
|
reaction diffusion processes are of wide interest in physics , chemistry and biology @xcite . in physics
they present a relatively simpler case of non equilibrium stochastic processes with non trivial behaviour .
traditionally they have been studied via mean field type methods ( e.g. law of mass action , local chemical kinetics ) @xcite . however , in sufficiently low spatial dimension , particle density fluctuations become dominant and mean field methods become invalid @xcite .
the role of fluctuations in these processes has thus been studied for a while , but has received renewed attention recently @xcite , as new exact results in one dimension @xcite and systematic renormalisation group studies have appeared @xcite .
one interest of these models is their relation to phase ordering kinetics via the `` coarsening '' of domain structures evolving towards equilibrium @xcite . in some cases ,
these can be seen as reaction diffusion processes for defects , for instance domain walls in one dimension or xy type vortices in two dimension , which diffuse and can annihilate or coalesce upon meeting .
these coarsening processes have also been much studied recently , especially in an effort to characterize their so - called persistence ( or survival or first passage ) properties for single spins , domains or global magnetization @xcite although many results are now available for reaction diffusion processes in homogeneous situations , comparatively little is known on their dynamics in the presence of quenched disorder , which is expected to play a role in many physical realisations .
it can be introduced in the models in several ways , e.g. in the reaction rates or in the single particle diffusion .
one can expect that it will strongly modify the behaviour of the system in some cases by amplifying the role of spatial density fluctuations .
these effects are interesting , but difficult to study analytically because of the present lack of methods , beyond mean field approximations or perturbation theory , to treat the dynamics of such disordered systems . even in the absence of quenched disorder
, there is an apparently unlimited variety of behaviours in reaction diffusion systems .
the more complex ones , such as oscillatory or chaotic behaviours , become possible for a large enough number of species @xcite . in simpler cases
, attempts have been made to identify possible universality classes and a wide class of models with finite reaction rates , amenable to field theoretical treatments , has been studied @xcite .
for instance , branching and annihilating random walks ( barw ) , i.e reactions such as @xmath13 and @xmath14 or @xmath15 exhibit transitions from inactive ( no particle ) to active states , which were found to be either in the universality class of directed percolation @xcite ( odd number of offsprings ) or in the so called parity conserving class ( even number of offsprings ) @xcite .
this was confirmed by exact results in one dimension @xcite .
related types of models describe epidemy propagation , such as @xmath16 ( with rate @xmath17 ) and either @xmath18 ( recovery ) or @xmath19 ( immune ) ( rate @xmath20 ) were also studied via rg @xcite ( see @xcite for review ) .
the effect of quenched disorder has been studied in this class of barw models , via random rates @xmath21 and @xmath22 but with limited success as the rg flows to strong coupling @xcite . as for directed percolation with disorder
, it is still a largely open problem @xcite .
there is a simpler class of homogeneous models without branching ( i.e without particle production ) , such as @xmath14 , @xmath23 @xcite , @xmath24 @xcite , etc .. which has still non trivial behaviour @xcite .
one interesting phenomenon is that in low enough dimension , the process becomes diffusion limited rather than reaction limited .
indeed particles in close proximity react quickly and the remaining particles are typically separated by a length related to the pure diffusion length @xmath25 .
this leads to a decay of specie density , e.g. @xmath26 for @xmath27 in the case of @xmath28 , slower than the mean field prediction @xmath29 valid for @xmath30 ( for @xmath24 a related argument leads @xmath31 for @xmath32 ) @xcite .
this type of results for such models in the pure case are well established from heuristic arguments , numerical simulations , perturbative rg @xcite and in some cases rigorous methods @xcite .
it is now interesting to investigate how disorder will modify these behaviours . with disorder ,
models in this class are easier to study than the barw type models , although it is still a difficult task .
the reactions @xmath14 and @xmath33 have been studied using perturbative field theoretic rg methods for particles diffusing in random flows , either in two dimension @xcite or in a special hydrodymamic flow @xcite . as can be expected from the study of single particle diffusion in such flows @xcite
, the behaviour should be qualitatively different in the case of _ potential _ disorder , which tends to segregate the particles and slow the reaction ( and the diffusion ) than for _ hydrodynamic _ flows which tend to mix the particles and increase the effective reaction rate ( and leads to hyperdiffusion ) .
the competition which arises when both components are present has been studied very recently in @xmath34 in @xcite and in @xmath35 @xcite .
remarkably , the one dimensional problem seems quite far from the reach of such perturbative rg methods and no generic result is available at present @xcite in that case , hence the interest of the present study .
indeed , in @xmath36 only potential disorder can exist and is known to lead to ultraslow single particle diffusion described by a strong disorder ( i.e zero temperature ) fixed point @xcite . to make progress in @xmath36 requires developing non perturbative techniques , which is the aim of the present work .
in this paper we study a broad class of reaction diffusion models where particles diffuse on a one dimensional lattice and can react or annihilate upon meeting . apart from their reactions the particles
are non interacting .
more specifically , each site of the lattice can be in one of several possible `` states '' , labeled @xmath37 .
@xmath38 corresponds to the empty state with no particle present at that site . @xmath39
correspond to the presence of particles of different types .
when two particles ( i.e states ) @xmath40 and @xmath41 meet they react and give a another state @xmath17 with a probability @xmath42 .
@xmath17 may be the empty state @xmath38 , corresponding to an annihilation .
the reaction is thus a stochastic process @xmath43 characterized by a fixed transition probability matrix which satisfies @xmath44 the matrix @xmath42 can be extended to include @xmath45 by defining @xmath46 for any @xmath47 , which is the property expected for an empty state ( @xmath48 with probability 1 ) .
one prominent example will be identical particles @xmath4 which react upon meeting as @xmath49 in this case there are only two states : @xmath38 corresponds to no particule present ( @xmath50 ) and @xmath51 to one particule present @xmath4 .
the transition matrix is then @xmath52 , @xmath53 and @xmath54 .
we will obtain results for processes within the above class ( [ general ] ) and study some specific examples .
we will restrict to symmetric reaction rates @xmath55 .
asymmetric rates , depending on the side from which the two species come in contact , can be defined in @xmath36 and can be studied by the same methods .
we will mostly consider reaction diffusion processes with only a _
finite _ number of possible states ( or species ) .
processes with unbounded number of states ( @xmath56 ) can also be studied by the present method , and we will give some examples . classifying the much larger variety of complex behavior possible in that case is beyond the scope of the present study .
other extensions include randomness in the reaction rates , which we will briefly discuss in the end .
up to now we have not specified the way in which particle diffuse , nor the reaction rates .
let us first concentrate on the process ( [ potts ] ) and recall the known results in the case of pure diffusion ( i.e homogeneous hopping rates ) , which has been extensively studied .
it is of particular interest in one dimension since it is also a model for zero temperature domain growth in the ferromagnetic @xmath57-states potts model ( with glauber dynamics ) where @xmath58 @xcite .
the case @xmath59 ( ising ) corresponds to walkers ( i.e domain walls ) always annihilating when they meet and @xmath60 to walkers always coagulating @xcite .
it is known that the reaction rate can be chosen infinite ( immediate reaction upon meeting ) without changing the universality class , and the same will hold here in presence of disorder , hence our general choice of model ( [ general ] ) . for all @xmath61 the concentration of particles @xmath4 is known to decrease as : @xmath62 where the ( @xmath63 ) coefficient is expected to be universal ( e.g @xmath64 , @xcite ) .
more detailed properties , such as persistence , have also been studied .
the probability @xmath65 that no particle @xmath4 ( domain wall ) has crossed a given point @xmath66 up to time @xmath8 has been shown to decay as : @xmath67 where @xmath68 is the characteristic length and @xmath69 the so - called persistence exponent .
@xmath65 also corresponds to the probability that a spin has never been flipped up to time @xmath8 in the potts model .
the exact expression of @xmath69 was obtained in @xcite .
the domain size distribution has also been computed for this process in @xcite .
two new independent exponents , @xmath11 and @xmath12 were introduced and studied in @xcite to characterize the persistence ( survival ) of domains for this model , but up to now no exact result is available for these exponents .
the concept of persistence properties was extended to other observables , finite temperature and studied in a variety of other models : persistence for global order parameter in @xcite , spin block persistence in @xcite , generalized persistence and large deviations in @xcite and persistence for fluctuating interfaces @xcite . in this paper
we study the case where the hopping rates are inhomogeneous with short range correlations , corresponding to random local bias .
the generic model for this type of disorder is the sinai model where each particle performs arrhenius diffusion in the same energy landscape @xmath70 where the local random forces @xmath71 are independent random variables , of zero average ( we restrict the present analysis to zero global bias ) .
various analytical results are known for the single particle sinai model @xcite .
diffusion is ultraslow as @xmath72 .
recently we have reexamined this model @xcite using a real space renormalization group method ( rsrg ) which yields exact results at long time . in the present paper
we apply the rsrg method to study reaction diffusion of the type ( [ general ] ) for particles in a sinai landscape .
some of the results have already appeared in @xcite .
although we give here a detailed treatment of the reaction diffusion rsrg , we will rely on ref .
@xcite for all details concerning the single particle diffusion aspects of the problem ( which we will only sketch , referring the reader to @xcite for details ) .
note that we consider here only models where all particles share the same diffusion property ( i.e. see the same landscape , and have same diffusion coefficient ) .
thus this does not include reaction diffusion models such as ising domain walls in a random field , for which a specific treatment is necessary and which are studied in @xcite .
similarly , relations to other problems such as disordered quantum spin chains @xcite or disordered free fermion models are discussed in @xcite . in particular
we have chosen to discuss our present results exclusively in terms of reaction diffusion dynamics , and not in their equivalent formulation as ( non hermitian ) 1d quantum models ( see @xcite for details of such relations in the pure case ) . finally , note that an exact rg has also been applied to the problem of coarsening of the pure 1d @xmath73 model at zero temperature for which persistence exponents have been computed @xcite .
as for the single particle problem , the rsrg method allows us to compute a number of quantities , and , remarkably , even some which are not known for the corresponding pure model ( e.g. the domain persistence exponents @xmath12 and @xmath11 ) .
this makes the disordered case all the more interesting to study .
we find that reaction diffusion processes in a sinai landscape are strongly controlled by the ultraslow diffusion , e.g the relevant length scale is the diffusion length @xmath74 , but that they still possess non trivial reaction properties .
we will characterize a broad set of universality classes , containing all reactions of type ( [ general ] ) . the reaction times ( provided they are finite )
do not affect any universal quantity , so that we can consider the reactions as instantaneous for practical purpose .
as discussed in @xcite there are other single particle diffusion models with short range correlated disorder in one dimension apart from sinai s model universality class , such as random barriers ( symmetric hopping rates ) or random wells .
for interesting behaviour to occur , however , algebraically broad distributions are required from the start .
some results for reaction diffusion processes with this type of single particle diffusion have been obtained in @xcite .
the outline of the paper is as follows . in section [ rgmethod ]
we detail the rsrg method , first recalling known results in the case of a single particle in [ rgsinai ] , then deriving the rsrg equation for reaction diffusion models [ rgreacdiff ] .
the fixed points of this equation , and some physical properties of the corresponding asymptotic states are studied in [ fixedpoints ] and [ physical ] respectively .
section [ asymptoticdynamics ] is devoted to a detailed analysis of the dynamics near attractive or repulsive fixed points and of the convergence towards the asymptotic states . throughout the paper
we apply our results to the process ( [ potts ] ) but in section [ examples ] we discuss some applications to other examples of processes . in section [ persistence ]
we study persistence properties .
section [ conclusion ] contains the conclusion .
some more technical but useful details are contained in the appendices .
the model for the diffusion of a single particle in one dimension can be defined , with no loss of generality @xcite , as the arrhenius diffusion in a `` zigzag '' potential @xmath75 represented in fig .
[ fig1 ] ( a ) .
it consists in a set of bonds , each bond ( between @xmath76 and @xmath77 ) being characterized by an energy barrier @xmath78 ( where @xmath79 ) and a length @xmath80 .
the energy landscape is chosen by choosing a pair of bond variables @xmath81 , independently from bond to bond , from a distribution @xmath82 normalized to unity . the rg procedure , which captures the long time behaviour in a given energy landscape , is illustrated in fig .
[ fig1 ] ( b ) and consists in the iterative decimation of the bond with the _ smallest barrier _
@xcite , say @xmath83 , and to replace the three bonds @xmath84 by a single renormalized bond with barrier @xmath85 and length @xmath86 .
the new variables remain _ independent _ from bond to bond . to write the corresponding rg equation it is convenient to introduce @xmath87 as the smallest remaining barrier at a given stage of the decimation and the rescaled variables @xmath88 and @xmath89 .
the rg equation for the probability distribution @xcite @xmath90 reads @xcite : @xmath91 and coincides with the one derived in @xcite for the closely related problem of disordered quantum spin chains .
the symbol @xmath92 denotes a convolution with respect to @xmath93 only and @xmath94 with respect to both @xmath95 and @xmath93 .
the probability distribution is normalized to unity as @xmath96 .
the landscape is characterized by the large scale variance of the potential : @xmath97 with @xmath98 the distance from site i to site j , which is exactly preserved by the rg .
thus we will set @xmath99 in the following . restoring
@xmath100 simply amounts to a rescaling of lengths , and in particular @xmath100 drops out of all ( universal ) ratios of lengths that we consider later .
as shown in @xcite , the rg equation ( [ rgbond ] ) leads at large @xmath87 ( using laplace transforms ) to the following fixed point @xmath101 : @xmath102 thus for large @xmath87 one finds that the average bond length @xmath103 and the number of bonds @xmath104 per unit length are @xcite , respectively : @xmath105 the renormalized landscape allows to study the dynamics of a single walker starting from a given point @xmath66 at @xmath106 .
the decimation of barriers smaller than @xmath107 corresponds to the elimination of ( logarithmic ) time scales shorter than the arrhenius time @xmath8 for the particle to cross the barrier .
we are choosing everywhere time units such that the ( non universal ) microscopic attempt time scale @xmath108 be set to unity ( arbitrary units can be recovered by setting @xmath109 in what follows @xcite ) .
since at long time ( i.e large @xmath87 ) the renormalized landscape consists entirely of deep valleys separated by high barriers , a good approximation to the long time dynamics is to place the walker _ at the bottom _ of the renormalized valley at scale @xmath110 which contains the starting point @xmath66 , since with high probability it will be near to that point @xcite . upon proper rescaling of space and time
this approximation becomes in fact _ exact _ as @xmath87 tends to @xmath111 .
this defines what we will call the `` effective dynamics '' in the following and is illustrated in fig .
this allows to recover the scaling @xmath112 for the single particle diffusion as well as many other exact results detailed in @xcite . since it is customary , when studying reaction diffusion processes , to compare densities of reactants with a characteristic scale of diffusion , we give here the the exact expression for the single particle r.m.s displacement , or
diffusion length at large time : @xmath113 to study reaction diffusion processes it will be necessary to consider `` valleys '' ( two consecutive bonds sharing a common potential minimum ) . we thus slightly generalize the above rg equation ( [ rgbond ] ) to follow the distribution of renormalized valleys .
the rg equation for the valley probability distribution @xmath114 at scale @xmath87 in rescaled variables @xmath115 reads @xmath116 p^{\gamma}(\eta_1,\eta_2 ) \nonumber \\ & & + p^{\gamma}(\eta_1 , . ) * _ { \eta_2 } p^{\gamma}(0 , . ) + p^{\gamma}(.,0 ) * _ { \eta_1 } p^{\gamma } ( .
, \eta_2 ) \label{rgvalley}\end{aligned}\ ] ] where we have omitted ( i.e integrated over ) the lengths for simplicity . the large time ( @xmath87 ) behaviour of this equation can be studied similarly .
valley distributions which have the decoupled form @xmath117 where @xmath118 satisfies the bond rg equation ( [ rgbond ] ) are of course solution of the rg equation for valleys ( [ rgvalley ] ) .
the subspace of such decoupled distributions ( called decoupled subspace in the following ) is thus preserved by rg .
since the initial condition is uncorrelated , the rg flow defined by ( [ rgvalley ] ) remains in this decoupled subspace , and converges towards the fixed point @xmath119 .
this convergence result extends to the case of small correlations between valley sides as will be discussed below .
it was shown in @xcite that the convergence towards the bond fixed point @xmath120 is like @xmath121 with eigenvector @xmath122 .
thus the convergence towards the the valley fixed point @xmath123 within the decoupled subspace is of the form : @xmath124 and is also as @xmath121 with eigenvector @xmath125 .
we now turn to diffusion reaction models of type ( [ general ] ) in one dimensional landscapes with random local biases . from the results on the dynamics of a single particle in sinai landscape recalled in the preceding section ,
it is clear that one can study most of the properties of the initial reaction diffusion problem by following its evolution under the effective dynamics .
it also becomes obvious that one must now consider valleys , and the species contained in these valleys . at the decimation time scale @xmath110 in some places in the system
, two valleys will merge into one and the reaction ( [ general ] ) governed by the rates matrix @xmath126 will take place .
this process is illustrated in fig .
the errors made by this approximation are expected to become again smaller at large time , as will be discussed later on .
the general method to study the process ( [ general ] ) is thus to associate to each valley the specie which it contains , which is one of several possible states @xmath17 ( @xmath38 being the empty state ) .
a convenient initial model is thus one where each valley and its content is statistically independent and characterized by a probability distribution @xmath127 with @xmath128 .
it remains so under the rg .
the effective dynamics is described by the rg equation : @xmath129 \label{rgreadiff}\end{aligned}\ ] ] where summation over repeated indices is implied .
the summation over @xmath17 yields back the valley rg equation ( [ rgvalley ] ) .
since the average length of a valley is @xmath130 , the total concentration @xmath131 of a given specie @xmath17 ( the total number of @xmath17-particles per unit of length ) is given as : @xmath132 where : @xmath133 and @xmath104 is the number of remaining bonds ( [ scaling ] ) per unit length at scale @xmath110 .
we stress that the rg equation ( [ rgreadiff ] ) is more complicated to analyze than ( [ rgvalley ] ) since it can not in general be factorized into bond distributions . however , it turns out that there is still a simple subspace of distributions which is exactly preserved by the rg equation ( [ rgreadiff ] ) .
it is the subspace of functions of the sum @xmath134 : @xmath135 where the functions @xmath136 satisfies : @xmath137 which conserved the normalization @xmath138 .
this subspace plays an important role in the following .
already one sees that both the fixed point @xmath139 and the leading eigenvector @xmath125 of the linearized landscape rg equation ( [ rgvalley ] ) belongs to this subspace .
we now determine the fixed point solutions of the rg equation for the valley distributions ( [ rgreadiff ] ) .
we already know that the sum @xmath140 converges towards the fixed point of equation ( [ rgvalley ] ) @xmath141 which describes the landscape .
it is thus natural to look for fixed points of ( [ rgreadiff ] ) of the following form @xmath142 where @xmath143 and @xmath144 by normalization .
plugging this form into ( [ rgreadiff ] ) leads to a consistent @xmath87 independent solution if the @xmath145 satisfy the condition : @xmath146 note that any solution of this equation satisfies a priori @xmath147 or @xmath148 , as a consequence of ( [ matrixw ] ) .
thus apart from the unphysical solution of ( [ eq ] ) where all @xmath145 vanish , all other solutions are automatically correctly normalized
. in general the equation ( [ eq ] ) has several solutions and thus there are several fixed points to the valley equations ( [ rgreadiff ] ) .
clearly some of these fixed points are attractive and correspond to possible large time asymptotic states for the reaction diffusion process while other fixed points are repulsive .
in some cases several attractive fixed points can coexist and lead to a non trivial phase diagram . the stability of each fixed point , as well as their convergence property , will be studied in detail in the next section .
here we just mention one important result .
the dynamics in the vicinity of a fixed point @xmath145 is determined by the following stability matrix : @xmath149 we denote by @xmath150 the eigenvalues of @xmath151 , and by @xmath152 the associated eigenvectors @xmath153 . for @xmath154 reacting species ( in addition to the empty state @xmath38 ) , @xmath151 is an @xmath2-dimensional matrix with non - negative elements .
one eigenvalue is trivial , being simply @xmath155 of eigenvector @xmath156 ( [ eq ] ) .
the other @xmath154 eigenvalues @xmath157 ( @xmath158 ) , which can be complex in general , have a smaller modulus , due to the perron - froebenius theorem .
we will focus in the following on simple enough processes where all @xmath157 are real , but we will also give example of more general reaction diffusion where complex eigenvalues arise ( e.g. reactions with cycles ) .
let us consider here only fixed points with all @xmath157 being real .
the result of the next section is that , in that case , a given fixed point if attractive is the @xmath154 eigenvalues @xmath159 for @xmath158 .
it is repulsive if at least one of these @xmath160 ( and is then repulsive along the corresponding eigendirection ) .
for example , in the case of the reaction - diffusion process ( [ potts ] ) , there are two solutions of ( [ eq ] ) and thus two fixed points .
one is the empty state e=(@xmath161 , @xmath162 ) and the other is s=(@xmath163 , @xmath164 ) .
the stability matrix associated to e is simply the @xmath165 identity matrix ( i.e. @xmath166 ) and this fixed point is thus repulsive .
the matrix associated to the state s reads @xmath167 with eigenvalues @xmath155 and @xmath168 .
the fixed point s is thus attractive and corresponds to the asymptotic state which represents the large time behaviour of the system .
in fact the reaction - diffusion process ( [ potts ] ) possesses an interesting property : the outcome of a sequence of reactions does not depend on the order it was performed .
we call these processes `` associative processes '' .
they have the property that @xmath169 , i.e @xmath170 .
some properties of these associative processes are detailed in appendix ( [ appassoc ] ) .
we now study some physical consequences .
each attractive fixed point corresponds to a possible large time behaviour of the system , i.e an asymptotic state .
if there are several attractive fixed point , the one chosen by the system will depend on the initial value of the parameters ( mainly the specie concentrations ) . from the results ( [ fixedpointeta ] ) and ( [ scaling ] ) we obtain that in an asymptotic state ( characterized by a set of @xmath171 solution of ( [ eq ] ) , the density of specie @xmath17 behaves at large time as : @xmath172 where we have restored the microscopic attempt time scale @xmath108 @xcite
note that this result ( [ density ] ) represents the leading large time contribution , subleading corrections ( which become dominant only if @xmath173 ) will be determined in the next section .
interestingly , this leading behaviour is independent of the initial concentration ( provided it is in the basin of attraction of the fixed point ) .
this universality property can be further characterized by computing universal amplitudes . in pure models
a commonly studied amplitude is the product of specie concentration by the diffusion volume . in a disordered model
one has more choices of definitions , but we will define an amplitude as in @xcite . here
we find the exact result for the following _ universal amplitude _
( associated to specie @xmath17 ) : @xmath174 this gives for instance @xmath175 for the process ( [ potts ] ) , i.e @xmath176 for @xmath28 studied in @xcite by perturbative methods @xcite . from the statistical independence of valleys in an asymptotic state , information about the spatial distribution of the species can also be obtained . for instance one can define `` domains '' , in the simplest case as intervals between particles ( i.e between non empty states @xmath177 , irrespective of their content ) we can now compute exactly the distribution of size of domains . from the above form of the fixed points the normalized distribution @xmath178 of domain sizes @xmath179 in an asymptotic state
takes a scaling form @xmath180 which can be computed as follows . the above rg equation for valley distributions ( [ rgreadiff ] )
can be readily extended to @xmath181 which takes into account the rescaled lengths @xmath182 , @xmath183 of the two bonds of the valley , extending ( [ rgbond ] ) .
the generalized fixed point ( [ fixedpointeta ] ) reads : @xmath184 where @xmath101 is the fixed point solution ( [ solu ] ) of the bond rg equation . a domain as defined above , is thus a set of consecutive empty valleys between two occupied valleys , together with one bond in each of the occupied valleys ( see fig .
[ fig2 ] ) . since in the asymptotic state valleys are statistically independent and either empty ( @xmath38 ) with probability @xmath185 ( where @xmath186 ) or contains a particle ( @xmath177 ) with probability @xmath187 , one easily obtains the laplace transform of @xmath188 as : @xmath189 where we have used the explicit form ( [ solu ] ) for the fixed point bond distribution . formula ( [ lengths ] ) can be inverted and yields the distribution of rescaled domain sizes : @xmath190 with @xmath191 .
this can be applied , e.g. in the case of the process ( [ potts ] ) .
substituting @xmath192 in ( [ distriblengths ] ) yields the distribution of distances between neighboring walkers a. note that the case where walkers a always coalesce upon meeting ( @xmath193 , @xmath194 ) corresponds to @xmath195 and in this limit ( [ distriblengths ] ) becomes : @xmath196 it is interesting to compare the result ( [ distriblengths ] ) concerning the disordered case with the result of derrida and zeitak @xcite for the case of homogeneous hopping rates . for small domain sizes ( @xmath197 )
, the distribution vanishes much faster in the disordered case ( as @xmath198 ) than in the pure case ( as @xmath199 ) . for large domain sizes ( @xmath200 ) , both have exponentially decaying behaviour ( except for @xmath201 , i.e @xmath193 in the pure case ) .
in addition , in the present case the consecutive domains lengths are statistically independent , which is not the case for the pure system . the above calculation is easily generalized to compute the distribution of relative distances between two walkers of a given species @xmath17 , simply by substituting @xmath202 in the above formula ( [ distriblengths ] ) .
in this section we study the dynamics near the possible fixed points of the valley rg equation ( attractive and repulsive ) .
we will first focus strictly on the effective dynamics exactly described by the rg equation ( [ rgreadiff ] ) , and mention some possible corrections in the real dynamics at the end of the section . for the effective dynamics we will solve the problem in two steps . as mentionned above the matrix @xmath151 in ( [ matrixm ] ) and its eigenvalues @xmath157 control the asymptotic dynamics .
interestingly they readily provide an approximation of the dynamics , which we will call the `` uniform approximation '' , which is interesting as it allows to classify the spectrum of eigenperturbations and , in the case of real eigenvalues , already allows to see whether a given fixed point is stable or unstable . in a second step we will obtain the exact results for the spectrum of eigenperturbations .
it is natural to define the total occupation probability of specie @xmath17 at scale @xmath203 as @xmath204 .
the difficulty of the problem comes from the fact that it does not satisfy a closed equation .
however , if one also introduces @xmath205 , i.e the occupation probability of specie @xmath17 of the valleys just being decimated at @xmath87 , one can obtain a closed coupled equation by integration of ( [ rgreadiff ] ) which reads : @xmath206 it is then tempting to set , as an approximation , @xmath207 .
this would be correct at any of the fixed points ( [ fixedpointeta ] ) , but since we are studying convergence to a fixed point , it is an approximation which amounts to neglect the dynamical correlations between the deviations in the specie concentration and in the distribution of barriers heigths . for this reason we call it the uniform approximation .
it yields the following approximate closed rg equation for the @xmath208 : @xmath209 which preserves the normalization condition @xmath210 .
this approximate flow has the same fixed points @xmath211 than the true one ( [ eq ] ) .
this equation , remarkably , is reminiscent of a
mean field type rate equation , except that the role of time would be played by the variable @xmath212 . the relaxation of ( [ approximate ] ) towards any of these fixed points
is studied by setting @xmath213 and linearizing for the small perturbation @xmath214 around the fixed point @xmath145 .
it yields , in terms of the matrix m introduced in ( [ matrixm ] ) : @xmath215 and thus the convergence towards the fixed point has components behaving as @xmath216 where the exponents are given in terms of the eigenvalues @xmath150 of the matrix m as @xmath217 with @xmath218 .
so this `` uniform approximation '' would indicate that a fixed point is stable if all @xmath219 for all @xmath218 , and unstable otherwise .
remarkably , this stability criterion coincides with the exact result _ when the eigenvalues are real _ as we will now show , even if the naive convergence eigenvalues @xmath220 are not correct ( they are `` renormalized '' to larger absolute values ) . up to now
we have studied the convergence of the landscape alone ( [ rgvalley ] ) , and the convergence of the @xmath221 within a uniform approximation assuming @xmath222 .
we now study the full dynamics near a fixed point solution of the full reaction - diffusion equations ( [ rgreadiff ] ) .
we will indeed find that there are some correlations between deviations in total occupation probabilities ( from the fixed point concentrations ) and deviations in the barrier distribution profile ( from the simple fixed point shape @xmath223 ) , resulting in deviations with respect to the uniform approximation .
we thus consider a perturbation around the fixed point of the form @xmath224 and linearize the equation for the perturbation @xmath225 @xmath226\end{aligned}\ ] ] where we have used the symmetry of the @xmath126 , and the definition of the matrix m ( [ matrixm ] ) . note that in the end we are interested into the behavior of the species proportions @xmath227 the normalisation condition of course implies that @xmath228 .
decomposing @xmath229 upon the eigenvectors @xmath230 corresponding to the eigenvalues @xmath150 of the matrix m as @xmath231 , we obtain decoupled equations for the coefficients @xmath232 @xmath233\end{aligned}\ ] ] for a given @xmath157 , we look for solutions behaving as @xmath234 and determine the exponent @xmath235 as a function of the eigenvalue @xmath150 . here , a priori both @xmath150 and @xmath235 can be complex . before we study this equation for general @xmath236 we will first study the simpler cases @xmath237 and @xmath238 .
note that for associative processes this will be sufficient .
this case is important for naively stable fixed points of associative processes ( which have all @xmath239 for @xmath240 ) .
the @xmath87 dependent equation ( [ relaxation ] ) with @xmath239 can be integrated out explicitly starting from its initial value at @xmath241 @xmath242 and thus the convergence of proportions @xmath208 towards @xmath171 is governed by ( [ epsilon ] ) @xmath243 so for generic initial perturbations @xmath244 that are not exponentially growing as @xmath245 , we obtain that the convergence towards the fixed point is exponential in @xmath87 .
note that here this exact result is very different from the naive approximation which would predict a convergence as @xmath246 with @xmath247 . to understand why this is so
, one can compute from the exact solution ( [ exactmu0 ] ) the ratio : @xmath248 which is found to grow with @xmath87 !
this is why the uniform approximation is particularly bad for this case where it predicts a power - law instead of the exponential convergence in @xmath87 . to study the dynamics ( [ relaxation ] ) in the case @xmath249 , it is useful to introduce the function @xmath250 ,
since it satisfies the closed simpler equation @xmath251 that gives after integration from an initial condition at @xmath241 @xmath252 and thus for initial conditions @xmath253 that are not exponentially growing at @xmath245 , we obtain that the the function @xmath254 converges towards 0 exponentially in @xmath87 .
this means that the perturbations @xmath255 converges exponentially in @xmath87 towards the decoupled subspace @xmath256 we now study the convergence towards the fixed point in that decoupled subspace to see if there are solutions behaving as @xmath257 : the equation for @xmath258 reads @xmath259 \psi_{\alpha } ( \eta ) + 2 \int_0^{\eta } d \eta ' \psi_{\alpha } ( \eta ' ) + \eta \psi_{\alpha } ( 0)\end{aligned}\ ] ] the only well - behaved solutions are found to be : @xmath260 the first solution corresponds to the convergence as @xmath261 of the landscape discussed previously .
this is not surprising since the linearized rg equation ( [ rgvalley ] ) is exactly ( [ relaxation ] ) with @xmath238 .
note en passant that ( [ hfinal ] ) shows that small correlations between barriers in the same valley decrease exponentially fast in @xmath87 towards the subspace of statistically independent distributions .
since this landscape eigenvector satisfies @xmath262 it does not affect the species proportions @xmath263 .
it is the second eigenvector in ( [ 2roots ] ) which is relevant for the reaction diffusion processes since @xmath264 .
it corresponds to the unstable eigenvalue ( growth as @xmath265 ) associated to a naively unstable fixed point ( e.g. of an associative process ) . note that in that case , the unstable eigenvalue found @xmath266 coincides with the naive value @xmath267 of the uniform approximation .
physically , this eigenvalue can be understood for e.g. the process ( [ potts ] ) .
the unstable fixed point is the empty state e with @xmath268 , @xmath269
. now if one starts at @xmath270 very close to the fixed point , there are very few @xmath4 and their number should not vary at first , as they will rarely meet .
this is indeed exactly what the above results says , namely : @xmath271 and using that e has @xmath272 .
we now study the case of a general @xmath157 .
it turns out that one can find solution of the original equation ( [ relaxation ] ) under the form @xmath273 where the function @xmath274 satisfies the differential equation @xmath275 together with the boundary condition @xmath276 .
the only well - behaved solution at @xmath277 is the confluent hypergeometric function @xmath278 , and the boundary condition at @xmath279 determines the possible exponents @xmath235 that should satisfy : @xmath280 . using the identity @xmath281
, this equation for @xmath235 reduces to @xmath282 we now discuss the behaviour of the solutions of this equation . one
must distinguish two cases : let us start with @xmath236 real . as mentionned above , one must have @xmath283 . for @xmath238
the equation ( [ eqphi ] ) reduces to @xmath284 which admits the two roots @xmath285 and @xmath266 and one thus recovers the eigenvalues ( [ 2roots ] ) .
the equation ( [ eqphi ] ) continues to admit two finite roots when @xmath236 belongs to the interval @xmath286 , which we denote @xmath287 and @xmath288 ( with @xmath289 and @xmath290 ) .
the behaviour of these roots as a function of @xmath236 is plotted in fig .
( [ figphi1 ] ) and ( [ figphi2 ] ) .
as @xmath236 is decreased from @xmath148 , @xmath291 increases and diverges when @xmath292 while @xmath288 increases from @xmath293 to @xmath294 . for @xmath295
the equation ( [ eqphi ] ) admits only one finite root @xmath296 which is positive and with @xmath297 and @xmath298 as @xmath299 .
again we can see from the exact solution above , why the uniform approximation is not valid . in terms of the dominant @xmath300 mode
the ratio : @xmath301 goes at large @xmath87 towards the finite limit @xmath302 found to be bigger than one when @xmath303 .
one can check that setting @xmath213 and @xmath304 into the equation ( [ intereadiff ] ) , one obtains that the relaxation towards the fixed point is now like @xmath305 with @xmath306 . since @xmath307 , the correct exponent @xmath308 is bigger than the naive exponent @xmath309 .
next we turn to complex @xmath236 .
the matrix @xmath151 being real , if it has complex eigenvalues they will come in pairs , @xmath157 and @xmath310 corresponding to complex conjugate eigenvectors @xmath311 and @xmath312 . in the uniform approximation
they will combine as @xmath313 and correspond to oscillatory ( growing or decaying ) solutions ( where @xmath314 is complex ) .
this situation happens in cyclic reactions , examples of which will be given below .
within the uniform approximation the fixed point is stable to this perturbation if @xmath315 and unstable otherwise .
this however turns out _ not _ to be correct .
indeed the correct exponent @xmath316 ( now complex ) is determined by the above equation ( [ eqphi ] ) associated to @xmath236 .
one has @xmath317 , and to each pair of @xmath236 , @xmath318 one can associate one ( or two in some cases ) pairs of @xmath316 , @xmath319 also corresponding to oscillatory solutions .
the oscillation frequency in the @xmath320 variable is now given by @xmath321 , and the stability being now determined by @xmath322 ( @xmath323 corresponds to a stable eigenperturbation , while @xmath324 correspond to an unstable one ) .
interestingly , @xmath325 is a decreasing function of @xmath326 as it is increased from @xmath327 .
thus the region of stability in the complex @xmath236 plane is _ different _ from the one inferred from the uniform approximation .
it is represented in ( [ figphistab ] ) .
one notes that for complex eigenvalues one can have a fixed point naively stable ( with @xmath315 ) which is in reality unstable ( with @xmath328 .
thus we have solved the problem of the dynamics near the asymptotic states of the form ( [ eq ] ) for arbitrary reaction diffusion process .
let us summarize the results .
_ when the eigenvalues @xmath150 of the matrix m ( [ matrixm ] ) are real _ , the stability of the fixed point is determined by the naive argument : a fixed point is stable if @xmath329 for all @xmath330 .
the decay exponents @xmath331 are obtained in terms of the @xmath150 as : @xmath332\end{aligned}\ ] ] where the function @xmath333 $ ] is defined implicitly by the single root of the equation @xmath334 , 2]=0 $ ] ( here @xmath335 and represented in fig .
( [ figphi1 ] ) . in terms of these exponents , the large time behaviour for the concentrations of the species
is found to be @xmath336 where the @xmath337 correction in @xmath104 comes from the convergence of the landscape @xcite .
in addition there are corrections to ( [ concentrationphi ] ) which decay must faster , exponentially in @xmath87 ( i.e algebraically in time ) .
the @xmath338 are constants , depending on the initial condition .
the formula ( [ concentrationphi ] ) can also be used to relate two late times .
if the system is very near the asymptotic state at @xmath270 , with @xmath339 , ( [ concentrationphi ] ) holds at @xmath8 with @xmath340 where we recall that the @xmath311 are the eigenvectors of @xmath151 and @xmath341 . for practical applications it is useful to note that ratios of concentrations of different species involve only the exponents @xmath331 . on the other hand , because of the factor @xmath104 , the relaxation of the concentration of a single specie @xmath17 to its asymptotic form is controlled ( provided @xmath342 ) by the exponent @xmath343 ( where @xmath316 is the minimum of all exponents @xmath331 appearing in the corresponding formula ( [ concentrationphi ] ) for @xmath131 ) .
the formula is even more interesting in the case @xmath344 ( i.e if the specie @xmath17 disappears in the reaction ) since then the first correction becomes the dominant decay and one has at large time that @xmath345 .
examples of such cases are studied in section [ coales ] .
let us stress again that the difference between the exact value @xmath331 and the uniform approximation value @xmath346 is due to the fact that near the asymptotic states the ratios @xmath347 differ from one , i.e the valleys to be decimated do not have the average distribution of species : there is a non trivial mixing between valley heights and concentration of species , missed by the naive argument , and responsible for the non trivial relaxation exponents found here .
_ when some eigenvalues @xmath150 of the matrix m
( [ matrixm ] ) are complex _ the fixed point is stable provided all @xmath157 lie in the part of the complex plane on the left to the curve represented in fig .
( [ figphistab ] ) .
the specie concentrations then relax with oscillations as : @xmath348 finally , in the case where @xmath349 ( or more generally on the line @xmath350 ) linear analysis is insufficient to determine the evolution of the system , and one must study the full non linear rg equation ( [ rgreadiff ] ) , which goes beyond the present study @xcite . to conclude this section , let us recall that the results derived above concern , strictly speaking , the effective dynamics described by the rg equation ( [ rgreadiff ] ) . as was discussed in great details in ref .
@xcite for the single particule diffusion , there are corrections in the real dynamics , with respect to the effective dynamics .
indeed , in the effective dynamics the whole thermal packet jumps at @xmath351 over a barrier @xmath87 , while in the real dynamics typically a fraction of a thermal packet ( which can be written as @xmath352 ) has not yet jumped at time @xmath8 .
since the distribution of barriers becomes broader and broader , this generates corrections which at large time are only subdominant for most quantities ( at most @xmath337 ) coming typically from rare events such as degeneracy of order @xmath353 between neighboring barriers .
they become dominant only for certain quantities , such as the width of the thermal packet , which have vanishing leading order in the effective dynamics . in ref .
@xcite the corrections to first order in @xmath337 were evaluated and found to originate from three rare events ( a ) valleys with degenerate minima , ( b ) almost degenerate barriers ( c ) valley just being decimated with a barrier @xmath354 ( see fig .
7 of @xcite ) . a similar detailed study of the rare events in presence of reaction processes can be performed but goes beyond the present paper . with similar arguments as in @xcite
we do not expect any correction to the leading order of the quantities computed in this paper . in principle , subdominant corrections could add to the subleading terms computed above , e.g. in ( [ concentrationphi ] ) .
they are certainly at most of order @xmath337 ( and thus can not affect any decay as @xmath355 with @xmath356 ) but it is likely that they are even of higher order .
indeed most of these corrections ( e.g. ( b ) ) come from single particle diffusion and can be reabsorbed into the global factor @xmath104 . other events ( such as ( a ) ) can not affect specie concentrations .
although this point deserve further study , it is likely that the corrections from the real dynamics to the convergence to asymptotic states obtained in this section using is subdominant .
up to now we have only applied the general results only to the process ( [ potts ] ) .
we give here several examples of other processes , starting with a process which exhibits a dynamical phase transition and to which the general results apply directly .
then we present other cases which raise interesting questions which go slightly beyond our present analysis .
let us consider the system involving two species a and b and the empty state 0 : @xmath357 the solutions of equation ( [ eq ] ) for the fixed points are the empty state @xmath358 , the a - phase @xmath359 , the b - phase @xmath360 , and the fixed point @xmath361 .
the a phase and the b phase are attractive fixed points with eigenvalues @xmath362 corresponding to asymptotic exponential decay in @xmath363 ( i.e a power law in time ) of the other specie .
the empty state is a repulsive fixed point with eigenvalues @xmath364 .
the critical point c is attractive for symmetric perturbation @xmath365 corresponding to eigenvalue @xmath366 , but instable with eigenvalue @xmath367 for any dissymetric perturbation @xmath368 .
this corresponds to the exponent @xmath369 ( the other root is @xmath370 ) .
since it is globally attractive over the critical manifold , this fixed point controls the dynamical transition from the @xmath4 phase to @xmath371 phase .
thus we conclude that if one starts with a system of @xmath4 and @xmath371 in almost equal concentrations , the difference @xmath372 ( or equivalently the relative concentrations of @xmath4 and @xmath371 ) grows with time as : @xmath373 or , equivalently , the differences of absolute concentrations decay as @xmath374 , i.e more slowly than the decay of both concentrations of @xmath4 and @xmath371 , which itself behaves as @xmath375 .
the system eventually reaches a broken symmetry state where either @xmath4 or @xmath371 predominates after a time @xmath376 which scales as : @xmath377 where @xmath270 is a ( shorter ) reference time scale and @xmath378 a ( @xmath270 dependent ) constant .
note that the uniform approximation would predict @xmath379 significantly smaller that the exact result .
finally , the asymptotic final decay of the minority phase is fast , exponential in @xmath87 ( as @xmath237 at either @xmath4 ot @xmath371 fixed points ) .
restraining from giving further examples among the large number of possible processes with similarly interesting behaviour to which our general results readily apply , we now turn instead to cases where open questions remain .
let us now consider the reaction : @xmath380 the fixed points are the empty state @xmath381 which is unstable with eigenvalues @xmath382 , the b phase @xmath360 which is fully attractive with eigenvalues @xmath362 , and the fixed point u @xmath383 which is attractive with @xmath366 along the axis @xmath384 and marginal with @xmath385 for any perturbation @xmath386 .
going beyond the linear approximation , we find in the uniform approximation that @xmath387 , i.e. a small initial proportion @xmath388 grows very slowly as @xmath389 and thus the time scale @xmath390 where the proportion of @xmath371 becomes finite grows like @xmath391 for small @xmath388 . in the real dynamics
, the exponent @xmath392 being @xmath327 we also expect a kind of marginal behaviour near the fixed point u. a full study of this behaviour is an interesting question which goes beyond the present paper .
let us study the reaction : @xmath393 the solutions of ( [ eq ] ) are the three pure phases @xmath394 , @xmath395 , @xmath396 and the mixed state @xmath397 .
the pure phase @xmath394 is stable ( @xmath237 ) in the direction @xmath398 , and unstable ( @xmath238 ) in the direction @xmath399 .
the mixed fixed point @xmath397 has complex eigenvalues @xmath400 , leading to purely imaginary naive exponents @xmath401 .
as can be seen on fig [ figphistab ] , the exact convergence exponents @xmath402 , solutions of ( [ eqphi ] ) , have a negative real part , and thus the fixed point @xmath397 is also _ unstable_. this shows that the asymptotic behaviour of the system is more complex than being described by a fixed point of type ( [ eq ] ) .
in fact , going back to the equations ( [ approximate ] ) of the uniform approximation , and eliminating @xmath403 , we obtain that the flow equations for the two variables @xmath404 take the `` divergence free '' form @xmath405 with @xmath406 . as a consequence ,
all starting points where the three concentrations @xmath407 are non - zero , belong to closed flow lines @xmath408 of constant value of @xmath409 .
thus in the uniform approximation the asymptotic behaviour is always cyclic .
this however does not carry to the real dynamics , beyond the uniform approximation , since that one can check that these cycles @xmath410 are _ not _ solutions of the rg equation ( [ rgreadiff ] ) .
thus the question of determining the asymptotic behaviour of this problem is still open .
a more complex cyclic solution , or a new non trivial fixed point are among the possiblities .
we close this section by noting that one can also expect from @xcite that reactions with a large enough number of species have chaotic solutions at the level of the uniform approximation .
it would be interesting to investigate whether such chaotic solutions could also exist in the rg and in the exact dynamics of these disordered reaction diffusion problems .
we now consider the much studied @xmath411 inert reaction , which , in the absence of disorder is known to lead to segregation @xcite of the two species .
@xmath412 we introduce the notations @xmath413 , @xmath414 and @xmath415 ( @xmath416 ) .
the possible contents for a valley are now the @xmath417 with @xmath418 and thus their number is infinite .
the reaction rules become with these notations @xmath419 so for the rg procedure , it is convenient to associate to each valley an auxiliary variable @xmath420 representing the contents of the valley , and to write the rg equation for the probability distribution @xmath421 where the rg rule for the auxiliary variable @xmath420 upon fusion of valleys simply reads @xmath422 . we can use the result of the appendix of ( @xcite ) ( for the same rg rule of an auxiliary variable ) and obtain the scaling @xmath423 thus we find that charges of order @xmath363 of both signs ( i.e groups of size of order @xmath363 of particles of the same specie ) will accumulate near the bottom ( in a packet of typical size @xmath424 ) of each renormalized valley .
these packets will be separated by a large distance of order @xmath425 .
the total number of particles will still decay , as @xmath426 .
this asymptotic state thus still presents strong features reminiscent of the segregation observed in the pure case @xcite .
by contrast with the pure model , here several packets of @xmath4 can also be found in successive neighboring valleys .
we now study persistence properties in the reaction diffusion models defined in ( [ secmodel ] ) .
as explained in the introduction one is interested in computing probabilities that some type of event has not occured between time @xmath327 and @xmath8 .
the decay with time @xmath8 of these probabilities usually involves new non trivial exponents which characterize the non equilibrium dynamics . since they integrate over time the behaviour of the system , they are usually hard to obtain analytically , even in the pure systems . for reaction diffusion models in a random environment the following types of persistence probabilities
can be defined , and will be studied in the following corresponding sections . a - the simplest persistence observable is the probability @xmath427 over runs and environments that a given point @xmath428 has not been crossed by any particle between time @xmath327 and @xmath8 in a given run .
the decay of @xmath427 : @xmath429 where @xmath430 is a characteristic length of the diffusion process , here given from ( [ scaling ] ) as @xmath431 , defines the persistence exponent @xmath0 .
this is the definition used in this paper , even when referring to the pure case , where @xmath432 , whereas another frequent definition is in terms of the power of @xmath8 . since here
the diffusion is logarithmic we choose everywhere in this paper the more general definition ( [ refdef ] ) both for pure and disordered problems .
b - in the presence of quenched disorder one can also study the probability @xmath433 over environments that a given point @xmath428 has not been crossed by any of the thermally averaged trajectories @xmath434 of the particles .
similarly the decay of @xmath435 defines the exponent @xmath1 .
one expects in general that @xmath436 and here we find that these two exponents are quite different .
c - more generally , one can defined the probability that a given pattern present at time @xmath327 has survived up to time @xmath8 .
we study the example of the survival of domains ( i.e. intervals between particles ) , which in the pure case was shown to lead to the definition of two new exponents @xcite called @xmath12 and @xmath11 : @xmath12 characterises the probability that a domain has survived up to time @xmath8 without merging with other domains , and @xmath11 characterises the probability that a domain has survived up to time @xmath8 with mergings with other domains .
d - finally we study the exponents @xmath437 and @xmath438 characterising the probability that a particle a has survived up to time @xmath8 , without any coalescence and with coalescences respectively .
we start by computing the probability @xmath427 that @xmath428 has not been crossed by _ any _ particle up to @xmath8 .
we consider a rather general reaction diffusion process with a vacuum state @xmath327 ( @xmath38 ) and occupied states ( with particles in them ) , @xmath439 . to solve this problem
we can consider separately the two half spaces @xmath440 and @xmath441 and study the problem of a semi infinite system ( @xmath440 ) with an absorbing boundary at @xmath428 ( absorbing for the states @xmath439 ) . for diffusion in a sinai landscape in presence of an absorbing boundary at @xmath428
one defines a new rsrg with slightly new rules : the boundary rsrg , explained in detail in ref .
the first bond is by definition always ascending with an infinite barrier ( and thus can never be decimated ) and represents an `` absorbing zone '' ( see figure ( [ figboundary ] ) ) .
if the smallest barrier in the system at @xmath87 is the third bond from the boundary or further the rules are identical to bulk rsrg . if the smallest barrier is the second bond , i.e the first descending bond , the procedure consists in eliminating the first valley ( i.e the second and third bond ) which is merged with the absorbing zone .
since the reaction rules of the species upon merging valleys are unaffected by the boundary , at a given @xmath87 all renormalized valleys in the bulk are distributed independently with @xmath127 which satisfies the equation ( [ rgreadiff ] ) .
we now explicitly check that the first renormalized valley has also the same distribution . indeed
the probability @xmath442 that the first renormalized valley has ( @xmath95 , @xmath443 , @xmath17 ) satisfies the rg equation : @xmath444 where the first term corresponds to the decimation of the second bond ( which results in the increase of the absorbing zone ) and the old second valley becomes the new first renormalized valley .
the third and fourth terms correspond to the decimation of the left bond of the second valley ( the loss term must be explicitly written on the l.h.s . to keep the distribution @xmath445 correctly normalized to unity ) .
one can now check that @xmath446 is a consistent solution . to compute the persistence exponent ,
we now define @xmath447 as the probability over all environments that the boundary at @xmath428 has never been crossed by a walker between time @xmath327 and @xmath87 _ and _ the first renormalized valley has ( @xmath95 , @xmath443 , @xmath17 ) at @xmath87 .
it satisfies the rg equation : @xmath448 similar to ( [ rgfunctionr ] ) except for the first term which carries the restriction that the second bond can be decimated _ only if _ the first renormalized valley is empty ( since , if it contains a particle , this particle gets absorbed by the wall , i.e crosses the origin ) .
a consistent solution is simply @xmath449 with @xmath450 we now use the fact that the system reaches for large @xmath87 an asymptotic state corresponding to an attractive fixed point ( [ fixedpointeta ] ) , and this leads to the asymptotic decay @xmath451 since the probability @xmath427 that the point @xmath428 has not been crossed by any particle up to time @xmath8 on the infinite line is the square of the corresponding probability for the semi infinite problem we obtain : @xmath452 with @xmath431 and the result for the persistence exponent : @xmath453 as an example , we explicit the result for the process ( [ potts ] ) where @xmath454 : @xmath455 for @xmath193 where the particles @xmath4 always merge and occupy all valleys , we recover the half - space exponent @xmath456 corresponding to the decay exponent of the probability of no return to the origin for a single sinai walker obtained in @xcite . a generalization of the persistence exponent @xmath0 can be defined for reaction - diffusion models on the semi - infinite line @xmath440 in the presence of an absorbing boundary at @xmath428 .
there one can compute the probability @xmath457 that exactly @xmath2 particules have been absorbed by the wall up to time @xmath8 . generalising the approach of ( [ rgfunctionv ] ) , we obtain that @xmath457 satisfies at large @xmath87 : @xmath458 \qquad n \ge 1 \\ & & \gamma \partial_\gamma q_{\gamma}(0 ) = - ( 1-p_0^ * ) q_{\gamma}(0)\end{aligned}\ ] ] the rg equation for the generating function @xmath459 @xmath460 thus leads to the decay @xmath461 . introducing the rescaled number of absorbed particules @xmath462 and using as in @xcite the saddle point method , we find after a legendre transform , that the probability distribution prob(g ) behaves as @xmath463 with the generalized persistence exponent : @xmath464 for @xmath465 , one recovers the persistence exponent of the half space @xmath466 , where @xmath0 is given by ( [ restheta ] ) .
@xmath467 has a zero minimum at @xmath468 which is thus the value that @xmath469 takes with probability one as @xmath470 @xmath471 as was discussed in detail in ref .
@xcite , thermally averaged trajectories of a single sinai walker follow the effective dynamics which we now use to study their persistence properties .
[ fig3 ] illustrates the difference in the persistence properties between the single run dynamics studied in the previous section and the effective dynamics of thermally averaged trajectories .
let us consider the case of a valley with a right bond of barrier @xmath87 such that the point @xmath428 lies to the left of a valley bottom and is separated from it by a barrier less than @xmath87 . in this case , @xmath428 will be crossed ( several times ) in a typical single run , while the thermal average @xmath472 will remain at the bottom of the valley until it jumps to the right over the barrier without crossing @xmath428 .
thus , as was found in @xcite for the return to the origin of a single walker , the exponents @xmath0 and @xmath1 should generically be different .
we now compute the probability @xmath433 that the point @xmath428 has not been crossed by any particle up to time @xmath8 within the effective dynamics .
let us define for each valley the auxiliary variables @xmath473 equal to the total number of sites in the descending ( @xmath474 ) and ascending ( @xmath475 ) bonds respectively , which have not been crossed by any particle between @xmath327 and @xmath8 .
we define the probabilities @xmath476 that a valley has a specie @xmath17 , bonds @xmath477 , @xmath478 and variables @xmath473 .
consider the decimation represented in fig .
let us denote the two valleys corresponding to bonds @xmath479 and @xmath480 containing respectively the species @xmath481 and @xmath482 .
upon decimation of bond @xmath483 the two valleys merge and the specie @xmath481 jumps to the bottom of the valley @xmath480 and thus goes over the bond @xmath484 and @xmath485 to react there with the specie @xmath482 . as a consequence , the auxiliary variable of the new renormalized bond @xmath486 evolves with the rule : @xmath487 this is a particular case of the auxiliary variables studied in appendix ( [ auxiliary ] ) with @xmath488 and @xmath489 .
the final result is that the fraction of sites that have never been crossed by any particle in the effective dynamics decays as @xmath490 where the persistence exponent @xmath1 is solution of the following equation involving the confluent hypergeometric functions @xmath491
@xmath492 for the process ( [ potts ] ) one has @xmath493 , and the resulting exponent @xmath494 is plotted in fig .
[ fig22 ] .
surprisingly we find that it is numerically extremely close for all @xmath61 ( to less than about one percent in relative difference ) of _ one half _ the result @xcite for the pure system which reads @xmath495 and is also plotted in fig .
[ fig22 ] .
the expansion for small @xmath61 gives @xmath496 and thus they are definitively different . in the case @xmath497 where particles always annihilate , we obtain @xmath498 which may be compared with @xmath499 .
the difference remains very small for all @xmath61 , as shown in fig .
[ fig23 ] .
we have also generalized the calculation presented in this section to compute the _ number of visits _ of thermally averaged trajectories of particles at a given point .
it leads again to a multifractal spectrum of exponents .
the calculation and the results are presented in the appendix [ auxiliary ] . here we define domains as intervals between particles . for the reaction ( [ potts ] ) , on which we now concentrate ,
when a domain dies ( the two particules a meet ) the two contiguous domains can either merge if the particles annihilate ( with probability @xmath61 ) or remain separate if the particles coalesce ( with probability @xmath500 ) . to characterize the statictics of the coarsening of domains in the pure case ( i.e potts domains with @xmath501 ) krapivsky and ben - naim
have introduced @xcite the following definition .
they define @xmath502 as the number of domains at time @xmath8 which have for ancestors @xmath416 initial domains .
this is illustrated on the figure [ figdomains ] .
the total number @xmath503 of domains remaining at time @xmath8 ( equal to the total number of @xmath4 at time @xmath8 ) is simply given by the sum @xmath504 and decays as @xmath505 , where @xmath506 is the characteristic length at time @xmath8 , given in ( [ scaling ] ) .
the fraction of initial domains which have a descendant that is still alive at @xmath8 is given by @xmath507 .
the decay of these quantities define two new independent persistence exponents @xmath12 and @xmath11 : @xmath508 and with these exponents , it is expected that @xmath502 takes the scaling form @xmath509 the scaling function is expected to behave for small @xmath510 as @xmath511 where the exponent @xmath100 is related to @xmath512 by the relation @xmath513 .
note that the inequalities @xmath514 imply that @xmath515 .
note that here , we have again defined the exponents with respect to the characteristic length @xmath430 at time @xmath8 .
thus in the pure case our definition differs from the one of @xcite by a factor of @xmath483 .
below , we obtain the exponents @xmath6 and @xmath7 exactly for the process ( [ potts ] ) . for a domain to survive up to time @xmath8 while keeping its variable @xmath516
, the two domains walls must avoid meeting each other up to time t , but they can meet other exterior domain walls , provided that upon meeting they coalesce and do not annihilate . since the two domains walls must not meet , given the properties of the effective dynamics in the rg , the decay of @xmath517 is governed by the events such that at some @xmath241 the two domains belong to two neighboring renormalized valleys . at all later times
they will still belong to two neighboring renormalized valleys and no decimation of the two renormalized bonds separating the two domains can occur . as a consequence to compute the exponent @xmath12 , we can consider separately the two corresponding half spaces . for a given half space , we introduce the probability @xmath518 that_the first bond has never been decimated _ and _ the valley is ( @xmath95 , @xmath443 ) _ and _ there is always one walker in the first valley_. the rg equation for this quantity reads @xmath519 where the @xmath520 term arises because the @xmath521 , unlike the @xmath522 is not normalized to 1 , and one must count the loss associated with the left bond of the second valley . integrating over
@xmath523 one finds that @xmath524 evolves with @xmath525 corresponding to the three forbidden cases : decimation of the first bond , decimation of the second or third bond when both valleys are full and annihilation occurs .
the exponent @xmath12 will be given by the decay of the half space probability @xmath526 , since the probability associated the two sides will decay as the square of the probility for one side , i.e as @xmath527 .
setting @xmath528 , and using @xmath529 , one finds : @xmath530 \rho(\eta,\eta ' ) + \\ & & \frac{1}{1+r } ( \int_0^\eta d\eta_1 \rho(\eta_1,0 ) + \int_0^{\eta ' } d\eta_2 \rho(\eta,\eta_2 ) ) \end{aligned}\ ] ] there is a solution of the form @xmath531 where @xmath532 satisfies : @xmath533 \tilde{\rho } ( \eta ) + \frac{1}{1+r } \int_0^\eta d\eta ' \tilde{\rho}(\eta ) \end{aligned}\ ] ] after derivation with respect to @xmath95 , one finds a standard hypergeometric differential equation which admits for only solution not growing exponentially at @xmath534 , the confluent hypergeometric function @xmath535 .
the boundary condition at @xmath536 then determines the exponent @xmath6 as the solution of the implicit equation : @xmath537 using the functional relation @xmath538 , the exponent @xmath6 is finally the solution of the equation @xmath539 the solution of this equation is plotted in fig .
[ figdeltar ] . in the case
@xmath193 , we find @xmath540 as expected . indeed , in that case where particles always coalesce , domains can not merge , and thus @xmath516 is the only possible value : @xmath541 and thus @xmath542 as in the pure case . for the case
@xmath497 , where particles always annihilate , we find @xmath543 which is remarkably close to the numerical result obtained in ( @xcite ) for the ising pure case : @xmath544 .
this puzzling feature also holds for other values of @xmath61 , as shown in fig [ figdeltar ] , with less than about one percent in relative difference . to compute @xmath11 we need to obtain the scaling behaviour of the average number of ancestors of the domain @xmath545
however this is a priori difficult , as the variable @xmath420 is associated to a domain which can extend over many renormalized bonds and is thus `` non local '' .
however we can circumvent this difficulty by decomposing @xmath420 upon the several renormalized bonds which make up a domain , in order to have a _
rule under rg for an auxiliary variable @xmath546 associated to bonds .
thus we write each variable @xmath420 for a domain made out of @xmath57 bonds , as the sum @xmath547 , of new auxiliary variables , each associated to a bond .
since @xmath57 does not grow with @xmath87 , the scaling of @xmath548 and @xmath549 with @xmath87 are identical .
we define rg rules for the local @xmath546 variables as follows .
we consider two neighboring valleys as in fig [ fig2 ] with bonds ( 1,2 ) containing specie @xmath481 and ( 3,4 ) containing specie @xmath482 , respectively .
one must think of the variable @xmath546 as counting the number of ancestors associated to a renormalized bond and thus upon decimation of bond ( 2 ) the variables @xmath546 , @xmath550 associated to the new bonds of barriers @xmath551 and @xmath552 become : @xmath553 the first case where the decimated valley is empty is obvious . in the second case , where a particle @xmath4 jumps from valley ( 1,2 ) to the empty valley ( 3,4 ) , the ancestors of the domain to the right of @xmath4 previously associated to the bonds ( 2 ) , ( 3 ) and ( 4 ) must now all be associated to the bond ( 4 ) . in the third case , where the two @xmath4 meet , the ancestors of ( 2,3 ) disappear of the problem in all cases ( i.e annihilation or coalescence of the @xmath4 particles ) . introducing the rescaled variables @xmath554 , where @xmath555 , the fixed point rg equation for the valley distribution @xmath556 reads : @xmath557 we introduce the two first moments ( @xmath38 and @xmath558 ) @xmath559 where @xmath560 . using the symmetry
@xmath561 , we find that they satisfy the closed system : @xmath562 + p_0^ * \int_0^{\eta ' } [ g_0(\eta , . )
+ g_a(\eta , . ) ] \\ & & 0 = ( ( 1 + \eta ) \partial_{\eta } + ( 1 + \eta ' ) \partial_{\eta ' } - \eta - \eta ' - \phi ) g_a(\eta,\eta ' ) \\ & & + p_a^ * \int_0^\eta [ g_a(.,0 ) + g_0(.,0 ) + g_0(0 , . ) ] + p_0^ * \int_0^\eta g_a(.,\eta ' ) \\ & & + p_a^ * \int_0^{\eta ' } [ g_a(\eta,.)+g_0(\eta,.)+g_0(.,\eta ) ] + p_0^ * \int_0^{\eta ' } g_a(0 , . ) \label{systemg}\end{aligned}\ ] ] the exponent @xmath316 is determined by the condition that the solutions @xmath563 and @xmath564 of this system should be well behaved at infinity ( i.e. should not be exponentially growing ) .
we found that setting @xmath565 allows to obtain the following closed system for the two functions @xmath566 and @xmath567 @xmath568 \\ & & 0=(2+z ) s_a'(z)-(z+\phi ) s_a(z ) + \int_0^z [ s_a(.)+p_a^ * s_0(z)]\end{aligned}\ ] ] i.e , independently of the antisymmetric part of @xmath563 which we will not need . to decouple this system , we introduce two linear combinations @xmath569 that satisfy closed equations @xmath570 where the eigenvalues , using @xmath571 , are : @xmath572 the only solutions of ( [ systpsi ] ) that are not exponentially growing at infinity are again given in terms of degenerate hypergeometric function @xmath573 the boundary conditions @xmath574 finally give @xmath575 since @xmath576
, the growth of @xmath577 will be governed by @xmath578 , and thus the final result is that the exponent @xmath7 is determined by the equation @xmath579 in particular we have the following expansion around @xmath193 @xmath580 for the case @xmath497 where particles always annihilate , we find @xmath581 which again is remarkably close to the numerical result obtained in @xcite for the ising pure case : @xmath582 , a property which hold again for all @xmath61 , and again to within less than about one percent in relative difference , as illustrated in fig [ figpsir ] .
in the end we note that one can generalize the bound : @xmath583 discussed in @xcite , to the disordered case , as : @xmath584 i.e that the exponent @xmath7 is always bounded by the persistence exponent @xmath494 of thermal averaged trajectories found in ( [ eqthetabarre ] ) .
this comes from the observation that a point that has never been crossed by any particle up to time @xmath8 for the effective dynamics has to belong to domain that has a descendant still living at time t. here the reverse inequality is clearly not true ( for @xmath585 ) since a surviving domain may not contain any persistent site , as it can shift from its initial position , as shown in figure [ figbound ] .
in particular we have found ( [ eqthetabarre2 ] ) : @xmath586 thus @xmath7 and @xmath494 differ already at first order in @xmath61 .
this is different from the case of the random field ising model , studied in @xcite , where it is found that @xmath587 , as the situation depicted in figure [ figbound ] does not occur .
we now come to the study of persistence properties associated to a particle . following the general framework presented in the previous section for the study of domain merging statistics ,
we now introduce the number @xmath588 of particles @xmath4 at time @xmath8 which have @xmath2 particles a for ancestors in the initial condition .
this is illustrated on the figure ( [ figpart ] ) .
this will lead us to introduce two new exponents , @xmath437 and @xmath438 , which have not been computed ( nor defined ) in the pure case and which we will compute here in the disordered model .
the total number @xmath503 of particles at time @xmath8 is simply given by the sum @xmath589 .
the fraction of initial particles which have a descendant still alive at @xmath8 is given by @xmath590 .
again , the decay of @xmath591 , i.e of the number of particles which have encountered no other particles , and of @xmath592 define two new independant exponents @xmath593 together with these exponents , it is expected that @xmath588 takes the scaling form @xmath594 where the scaling function behaves for small @xmath510 as @xmath595 for small @xmath510 where the exponent @xmath596 is related to @xmath597 by the relation @xmath598 .
we will compute these exponents via the rsrg by two methods .
the first one is direct , while the second one , presented at the end , will rely on results previously established in section [ asymptoticdynamics ] . in the first method
, we introduce an auxiliary variable @xmath2 for each valley containing a particle a that counts the number of ancestors of this particle .
we then introduce the probability @xmath599 that a valley at scale @xmath87 has @xmath523 and contains a particle a having @xmath2 ancestors in the initial condition .
it satisfies the rg equation @xmath600 p^{\gamma}_a(\eta,\eta',n ) \\ & & = p^{\gamma}_{a}(\eta,.,n ) * _ { \eta ' } p^{\gamma}_{0}(0 , . )
+ p^{\gamma}_{a}(.,0,n ) * _ { \eta } p^{\gamma}_{0 } ( . , \eta ' ) \\ & & + p^{\gamma}_{0}(\eta , .
) * _ { \eta ' } p^{\gamma}_{a}(0 , .,n ) + p^{\gamma}_{0}(.,0 ) * _ { \eta } p^{\gamma}_{a } ( . ,
\eta',n ) \\ & & + ( 1-r ) \left [ p^{\gamma}_{a}(\eta , . , .
) * _ { \eta',n } p^{\gamma}_{a}(0 , . , . )
+ p^{\gamma}_{a}(.,0 , . ) * _ { \eta , n } p^{\gamma}_{a } ( . , \eta ' , . ) \right]\end{aligned}\ ] ] where @xmath601 is the probability that a valley at scale @xmath87 has @xmath523 and contains no particle , i.e. it satisfies ( [ rgreadiff ] ) . at this stage ,
the variable @xmath2 is an integer @xmath602 , and the convolution on @xmath2 is a discrete convolution . at large @xmath87 , we know the fixed point @xmath603 and @xmath604 where @xmath605
thus setting @xmath606 , we find that the function @xmath607 satisfies @xmath608 \delta^{\gamma}(z , n ) = 2 p_0^ * \int_0^z dz'\delta^{\gamma } ( z',n ) + ( 1-r ) p_a^ * \delta^{\gamma } ( . , .
) * _ { z , n } \delta^{\gamma } ( .,.)\end{aligned}\ ] ] and we now will compute successively @xmath9 , @xmath10 for all @xmath61 and the scaling function for @xmath193 . to compute the exponent @xmath437 giving the decay of the number of particles a that have only one ancestor in the initial condition , we need to solve the previous equation for the particular value @xmath609 .
@xmath610 decouples from the other values of @xmath2 , and satisfies the linear equation @xmath608 \delta^{\gamma}(z,1 ) = 2 p_0^ * \int_0^z dz'\delta^{\gamma } ( z',1 ) \end{aligned}\ ] ] since @xmath611 , we set @xmath612 .
the function @xmath613 is then the solution of the differential equation @xmath614 together with the boundary condition at @xmath279 @xmath615 we thus find @xmath616 , and the exponent @xmath12 is determined by the implicit equation @xmath617 using the functional relation @xmath538 , and @xmath454 , the exponent @xmath9 is finally the solution of the equation @xmath618 for the particular case @xmath497 where particles always annihilate upon meeting , we have @xmath619 as it should since in this case the particles can have only one ancestor @xmath620 . in the limit @xmath621 ,
where particles always coalesce upon meeting , we have @xmath622 : indeed in this case at large @xmath87 , all valleys contain a particle a @xmath623 , and the probability to have @xmath609 decays exponentially with @xmath87 , since it requires that four consecutive bonds ( the two bonds of the valley and the two neighbors ) are not decimated . to compute the exponent @xmath438 , we introduce the rescaled variable @xmath624 .
the fixed point solution @xmath625 of the rescaled variables has to satisfy @xmath626 \delta ( z,\nu ) \\ & & + 2 p_0^ * \int_0^z dz'\delta ( z ' , \nu ) + ( 1-r ) p_a^ * \delta ( . , .
) * _ { z,\nu } \delta ( . , .
) = 0\end{aligned}\ ] ] in particular , using @xmath627 , we find that its first moment @xmath628 satisfies the differential equation @xmath629 with the boundary condition at @xmath279 @xmath630 so finally @xmath631 , where the exponent @xmath438 is the solution of the implicit equation @xmath632 using again the functional relation @xmath538 , and @xmath454 , the exponent @xmath10 is finally the smaller solution of the equation @xmath633 for the particular case @xmath497 where particles always annihilate upon meeting , we have @xmath634 as it should , since in this case the particles can have only one ancestor . in the case
@xmath635 , where particles always coalesce upon meeting , we have @xmath636 : indeed the probability for an initial particle to have a descendant living at @xmath87 is one , and thus @xmath592 is constant and not decaying .
the distribution @xmath637 of the rescaled variable @xmath638 can in principle be obtained in terms of the solution @xmath639 of ( [ eqdelta ] ) as @xmath640 in laplace with respect to @xmath641 , we have that @xmath642 satisfies @xmath643 \hat{\delta}_r(z , q ) \\ & & + \frac{2r}{1+r } \int_0^z dz ' \hat{\delta}(z',q ) + \frac{1-r}{1+r } \hat{\delta}_r(.,q ) * _ { z } \hat{\delta}_r(.,q ) = 0\end{aligned}\ ] ] in the case @xmath635 , where particles always coalesce upon meeting , the number @xmath2 of ancestors should have the same statistical properties as the length of a valley , and thus using the fixed point solution ( [ solu , fixedpointeta2 ] ) , we should have @xmath644 indeed we find : @xmath645 is the solution of ( [ eqscalingfunction ] ) for @xmath646 , and thus in this case the scaling function @xmath637 reads @xmath647 to compute @xmath591 , i.e the probability that a given particle @xmath4 has met no other particles up to time @xmath8 , we can consider this particle @xmath4 as a tagged particle , give it a name , say x , and consider it as a new specie in very small concentration .
it satifies the following reaction rule : @xmath648 and of course @xmath649 and the same reactions for the @xmath4 as before .
we need only to work to linear order in @xmath650 and we are back exactly in the case studied in the section [ asymptoticdynamics ] of the dynamics near the asymptotic state , of a new reaction diffusion ( whose fixed point has @xmath651 ) .
the corresponding eigenvalue of the matrix @xmath151 introduced in ( [ matrixm ] ) is @xmath652 . here
@xmath653 which corresponds to an attractive fixed point at @xmath654 ( since the @xmath655 disappear ) and with @xmath656 where @xmath316 is solution of ( [ eqphi ] ) for @xmath657 .
since @xmath658 we recover the equation ( [ eqdeltaa ] ) . to compute the exponent @xmath10 we need to consider similarly the reaction for the tagged particle @xmath655 : @xmath659 and of course @xmath649 and the same reactions for the @xmath4 .
in this case @xmath660 and @xmath661 which corresponds to an _
unstable _ fixed point at @xmath654 .
one finds @xmath656 where @xmath316 is solution @xmath662 of ( [ eqphi ] ) for @xmath663 .
since @xmath664 we recover the equation ( [ eqpsia ] ) which determines @xmath10 .
it is interesting to study the stability of our results to an additional quenched disorder in the reaction probabilities given by the matrix @xmath126 ( i.e spatial inhomogeneities ) .
we continue to consider only the rule that species react immediately when they encounter , but the analysis in fact also covers - in an effective way - the case where reaction rates are finite and with quenched disorder .
we sketch in this section a possible way of applying the present rsrg procedure to this case .
let us consider a model where the reaction probabilities are themselves functions of the position @xmath665 .
a simple example is to allow the parameter @xmath61 to depend on @xmath666 as @xmath667 in the process ( [ potts ] ) .
let us examine what happens at a decimation time scale @xmath363 .
the particle in state @xmath481 in the decimated valley jumps over the barrier to a valley containing @xmath482 .
since @xmath482 is typically at equilibrium at the bottom of its valley , the reaction is most likely to take place at the bottom of the valley within a @xmath424 distance of it ( since this is where all the weight of the particule @xmath482 is concentrated ) .
thus as time increases , the total number of sites in the system where reactions can typically occurs decays as @xmath668 . in each renormalized valley at @xmath87 there is typically a `` finite '' number ( i.e not growing with @xmath87 ) of sites @xmath666 where reactions occur and thus a `` finite '' number of possible values @xmath669 ( a notation for the set of @xmath670 ) . for each valley
these form a given set fixed in time
. there are thus a priori two competing effects : the several values taken by @xmath126 in a valley result in an `` averaging '' effect for the effective @xmath126 of this valley . on the other hand the fact that this set is finite and fixed in time implies non trivial correlations between two reactions occuring at different times in the same valley . here
we will restrict to consider a toy model where we assign a single transition matrix @xmath126 to each renormalized valley with probability @xmath671 .
it would be accurate in the case where in the initial distribution the @xmath126 s are correlated over distances much larger than the typical thermal width of a packet @xmath672 , but still small compared to @xmath265 .
this problem can thus be treated by introducing the probability @xmath673 that a valley has rescaled barriers @xmath477 , @xmath478 and an associated rate @xmath126 .
when two valley merge upon a decimation the new one simply keeps the @xmath126 of the lowest one .
one notes that the statistical independence is again preserved by rg .
the rg equation is simply @xmath674 \label{rgdisorder}\end{aligned}\ ] ] where @xmath675 and summation over repeated indices is implied .
we also note that the distribution of @xmath126 , @xmath676 is preserved by the rg rule , thus @xmath677 .
thus we have a `` marginal '' problem , since in this toy model @xmath671 does not flow by rg @xcite .
one can now look for fixed points of this rg equation under the form : @xmath678 where the @xmath679 must satisfy : @xmath680 in the case of the model ( [ potts ] ) with a distribution @xmath681 of @xmath61 , one can show that a solution is : @xmath682 where @xmath683 and @xmath684 are the equilibrium occupation probabilities for the problem with a uniform @xmath61 , solution of the equation @xmath685 .
such a simple solution holds in that case because of the form of the matrix @xmath151 ( [ matrixm ] ) which is simply a projector onto the vector @xmath145 . in general
this does not hold and one has to solve the above equation .
it is then possible in principle to perform , for an arbitrary @xmath671 , the same study as the one done here , such as stability eigenvalues around the fixed point etc .. , which is left for the future .
to summarize , the above result indicates that within the toy model and the effective dynamics , quenched disorder in the transition matrix @xmath669 will lead to a modification of the large time properties .
these properties can be computed using the rg by assigning a an effective reaction probability matrix of each valley .
they depend in a continuous way on the asymptotic distribution @xmath671 .
there is thus an infinite dimensional manifold of fixed points in the rg sense , and the problem is _
marginal_. like all marginal problems , it is very sensitive to corrections which may make disorder marginally irrelevant or marginally relevant ( or remain strictly marginal ) .
the averaging effect may in the end make the disorder marginally irrelevant , but to decide , within the real dynamics @xcite and within a model with shorter disorder correlation length , whether disorder is actually marginally irrelevant , and how it flows , requires a more detailed study which goes beyond this work .
in this paper we have studied the problem of various species of particles diffusing in presence of quenched random local bias ( sinai landscape ) and reacting upon meeting .
we have shown that the real space renormalization group method ( rsrg ) , which has proved to be a powerful tool to study single particle diffusion in sinai landscape @xcite , can be extended in a simple and natural way to study a large class of reaction diffusion models . since here also the physics is controlled by infinitely broad disorder fixed points , this method , as in the single particle problem , is expected to yield the _ exact _ large time behaviour . focusing on renormalized valleys as well as on the particles ( and species )
contained in these valleys , and following the evolution of their distribution by decimation upon increase of the time scale , allowed us to obtain many new exact results on this problem .
we have obtained a detailed description of the asymptotic states , such as the large time decay of the density of each specie , @xmath131 , and the spatial distribution of particles .
it confirms that in @xmath36 sinai landscape the reaction is subdiffusion limited .
the first step was to identify simple fixed points of the valley distribution rg equation , which correspond - for a given reaction process described by a transition matrix - to possible asymptotic dynamical states .
each of these states is characterized by fixed fractions @xmath171 for each specie , the physical picture being the following . at time
scale @xmath110 the system consists of a set of successive renormalized valleys , which can be either empty , with probability @xmath686 or contain a particle of specie @xmath17 , with probability @xmath171 .
the separation between particles grows as the characteristic length @xmath687 , and thus @xmath688 .
the decay of concentration , when compared to the single particle diffusion length , leads to define universal amplitudes , which we obtained exactly .
also , from the exact statistical independence of the successive valley lenghts , the distribution of interval between particles ( domains ) was derived ( and compared with some pure case results ) . to confirm that a given fixed point is indeed an asymptotic state , actually reached by the system at large time
, it is necessary to study its linear stability .
we have thus obtained analytically the spectrum of stability eigenvalues around any simple fixed point , as a function of the reaction transition matrix , thereby solving the stability problem .
the convergence towards these asymptotic states ( i.e the attractive rg fixed points ) was studied .
the leading convergence towards these asymptotic states was found to be generically as @xmath689 with a non trivial @xmath316 , solution of a hypergeometric equation ( with in addition , an amplitude periodic in @xmath690 in the case of complex eigenvalues ) . in some cases ,
corresponding to @xmath691 , the convergence is faster as a power law in time with non universal exponent depending on details of the initial model .
eigenvalues corresponding to unstable fixed points , which were also determined , are of particular interest for reactions which lead to several distinct dynamical phases ( i.e several possible asymptotic states ) . the transitions between different dynamical phases being controlled by such unstable rg fixed points
, we have thus obtained the corresponding critical exponents . as an example
, a process with a non trivial phase diagram was studied .
we have also studied persistence properties associated to a given asymptotic state . as in pure systems , where it was originally defined ,
persistence can be studied for various types of patterns ( single particles , domains etc .. ) . remarkably , for the disordered models at hand we are able to derive a _ much larger _ set of exact results than exists at present for the corresponding pure systems .
we have first obtained , for a generic process , the decay exponent @xmath0 for the probability of no crossing of a given point by single particle trajectories .
as noted in @xcite , in a disordered system , persistence of thermal averages can be quite different from single particle persistence .
thus we have also computed the probability of no crossing of a given point by thermally averaged packets , which yields the decay exponent @xmath1 .
the properly generalized persistence exponents associated to @xmath2 crossings have been defined , and computed .
next , we have characterized the statistics of domains , which can disappear or merge as time increases .
restricting , for simplicity , to the process @xmath3 or @xmath4 with probabilities @xmath5 , we have obtained exactly the exponents @xmath6 and @xmath7 characterizing the survival up to time @xmath8 of a domain without any merging or with mergings respectively .
we have also introduced new exponents which similarly characterize the statistics of the coalescence of particles .
we have then computed them , namely @xmath9 and @xmath10 characterizing the survival up to time @xmath8 of a particle @xmath4 without any coalescence or with coalescences respectively .
we have found these new exponents as solutions of hypergeometric equations . for comparison ,
the only known analytical result in the pure case is for the exponent @xmath692 for the process ( [ potts ] ) .
a surprising outcome was that several exact exponents of the model with disorder were found to be numerically very close , for all values of @xmath61 , to some exponents for the pure system , although they are associated to completely different diffusion length ( @xmath693 while @xmath506 ) .
indeed we found that @xmath694 although they are definitely distinct , and furthermore that @xmath695 and @xmath696 where @xmath697 , @xmath698 - not known analytically - are extracted from the numerical simulation of @xcite .
the agreement in relative values is better than about one percent for all @xmath61 . it may be that this observed numerical coincidence could be traced to the exact statistical independence of valley lengths in the disordered problem , while the so called `` independent interval approximation '' , gives reasonable approximation in the pure case ( but , surprisingly , poorer than the one provided by these new exponents )
this however is far from an explanation and further investigation may be of interest .
the effect of additional disorder in the reaction rates was also discussed . in a simple case
it was found to be marginal , and thus yield non trivial modifications , continuously varying with the disorder distribution .
the question of whether this result is stable to corrections resulting from the real dynamics or from disorder with shorter correlation length remains to be further investigated although we have not considered explicitly branching barw processes , with additional creation of particles , it is clear that for at least some of them the physics should not be qualitatively too different from the one obtained here .
indeed , since in sinai disorder particles are essentially confined to the bottom of large renormalized wells , as long as the process is such that particles are not created out of the vacuum and that the annihilation reactions are sufficient to maintain the number of particles small when at local equilibrium in a well , the reaction can be treated very similarly via rsrg as for the model studied here .
we have thus characterized a broad set of reaction diffusion models with disorder .
finally , it is worthwile to mention that we have also identified cases which clearly require a more complicated analysis going beyond the present paper .
for instance we have given an example of a marginal reaction , which requires a non linear stability analysis .
also , we have given an example of a cyclic reaction for which all simple rg fixed points are shown to be unstable .
the question of the determination of the true asymptotic states of this reaction is thus still open .
another open interesting , and maybe related , question is whether reactions with large enough number of species , which can lead to chaotic attractors in pure cases @xcite , will also lead to chaotic behaviour in presence of disorder .
2.pt [ cols="^,^,^",options="header " , ] : summary of the results obtained in the text for persistence exponents associated to the decay of the probability of each indicated event , in the case of the reaction diffusion process @xmath699 ( proba @xmath61 ) @xmath700 ( proba @xmath500 ) .
* acknowledgements * we thank d. s. fisher for fruitful discussions as well as s. fauve , u. tauber and p. chauve for helpful remarks .
in this appendix we study auxiliary variables @xmath701 associated to bonds that evolve upon decimation as follows .
consider the decimation of bond @xmath484 on fig .
[ fig2 ] : the two valleys corresponding to bonds @xmath479 and @xmath480 and containing respectively the species @xmath481 and @xmath482 merge , and the specie @xmath481 jumps to the bottom of valley @xmath480 and thus goes over the bond @xmath484 and @xmath485 to react there with the specie @xmath482 .
it is thus natural to consider an auxiliary variable @xmath420 which , upon decimation of the barrier @xmath486 , evolves with the general rule : @xmath702 where the coefficients @xmath703 depend on the specie @xmath17 which diffuses upon the corresponding decimation .
we now write the valley rg equation for @xmath704 : @xmath705\end{aligned}\ ] ] integrating this equation over the @xmath420 variables one recovers of course the specie valley rg equation ( [ rgreadiff ] ) .
we now define the first moment : @xmath706 since we are looking at the symmetric case , we have that @xmath707
. we can thus write the following closed equation for @xmath708 @xmath709 in the asymptotic state we use the fixed point solution @xmath710 , and write @xmath711 and obtain the equation for the new function @xmath712 : @xmath713 g_k ( \eta_1,\eta_2 ) \\ & & + w^k_{k_1,k_2 } [ p_{k_2}^ * \int_0^{\eta_2 } g_{k_1}(\eta_1 , . ) + p_{k_1}^ * a_{k_1 } \int_0^{\eta_1 } g_{k_2}(.,\eta_2 ) \\ & & + p_{k_2}^ * b_{k_1 } \int_0^{\eta_1 } g_{k_1}(0 , . )
+ p_{k_2}^ * d_{k_1 } \int_0^{\eta_1 } g_{k_1}(.,0)\end{aligned}\ ] ] since the @xmath420 variable is associated to bonds , it is natural to look for solutions where the function @xmath714 is a function of @xmath477 alone .
we thus try solutions of the form @xmath715 , where the exponent @xmath11 characterizes the scaling of the @xmath420 variable @xmath716 . for this to work we
obtain , in terms of the matrix @xmath151 defined in ( [ matrixm ] ) , the necessary condition : @xmath717 together with the differential equation for @xmath718 : @xmath719 g_k(\eta ) + ( m_{k , k_1 } d_{k_1 } + w^k_{k_1,k_2 } p^*_{k_2 } a_{k_2 } ) \int_0^{\eta } g_{k_1 } ( . )
+ m_{k , k_1 } b_{k_1 } g_{k_1}(0 ) \eta \end{aligned}\ ] ] one can then try @xmath720 , which automatically satisfies the necessary condition ( [ condition1 ] ) above ( since @xmath145 is by construction an eigenvector of the @xmath151 matrix of eigenvalue 1 ) , and then the second equation gives the conditions involving two numbers @xmath721 @xmath722 together with the differential equation for @xmath723 : @xmath724 g(\eta ) + \lambda_1 \int_0^{\eta } \psi ( . )
+ \lambda_2 \psi(0 ) \eta\end{aligned}\ ] ] we now give two applications of this general analysis .
we now study the case @xmath488 and @xmath489 corresponding to the auxiliary variable ( [ rulempersistence ] ) needed to compute the persistence exponent @xmath1 the conditions ( [ condition2 ] ) become : @xmath725 since the rates involving the empty state @xmath726 are given by definition by @xmath727 , we have @xmath728 .
the conditions above are thus satisfied with @xmath729 and @xmath730 and thus the problem reduces to studying the the equation ( [ eqgpersist ] ) for @xmath723 : @xmath731 g(\eta ) + ( 1 + p^*_0 ) \int_0^{\eta_1 } g ( . ) + p^*_0 g(0 ) \eta_1\end{aligned}\ ] ] differentiating with respect to @xmath95 one gets : @xmath732 with the boundary condition @xmath733 .
the solution of this confluent hypergeometric equation that is well behaved solution at infinity ( i.e not growing exponentially ) reads : @xmath734 the boundary condition at @xmath536 then leads to the following equation for the exponent @xmath11 governing the scaling of the m variable @xmath716 , as a function of @xmath735 : @xmath736 using functional relations of the confluent hypergeometric functions @xmath737 , we finally obtain that the fraction of sites that have never been crossed by any particle in the effective dynamics decays as @xmath738 where the persistence exponent @xmath739 is solution of the equation @xmath740 we introduce the bond variable @xmath741 which is the number of sites on the bond which have been crossed exactly @xmath2 times by a particle ( any non empty state ) in the effective dynamics ( i.e by a thermally averaged trajectory ) .
it satifies upon decimation of bond @xmath484 with the same conventions as above : @xmath742 introduces the generating function @xmath743 , the rule becomes : @xmath744 which , for fixed @xmath510 , is the same rule as above with now @xmath745 and @xmath746 . the conditions ( [ condition2 ] ) become @xmath747 and thus using again @xmath727 , these conditions are satisfied with @xmath748 and @xmath749 . , i.e. we only need to perform the substitution @xmath750 in the previous solution to obtain the equation for the exponent @xmath751 governing the scaling of the ratio @xmath752 : @xmath753 the probability that a given point has been visited by @xmath2 thermally averaged trajectories up to time @xmath8 , is thus obtained in the rescaled variable @xmath754 as : @xmath755 it decays with the exponent @xmath756 obtained through the legendre transform @xmath757 where @xmath758 is the solution of @xmath759
. one can compute simply the value @xmath760 that @xmath469 takes with probability one as @xmath470 .
it is given by @xmath761 .
this gives @xmath762 , and thus differentiating ( [ thetaz ] ) with respect to @xmath510 and taking @xmath763 we finally get @xmath764 where we have used the notations @xmath765 and @xmath766 .
it turns out to be useful to introduce the notion of associative processes " : these are processes such that the outcome of a sequence of reactions does not depend on the order in which it was performed , i.e. such that the rates @xmath126 satisfy : for all @xmath768 ( contraction over @xmath300 and @xmath769 is implied ) .
this means that the probability of @xmath770 is identical to the probability of @xmath771 ( @xmath772 denotes the result of the reaction of @xmath773 and @xmath546 ) . for example , the process defined in ( [ potts ] ) is associative .
the bond rg equation ( [ rgbonds ] ) can in fact be interpreted to characterize the following modified reaction diffusion process , that we call the bond - reaction diffusion process " : one associates to the bottom of each bond ( i.e the point of lowest energy ) a specie in one of the possible `` states '' and define probability distributions @xmath778 for the bonds .
we consider two consecutive valleys made with the bonds ( 1,2 ) and ( 3,4 )
. initially the bonds ( 1,2,3,4 ) respectively contain the species @xmath779 .
upon decimation of bond ( 2 ) , the bond diffusion process is defined as follows in three steps ( i ) first the two species @xmath780 on bonds ( 1,2 ) react to give another state @xmath781 with the rates @xmath782 ( ii ) the new specie @xmath781 diffuses towards the bottom of the bond 3 .
( iii ) the species @xmath783 react at the bottom of bond 3 to give a new species @xmath784 with the rates @xmath785 ) . for comparison
, it is useful to recall the corresponding real dynamics with valleys : initially , the valley @xmath479 contains some species @xmath781 , the valley @xmath480 some species @xmath786 . upon decimation of bond
( 2 ) , the species @xmath781 diffuses towards the bottom of the valley @xmath480 and reacts there with @xmath786 to give @xmath17 with probability @xmath787 .
thus , in the end , the physical content ( the specie ) of the renormalized valley for the bond - diffusion process is @xmath17 with probability @xmath788 , whereas in the original valley process , the final result is @xmath17 with probability @xmath789 .
the two descriptions are thus equivalent in that sense only if the rates satisfy the associativity condition ( [ associative ] ) .
bray , b. derrida and c. godreche europhys .
* 27 * 175 ( 1994 ) .
a.d . rutenberg and a.j .
bray phys .
e * 50 * 1900 ( 1994 ) .
bray and b. derrida phys .
e * 51 * r1633 ( 1995 ) .
b. derrida , c. godreche and i. yekutieli europhys .
* 12 * 385 ( 1990 ) and phys . rev . a * 44 * 6241 ( 1991 ) .
the question of the extrapolation to @xmath36 models , via a @xmath790 expansion is subtle , as can be seen already for the single particle diffusion . within the context of single particle diffusion with short range gaussian disorder , there are in @xmath34 _ three _ distinct rg fixed points @xcite : ( i ) purely divergence - free ( hydrodynamic ) flows , ( ii ) purely potential disorder ( iii ) mixed disorder . since only pure potential disorder can exist in @xmath36 , the most natural approach to describe the sinai model in @xmath36 would be to extrapolate while remaining within the class of potential disorder models ( ii ) .
this is delicate however , as discussed in @xcite , as models ( ii ) have a vanishing beta function in @xmath34 : they are described by a line of fixed points in @xmath34 while presumably by a strong disorder ( non perturbative ) fixed point for any @xmath27 .
another possiblity would be to extrapolating down to @xmath36 starting from the mixed disorder fixed point ( iii ) ( as was done e.g. in ref .
@xcite ) as it is natural to assume that as @xmath791 the fixed points ( ii ) and ( iii ) merge .
however , since ( ii ) and ( iii ) have a completely different physics and symmetries in @xmath34 , one may expect that , although workable , this expansion would be rather poor , as can be seen already as @xmath792 while @xmath793 in @xmath36 .
the increase of reaction rate observed at ( iii ) in @xmath34 in ref .
@xcite ( while we prove the opposite effect for sinai disorder in @xmath36 ) is evidence of the different physics between ( iii ) and ( ii ) .
finally , concerning the random barrier or trap models ( with broad distributions of waiting times ) in @xmath36 @xcite , they are truly a different universality class and we do not see any reason why they could be reached by extrapolating gaussian disorder in @xmath34 .
the mapping extends to random bonds , each walker ( domain wall ) seeing a potential equal to the local bond strength @xmath794 .
diffusion of the potts domains in a sinai landscape then would correspond to glauber dynamics in a potts model with _ correlated _ bond randomness ( of correlator growing proportional to the distance ) .
in addition to not being very physical , such randomness would also lead - at large enough scale - to sign changes in bonds , invalidating the assumption @xmath795 necessary to treat the glauber dynamics by the present method .
the process ( [ potts ] ) however , remains interesting and fully well defined , as a reaction diffusion problem in presence of disorder , and results can be compared to the non trivial ones obtained for the pure case .
this is thus the point of view that we choose in this paper . keeping in mind the cautionary remarks in @xcite , we can still attempt to compare with the @xmath790 expansion formula ( 14 ) from @xcite , which reads @xmath797 and leads to @xmath798 by setting @xmath799 . restoring the microscopic time unit @xmath108 ,
one notes that the coefficient @xmath800 of the @xmath337 correction in @xmath104 can be set to @xmath327 by a redefinition of the microscopic time scale @xmath108 @xcite .
this redefinition does not affect the formula for the @xmath801 as it corresponds to the landscape eigenvector with @xmath802 .
the case of a non diagonalisable block of degenerate eigenvalues also deserves a separate study , not performed here .
it is likely to lead to additional logarithmic corrections ( @xmath320 and powers of it ) to the large time behaviour . within this toy model
there will be corrections due to rare events which split the thermal packet between distants wells .
for instance events ( a ) in fig . 7 of @xcite tend to average the disorder in the @xmath126 .
however they occur with probability at most of order @xmath337 and should not affect the leading behaviour .
a detailed study of these corrections goes beyond this paper .
|
we study a large class of one dimensional reaction diffusion models with quenched disorder using a real space renormalization group method ( rsrg ) which yields exact results at large time .
particles ( e.g. of several species ) undergo diffusion with random local bias ( sinai model ) and may react upon meeting .
we obtain a detailed description of the asymptotic states ( i.e attractive fixed points of the rsrg ) , such as the large time decay of the density of each specie , their associated universal amplitudes , and the spatial distribution of particles .
we also derive the spectrum of non trivial exponents which characterize the convergence towards the asymptotic states . for reactions which lead to several possible asymptotic states separated by unstable fixed points
, we analyze the dynamical phase diagram and obtain the critical exponents characterizing the transitions .
we also obtain a detailed characterization of the persistence properties for single particles as well as more complex patterns .
we compute the decay exponents for the probability of no crossing of a given point by , respectively , the single particle trajectories ( @xmath0 ) or the thermally averaged packets ( @xmath1 ) .
the generalized persistence exponents associated to @xmath2 crossings are also obtained .
specifying to the process @xmath3 or @xmath4 with probabilities @xmath5 , we compute exactly the exponents @xmath6 and @xmath7 characterizing the survival up to time @xmath8 of a domain without any merging or with mergings respectively , and the exponents @xmath9 and @xmath10 characterizing the survival up to time @xmath8 of a particle @xmath4 without any coalescence or with coalescences respectively .
@xmath1 , @xmath11 and @xmath12 obey hypergeometric equations and are numerically surprisingly close to pure system exponents ( though associated to a completely different diffusion length ) .
the effect of additional disorder in the reaction rates , as well as some open questions , are also discussed .
1.5 cm -0.5 cm
|
one of the most intriguing questions of astrophysics and astronomy is the origin of the elements in the early universe , and how this relates to the first formation of structure .
about 200,000 years after the big bang the epoch of structure formation emerged and the first stars were born from the initial , pristine baryonic matter . without any elements heavier than helium to provide cooling , the first stars that formed from the baryonic matter trapped in the emerging mini - dark matter halos were probably very massive , greater than 30@xmath0 ( e.g. * ? ? ?
these massive stars burned through their available fuel in about one to two million years , exploded as supernovae , and dispersed the first elements heavier than helium into the nascent universe , or collapsed into black holes .
these first events of stellar evolution influenced their early universe neighborhood , and determined under which conditions and with which initial abundance low - mass stars with masses like the sun eventually formed .
these take about 100 to 1000 times longer to form than more massive stars .
thus , by the time the first low - mass stars formed there was already a small amount of heavier elements present .
such low - mass stars are really the second stars , stars which formed from a non - primordial abundance distribution .
these second stars are important because while the massive first stars have long since vanished , some second stars are still around .
their importance lies in their capacity as carriers of the information about the formation and evolution of those long gone first generations of stars .
the second stars are also important because of the nuclear production they are capable of by themselves ( for example the first s - process in the second stars ) .
observations of such second stars are now emerging in increasing quantity and with high - resolution spectroscopic abundance determination @xcite .
there are a number of strategies to find ultra and extremely metal - poor stars .
the most effective appears to be spectroscopic surveys .
most of the emp and ump stars have been discovered in the hk survey @xcite and the hamburg / eso survey ( he s * ? ? ?
* ) . here ,
i designate somewhat arbitrarily ultra metal - poor ( ump ) stars as those with @xmath1 } } } \leq -3.5 $ ] ( @xmath2 ) and extremely metal - poor ( emp ) stars as those with @xmath1 } } } \leq -1.8 $ ] ( @xmath3 ) .
emp stars include the most metal - poor globular clusters .
ump stars are rare and hard to find in the spectroscopic surveys .
only approximately a dozen have been reported so far , with the present record holder he 0107 - 5240 at @xmath1 } } } = -5.4 $ ] @xcite . at and below this metallicity spectral features
are so weak that even the 10 m class telescopes are barely suitable for the job .
the formation mode of the second stars depends on a complex history of events , each of which may leave signatures in the observed abundances of emp and in particular the ump stars .
these events include the ionizing radiation of the very first primordial stars and the expansion of the associated h ii regions , their effect on possibly primordial star formation in neighboring mini - dark matter halos , as well as possibly the metal enrichment from a pair - instability supernova , depending on the mass of the first primordial stars .
these processes and the cosmological assumption that fix the underlying dark - matter structure will eventually determine the abundance distribution of the emp and ump stars now discovered and analyzed .
the most recent studies confirm that the very first primordial stars are massive @xcite .
however in these calculations the effect of the ionization from star formation in nearby mini - dark matter halos is taken into account , and it is found that the following generation of primordial stars may include intermediate mass stars .
the most metal - poor stars are potentially an extremely powerful tool to study star formation and evolution in the early universe @xcite .
these stars with metallicities reaching in excess of 5 orders of magnitude in metallicity below the solar metallicity often show abundance distributions that are very different from the solar abundance distribution .
the solar abundance distribution is the result of a long galactic chemical evolution , reflecting the cumulative result of mixing and nuclear processing of many generations of stars of a wide range of masses .
the most metal - poor stars are very old and have formed very early , at the onset of galactic chemical evolution .
it is therefore reasonable to assume that the abundance pattern in these second stars are the result of only few different sources and corresponding stellar nuclear production sites .
the goal is to identify these nuclear production sites and the mechanisms and processes through which the corresponding stellar ejecta have been brought together in the star formation cloud of the second stars .
an important part of the problem is to account for self - pollution or external pollution events that may have altered the stellar abundances between the time of second star formation and observation .
as the number of emp and ump stars with detailed abundance continuously grows one can discern certain abundance patterns .
a very unusual sub - group are the carbon - rich emp and ump stars .
observations indicate that many of these stars are polluted by asymptotic giant branch ( agb ) binary companions .
this implies that reliable models of agb stars including their heavy element production through the s - process at extremely low metallicity are required to understand these stars , and to disentangle the agb contribution from other sources that have contributed to the observed abundance patterns . in the following section
the carbon - rich emp and ump stars will be discussed in more detail . in [ sec : improve ] current models of emp intermediate mass stars and their uncertainties are presented .
such models have important applications for the emerging field of near - field cosmology , which uses galactic and extra - galactic stellar abundances to reconstruct the processes that lead to the formation of our and other galaxies ( [ sec : nfc ] ) .
then , ongoing work both observational and theoretical to determine the nucleosynthesis of nitrogen at extremely low metallicity is described ( [ sec : nitrogen ] ) and [ sec : hingba ] covers the hydrogen - ingestion flash in born - again giants , because this type of events may be important for emp agb stars as well , and may indeed be a major source of nitrogen in these stars .
[ sec : spr ] deals with the , and how it can be used as a diagnostic tool to improve the physics understanding of stellar mixing
. finally , and before some forward lookgin concluding remarks ( [ sec : concl ] ) the importance of nuclear physics input for the question of c - star formation will be discussed in [ sec : enrt ] .
about @xmath4 of all emp and ump stars show conspicuous enrichment of the cno elements , most notably c. for example , cs 29497 - 030 @xcite has @xmath1 } } } = -2.9 $ ] , @xmath5 } } } = 2.4 $ ] , @xmath6 } } } = 1.9 $ ] , @xmath7 } } } = 1.7 $ ] , and @xmath8 } } } = 0.5 $ ] .
others , like lp 625 - 44 ( @xmath1 } } } = -2.7 $ ] * ? ? ? * ) or cs 22942 - 019 ( @xmath1 } } } = -2.7 $ ] * ? ? ?
* ) , have a similar overabundance of c , but smaller n ( @xmath6 } } } = 1.0 $ ] for lp 625 - 44 ) and larger na ( @xmath8 } } } = 1.8 $ ] for lp 625 - 44 ) .
many of the heavier elements , like ti , cr , mn or zn in cs 29497 - 030 are either not overabundant or rather somewhat underabundant compared to the solar distribution .
large overabundances of n and na are not predicted by standard models of massive stars , neither for pop iii @xcite nor at larger metallicity @xcite
. primary production of n requires to expose the he - burning product c to h - burning again .
such a condition is encountered in massive agb stars during hot - bottom burning @xcite .
na is made in c - burning in massive stars but its final abundance is sensitive to the neutron excess , and thus scales less than linear with metallicity . if primary @xmath9 is exposed to he - burning it will lead to the production of @xmath10 .
either an additional proton collected in hot - bottom burning , or a neutron from the @xmath11 reaction in the he - shell flash of agb stars can then lead to primary @xmath12 production .
this process leads to significant na overabundances in standard agb stellar models of very low or zero metallicity @xcite .
the overabundances of c , n and na are not the only indication that the abundances in c - rich emp stars are at least partly caused by agb stellar evolution .
all of the objects mentioned above , and in fact most c - rich emp stars show in some cases significant overabundances of the s - process elements .
in addition they reveal their binarity through radial velocity variations , corresponding to binary periods of a few days in the case of he 0024 - 2523 @xcite to more typically one to a dozen years .
in fact , in their statistical analysis @xcite find that observations of radial velocities of cemp - s stars ( c - rich emp stars with s - process signature ) obtained to date is consistent with a @xmath13 binarity rate of cemp - s stars .
he 0107 - 5420 is currently the intrinsically most metal - poor star with @xmath1 } } } = -5.3 $ ] .
however , the total metal abundance in terms of elements heavier than he is quite large . in particular
the cno elements are significantly overabundant compared to the solar abundance ratios , with @xmath5 } } } = + 4.0 $ ] , @xmath6 } } } = + 2.4 $ ] @xcite , and @xmath7 } } } = + 2.3 $ ] @xcite .
the large overabundance alone indicate that only a few nuclear sources were involved in creating this abundance pattern . together with the low fe abundance the large cno abundances require a primary production of the cno elements . a primary nucleosynthesis production chain in a star is based only on the primordial elements h and he .
a secondary production in contrast requires heavier elements already to be present in the initial element mix at the time of star formation .
for example , standard models of massive stars predict a secondary production of n from cno cycling of c and o that is initially present from earlier generations of stars .
he 0107 - 5420 is a good example to discuss the possible nuclear production sites that may be important at the lowest metallicities
. certainly the production of the first massive pop iii stars plays an important role . @xcite
have discussed the observed abundance pattern within the framework of mixing and fallback of the ejecta during the explosion of supernovae that eventlually form a black hole .
they show that the mixing and fallback parameters can be chosen in a way that accounts for much of the observed pattern .
for example , their model correctly reflects the large observed fe / ni ratio , and the large c / fe and c / mg ratios .
the model does predict a significant n overabundance but it quantitatively falls short of matching the observed value .
their model predicts the charactereistic odd - even effect that reflects the small ( or missing ) neutron - excess of the primordial nuclear fuel .
one of the observed species that hints that he 0107 - 5420 may not have such a pronounced odd - even pattern is na .
the observerved overabundance is @xmath8 } } } = 0.8 $ ] , more than 1.5 dex above the model prediction from the mixing and fallback supernova .
unfortunately , for the heavy trans - iron elements only upper limits could be derived so far , which prohibits definite access to the s- and r - process abundances in he 0107 - 5420 . at this point na and to a lesser extent n are the major indicators that some additional sources are part of the nuclear production site inventory that caused the abundance pattern in this most metal - poor star . without better guidance from ab - initio coupled structure and star formation simulations such as shown in fig.[fig : dmhalo ] several scenarios have to be considered .
material from the first supernova could be ejected from the corresponding halo and injected into a neighbouring halo .
this metal enrichment could lead to the formation of an intermediate mass star , for example a very massive agb star , maybe even a super - agb star that eventually ends in a onemg core - collapse supernova .
as shown in
[ sec : improve ] a contribution from such stars can account for both na as well as n , but more detailed models of this particular scenario need to be done for quantitative comparison . a low - mass star like he 0107 - 5420
could then form from the combined nuclear production of these two sources , a roughly @xmath14 black hole forming supernova and a massive agb or super - agb star . alternatively the low - mass star forms from the ejecta of just one supernova , and this initial abundance pattern is modified by self - polution .
he 0107 - 5420 is in fact a giant on the first ascent .
observations , in particular of globular cluster red giants show that mixing and nuclear processing of c into n below the deep convective envelope lead to secondary n production @xcite .
this could explain the quantitative discrepancy between the observed n abundance and the model by @xcite .
however , the rather large carbon isotopic ratio @xmath15 implies that self - polution can not account for the observed na abundance , in particular assuming the absence of @xmath10which is a secondary nucleosynthesis product in massive stars .
this leads to the external polution scenario that is frequently considered in this context . as mentioned above many cemp stars
are in fact in binary systems , and the abundance signatures seen in these c - rich very metal - poor stars could be in part the result of mass transfer from the agb progenitor of a white dwarf companion . for he 0107 - 5420 @xcite discuss this possibility in detail .
for the evolution of primordial agb stars they consider the possibility of the h - ingestion flash , a process in which the he - shell flash convection zone reaches outward into the h - rich envelope , resulting in a peculiar nucleosynthesis and mixing regime .
such events are known at solar metallicity to cause a significant fraction of young white dwarfs to evolve for a short period of time back to the agb ( born - again evolution , [ sec : hingba ] ) , and the most prominent representative of this class of objects is sakurai s object @xcite .
as discussed in the previous section the evolution of low- and intermediate mass stars at zero or extremely or ultra metal - poor abundance is essential for studying the most metal - poor stars and their cosmological origin and environment . this section deals first with the evolution of agb stars that may have formed from big bang material . a possibly important evolutionary phase -
the h - ingestion flash triggered by the he - shell flash - has been observed in models of pop iii and even in models of non - zero but ultra - low metal content ( [ sec : hflash ] ) . finally the evolution and yields of extremely metal - poor tp - agb stars is described .
the evolution of pop iii massive stars is discussed , for example , by @xcite .
simulations show that many of the overall properties of pop iii intermediate mass stellar models @xcite probably apply to ultra metal - poor stars as well . at @xmath16
the initial thermal pulses may show peculiar convective mixing events at the core - envelope boundary during or after the actual he - shell flash .
but eventually the models enter a phase of rather normal thermal pulse agb evolution , with regular thermal pulse cycles , third dredge - up , hot - bottom burning and mass loss . many uncertainties of the pop iii stellar evolution is related to the possibility of flash - burning that is not yet well understood ( [ sec : hflash ] ) .
@xmath16 stellar evolution is different from evolution at higher metal content because cno catalytic material for h - burning is initially absent .
burning hydrogen via the pp - chain requires a higher temperature than h - burning via the cno cycle .
eventually the triple-@xmath17 process will provide some c , and a mass fraction as low as @xmath18 is sufficient to switch h - core burning to cno cycling . during he - core burning
some c is produced outside the convective core and a tail of carbon reaches from the convective core out toward the h - shell .
after the end of he - core burning , and depending on the initial mass , this carbon and some nitrogen is mixed into the envelope . according to @xcite this second dredge - up raises the envelope c - abundance above @xmath19 for @xmath20 .
for such a c - abundance the h - shell is fully supported by cno cycling . as a result
the thermal pulse agb evolution is like that of the ultra metal - poor cases .
models with initial masses below @xmath21 show a peculiarity that is unknown in more metal - rich models .
after an initial series of weak thermal pulses the h - shell forms a convection zone when it re - ignites after the he - shell flash @xcite .
the lower boundary of this convection zone is highly unstable , because it coincides with the opacity discontinuity that marks the core - envelope interface .
even small amounts of mixing will drive flash - like burning and deep mixing , as protons enter a @xmath22-rich zone .
it seems that the details of the evolution of the h - shell convection zone is not important as this one - time event leads to deep dredge - up and enrichment of the envelope with a sufficient amount to support regular cno cylce burning in the h - shell thereafter .
accordingly , these lower - mass cases will evolve like extremely metal - poor stars too , and the @xmath16 models discussed here can be used as proxies for what is in reality born as an ultra metal - poor star .
the effect of rotationally induced mixing processes may alter the evolution of low- and intermediate - mass stars and very low metallicity even before the the agb and lead to a qualitatively different thermal pulse evolution . @xcite
describe models for metallicity @xmath23 in which @xmath22 and @xmath24 is mixed out of the he - burning core and up to the location of the h - shell .
this leads to an increase of cno material , in particular @xmath9 in the envelope that could result in a rather normal thermal pulse evolution without any peculiar hydrogen convection zone . whether the rotationally induced
mixing processes before the agb or the subsequent thermal pulse agb evolution with the associated dredge - up events dominate the final cno yields is not yet clear .
the difference between a real pop iii and an extremely metal - poor thermal pulse agb star is the initial absence of fe and other elements heavier than al in the pop iii star .
this does not preclude that a pop iii agb stars can not generate -elements by the slow neutron capture process .
@xcite have used the -framework established for solar - like metallicities ( [ sec : spr ] ) and ran network calculations with an initial abundance appropriate for the interior of a @xmath16 tp - agb star .
they used the models of @xcite as stellar evolution input .
the major uncertainty was the mixing for the formation of the neutron donor species @xmath25 , and the unknown feedback of such mixing to the thermodynamic structure evolution . @xcite and @xcite have shown that this feedback is likely very important . nevertheless , the -models for @xmath16 are instructive , because they show that if a sufficient neutron source is available , the can be based on seed nuclei with lower mass number than fe , for example c. c is produced in a primary mode in tp - agb stars .
for such a @xmath16 the fe / ni ratio should be markedly sub - solar , approximately close to the quasi steady - state value of @xmath26 , that is given by the ratio of the neutron cross sections of ni and fe ( the isotopic minimum for each element ) .
he 0107 - 5240 with @xmath1}}}=-5.3 $ ] @xcite shows a larger than solar fe / ni ratio , which is evidence for the contribution of supernovae to the abundace distribution of this star .
the peculiar h - convective episode described in the previous section was observed in agb models with of @xmath27 and @xmath16 by @xcite . during the following thermal pulse the he - shell flash convection zone reachea out into the h - rich envelope .
this leads to the h - ingestion flash ( hif ) .
protons are mixed on the convective time scale into the he - shell flash region which is with depth increasingly hot .
the hydrogen flash - burning leads to a separate convection zone as shown in fig.[fig : ff08348 ] .
convective h - burning in this layer is characterized by a large abundance of @xmath22 and protons .
the protons have been mixed into this region from the envelope . accordingly
the @xmath9 abundance in this layer is very large , up to @xmath28 by mass in a layer of @xmath29 .
the models show that after both the he - shell flash and the hif have subsided a deep dredge - up episode will mix this @xmath9-rich layer into the envelope . this may account for a substantial @xmath9 production and part of the observational pattern observed in emp stars that are believed to be poluted by emp agb stars ( [ sec : nitrogen ] ) .
a hif was also found in the @xmath30 , @xmath16 model , but @xcite did not find the hif in models of @xmath31 .
agb stars are highly non - linear systems .
the occurrence of the hif depends on many details of the models , including the physics of mixing , opacities , the numerical resolution and nuclear reaction rates .
small changes in any of these ingredients , that may go unnoticed during a more robust stellar evolution phase ( like the main - sequence ) , can be the deciding factor here whether a hif occurs or not .
currently , theoretical models do not agree if and in which initial mass and metallicity regime the hif occurs @xcite .
however , it is important to better understand this situation because the hif may play an important role in the in extremely metal - poor stars @xcite .
hydrogen ingested into the he - shell flash or pulse - driven convection zone ( pdcz ) may lead to @xmath25 production which could release neutrons . @xcite
propose that the abundance pattern of the currently most metal - poor star
he 0107 - 5420 is in part due to the nucleosynthesis in a thermal pulse agb star with hif .
the details of the nucleosynthesis in a hif depend sensitively on the physics of simultaneous rapid nuclear burning and convective mixing .
the hydrogen - ingestion flash can also be a source of lithium , distinctly different from conditions during hot - bottom burning @xcite .
the evolution of born - again stars , like sakurai s object , provide valuable additional constraints with regard to the hif in agb stars ( [ sec : hingba ] ) .
the very late thermal pulse is associated with a hif @xcite . by reproducing the evolution of stars like sakurai s object one can gain some confidence in models of the hif in metal - poor agb stars .
@xcite showed that models in which the convective hydrogen ingestion occurs with lower velocities than predicted by the mixing - length theory @xcite feature a faster born - again evolution than mlt models , quantitatively in agreement with observations .
the physical interpretation is that rapid nuclear burning on the convective time scale releases energy and adds buoyancy to down - flowing convective bubbles , leading to additional breaking of the plumes .
the temperature and time scales for nucleosynthesis during the hif would be significantly different .
this has not yet been applied to hif models at @xmath16 or extremely low metallicity . by definition
all yields of pop iii stars are primary , because the initial composition contains only big bang material .
nuclear production is secondary when nuclei heavier than h and he are already present in the initial abundance distribution and transformed in other species .
pop iii or emp agb stars can produce a large number of species in primary mode , including the cno elements , ne , na , mg and al .
the production of heavier elements depends on the availabilitity of neutrons .
some of these primary species , like oxygen , are usually not considered a product of agb evolution at moderate metal - deficiency or solar metallicity . however , the initial envelope mass fraction of oxygen for models with @xmath32 is @xmath33 .
@xmath24 in the intershell material ( cf .
fig.[fig : kipp ] ) that is dredged - up is primary and even at this low metallicity of the order @xmath34 by mass .
dredge - up of such material will enhance the envelope abundance significantly .
a set of low- and intermediate mass agb stellar evolution models with @xmath32 ( [ fe / h ] = -2.3 ) with detailed structure , nucleosynthesis and yield predictions has been presented by @xcite . the oxygen overabundance in these models is @xmath35 , depending on mass . the abundance evolution for c , n , o and na is shown in fig.[fig : nacno ] for two initial masses .
the evolution of c and n is qualitatively different in the two cases .
the lower mass sequence shows the increase of c and some o due to the repeated dredge - up events .
no significant amount of @xmath9 is dredged - up .
the result is a large c / n ratio .
the @xmath27 track shows the opposite behaviour .
the much lower c / n ratio is the result of hot - bottom burning @xcite .
c and o are dredged - up after the thermal pulse as in the lower mass case , but during the following quiescent intperpulse phase the envelope convection reaches into the region hot enough for h - burning and the envelope c is transformed into n. the initial - mass transition between hot - bottom burning models with large n abdundances and c dredge - up models without hot - bottom burning is very sharp .
this precludes the notion that emp stars with simultaneously large n and c overabundances , like cs 29497 - 030 with @xmath5}}}=2.4 $ ] and @xmath6}}}=1.9 $ ] @xcite , may be poluted by such agb stars in the transition regime where hot - bottom is only partially efficent ( see [ sec : nitrogen ] ) .
the signature of hot - bottom burning in fig.[fig : nacno ] is a gradual change of abundance between the steps caused by dredge - up . from this
it can be seen that na has a different nuclear production site depending on initial mass . in the @xmath30 case
na is produced in the he - shell flash , by n - captures on @xmath10 .
the neutrons come from the @xmath36 reaction .
@xmath12 is the dredged - up but not produced during the interpulse phase .
the neutron - heavy mg isotope @xmath37 is produced in a similar way . in the @xmath27 case
na is mainly produced in hot - bottom burning , by p - capture on dredged - up @xmath10 . in the previous section
it was argued that due to the large primary production of cno and many other species the @xmath16 model predictions would also apply approximately to stars with ultra - low metallicity as well .
neutron capture reactions play an important role , and need to be included in yield - predictions , together with a neutron - sink approximation for species not explicitely included in the network .
this has been done in a qualitatively similar way by both @xcite for the @xmath32 models and @xcite for their @xmath16 models , and the abundance evolution is quantitatively similar .
both model sets predict for non - hot - bottom burning cases large primary production of c and o , while the @xmath9 abundance does not change .
@xmath10 is produced in both cases in the he - shell flashes , and in both cases
the final @xmath10 exceeds that of @xmath9 .
the evolution of @xmath12 , @xmath38 and @xmath37 is qualitatively the same too .
in particular the ratio of these three isotopes is quantitatively the same in the last computed model of the @xmath16 sequence and the final agb model at @xmath32 .
this is in particular interesting as these low - mass agb models predict that @xmath39 , contrary to the signature of hot - bottom burning .
one of the largest uncertainties is the adopted mass loss .
while mass loss in agb stars of solar metallicity can be constraint observationally , this has not been possible for extremely low metallicity .
other uncertainties relate to mixing in emp agb stars .
@xcite has shown that convection induced extra - mixing , like exponential overshoot that is used to generate a @xmath25 pocket for the in higher - metallicity models , may lead to vigorous h - burning during the third dredge - up . depending on the efficiency of such mixing the third dredge - up
may turn into a flame - like burning front , leading to very deep core penetration .
this may significantly impact the formation and effectiveness of a @xmath25 pocket for the @xcite .
accurate stellar yields for intermediate mass stars are requested by another emerging field , near - field cosmology @xcite .
the baryon ( stellar ) halo of the milky way retains a fossil imprint of the merging history of the galaxy .
different merging components , like the infalling dwarf galaxies with a range of masses , have different star formation histories that translate into different abundance signatures of the member stars of these components .
the current surge of spectroscopic multi - object capabilities at large telescopes will likely significantly enhance the importance of this approach .
an example is the recent work by @xcite in which abundances of halo stars are compared with abundances of stars in satellites , dwarf galaxies trapped in the potential well of the galaxy . here
, the basic idea is that dwarfs are merging with a galaxy at different times at which point star formation and chemical evolution stops .
satellites that survive until the present day should show the signature of more evolved chemical evolution , for example including the elements associated with the long - lived low - mass stars .
in contrast , halo stars that are the dispersed members of dwarf galaxies that have merged at an earlier time in the evolution of the galaxy , have less evolved chemical evolution patterns , for example showing more clearly the patterns of nucleosynthesis in massive star evolution .
potentially , the implications that can be derived from comparison of halo stars and satellite members may include some fundamental questions of nucleosynthesis itself .
the data presented in @xcite show that @xmath40}}}$ ] is systematically smaller in stars belonging to satellites compared to their halo counterparts
. this can be explained in the @xmath41 cold dark matter model that predicts that most halo stars have formed in rather massive dwarf galaxies , which merged with the galaxy a long time ago @xcite .
these stars show the signature of truncated chemical evolution , dominated by the yields of supernovae type ii and their significant @xmath17-element contribution and moderate fe ejecta .
stars in present , less massive satellite galaxies show the signature of chemical evolution components that take more time , like the sn ia .
these events add fe but little @xmath17-elements , and their @xmath40}}}$ ] ratio is therefore smaller .
abundances of other elements could be compared too .
for example , @xcite compare among others the abundances of y , ba and eu in halo stars and satellite dwarf galaxies .
ba is typically considered an main - component element ( at least at @xmath1 } } } > -2 $ ] ) because the elemental solar abundance has a @xmath42 contribution @xcite .
only the smaller remaining fraction is made in the r - process .
as discussed above , the nuclear production site of the main - component of the s - process are low- and intermediate mass agb stars , in the initial mass range @xmath43 .
this component of galaxy chemical evolution needs even more time to contribute than the sn ia . and
indeed , [ ba / fe ] behaves differently from @xmath40}}}$ ] in halo stars and satellite stars .
[ ba / fe ] is on average higher in the satellite stars than in halo stars , with a considerable spread .
this is consistent with the framework of truncated chemical evolution of systems that were early disrupted in merger events , and with the understanding of stellar evolution and nucleosynthesis that @xmath17-elements are predominanty made in short - lived massive stars , while ba originates in rather old populations .
it is then extremely interesting to consider the r - process element eu , which in the solar abundance distribution has a very small contribution of only @xmath44 .
this element does not behave like @xmath17-elements . instead , [ eu / fe ] is on average the same in halo stars and dwarf galaxy stars , however with a larger spread in the latter . in order to remain in the proposed scenario of why abundances in the halo differ from those in the nearby dwarf galaxies one
would then have to assume that at least an important fraction of the r - process elements does not originate in sn ii , which eject their yields on a short time scale .
instead , one would have to assume that eu in the satellite stars comes from a source that releases the ejecta on a time scale comparable to or longer than the fe production in sn ia .
such sources could be the accretion induced collapse @xcite or the collapse of a super - agb star in the initial mass range @xmath45 , if the core mass grows to the chandrasekhar limit @xcite .
it would certainly go to far at this point to draw any further conclusion , for example in conjunction with the two r - porcess source model proposed by @xcite .
it is clear , that the full potential of using the abundances patterns of stars to reconstruct the formation of galaxies in their cosmological context can only be reached with detailed yield predictions including all masses and reaching down to extremely low metallicity .
y is another example that reinforces this notion . in the solar abundance distribution
it has a @xmath46 fraction from the main - component originating in low - mass stars .
however , it is also the termination point of the weak from massive stars , and it does have some r - porcess contribution . in the data presented by @xcite
it behaves with regard to the difference between halo and satellite stars like the @xmath17-elements , which in this context would imply an important contribution from massive stars .
nitrogen is the fifth most common element in the universe . in stellar evolution of very low metallicity
the primary production is of particular importance . strictly from a nucleosynthesis point of view nitrogen is always made in a well established sequence of events , involving first the triple-@xmath17 process that makes @xmath22 and then two subsequent proton captures that are part of the cn - cylce . in a stellar model sequence the problem is to identify the mixing processes that can account for this sequence of events . how can @xmath22 , that is made in the he - burning that requires higher temperatures , be brought into layers of lower temperature where h - burning still takes place .
or , how can protons be mixed down into the he - burning layers and be captured by @xmath22 to eventually form @xmath9 , but then not be exposed to @xmath17-captures that would convert @xmath9 into @xmath10 . because the primary production of nitrogen depends so much on stellar mixing , this element assumes a key role in understanding stellar nucleosynthesis at extremely- and ultra - low metallicity .
nitrogen is certainly present in the most metal - poor stars in the milky way @xcite . according to standard stellar evolution and nucleosynthesis models primary
n comes from intermediate mass ( i m ) agb stars @xcite , with perhaps a small contribution from rotating massive stars @xcite . in agb
stars the repeating sequence of thermal pulses induce convective mixing events that lead to the prodution of n. c is made in the he - shell flash .
dredge - up after the flash mixes that c into the envelope , and in massive agb stars the c in the envelope is then transformed into n. the low [ n/@xmath17 ] abundances observed in some damped lyman-@xmath17 systems @xcite have prompted suggestions that the imf may be biased in favor of massive stars in some systems @xcite , or that these systems represent the earliest stages of chemical evolution of massive stars only @xcite .
observations of n and c abundances in c - rich emp stars with -signature provide another way to study the n production at low metallicity .
about 30@xmath48 of all extremely metal - poor stars ( [ fe / h ] < -2 ) are strongly c- and to a lesser extent n - enhanced ( [ sec : cemp ] ) .
in particular the overabundance of c and of n in addition to their s - process signature has lead to the assumption that the agb star progenitor of the current white dwarf companion to the cemp - s star is responsible for the observed abundance pattern .
because of the established binary nature and the -signature the cemp - s stars are assumed to be poluted by the individual agb stars that are the progenitors of their present white dwarfs companions . in fig.[fig :
nitrogen ] the [ c / n ] and [ c / fe ] ratios of literature data are shown , together with the agb model prediction of @xmath49 - @xmath21 tracks .
the literature data show systematically lower [ c / n ] ratios than what is expected by @xmath49 or @xmath50 models that dredge - up c , but do not produce n. this poses the question of the primary origin of the n in theses stars .
models of initial masses between 4 and 6 feature efficient hot - bottom burning ( hbb ) which turns most dredged - up primary carbon into nitrogen resulting in [ c / n]<-1 and [ c / fe]<1.5 .
however , none of the literature data seem to show the very low [ c / n ] ratio that would be expected for an emp star that happened to have an agb companion in the @xmath51 initial mass range .
we have carried out observations of emp targets with 0.5<[c / fe]<1 using the ch and the nh band for abundance determination @xcite .
specifically we wanted to find out whether the paucity of emp stars with [ c / n]<-1 is a selection effect or a systematic observational bias imposed by the abundance indicators employed in previous studies .
while c - strong stars can be identified by the g - band at @xmath52 , the strong nh band at @xmath53is never observed in the medium - resolution surveys that provide the targets for detailed abundance studies ( e.g. * ? ? ?
only the cn bands at @xmath54and @xmath55are included , and cn lines are not strong in n - rich stars unless c is also enhanced by a large amount .
therefore , studies of c and n abundances have favored c - rich stars that are easily identified , and may have missed stars that have @xmath5}}}\sim 0 - 1 $ ] , but are more rich in n. thus the literature data in fig.[fig : nitrogen ] , drawn from high - resolution follow - up of the medium - resolution candidates , may be the result of a strong observational bias against finding n - rich stars . in order to overcome this bias we obtained spectra for 18 new stars . in fig.[fig : nitrogen ] , we plot the preliminary estimates of the [ c / n ] ratios of 18 stars , based on observations of the nh bands during 6 nights at ctio / kpno
. we did not find any stars with low [ c / n ] , and the analysis of additional data is underway to put our findings on a more robust statistical basis .
this leaves us with two important open questions : ( 1 ) where are the emp stars polluted by massive agb stars ? and ( 2 ) where does the n in the cemp stars come from that we do observe ?
while we do not have any idea at this point of what the answer to the first question could be , there are a number of possibilities to address the absence of primary n. these include rotationally induced mixing before the agb phase ( [ sec : z00 ] ) , the h - ingestion flash ( [ sec : hflash ] ) , and extra - mixing in agb stars @xcite as envoked in models of rgb stars to account for abundance anomalies in globular cluster member stars @xcite . in any of these cases
the interpretation of n abundances in metal - poor systems of all kinds may have to be re - evaluated .
nuclear production in the agb stellar interior can become observable at a later time in the evolution , during the post - agb phase , on the surface of those central stars of planetary nebulae that become h - deficient . here
h - deficient means in most cases that the mass fraction of h as observed in the stellar spectrum is less than @xmath56 .
this leads to additional important constraints on the evolution of agb stars , which is in particular important to study agb evolution at extremely low metallicitiy .
h - deficient central stars of planetary nebulae ( cspn ) as well as very young white dwarfs have been analysed in detail ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and new important information about the abundances of these stars is yet to be revealed .
stellar evolution predicts that about @xmath57 of all post - agb stars will eventually loose all their small remaining h - rich envelope mass of the order @xmath58 and expose the bare h - free cores .
the carbon , oxygen and helium rich nature of these objects is evident from their wolf - rayet or pg1159 type emission line spectra , and reflects nuclear processed material that has been built up during the stars progenitor evolution .
although there is considerable spread in the observed abundances of these stars , the pattern can be summarized in mass fraction as he @xmath59 c @xmath60 and @xmath61 .
these h - deficient cspn are important for agb evolution modeling , because they provide a unique opportunity to study directly the nuclear processing shells in agb stars . in order to explain the evolutionary origin @xcite introduced the born - again evolution scenario .
the star evolves off the agb , and becomes a hydrogen - rich central star and eventually a very hot , young white dwarf .
however , the he - shell may still be capable to ignite a late he - shell flash , and in that case the star retraces its evolution in the hrd , back into the giant region ( fig.[fig : hrdsak ] ) . as a result of hydrogen - ingestion into the he - convection and rapid burning , or by mixing from the emerging convective envelope ( or because of both )
the surface abundance of such born - again stars will be extremely h - deficient or even h - free .
stellar models of this evolutionary origin scenario connect the surface abundance of the wolf - rayet central stars and the pg1159 stars with the intershell abundance of the progenitor agb star .
the intershell layer between the he- and the h - shell is well mixed during each he - shell flash .
this zone contains the main nuclear production site of the progenitor agb star , and the abundance of this zone reflects contributions from both he- and h - shell burning ( fig.[fig : kipp ] ) . as a result of the born - again evolution
these layers become visible at the surface of the resulting h - deficient cspn .
the initial models of the h - deficient central stars by iben and collaborators showed qualitatively that the born - again scenario could account for high he and c abundances , as these elements were abundant in the intershell of the agb progenitor model they used .
however , they could not account for the high observed oxygen abundance .
@xcite were the first to propose that the solution could be non - standard mixing during the agb evolution .
models with overshooting at the bottom of the pulse - driven convection zone do not only feature higher temperatures at the bottom of that layer ( [ sec : spr - mix ] ) , but they also show higher c and in particular o abundances in the pdcz .
subsequently , @xcite has explored in detail how the various abundances depend on the overshoot efficiency and other details .
in essence , overshooting brings agb intershell abundances of he , c and o in very good quantitative agreement with the observed abundances of wolf - rayet type central stars and pg1159 stars , in the framework of the born - again evolution .
more effort is needed to consolidate these constraints on the intershell abundance with the possibly tight upper limits on overshooting at the bottom of the pdcz that the may provide ( [ sec : spr - mix ] ) .
the connection of the surface abundances of the hot pg1159 stars and the progenitor agb intershell abundance has been reinforced recently by several new observational findings .
these include a substantial overabundance of ne @xcite , f abundances ranging from solar up to 250 times solar @xcite in good agreement with agb nucleosynthesis predictions by @xcite , and fe - deficiencies of at least 1 dex compared to solar @xcite , which may reflect the depeletion of fe in the agb intershell due to n - captures . in that case
the fe / ni ratio should be low , as in fact observed in sakurai s object .
this born - again star is the smoking gun of the very late thermal pulse scenario and links this evolutionary scenario with the h - deficient pg 1159 and wolf - rayet type central stars @xcite .
much of the observed abundance pattern of sakurai s object as well as the rapid evolutionary time scale has been reproduced quantitatively with stellar models @xcite .
similar to the pg 1159 stars , sakurai s object shows fe depletion of about 1dex compared to solar .
the low ratio of @xmath62 , shows that sakurai s object shows surface material that has been directly irradiated with neutrons @xcite .
new radio observations show that sakurai s object has started reheating again , on its second evolution into the cspn @xcite .
this marks a new phase of the evolution of this star , and confirms the concept of convective mixing efficiency modified by nuclear burning @xcite .
the is the origin of half of all elements heavier than iron @xcite .
it is also important for isotopic ratios and in some cases elemental abundances of lighter elements - in particular at extremely low metallicity .
the heavy elements are made by the through neutron captures that are slow compared to the competing @xmath63-decay .
starting from the abundant iron group elements it follows closely the valley of stability in the chart of isotopes .
it is characterized by neutron densities @xmath64 .
observations and theory agree that the nuclear production site of the main and strong component of the ( @xmath65 ) originates in low mass asymptotic giant branch ( agb ) stars .
the weak component below the first peak at @xmath66 is produced during he and c burning in massive stars .
here we deal only with the in agb stars .
the is important in the context of the second stars for two reasons .
first , the observed elemental abundance distribution of the trans - iron elements is at any given metallicitiy a mix of s- and r - process contribution . in order to disentangle the two contributions in very - low metallicity stars ,
the in this metallicity regime needs to be understood .
this may than enable progress in identifying the conditions for the r - process .
second , -branchings can be used to probe the physics for stellar mixing ( [ sec : spr - diag ] ) , which is particularly usefull when modeling the evolution of emp stars .
the in low - mass agb stars has two neutron sources .
the main source is @xmath25 , that forms via the @xmath67 reaction . at the end of the dredge - up phase ( fig.[fig : kipp ] ) after the he - shell flash , when the bottom of the h - rich convective envelope has penetrated into the @xmath22 rich intershell layer , partial mixing at this interface would create a thin layer providing simultaneously the required protons and @xmath22 .
the in the @xmath25 pocket is characterized by low neutron densities ( @xmath68 ) that last for several thousand years under radiative , convectively stable conditions during the quiescent interpulse phase .
the physics of mixing at the h/@xmath22 interface at the end of the third dredge - up phase has not yet been clearly identified ( see below ) .
most likely it is some type of convection induced mixing beyond the convection boundary .
the @xmath11 reaction requires the high temperatures that can be found at the bottom of the pulse - driven convection zone ( pdcz ) during the he - shell flash ( @xmath69 ) .
the neutrons are released with high density ( @xmath70 ) in a short burst @xcite .
these peak neutron densities are realised for only about a year , followed by a neutron density tail that lasts a few years , depending on the stellar model assumptions .
the current quantitative modeling of the uses the thermodynamic output from a stellar evolution calculation including mass loss as input for nucleosynthesis calculations with a complete network @xcite .
the post - processing accounts for both the @xmath25 neutron source as well as for the @xmath10 source , and mixes the different contributions according to the information provided by the stellar evolution calculations .
the free parameter of the model is the @xmath25 abundance in the @xmath25 pocket that is proportional to the neutron exposure that results from burning the @xmath25 in the @xmath71 reaction .
physical mixing processes which are responsible for bringing protons down from the envelope into the @xmath22-rich core to enable @xmath25 formation , are not explicitely included in this model .
observationally , models have to account for observed spread in observables that are related to the neutron exposure ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this spread is not only evident from stellar spectroscopy , but also from sic grain data , that indicate that a spread by a factor of five is necessary for the neutron exposure for a given mass and metallicity @xcite . currently , this spread is accounted for by a range of different cases in each of which the @xmath25 abundance in the pocket is assumed to be different @xcite .
however , the physics of the mixing that is associated with the range of neutron exposures has yet to be identified .
it is useful to distinguish -mixing for solar and moderately metal - poor stellar evolution , and the extremely and ultra metal - poor cases . in the first , the assumption that -mixing does not feedback strongly into the thermodynamic evolution is generally valid , in the latter it is generally not .
many more observational constraints exist for the solar and moderately metal - poor , in particular the isotopic information from the pre - solar sic grains .
mixing for the for the solar and moderately metal - poor has to satisfy two general constraints : ( 1 ) how is the partial mixing zone of h and @xmath22 generated that eventually forms the neutron source species @xmath25 , and ( 2 ) what is the origin of the observed spread in neutron exposures .
mixing for the @xmath25 pocket is probably related to the penetrative evolution of the bottom of the convective envelope during the third dredge - up .
possible mechanisms include exponential diffusive overshooting @xcite , mixing induced by rotation @xcite , and mixing by internal gravity waves @xcite .
each of these effectively leads to a continuously and quickly decreasing mixing efficiency from the h - rich convection zone into the radiative @xmath22-rich layer , and each of these will lead to the formation of two pockets which are overlapping @xcite . at low h/@xmath22 ratios h - burning is proton - limited , and protons will make @xmath25 via the @xmath72 reaction but no ( or little ) @xmath9 . at larger h/@xmath22 ratio a @xmath9pocket forms
the maximum abundance of both @xmath25 and @xmath9 depends on the @xmath22 abundance in the intershell . therefore , the conditions in the @xmath25 pocket are not independent of , for example , the mixing at the bottom of the pdcz .
@xcite derive the relationship @xmath73 , where @xmath74 is the intershell @xmath22 mass fraction and @xmath75 is the maximum neutron exposure reached in the @xmath25pocket .
the observations of h - deficient post - agb stars described in
[ sec : hingba ] require that the @xmath22 abundance in the pdcz of agb stars is about @xmath76 .
this value is reproduced in agb stellar evolution models with overshooting at the bottom of the pdcz , and about twice as large as in models without overshooting .
apart from the maximum amount of @xmath25 in the pocket the mixing process must produce a partial mixing layer of the right mass @xmath77 which eventually contains enriched material . in the exponential overshooting model of @xcite
the mixing coefficient is written as @xmath78 , where @xmath79 is the mixing - length theory mixing coefficient at the base of the convection zone , @xmath80 is the geometric distance to the convective boundary , @xmath81 is the pressure scale height at the convective boundary , and @xmath82 is the overshooting parameter . if applied to core convection @xmath83 reproduces the observed width of the main sequence .
@xcite finds that @xmath84 at the bottom of the convective envelope generates a large enough @xmath25 pocket .
however , the maximum neutron exposure in the @xmath25 pocket of the overshooting model is @xmath85 , while in the non - overshooting models this value is @xmath86 .
correspondingly @xcite obtained only negative values for the logarithmic ratio [ hs / ls ] , while the overshooting model with the larger neutron exposure predicts [ hs / ls]@xmath87 .
this compares to an observed range of @xmath88}}}<0.0 $ ] for stars of solar metallicity ( see * ? ? ?
* for a compilation of observational data ) .
models with overshooting at all convective boundaries can reproduce only the largest observed hs / ls ratios , indicating that the neutron exposure in the @xmath25 pocket in these models is at the maximum of the observationally bounded range .
overshooting alone is not able to account for all features of -mixing .
in particular there is no mechanism to acount for the spread in neutron exposures within the overshooting framework .
rotation , however , may induce a range of mixing efficiencies for a sample of stars with otherwise identical parameters .
models of rotating agb stars were presented by @xcite , and the was analyzed in detail by @xcite and @xcite .
the implementation of rotation for the agb models was the same as the one that had been used previously to construct rotating models of massive stars @xcite .
this implementation yields a @xmath25 pocket generated by shear mixing below the envelope convection base , however an order of magnitude smaller than what is needed in the partial mixing zone of a non - rotating model to reproduce the observed overproduction in stars .
the second important finding is that shear mixing which initially generates the small @xmath25 pocket , prevails throughout the interpulse phase , even when the base of the convection is receding in mass after h - shell burning has resumed .
when the dredge - up ends , the low - density slowly rotating convective envelope and the fast rotating compact radiative core are in contact and mixing is induced through shear at this location of large differential rotation .
this radial velocity gradient remains as a source for shear mixing at exactly the mass coordinate of the @xmath25 pocket with important consequence for the .
shear mixing during the interpulse phase swamps the @xmath25 pocket with @xmath9 from the pocket just above . by the time the temperature has reached about @xmath89 and @xmath25 starts to release neutrons via @xmath90 , @xmath9 is in fact more abundant than @xmath25 in all layers of the @xmath25 pocket .
@xmath9 is a very efficient neutron poison .
it has a very large @xmath91 rate and simply steels neutrons from the iron seed . as a result ,
the neutron exposure is only @xmath92 , about a factor of ten too small to generate the observed abundance distribution .
the detailed post - processing models of current rotating agb stars showed that they are not capable to account for the observed overabundances .
parametric models show that weaker shear mixing during the interpulse leads to a weaker poisoning effect .
for very small poisoning effects the neutron exposures are still large enough to reproduce some observations .
while neither rotation or overshooting alone provide the right amount of -mixing it is interesting to consider a combination of these .
in essence the idea is that overshooting would provide mixing for the formation for the @xmath25 pocket and for a larger @xmath22abundance in the pdcz to obtain a large neutron exposure in the @xmath25pocket .
then , shear mixing during the interpulse could add some poison and result in the observed spread .
magnetic fields have not yet been considered in agb stellar evolution .
however , one may assume that qualitatively magnetic fields will add coupling between the fast rotating core and the slowly rotating envelope and provide additional angular momentum transport .
models including this effect could result in smaller shear mixing than predict by the current rotating agb stellar models . it is interesting to note that such an effect of magnetic fields may also help to reconcile the predicted rotation rates of agb cores of @xmath93 with the rotation rate determinations of white dwarfs .
spectroscopic determinations of rotation rates of white dwarfs of spectral type da can not rule out such values , but most are also consistent with zero or very low rotation @xcite . however , asterioseismological measurements of wd rotation rates clearly yield smaller values in the range @xmath94 ( see references in * ? ? ? * ) .
the high precision information on the pre - solar meteoritic sic grains provide isotope ratio measurements that allow to probe the conditions at the nuclear production site in more detail .
one example is the highly temperature dependent nucleosynthesis triggered by the release of neutrons from @xmath10 at the bottom of the pdcz .
the temperature at the bottom of the pdcz correlates with the efficiency of extra mixing like overshooting at the bottom of this convection zone @xcite .
neutrons in the pdcz are generated by the @xmath11 reaction . for larger temperatures
the neutron density is higher .
isotopic ratios that enclose a branch point isotope in the path will be more neutron heavy for higher neutron densities .
an example is the @xmath95/@xmath96-ratio , that is set by the branching at the radioactive @xmath97 ( fig.[fig : zr ] ) . for low neutron densities
@xmath97 decays . for @xmath98 @xmath99
becomes significant , hence @xmath95 is produced .
if the temperature is larger the neutron density is larger and the @xmath95/@xmath96 ratio , which can be measured in sic grains , is larger as well .
@xcite have studied the measured isotopic ratios of mo and zr as well as sr and ba from sic grains .
they evaluate the sensitivity of their results in terms of nuclear reaction rate uncertainties .
all branchings that are activated by the @xmath10 neutron source depend on the still uncertain @xmath11 rate .
in addition , the @xmath95/@xmath96 ratio depends on the neutron cross - section of the unstable isotope @xmath97 .
like for almost all radioactive nuclei the @xmath100 rate of @xmath97 is not measured .
the theoretical estimates vary from maxwellian - averaged cross sections of @xmath101 @xcite to @xmath102 ( jendl-3.2 ) . in order to make full use of the potent method of using the as a diagnostic tool it is critical that n - cpature rates of the radioactive -branch point nuclides are measured .
different -branchings are sensititv to different mixing processes , including those that may be induced by rotation
. the as a diagnostic tool can provide information on mixing processes that are potentially relevant for the evolution of stars of all masses , including the progenitors of supernovae .
in particular the details of the initial model for a supernova calculations determines important properties of the explosion , like asymmetries , or the final fate as black hole or neutron star @xcite .
in order to understand c - rich extremely metal - poor ( cemp ) stars the formation of c - rich agb stars has to be understood in a quantitative way .
agb stars become c - rich because of the third dredge - up , which mixes c - rich material from the intershell into the envelope ( fig.[fig : kipp ] ) .
the evolution of c and o in agb stars and the problems related to modelling the third dredge - up are discussed in @xcite . in summary ,
the third dredge - up is now obtained in 1d stellar evolution calculations in sufficient amount and at sufficient low core mass .
these calculations take into account convection - induced mixing into the stable layers in a time- and depth dependent way , and use high numerical resolution .
uncertainty is introduced by the mixing length parameter @xcite .
understanding the dredge - up properties of agb stars is important , because the dredge - up dependnent yield predictions for low and intermediate mass stars enter models for galaxy chemical evolution .
agb stars serve as diagnostics for extragalactic populations , and for this purpose the conditions of the o - rich to c - rich transitions needs to be known .
c - rich giants are the brightest infrared population in extra - galactic systems .
finally , the envelope enrichment of agb stars with the s - process elements is intimately related to the dredge - up properties of the models .
recently , it has been shown that uncertainties in nuclear reaction rates propagate in a significant way into the dredge - up and thereby yield predictions .
@xcite calculated an extensive grid of @xmath30 , @xmath103 tracks for combinations of rates for the @xmath104 , the triple-@xmath17 and the @xmath105 rate within the errors given in the nacre compilation @xcite .
the main result was that dredge - up and c yields are larger for lower @xmath106 rate and for larger triple-@xmath17 rate .
the @xmath107 rate plays a less important role .
it was also found that nuclear physics work since the 1999 nacre compilation require a downward revision of the @xmath108 rate , by almost a factor of 2 .
fig.[fig : enrt ] shows the revised range for this rate and the yields from the calculations assuming the uncertainty range of the nuclear reaction rates .
extremely metal - poor stars are an emerging field of astrophysical and astronomical research , pushing the limits of observations , and theory and numerical simulation .
potentially , much can be learned about challenging questions of astrophysics .
how do galaxies like our milky way form ?
how did the first stars and their cosmological environment form and evolve ?
some exciting clues about the evolution of stars come from the smallest astrophysical bodies , pre - solar stardust extracted from primitive meteorites .
thus stellar evolution and nucleosynthesis connects nearby phenomena of planetary system formation with star formation and evolution in the earliest time of the universe as recorded in the element distribution patterns of the most metal - poor stars . in the future
the quantity and quality of spectroscopic data of stars , in particular the valuable most metal - poor stars , will increase dramatically , due to multi - object spectroscopy and large spectroscopic surveys .
it is already becoming clear that this data can be put to full use only with qualitatively and quantitatively improved simulations of nuclear production in stars , including low and intermediate mass and massive stars .
i would like to thank d. schoenberner for nominating me for the ludwig - biermann award , and for directing my interest to stellar evolution and nucleosynthesis as my _
doktorvater_. i am very grateful to n. langer and d. vandenberg for their continuing support .
i would also like to thank my colleagues at the theoretical astrophysics group ( t-6 ) at los alamos national laboratory , in particular c. fryer , a. heger , and f. timmes , as well as r.reifarth at lansce-3 , who contribute to a very stimulating atmosphere .
finally , i would like to thank a. font and b. oshea for stimulating discussions , that have contributed to some views expressed in this article .
this work was funded under the auspices of the u.s.dept . of energy , and supported by its contract w-7405-eng-36 to los alamos national laboratory . , w. , ando , h. , honda , s. , iye , m. , izumiura , h. , kajino , t. , kambe , e. , kawanomonoto , s. , noguchi , k. , okita , k. , sadakane , k. , sato , b. , shelton , i. , takada - hidai , m. , takeda , y. , watanabe , e. , & yoshida , m. 2002 , pasj , 54 , 427 , t. , bonifacio , p. , molaro , p. , cayrel , r. , spite , m. , spite , f. , plez , b. , andersen , j. , barbuy , b. , beers , t. c. , depagne , e. , hill , v. , franois , p. , nordstr " om , b. , & primas , f. 2004 , a&a , 413 , 1073
|
_ the ejecta of the first probably very massive stars polluted the big bang primordial element mix with the first heavier elements .
the resulting ultra metal - poor abundance distribution provided the initial conditions for the second stars of a wide range of initial masses reaching down to intermediate and low masses . the importance of these second stars for understanding the origin of the elements in the early universe are manifold . while the massive first stars have long vanished the second stars are still around and currently observed .
they are the carriers of the information about the first stars , but they are also capable of nuclear production themselves .
for example , in order to use ultra or extremely metal - poor stars as a probe for the r - process in the early universe a reliable model of the s - process in the second stars is needed .
eventually , the second stars may provide us with important clues on questions ranging from structure formation to how the stars actually make the elements , not only in the early but also in the present universe .
in particular the c - rich extremely metal - poor stars , most of which show the signature , are thought to be associated with chemical yields from the evolved giant phase of intermediate mass stars .
models of such agb stars at extremely low metallicity now exist , and comparison with observation show important discrepancies , for example with regard to the synthesis of nitrogen .
this may hint at burning and mixing aspects of extremely metal - poor evolved stars that are not yet included in the standard picture of evolution , as for example the hydrogen - ingestion flash .
the second stars of intermediate mass may have also played an important role in the formation of heavy elements that form through slow neutron capture reaction chains ( s - process ) .
comparison of models with observations reveal which aspects of the physics input and assumptions need to be improved .
the s - process is a particularly useful diagnostic tool for probing the physical processes that are responsible for the creation of elements in stars , like for example rotation . as new observational techniques and strategies continue to penetrate the field , for example the multi - object spectroscopy , or the future spectroscopic surveys
, the extremely metal - poor stars will play an increasingly important role to address some of the most fundamental and challenging , current questions of astronomy . _ * the second stars * + falk herwig + los alamos national laboratory + los alamos , theoretical astrophysics group , nm 87544 , usa + [email protected]
|
in this section , we show the derivation of the exact bethe - schwinger cavity perturbation formula , analogous to the approach by @xcite .
afterwards , we discuss how small approximations result in equation ( [ eq : bs ] ) of the main text .
we note , that the calculation of the different contributions to the eigenfrequency change in our simulations did not involve the approximated equation ( [ eq : bs ] ) , but the exact formula as shown in equation ( [ eq : bethe ] ) .
we consider the modes of an open cavity described by a spatial distribution of permittivity @xmath74 and permeability @xmath75 ( in the following , the spatial dependence of both is implicitly assumed ) .
the eigenmodes of the system are found by solving maxwell s equations in all of space in the absence of external drives .
this yields solutions of the form @xmath76 which satisfy maxwell s equations for some complex frequency @xmath77 where @xmath7 and @xmath78 denote the ( real ) resonance frequency and energy decay rate , respectively .
typically , we will consider systems where @xmath79 .
important relationships between @xmath80 and @xmath81 include @xmath82 we now consider that the cavity is perturbed , such that in a finite volume @xmath83 , the relative permittivity and permeability are changed to new values @xmath84 this perturbed system has different eigenmodes written as @xmath85 where @xmath86 and @xmath87 describe the spatial dependence and resonance frequency of the perturbed cavity mode .
defining @xmath88 , @xmath89 and @xmath90 yields @xmath91 combining the above equations for the curl of the fields gives @xmath92 we next take the dot product of @xmath93 and @xmath94 with the curls of @xmath39 and @xmath40 , respectively and rewrite both making use of the vector identity @xmath95 subtracting the obtained expressions gives us @xmath96 finally , we take the integral of both sides over a ( very large ) volume @xmath97 , and apply gauss s theorem to arrive at the _ bethe - schwinger equation _ : @xmath98 = -\left({\omega}+{\delta\omega}\right ) & \int_{{\delta v } } dv \left[{\epsilon_0}{\delta \epsilon}{{\mathbf{e}}_0}^\ast\cdot{\mathbf{e}}_p+{\mu_0}{\delta \mu}{{\mathbf{h}}_0}^\ast\cdot{\mathbf{h}}_p\right ] \nonumber \\ & - { \mathrm{i}}\int_{\partial v } da \left[\left({\delta\mathbf{e}}\times{{\mathbf{h}}_0}^\ast\right)\cdot\hat{\mathbf{n}}+\left({{\mathbf{e}}_0}^\ast\times{\delta\mathbf{h}}\right)\cdot\hat{\mathbf{n}}\right ] , \label{eq : bethe}\end{aligned}\ ] ] where @xmath99 denotes the boundary of @xmath97 and @xmath100 the unit vector normal to that surface .
the above equation ( [ eq : bethe ] ) yields an _ exact _
expression for the ( complex ) cavity shift @xmath101 to arrive at equation ( [ eq : bs ] ) of the main paper , we start by making the following approximations : if the effect of the perturbation is small , then @xmath102 , and when the total volume is large compared to the volume of the perturbation , it will be allowed to omit the contribution of @xmath103 in the integral on the left side of equation ( [ eq : bethe ] ) , which can now be written as @xmath98 \approx { \delta\omega}\int_v dv \left[{\epsilon_0}{\epsilon}\left|{{\mathbf{e}}_0}\right|^2 + { \mu_0}\mu\left|{{\mathbf{h}}_0}\right|^2\right]=4{\delta\omega}u_0,\end{aligned}\ ] ] where @xmath2 denotes the energy stored in the unperturbed cavity .
now the bethe - schwinger equation becomes @xmath104\\ -\frac{{\mathrm{i}}}{4u_0 } \int_{\partial v } da \left[\left({\delta\mathbf{e}}\times{{\mathbf{h}}_0}^\ast\right)\cdot\hat{\mathbf{n}}+\left({{\mathbf{e}}_0}^\ast\times{\delta\mathbf{h}}\right)\cdot\hat{\mathbf{n}}\right ] .
\label{bethenew}\ ] ] if the perturbation is caused by a variation of the permittivity ( @xmath105 ) and if the dimensions of its volume @xmath83 are small compared to the scale over which the cavity field @xmath80 varies , one can make a useful approximation by considering the perturbation as a polarizable ( point ) dipole .
its total dipole moment @xmath106 is given by @xmath107 where we introduced the dipole moment per unit volume , which we identify as @xmath108 taking into account that @xmath80 can be considered constant over @xmath83 , we can now write the integral over @xmath83 in equation ( [ bethenew ] ) as @xmath109 where we have introduced the polarizability @xmath110 defined in this case through @xmath111 if @xmath110 is diagonal , and we denote the magnitude of the polarizability in the direction of the cavity field by the scalar @xmath14 , the integral over @xmath83 reduces to @xmath112 . in this approach , we arrive at a new version of the bethe - schwinger equation for the perturbation of the cavity frequency : @xmath113 , \label{bethealpha}\ ] ] here we see how the real part of the particle s polarizability contributes to a frequency shift of the cavity , but so does the imaginary part of the surface integral .
likewise , the change of the cavity linewidth is related to the imaginary part of @xmath14 and the real part of the integral over @xmath99 .
gold nanoantennas are fabricated in an array on a 170 @xmath27 m thick glass coverslide .
to start , a 130 nm layer of zep-520 resist is spin - coated on top of the coverslide .
the nanoantennas are patterned into the resist using electron - beam lithography . after development , thermal evaporation of gold and a lift - off step yield the desired antennas . the antenna width and thickness
were designed to be 120 and 40 nm and the length was varied between 400 and 360 nm .
the pitch along the short - axes of the antennas was varied between 750 and 1500 nm , and the pitch along the long - axes was fixed at 800 nm .
a high - q silica microtoroid ( diameter @xmath23 36 @xmath114 ) is fabricated on the edge of a silicon sample , largely following methods as previously reported @xcite . in this work ,
spin - coating ( ma - n 2410 ) and subsequent cleaving of the sample enabled targeted e - beam lithography of the disks .
+ + figure [ fig : setup ] shows the experimental setup used in this work .
we use a tunable fiber - coupled external cavity diode laser ( new focus tlb-6728 , @xmath115100 khz linewidth ) to probe the microcavity .
the laser frequency is controlled by an external voltage from a function generator .
coupling of light into the cavity is achieved using a tapered optical fiber , of which the position is controlled using piezo positioners ( not shown ) .
the electro - optical modulator ( eom , eospace ) is used to generate sidebands of the cavity mode at known rf frequency , which allows for calibration of our optical frequency axis .
fiber polarization controllers ( fpc ) ensure effective coupling into the optical cavity mode of interest . before detection , the light that is transmitted through the tapered fiber
is amplified using an erbium doped fiber amplifier ( edfa ) .
radiation leakage from the microcavity is collected using an objective ( nikon , cfi apo tirf 100x ) to obtain information regarding polarization and mode profile ( through fourier - space imaging and polarization analysis of the back focal plane ( bfp ) , fig .
[ fig : mode ] ) of the cavity mode . using this method
we identify a fundamental cavity mode , of which the polarization is aligned with the long axes of the antennas . due to limitations with regards to the numerical aperture ( na ) of our objective ( na @xmath23 1.33 )
, imaging of the complete cavity mode - profile is impossible , thus resulting in a cut - off of the imaged mode profile . considering this cut - off at na=1.33 and the refractive index of silica ( 1.46 ) , we estimate an effective mode index of around 1.35 .
we check the frequency stability of the laser itself using a fiber loop reference cavity ( flc ) . ) .
the blue solid lines serve as a guide to the eye.,title="fig : " ] + , we average all the points clearly on the _ plateau _ and subtract the average reference value on glass ( @xmath116 ) . ]
figure [ fig : sweep]a shows that the experimental data is obtained by stepwise scanning the microcavity , which is placed on a piezo - electric stage , along an antenna array and a stretch of bare substrate at a distance of @xmath251.1 @xmath114 ( based on simulations ) . during each scan ,
the cavity - fiber distance is kept constant , which is checked before and after each experiment when the cavity is positioned at opposite positions adjacent to the antenna arrays .
this check is performed by monitoring the cavity linewidth ( broadened by the presence of the substrate ) , which depends strongly on the distance to the substrate .
each scan consists of 80 - 120 steps and multiple scans ( 2 - 5 ) are performed on each array .
the change in resonance frequency ( @xmath34 ) and linewidth ( @xmath117 ) due to the antennas is obtained by subtracting the cavity resonance frequency and linewidth without the array placed in the near - field of the cavity , from the cavity resonance frequency and linewidth with the antenna array placed in the evanescent field of the cavity mode .
example data for the change in linewidth is shown in fig .
[ fig : sweep]b . to correct for
any remaining sample tilt and/or slow drift , reference measurements on glass taken at the start and end of a scan are averaged .
the values of @xmath34 and @xmath117 are calculated by excluding data points taken during the transition from glass to array .
even though we check that all measurements are performed at the same cavity - sample distance , we would like to eliminate any errors due to small variations of that distance .
to this end , we retrieve parameters that are robust towards small distance changes by considering the relative change in linewidth(resonance frequency ) due to the antennas : @xmath118 , where @xmath119 is the broadening the cavity mode experiences due to the glass substrate ( @xmath120 , with @xmath121 the decay rate of the bare cavity ) .
figure [ fig : approach ] shows that this method results in a relative change of the resonance frequency @xmath7 and linewidth @xmath8 which is independent of the distance to the substrate .
it is obvious that this quantity now allows straightforward averaging over multiple scans on the same array and comparison between different antenna arrays .
errors on the mean are calculated for @xmath34 and @xmath45 , while the error ( one standard deviation ) on @xmath122 ( and thus the detuning ) is retrieved from a gaussian fit to the sum of squared residuals , obtained by displacing the fit result with respect to the normal - incidence tranmission spectra of the antenna arrays .
to gain more insight in our experimental results , and to calculate the different contributions to the complex frequency shift @xmath22 , we perform 2d eigenfrequency simulations using the comsol software package ( v5.0 ) .
the cavity ( diameter @xmath123 36 @xmath114 ) has a refractive index of 1.35 , comparable to the experimentally determined effective index of our cavity mode and is placed 1.1 @xmath114 from the substrate , which has a refractive index of 1.5 .
the antenna is assigned a varying permittivity to mimic a change in detuning between cavity and antenna and for the surrounding we take a refractive index of 1 .
the simulation includes the use of a perfectly matched layer ( pml ) . in our experiment , changing the length of the antennas results in a changing plasmon resonance , in turn causing a varying detuning between cavity and antenna resonance frequency . to mimic this in our simulation ,
we assign a tunable permitivitty , which has a lorentzian lineshape , to the perturbing antenna(s ) : @xmath124 sweeping the variable @xmath125 now allows to calculate the change in cavity eigenfrequency as a function of detuning .
the parameter s can be used to control the scattering strength of the antenna , while the antenna linewidth is held fixed at @xmath126 . to determine a sensible value of s
, we analyze the eigenfrequency shift of the cavity when perturbed by 11 antennas ( pitch 1500 , fig . [
fig : comsol_11scatt]a ) for different values of s and choose a value ( s=340 ) which yields the best resemblance to our experiment ( fig .
[ fig : comsol_11scatt]b ) .
the values for the relative change in linewidth and resonance frequency are obtained via similar analysis as performed in the experimental situation , thus involving separate simulations for the bare cavity , and cavity perturbed by just a glass substrate . and @xmath17 , which are used in ( b ) .
b ) the blue circles and solid lines are experimental data points and a fit using the coupled - mode model , respectively .
the orange ( top ) and pink ( bottom ) filled circles are the relative frequency and linewidth change that are obtained from a simulation with s=340 .
the rescaled horizontal axis is obtained from the fit in ( a ) . ]
it is important to realize that the values @xmath125 and @xmath127 are input parameters in the simulation , which do not include the effect of radiation losses on the antenna linewidth and resonance frequency @xcite . to allow for a fair comparison with our experiment
, we fit the results retrieved from the simulation with a fano lineshape ( black lines in fig .
[ fig : comsol_11scatt]a ) , to obtain the radiation corrected values @xmath19 and @xmath17 . using these corrected values of @xmath19 and @xmath17 , we rescale our horizontal axis and obtain fig .
[ fig : comsol_11scatt]b .
note that this approach resembles the experimental situation , where the radiation corrected values of @xmath19 and @xmath17 are retrieved from the normal - incidence transmission spectra .
it can be observed from fig .
[ fig : comsol_11scatt]b that the orange ( resonance frequency ) and pink ( linewidth ) circles are a good , but not perfect , match to the experimentally obtained cavity frequency and linewidth change ( blue circles ) .
we attribute the discrepancy to the intrinsic difference between a 2d simulation , where the cavity is modeled as an infinite cylinder , and the experimental situation , which employs a toroidal cavity .
this will necessarily change the radiation profile determining overlap and phase .
nonetheless , the 2d simulation reproduces all the main features observed in the experiment . when calculating the complete bethe - schwinger formula , it is useful to first consider the situation where a cavity is perturbed by a single perturbing particle , in the absence of a substrate , as this should not allow for radiation interaction taking place , and thus result in a perturbation of the cavity eigenfrequency which is solely governed by the polarizability of the perturbing particle .
the results of this simulation are shown in fig .
[ fig : comsol_eigenvalues]a .
it is clear that the cavity indeed lacks radiation interaction ( green stars ) with the antenna and experiences linewidth and resonance frequency shifts that are solely related to the polarizability of the particle ( black crosses ) , such that the total eigenfrequency change ( open blue circles ) is completely dominated by local effects .
we remark that the bethe - schwinger cavity perturbation formula , which needs fields as it input , perfectly reproduces the cavity eigenfrequency shifts as predicted by the eigenvalues ( orange(pink ) filled circles ) directly obtained from comsol .
as discussed in the main text , introducing a glass substrate enables and extra coupling channel between antenna and cavity due to overlapping radiation profiles .
[ fig : comsol_eigenvalues ] shows results quantifying this effect for a cavity perturbed by a single antenna . here
the radiative interactions ( green stars ) are of importance when calculating the total ( open blue circles ) eigenfrequency shift of the cavity using the bethe - schwinger equation , dominating the effect of local perturbation ( black crosses ) .
note that the calculation of complex @xmath22 using the full bethe schwinger equation , yields almost exactly the same results when compared to the cavity eigenfrequency changes based on the eigenvalues calculated in the simulation ( orange(frequency ) and pink(linewidth ) filled circles ) .
we attribute the small difference that is observed mostly to the way joined data sets are handled in comsol .
the use of joined datasets is necessary to obtain the fields @xmath39 , but causes an error when integrating the closed - surface flux integrals ( last terms of equation ( [ eq : bs ] ) ) .
the reason is that for joined datasets comsol projects full datasets on a less dense intermediate mesh .
we performed mesh convergence studies to see if this effect would diminish with increasing mesh density ( up to 60 elements/@xmath128 ) .
although we observe that both the eigenvalues and the fields of our solutions are clearly converged , the small difference in the cavity eigenfrequency shift that we obtain using both methods does not disappear .
nonetheless , the discrepancy is small enough to reproduce all observed trends . figure [ fig : bs_comsol]a of the main paper shows the field profile @xmath129 of a cavity perturbed by an antenna on a glass substrate . to increase the visibility of the radiation pattern , that figure and fig .
[ fig : comsol_field_profiles ] show the field amplitude for a cavity placed 0.3 @xmath114 from the substrate and antenna . to calculate the full cavity perturbation formula ,
it is necessary to find the unperturbed cavity field profile ( fig .
[ fig : comsol_field_profiles]a , @xmath130 ) and subtract this from @xmath129 in order to obtain the field @xmath131 ( fig .
[ fig : comsol_field_profiles]b ) .
the teal line surrounding the field profiles represents a perfectly matched layer ( pml ) , which is used in all calculations discussed in this paper , while the dashed black line is the boundary which we used for the calculation of the surface integral of equation ( [ eq : bs ] ) .
a pitfall in subtracting fields from different eigenfrequency simulations lies in the arbitrary choice of amplitude and phase of the solutions . to overcome this problem
, we monitor the amplitude and phase in both simulations ( with and without antenna ) in a point where modification of the field profile due to the perturbing antenna is expected to be negligible .
this can be , for example , a point in the field maximum of the cavity mode far away from the perturbation , thus allowing to correct for possible differences in amplitude and phase between the calculations . to obtain the data in fig [ fig : bs_comsol]b , a single phase correction was performed , to match the phase at @xmath132 to the unperturbed phase , in a mode maximum at large distance from the perturbing particle . for a cavity in the absence of perturbing particles .
b ) the field profile @xmath39 of the perturbing antenna .
this profile is obtained by subtracting the fields in ( a ) from those shown in fig .
[ fig : bs_comsol]a of the main paper . in both ( a ) and ( b )
, the teal line surrounding the field profiles represents a perfectly matched layer ( pml ) . ]
we will describe the interactions between cavity and antenna using coupled mode theory @xcite , essentially modeling the system as coupled harmonic oscillators with lorentzian response .
we note that while it is possible to treat the antenna as a harmonic oscillator without making the lorentzian approximation , all essential physics is captured in the simpler lorentzian model .
we treat the decay in a common ( overlapping ) radiation continuum , and its possible backaction on both antenna and cavity , by considering the ` environment ' as a third mode coupled to both antenna and cavity as well as to an independent decay port .
the complex mode amplitudes of cavity ( @xmath133 ) , antenna ( @xmath134 ) and environment ( @xmath135 ) are normalized such that their magnitude squared equals the energy in each degree of freedom . they are collected in vector @xmath136 .
the equations of motion of the system ( in absence of driving fields ) can be written as @xmath137 where @xmath138 is a hermitian matrix containing the resonance frequencies and coupling rates .
these coupling rates can in general be complex , and we will write them later as @xmath139 with real @xmath140 and @xmath141 .
the matrix @xmath142 is also hermitian and contains the damping rates to the different independent coupling channels , which are associated with an individual mode @xmath133 , @xmath134 , and @xmath135 . as such , the output fields @xmath143 in these ports can be written as @xmath144 , where @xmath145 these equations satisfy time reversal symmetry and conservation of energy , assured by the hermiticity of @xmath146 and @xmath147 together with the condition @xmath148 .
we now introduce the detunings @xmath149 , @xmath150 and @xmath151 .
the eigenmodes can be found by solving the system @xmath152 where @xmath153 since we assume the cavity and antenna are weakly coupled and we are only interested in the perturbed cavity mode , we can neglect @xmath154 in the second and third row of @xmath155 .
the solution is @xmath156 where we have introduced @xmath157 .
it is instructive to consider the case where the cavity mode is absent ( @xmath158 ) . in that case , the power radiated via the environment into port @xmath159 is equal to @xmath160 .
we therefore define the rate @xmath161 which is the rate at which the antenna decays into the common radiation continuum if the cavity were not there .
likewise , we consider the case without the antenna ( @xmath162 ) , and define @xmath163 which is the rate at which the cavity would decay into this channel in absence of the antenna . in the latter case
, the eigenmode would have detuning @xmath164 where @xmath165 is the _ total _ decay rate of the cavity without the antenna .
we now specifically write the general solution as @xmath166 , such that @xmath167 is the _ change _ of the eigenmode frequency due to the presence of the antenna and @xmath168 . taking the limit of @xmath54 to infinity , which effectively amounts to setting @xmath169 , it reads @xmath170 where we have introduced @xmath171 ; a number between 0 and 1 that is related to the overlap of the radiation of both modes .
this is expression is equivalent to equation ( [ eq:3co_domega ] ) in the main text .
next , let us turn to the relation of this model to the phase difference @xmath46 as it appears in the bethe - schwinger equation .
to this end , we express the field in the environment @xmath135 , compared to the field in the cavity @xmath133 .
its full expression ( still taking @xmath169 ) is @xmath172 the phase @xmath46 must be equal to the phase difference of the above in absence of the particle ( @xmath173 ) and its phase in absence of direct coupling between cavity and environment ( @xmath174 ) . performing this subtraction yields @xmath175 showing that the phase difference of the two decay paths is directly related to the ( relative ) phases of the coupling rates via @xmath67 and the phase response of the antenna ) when analyzing comsol data . ] .
it becomes clear that the standard case , where the two paths interfere destructively on resonance ( @xmath20 ) , occurs for @xmath176 , such that @xmath71 .
finally , we note that it is straightforward to eliminate the environment mode @xmath135 from the model to construct an equivalent system of equations @xmath177 . in that case the total damping rates @xmath8 and @xmath17 of the uncoupled cavity and antenna appear on the diagonal : @xmath178 we can choose the phase @xmath179 without loss of generality , such that @xmath180 note that the off - diagonal matrix elements of @xmath181 , representing coupling between cavity and antenna , are now no longer hermitian .
the text of the main paper shows the results and analysis of antenna arrays with pitch 800(1500 ) nm along the long(short ) axis of the antennas . more pitches are , however , experimentally investigated and fig .
[ fig : varying_pitch ] shows the main results which we obtain when varying the pitch along the short axis ( 750 - 1500 nm ) of the antennas , while the pitch along the antenna long axis is kept constant at 800 nm . in this figure , the points which are surrounded by a dashed circle correspond to the arrays used to generate fig . [
fig : bs_exp]a of the main text .
the main conclusion that can be drawn from fig .
[ fig : varying_pitch ] is that , qualitatively , the different arrays behave the same : all arrays induce blueshifts of the cavity resonance frequency ( top panel ) and linewidth narrowing ( bottom panel ) of the cavity on approaching @xmath44 . with changing pitch , the value of @xmath28 where the `` relative linewidth change '' equals 0 , slightly varies .
we attribute this effect to the influence of the array on the single particle polarizability
. remarkably , for the largest possible detuning we measure linewidth narrowing up to 30% for a pitch of 1100 nm . .
modification of the single particle polarizability due to the array is the most likely cause of this behaviour .
the values on the vertical axis are corrected for the unit - cell area and the points which are surrounded by a dashed circle correspond to the arrays used to generate fig .
[ fig : bs_exp]a of the main text .
errors on the mean ( vertical axis ) fall mostly within the plot markers and the error on the horizontal axis is one standard deviation , retrieved from a gaussian fit to the sum of squared residuals ( obtained by displacing the fit result ) . due to limited detector range , a reliable horizontal error for the data at p1100,@xmath182 could not be obtained using this method .
the error on this data point was thus set equal to the error for array p1100,@xmath183 . ] as in the main paper , the values of @xmath28 are retrieved from a fit of the normal - incidence transmission spectra using a lorentzian lineshape .
lattice sum calculations @xcite are performed to exclude strong modification of @xmath19 due to the change of effective angle of incidence ( in the experiment ) with respect to the normal - incidence transmission spectra .
in addition , total internal reflection measurements on the samples with a 1500 nm pitch showed no significant change of @xmath19 with respect to the values obtained from the normal - incidence measurements .
10ifxundefined [ 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 10.1049/pi - c.1960.0041 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1038/nature01371 [ * * , ( ) ] link:\doibase 10.1038/nphys1425 [ * * , ( ) ] http://books.google.com/books?hl=en&lr=&id=rhc_awaaqbaj&pgis=1[__ ] ( , , ) link:\doibase 10.1109/jqe.2004.834773 [ * * , ( ) ]
link:\doibase 10.1103/revmodphys.79.1267 [ * * , ( ) ]
|
the influence of a small perturbation on a cavity mode plays an important role in fields like optical sensing , cavity quantum electrodynamics and cavity optomechanics .
typically , the resulting cavity frequency shift directly relates to the polarizability of the perturbation .
here we demonstrate that particles perturbing a radiating cavity can induce strong frequency shifts that are opposite to , and even exceed , the effects based on the particles polarizability .
a full electrodynamic theory reveals that these anomalous results rely on a non - trivial phase relation between cavity and nanoparticle radiation , allowing back - action via the radiation continuum .
in addition , an intuitive model based on coupled mode theory is presented that relates the phenomenon to retardation .
because of the ubiquity of dissipation , we expect these findings to benefit the understanding and engineering of a wide class of systems .
the fact that a small perturbation of a potential can influence the distribution of a system s energy levels is a well - known principle permeating various branches of physics . in quantum mechanics , for example , the effect of a perturbing potential @xmath0 on an eigenstate @xmath1 is that it modifies its unperturbed energy @xmath2 by an amount @xmath3 , where @xmath4 is the new eigenstate and we assume @xmath5 . in electrodynamics ,
a local change of potential ( _ i.e. _ permittivity ) can impact the frequency of a resonant cavity .
this is at the basis of many applications that use the influence of a perturbing atom , molecule , or dielectric body to establish an interaction that can be exploited for optical sensing or control @xcite .
the shift of a mode s complex eigenfrequency @xmath6 , with cavity resonance frequency @xmath7 and linewidth @xmath8 , due to a local permittivity perturbation @xmath9 contained in a volume @xmath10 is given by @xmath11/4u_0 $ ] . here
@xmath12 and @xmath2 represent the field and total energy of the unperturbed cavity mode , respectively , and @xmath13 is the perturbed field @xcite . in particular , if @xmath10 is small enough such that the perturbing particle can be described in the dipole approximation , the complex frequency shift is directly related to the particle _ polarizability _ @xmath14 , reading @xmath15 @xcite . however , this widely employed result of bethe - schwinger perturbation theory @xcite is strictly only valid when radiation to the far field is negligible .
in several recent developments , radiation loss proved decisive in determining a system s eigenmode , _
e.g. _ for so - called ` states bound in the continuum ' @xcite and in describing complex plasmonic resonators @xcite . in that context , the question arises to what extent the conventional paradigm to determine perturbed cavity frequencies holds in practical , open , systems . here , we study the eigenfrequencies of a radiating optical cavity as it is perturbed by carefully designed resonant plasmonic nanoparticles . for a resonant perturbation with center frequency @xmath16 and linewidth @xmath17
, the value of @xmath14 strongly depends on the detuning @xmath18 . in absence of radiation , the cavity mode thus redshifts(blueshifts ) at negative(positive ) values of @xmath19 and the linewidth broadens near @xmath20 , in direct response to re[@xmath14 ] and im[@xmath14 ] , respectively ( fig [ fig : bs_system]a ) .
tuning @xmath16 by varying the length of the ( plasmonic ) resonators allows to systematically study the induced cavity response as a function of detuning .
importantly , our experiment is designed such that radiation losses from the cavity mode and plasmonic resonators overlap .
this , as we will show , leads to a strong additional contribution to @xmath22 which is not captured by @xmath14 .
we study this ` radiation interaction ' , as we will call the effect , in detail and show that it can induce strong eigenfrequency shifts that are opposite to , and even exceed , the effects based on the particles polarizability . .
b ) an array of perturbing gold nanoantennas is placed in the near field of a toroidal microcavity .
light transmission through the tapered fiber is detected to determine the cavity eigenfrequency shift .
c ) scanning electron microscopy ( sem ) image of a high - q silica microtoroid at the edge of a chip .
d ) sem image of gold nanoantennas on a glass substrate .
e ) transmittance through the tapered fiber around a cavity mode resonance . a lorentzian fit ( black solid line ) is used to determine the optical linewidth ( @xmath2330 mhz ) and frequency of the cavity .
the sidebands result from a 150-mhz phase modulation used to calibrate the frequency axis .
f ) normal - incidence transmission spectra of gold nanoantenna ( design length @xmath24 ) arrays .
lorentzian fits ( sold lines ) give the antenna resonance frequency and linewidth .
the dashed line indicates the cavity frequency . ] a sketch of the system is shown in fig .
[ fig : bs_system]b .
the experiments are performed using a fundamental cavity mode ( 194.4 thz , [email protected]@xmath26 , te polarized ) of a toroidal silica microcavity fabricated on the edge of a silicon chip ( @xmath2336 @xmath27 m diameter , fig .
[ fig : bs_system]c ) @xcite .
the cavity is perturbed by gold nanoantennas deposited on a glass substrate ( fig .
[ fig : bs_system]d ) , which are controllably placed in the evanescent field of the cavity . the antennas ( nanorods of length @xmath24 , width 120 nm and thickness 40 nm ) are aligned with their ( long ) principal dipole axis to the polarization of the cavity mode . a frequency - swept narrowband laser source ( @xmath250.7 @xmath27w )
is coupled into a tapered fiber that is brought close to the cavity .
the transmission spectrum through the fiber shows a lorentzian dip around the cavity resonance frequency ( fig .
[ fig : bs_system]e ) , from which we determine the ( perturbed ) resonance frequency @xmath7 and linewidth @xmath8 of the cavity mode . independent normal - incidence transmission measurements ( fig .
[ fig : bs_system]f ) on the fabricated antenna arrays , in absence of the cavity , yield the normalized cavity - antenna detuning @xmath28 .
notably , the presence of the glass substrate allows the cavity mode to radiate into the glass at a well defined angle just beyond the critical angle and , similarly , allows scattering of antenna radiation into the substrate @xcite .
as the antennas are coherently excited by the cavity field , from which also the cavity radiation originates , the zeroth - order diffraction by the array is expected to overlap with the cavity radiation . , antenna arrays induce linewidth broadening ( top ) and narrowing ( bottom ) , while simultaneously inducing the cavity resonance frequency to blueshift .
the blue and red horizontal bars represent the linewidths of fitted lorentzian lineshapes ( black lines ) .
b ) shift in resonance frequency and linewidth of the cavity mode , normalized to its linewidth in absence of the antennas @xcite , due to perturbation by antenna arrays with a constant pitch size and varying antenna length .
top : for all detunings a blueshift of cavity resonance frequency is observed .
bottom : approaching @xmath20 induces linewidth narrowing .
both the blueshift and linewidth narrowing are contrary to the expectation based on the particles polarizability ( fig .
[ fig : bs_system]a ) .
error bars depict standard deviation , vertical error bars fall within the plot markers .
the solid lines represent a coupled mode theory fit . ] to measure the change of the complex cavity resonance frequency due to the nanoantennas , we compare the frequency and linewidth of the cavity mode with and without the antenna array placed in the near - field of the cavity .
our measurement procedure @xcite ensures that in both cases the sample is positioned at the same distance to the cavity .
figure [ fig : bs_exp]a shows the results for two different arrays of nanoantennas : in the top figure , an antenna array ( @xmath29 , pitch 800(900 ) nm along the long(short ) axes of the antennas ) is seen to induce a broadening of the cavity linewidth , together with a _
blueshift _ of the frequency .
the latter is clearly surprising , given the fact that @xmath30 for this sample @xcite , where one expects a _ redshift _ if one only considers the resonant particle s polarizability ( fig .
[ fig : bs_system]a ) .
the bottom panel in fig .
[ fig : bs_exp]a shows that for another array ( @xmath31 , pitch 800(1100 ) nm ) with slightly positive detuning ( @xmath32 ) again a blueshift is observed , but this time accompanied by _ narrowing _ of the cavity mode .
such a reduction of damping can obviously not be ascribed to the particle s polarizability alone , as its imaginary part is necessarily positive ( fig .
[ fig : bs_system]a ) . to systematically study these effects
, we gradually tune the resonance frequency @xmath33 by varying the length of the antennas ( fig .
[ fig : bs_system]f ) .
the array pitch is kept fixed at 800(1500 ) nm along the long(short ) antenna axes , chosen such that inter - antenna coupling is negligible .
moreover , in all examples we show , bragg - scattering between clockwise and counter - clockwise modes @xcite is small enough such that mode splitting induced by the nanoantennas is smaller than the cavity linewidth .
figure [ fig : bs_exp]b shows the resulting cavity frequency and linewidth changes , normalized to the cavity linewidth on glass to allow averaging multiple scans @xcite , as a function of @xmath28 . for these negative detunings , a consistent increase of the resonance frequency ( i.e. blueshift ) is observed that slightly rises as the antenna frequency approaches that of the cavity . as such ,
both the sign and the trend of @xmath34 are incongruous with the expectation based on the polarizability ( fig .
[ fig : bs_system]a ) .
moreover , the broadening of the cavity , observed for a detuning of @xmath35 , quickly reduces as the strongly scattering
antennas are tuned closer to the cavity resonance .
we note that these trends are consistently observed also at other periodicities @xcite and that the shifts can not be explained by thermal heating , negligible at the employed powers .
to explore the origin of these surprising results , it is necessary to consider the complete bethe - schwinger cavity perturbation equation @xcite , derived without neglecting radiation : @xmath36 .
\label{eq : bs}\end{aligned}\ ] ] this full expression , which is also valid in any open , non - hermitian system @xcite , contains an additional integral term involving radiated fields at a surface @xmath37 ( with normal unit vector @xmath38 ) , enclosing the same volume that was used to evaluate the total energy @xmath2 . here ,
@xmath39 and @xmath40 are the difference between the fields in the presence ( @xmath13 , @xmath41 ) and absence ( @xmath12 , @xmath42 ) of the perturbation , and can as such be associated with the field scattered by the perturbation .
thus , the cavity eigenfrequency is additionally modified by an energy flux that is evaluated by _ combining _ the fields of the perturbed and unperturbed eigenmodes ; _
i.e _ , an overlap of scattering by the perturbation and direct cavity radiation .
remarkably , the integral is independent of the distance at which @xmath37 is chosen , consistent with the fact that it can be associated with radiation of the system . to date
, this far - field contribution has been omitted in practically all analyses of cavity perturbation . in a select number of experiments , a reduction of cavity linewidth
was observed @xcite and tentatively attributed to interference of radiation of the cavity and scattering by the perturbation .
importantly , the integral suggests that in principle also the resonance frequency ( i.e. , the real part of @xmath22 ) can be affected by the same mechanism , if the phase difference between cavity and scattered radiation in the far field ( captured in @xmath12 and @xmath39 , respectively ) differs from 0 or @xmath43 .
so far , the frequency has been expected to rely only on the local variation of the applied potential @xcite as it is contained in the first term of ( [ eq : bs ] ) . ) ) .
the arrow points to the position of the antenna .
b ) top(bottom ) : the change in resonance frequency(linewidth ) calculated using ( [ eq : bs ] ) .
black crosses ( ` a ' ) show the contribution of the polarizability to the eigenfrequency shift @xmath22 , which resembles the lineshape as we expect it from a resonant particle ( fig .
[ fig : bs_system]a ) .
the radiative contribution ( green stars , ` b ' ) shows distinct blueshifts and linewidth narrowing near @xmath44 . adding both terms ( blue circles , ` a+b ' ) yields the complete cavity eigenfrequency shift , which matches the experimental trends . ] to verify the importance of radiation in a complete description of the induced cavity response , we perform finite - element method eigenfrequency studies with and without antenna @xcite in a similar two - dimensional system ( fig . [ fig : bs_comsol]a ) , from which we extract the fields @xmath12 , @xmath42 , @xmath39 and @xmath40 .
this allows the calculation of the individual terms of ( [ eq : bs ] ) , displayed as a function of detuning in fig .
[ fig : bs_comsol]b .
the first term ( black crosses , ` a ' ) , contributes to @xmath22 according to the expected behaviour for a perturbing resonator sketched in fig .
[ fig : bs_system]a : it produces a dispersive detuning dependence of @xmath34 and a dissipative trend for @xmath45 .
this immediately disqualifies this term as an explanation for the measured changes in cavity resonance and linewidth
. the second term , _ i.e. _ the contribution to the change in eigenfrequency related to the radiation interaction ( green stars , ` b ' ) , shows a dramatically different behavior : it causes distinct blueshifts and linewidth narrowing around @xmath44 .
surprisingly , the magnitude of the blueshifts induced by this term can even exceed the contribution due to @xmath14 .
this is a key point of our observation , stressing the importance of this new contribution in a model system .
the sum of both terms ( blue circles , a+b ) yields the complete eigenfrequency shift of the perturbed cavity , which now qualitatively matches our experimental results ; blueshifts and linewidth narrowing around @xmath44 .
these calculations thus confirm that radiation interactions in open systems can lead to both cavity blueshifts and linewidth narrowing and importantly , that this effect can even _
dominate the total change in eigenfrequency _ of a cavity mode .
, which is the near - field interaction scaling with the polarizability of the antenna , and can radiate into the continuum with a certain delay ( @xmath46 ) with respect to each other .
the interference between both radiation profiles and the overlap with the cavity mode results in a back - action effect ( single - headed dashed arrow ) on the cavity mode .
b ) coupled - mode theory .
the cavity and antenna are coupled with rate @xmath47 and couple to the continuum with complex rates @xmath48 and @xmath49 respectively .
the difference between @xmath50 and @xmath51 , summed with the phase response of the antenna , is now equal to the phase difference @xmath46 . @xmath52 and @xmath53 are the losses of cavity and antenna into non - coupled channels .
the environment is modeled as an oscillator with decay rate @xmath54 . ]
it is essential to realize that a modification of the cavity resonance frequency ( in contrast to the alteration of the linewidth ) due to back - action via the radiation continuum only occurs when the phase @xmath46 between cavity and antenna radiation is different from 0 or @xmath43 ( fig .
[ fig : bs_coupled_oscillators]a ) .
the exact strength of this back - action , and as such the sum of direct ( _ i.e. _ due to @xmath14 ) and indirect ( _ i.e. _ via the radiation continuum ) coupling between cavity and antenna , will depend on the overlap between their radiation profiles and on the value of @xmath46 .
this phase difference and its physical connection to back - action can be captured in a simple model ( fig . [ fig : bs_coupled_oscillators]b ) based on temporal coupled - mode theory @xcite .
let us assume for now that the cavity and antenna are coupled with complex coupling rate @xmath55 and can both interact with the environment ( glass ) with coupling rates @xmath48 and @xmath49 , respectively .
we treat the joint environment as a separate mode to allow backaction , taking the limit of large decay rate @xmath54 to mimic a broadband continuum .
the cavity and antenna , considered separately , each decay via this continuum at a rate @xmath56 and @xmath57 , respectively .
this is distinguished from decay into all other , non - overlapping , modes at rates @xmath52 and @xmath53 , such that @xmath58 and @xmath59 .
importantly , the phase difference between cavity radiation ( blue path , fig .
[ fig : bs_coupled_oscillators]b ) and antenna scattering ( orange path , fig .
[ fig : bs_coupled_oscillators]b ) in this model now directly relates to @xmath46 via @xmath60 $ ] , where @xmath61 , and @xmath62 $ ] is the phase response of the antennas @xcite .
the natural mode of interest , with complex frequency @xmath63 , of this hybrid system can be found by equating the determinant of the coupling matrix @xcite @xmath64 to zero . solving for complex @xmath63 yields @xmath65 which depends on the radiation overlap @xmath66 , ( constant ) phase @xmath67 and the coupling rate @xmath68 . because the overlap and phase @xmath67 can not be independently chosen ,
it is impossible to directly fit ( [ eq:3co_domega ] ) to the experimental data .
therefore , we retrieve @xmath46 from the finite element simulation and , using the relation between @xmath46 and @xmath67 , obtain @xmath69 . note that this value of @xmath67 implies that @xmath70 at @xmath20 , completely opposite to the conventionally assumed case where destructive interference ( @xmath71 ) only contributes to the linewidth of the cavity mode .
fixing @xmath67 at this value and constraining the experimentally found values of @xmath8 , @xmath19 and @xmath17 ( retrieved from independent spectroscopic measurements , fig .
[ fig : bs_system ] ) , a fit of ( [ eq:3co_domega ] ) yields excellent correspondence to our data in fig .
[ fig : bs_exp]b ( solid lines ) , resulting in an overlap of @xmath72 and coupling rate @xmath73 ghz . concluding , we have shown how coupling through the radiation continuum leads to cavity blueshifts and linewidth narrowing in a coupled cavity - antenna system .
surprisingly , the non - local , radiative effect on the cavity is larger than the induced cavity response due to local perturbations , and is not just an artefact of large radiation losses but can even manifest itself in a high - q cavity such as studied here .
therefore , similar effects are expected in many other systems where cavity and perturbation can radiate into the same channels , such as photonic crystals @xcite and fabry - prot etalons @xcite . in addition
, it was shown that there is a direct link between the phase @xmath46 in the bethe - schwinger equation and the contributions to back - action in a coupled - mode model .
interestingly , these results could also shed new light on radiative interactions in strongly radiating systems such as metamaterials and complex plasmonic resonators . in @xcite
, it was postulated that radiation can be included in the coupling between two resonators by using a complex coupling rate , while in @xcite , a complex coupling rate was argued to originate from a complex - valued extinction cross - section . in fact , such a complex coupling rate between two resonators is easily derived from the coupled - mode model presented in this work @xcite . furthermore , it will be interesting to see how the concepts and theory established here relate to recently developed methods to relying on the normalization of leaky modes to describe the response of complex photonic systems @xcite . in a different context , recent advances in optomechanics using simultaneous dispersive and dissipative ( _ i.e _ radiative ) coupling @xcite
have been reported to enable on - resonance cooling @xcite and sensitive readout of nanomechanical motion @xcite .
as such , we expect our work to be directly important to the design of novel optical devices such as sensors , for example , and benefit the understanding of the physics of open systems in optics and beyond . + this work is part of the research programme of the foundation for fundamental research on matter ( fom ) , which is part of the netherlands organisation for scientic research ( nwo ) .
ev gratefully acknowledges support from an nwo vidi grant .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.73.565 [ * * , ( ) ] http://link.aps.org/doi/10.1103/revmodphys.87.347 [ * * , ( ) ] link:\doibase 10.1038/nmeth.1221 [ * * , ( ) ] link:\doibase 10.1016/j.aca.2008.05.022 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.86.1391 [ * * , ( ) ] http://www.nature.com/nphoton/journal/v4/n1/abs/nphoton.2009.237.html [ * * , ( ) ] link:\doibase 10.1126/science.1237125 [ * * , ( ) ] link:\doibase 10.1038/ncomms4808 [ * * , ( ) ] link:\doibase 10.1049/pi - c.1960.0041 [ * * , ( ) ] @noop _ _ ( , , ) link:\doibase 10.1103/physreva.32.3231 [ * * , ( ) ]
link:\doibase 10.1038/nature12289 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.213903 [ * * , ( ) ] link:\doibase 10.1021/nl2039748 [ * * , ( ) ] link:\doibase 10.1038/nmat2495 [ * * , ( ) ] link:\doibase 10.1038/nature01371 [ * * , ( ) ] link:\doibase 10.1038/nphys1425 [ * * , ( ) ] http://books.google.com/books?hl=en&lr=&id=rhc_awaaqbaj&pgis=1[__ ] ( , , ) @noop link:\doibase 10.1103/physrevlett.99.173603 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.153904 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.043902 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.123901 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.123902 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.041102 [ * * , ( ) ] link:\doibase 10.1364/ol.28.000272 [ * * , ( ) ] link:\doibase 10.1109/jqe.2004.834773 [ * * , ( ) ] link:\doibase 10.1021/nl903555u [ * * , ( ) ] link:\doibase 10.1364/ol.35.000953 [ * * , ( ) ] link:\doibase 10.1038/ncomms1487 [ * * , ( ) ] link:\doibase 10.1002/lpor.201100041 [ * * , ( ) ] link:\doibase 10.1021/ph500356n [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.237401 [ * * , ( ) ] link:\doibase 10.1021/acs.nanolett.5b00771 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.114.043601 [ * * , ( ) ] link:\doibase 10.1364/optica.2.000491 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.107.213604 [ * * , ( ) ]
|
the noncovalent cooperative effects in structure formation processes on mesoscopic scales let linear polymers be very interesting objects for studies of the statistical mechanics and thermodynamics of nucleation processes , even on a fundamental level .
structural properties of polymers can typically be well described by means of simple , coarse - grained models of beads and sticks ( or springs ) representing the monomers and the covalent bonds between adjacent monomers in the chain , respectively .
contemporary , sophisticated generalized - ensemble monte carlo simulation techniques as well as large - scale computational resources enable the precise and systematic analysis of thermodynamic properties of all structural phases of coarse - grained polymer models by means of computer simulations . among the most efficient simulations methods
are multicanonical sampling @xcite , replica - exchange techniques @xcite , and the wang - landau method @xcite . the high precision of the numerical data for quantities that are hardly accessible in analytic calculations one of the most prominent and , as it will turn out in the following , most relevant system - specific quantities is the density ( or number ) of states with energy @xmath0 , @xmath1 opens new perspectives for the physical interpretation and classification of cooperative processes such as phase transitions .
this is particularly interesting for small systems , where conventional statistical analyses are often little systematic and a general concept seems to be missing .
this is apparently reflected in conformational studies in the biosciences , where often novel terminologies are invented for basically the same classes of transitions .
the introduction of a unifying scheme appears insofar difficult as finite - size effects in transitions of small systems influence the thermal fluctuations of transition indicators like order parameters .
maximum energetic fluctuations , represented by peaks in the specific heat , do not necessarily coincide with peak temperatures of fluctuations of structural quantities such as the radius of gyration @xcite .
for extremely large systems , which are well described by the theoretical concept of the `` thermodynamic limit '' , this _
canonical _ approach is appealing as the fluctuation peaks scale with system size and finally collapse at the same temperature , allowing for the definition of a unique transition point .
however , for many systems , in particular biomolecules , the assumption of the thermodynamic limit is nonsensical and the explicit consideration and understanding of finite - size effects is relevant .
the _ microcanonical _ analysis @xcite allows for such an in - depth analysis of smallness effects .
it is completely based on the entropy as a function of the system energy , which is related to the density of states via @xmath2 , where @xmath3 is the boltzmann constant .
a major advantage of this quantity is the possibility to introduce the temperature as a _ derived _
quantity via @xmath4^{-1}\ ] ] which is commonly refered to as the `` microcanonical temperature '' .
this terminology is misleading since @xmath5 and thus @xmath6 do not depend on the choice of any statistical ensemble associated to a certain thermal environment of the system .
thus , the physical meaning of @xmath5 and @xmath6 is not restricted to systems well - described by the microcanonical ensemble only ( i.e. , for systems with constant energy ) .
the introduction of the temperature via eq .
( [ eq : te ] ) is also useful for another reason .
it applies independently of the system size and it is not coupled to any equilibrium condition . to make use of theoretical concepts like
the thermodynamic limit or quasiadiabatic process flow is not necessary .
the typically used heatbath concept for introducing the temperature in the context of the thermodynamic equilibrium of heatbath and system is useful for large systems when thermal fluctuations become less relevant and finite - size effects disappear .
however , as it has already been known from the early days of statistical mechanics , the statistical ensembles turn over to the microcanonical ensemble in the thermodynamic limit .
this is easily seen for the example of the fluctuations about the mean energy in the canonical ensemble . in this case , the canonical statistical partition function , linked to the free energy @xmath7 , can be written as an integral over the energy space , @xmath8 where the canonical system temperature @xmath9 corresponds in equilibrium to the heatbath temperature @xmath10 ( which is an adjustable external thermal control parameter in experiments ) : @xmath11 therefore , in equilibrium , the mean energy at a given heatbath temperature can simply be calculated as : @xmath12 the heat capacity @xmath13 must always be nonnegative because of the thermodynamic stability of matter , i.e. , @xmath14 increases monotonously with @xmath10 . for this reason ,
the dependence of @xmath14 on the temperature can trivially be inverted , @xmath15 , where we have made use of the equilibrium condition ( [ eq : equ ] ) . in complete analogy to the microcanonical definition of the temperature in eq .
( [ eq : te ] ) , we introduce the canonical entropy via the relation @xmath16_{n , v}^{-1},\ ] ] where particle number @xmath17 and volume @xmath18 are kept constant .
the canonical entropy can explicitly be expressed by the celebrated equation that links thermodynamics and statistical mechanics : @xmath19.\ ] ] instead of considering the canonical ensemble by fixing @xmath10 , @xmath18 , and @xmath17 as external parameters , we have turned to the caloric representation , where @xmath14 , @xmath18 , and @xmath17 are treated as independent variables .
if the fluctuations of energy about the mean value @xmath14 vanish , the canonical ensemble thus corresponds to the microcanonical ensemble .
this is obvious in the thermodynamic limit , where the relative width of the canonical energy distribution , @xmath20 vanishes , @xmath21 , since @xmath22 and @xmath14 are extensive variables as they scale with @xmath17 . thus , @xmath23 and @xmath24 in the thermodynamic limit .
microcanonical and canonical temperatures do typically not coincide , if finite - size effects matter .
this is particularly apparent under conditions where cooperative changes of macrostates , such as conformational transitions of finite molecular systems , occur . in transitions with structure formation ,
the conformational entropies associated to volume and surface effects in the formation of compact states of the system compete with the energetic differences of particles located at the surface or in the interior of the structure .
examples for such morphologies of finite size are atomic clusters , spin clusters , the interface of demixed fluids , globular polymers or proteins , crystals , etc .
since many interesting systems such as heterogeneous biomolecules like proteins are `` small '' in a sense that a thermodynamic limit does not exist at all , it is useful to build up the analysis of transitions of small systems on the most general grounds .
these are , as we have argued above , best prepared by the microcanonical or caloric approach .
the folding of a protein is an example for a subtle structure formation process of a finite system , where effects on nanoscopic scales ( e.g. , hydrogen bonds being responsible for local secondary structures such as helices and sheets ) and cooperative behavior on mesoscopic scales ( such as the less well - defined hydrophobic effect which primarily drives the formation of the global , tertiary structures ) contribute to the stable assembly of the native fold which is connected to the biological function of the protein .
proteins are linear polypeptides , i.e. , they are composed of amino acids linked by peptide bonds .
there are twenty types of amino acids that have been identified in bioproteins , all of them differ in their chemical composition and thus possess substantial differences in their physical properties and chemical reactivity .
the mechanism of folding depends on the sequence of amino acids , not only the content , i.e. , a protein is a disordered system .
it is one of the central questions , which mutations of a given amino - acid sequence can lead to a relevant change of morphology and thus the loss of functionality .
the atomistic interaction types , scales , and ranges are different as well .
for this reason , in compact folds , frustration effects may occur .
disorder and frustration cause glassy behavior , but how glassy is a single protein and can this be generalized ? since details seem to be relevant for folding , it has commonly been believed that the folding of a certain protein is a highly individual process .
thus , it was a rather surprising discovery that the search for a stable fold can be a cooperative one - step process which can qualitatively and quantitatively be understood by means of the statistical analysis of a single effective , mesoscopic order parameter .
the free - energy landscape turned out to be very simple ( exhibiting only a single barrier between folded and unfolded conformations ) for this class of `` two - state folders '' @xcite .
of course , folding pathways for other proteins can be more complex as also intermediate states can occur @xcite .
however , this raises the question about cooperativity and the generalization of folding behavior in terms of conformational transitions similar to phase transitions in other fields of statistical mechanics . in the following
, we are going to discuss a structure formation process , the aggregation of a finite system of heteropolymers , by a general microcanonical approach in order to show how the conformational transition behavior of a small system is indeed related to thermodynamic phase transitions .
as an example for the occurrence of hierarchies of subphase transitions accompanying a nucleation process , we are going to discuss molecular aggregation @xcite by means of a simple coarse - grained hydrophobic - polar heteropolymer model @xcite .
in this so - called ab model , only hydrophobic ( a ) and hydrophilic ( b ) monomers line up in linear heteropolymer sequences . in the following , we consider the aggregation of four chains with 13 monomers @xcite .
all chains have the same fibonacci sequence @xmath25 @xcite .
folding and aggregation of this heteropolymer have already been subject of former studies @xcite . in the model used here ,
bonds between adjacent monomers have fixed length unity .
nonbonded monomers of individual chains but likewise monomers of different chains interact pairwisely via lennard - jones - like potentials .
the explicit form of the potentials depends on the types of the interacting monomers .
pairs of hydrophobic and unlike monomers attract , pairs of polar monomers repulse each other .
this effectively accounts for the hydrophobic - core formation of proteins in polar solvent .
details of our aggregation model and of the implementation of the multicanonical monte carlo simulation method are described in ref .
@xcite .
microcanonical entropy @xmath26 , the gibbs hull @xmath27 , and the deviation @xmath28 as functions of the energy per monomer.,width=340 ] the multicanonical computer simulations enabled us to obtain a precise estimate for the microcanonical entropy @xmath5 of the multiple - chain system @xcite , as shown in fig .
[ fig : mic ] as a function of the energy per monomer , @xmath29 .
the entropy curve is convex in the energetic aggregation transition region as expected for a first - order - like nucleation transition of a finite system @xcite .
the gibbs tangent @xmath27 , connecting the two coexistence points where concave and convex behavior change , provides the least possible overall concave shape of @xmath26 in this region .
the difference between the gibbs hull and the entropy curve , @xmath28 , is also shown in fig .
[ fig : mic ] .
not only the entropic suppression in the transition region is clearly visible , it is also apparent that the transition possesses an internal structure . in order to better understand the subphases , we discuss in the following the inverse derivative of the entropy , the microcanonical temperature ( [ eq : te ] ) , @xmath30 , which is plotted in fig . [ fig : tcal ] .
the slope of the gibbs tangent corresponds to the maxwell line in fig .
[ fig : tcal ] at the aggregation temperature @xmath31 .
therefore , the intersection points of the maxwell line and the temperature curve define the energetic phase boundaries @xmath32 and @xmath33 , respectively , as both phases coexist at @xmath34 .
microcanonical temperature @xmath30 as a function of energy per monomer , @xmath29 , where @xmath17 is the total number of all monomers in the system .
the horizontal maxwell line marks the aggregation temperature @xmath34 , obtained by the gibbs construction .
vertical dashed lines separate the different phases and subphases , respectively.,width=340 ] for energies @xmath35 , conformations of a single aggregate , composed of all four chains , dominate .
on the other hand , conformations with @xmath36 are mainly entirely fragmented , i.e. , all chains can form individual conformations , almost independently of each other .
the entropy is governed by the contributions of the individual translational entropies of the chains , outperforming the conformational entropies .
the translational entropies are only limited by the volume which corresponds to the simulation box size . the energetic difference @xmath37 , serving as an estimator for the latent heat , is obviously larger than zero , @xmath38 .
it thus corresponds to the total energy necessary to entirely melt the aggregate into fragments at the aggregation ( or melting ) temperature .
representative conformations in the different structural phases . ,
width=340 ] the relation between energy and microcanonical temperature in the aggregate phase and in the fragment phase is as intuitively expected : with increasing energy , also @xmath30 increases .
however , much more interesting is the behavior of the system in the energy interval @xmath39 , which represents the energetic nucleation transition region .
figure [ fig : tcal ] clearly shows that in our example @xmath30 changes the monotonic behavior three times in this regime .
representative conformations in the diffferent structural phases are shown in fig .
[ fig : pics ] . the change of monotony of the microcanonical temperature curve
is called backbending effect , because the temperature decreases while energy is increased .
this rather little intuitive behavior is a consequence of the suppression of entropy in this regime ( @xmath5 is convex in the backbending region ) .
the surface entropy per monomer vanishes in the thermodynamic limit @xcite . if two chains aggregate ( subphase 1 in fig .
[ fig : tcal ] ) , the total translational entropy of the individual chains is reduced by @xmath40 , where @xmath18 is the volume ( corresponding to the simulation box size ) , whereas the energy of the aggregate is much smaller than the total energy of the system with the individual chains separated .
thus , the energy associated to the interaction between different chains , i.e. , the cooperative formation of inter - chain contacts between monomers of different chains , is highly relevant here .
this causes the latent heat between the completely fragmented and the two - chain aggregate phase to be nonzero which signals a first - order - like transition .
this procedure continues when an additional chain joins a two - chain cluster .
energetically , the system enters subphase 2 .
qualitatively , the energetic and entropic reasons for the transition into this subphase are the same as explained before , since it is the same kind of nucleation process . in our example of four chains interacting with each other ,
there is another subphase 3 which also shows the described behavior .
the energetic width of each of the subphase transitions corresponds to the respective latent heat gained by adding another chain to the already formed cluster .
the subphase boundaries ( vertical dashed , gray lines in fig . [
fig : tcal ] ) have been defined to be the inflection points in the raising temperature branches , thus enclosing a complete `` oscillation '' of the temperature as a function of energy .
the energetic subphase transition points are located at @xmath41 and @xmath42 , respectively .
therefore , the latent heat associated to these subphase transitions is in all three cases about @xmath43 ( @xmath44 , @xmath45 ) , thus being one third of the total latent heat of the complete nucleation process .
this reflects the high systematics of subphase transitions in first - order nucleation processes .
the most interesting result from this heteropolymer aggregation study is that first - order phase transitions such as nucleation processes can be understood as a composite of hierarchical subphase transitions , each of which exhibits features of first - order - like transitions .
since with increasing number of chains the microcanonical entropy per chain converges to the gibbs hull in the transition region , the `` amplitudes '' of the backbending oscillations and the individual latent heats of the subphases become smaller and smaller @xcite . thus , in the thermodynamic limit , the heteropolymer aggregation transition is a first - order nucleation process composed of an infinite number of infinitesimally `` weak '' first - order - like subphase transitions .
supercomputer time has been provided by the forschungszentrum jlich under project nos .
hlz11 , jiff39 , and jiff43 . b. a. berg and t. neuhaus , phys .
b * 267 * , 249 ( 1991 ) ; phys .
lett . * 68 * , 9 ( 1992 ) .
w. janke , physica a * 254 * , 164 ( 1998 ) ; b. a. berg , fields inst .
* 26 * , 1 ( 2000 ) . k. hukushima and k. nemoto , j. phys . soc .
jpn . * 65 * , 1604 ( 1996 ) . c. j. geyer , in _ computing science and statistics _ ,
proceedings of the 23rd symposium on the interface , ed . by e. m. keramidas ( interface foundation , fairfax station , 1991 ) , p. 156 .
f. wang and d. p. landau , phys .
86 * , 2050 ( 2001 ) .
m. bachmann and w. janke , in : _ rugged free energy landscapes : common computational approaches to spin glasses , structural glasses and biological macromolecules _ , edited by w. janke , lect .
notes phys .
* 736 * ( springer , berlin , 2008 ) , p. 203 .
m. bachmann and w. janke , j. chem .
120 * , 6779 ( 2004 ) .
d. h. e. gross , _ microcanonical thermodynamics _ ( world scientific , singapore , 2001 ) .
s. e. jackson and a. r. fersht , biochemistry * 30 * , 10428 ( 1991 ) .
a. r. fersht , _ structure and mechanisms in protein science : a guide to enzyme catalysis and protein folding _
( freeman , new york , 1999 ) . s. schnabel , m. bachmann , and w. janke , phys . rev .
lett . * 98 * , 048103 ( 2007 ) .
s. schnabel , m. bachmann , and w. janke , j. chem .
* 126 * , 105102 ( 2007 ) . c. junghans , m. bachmann , and w. janke , phys .
lett . * 97 * , 218103 ( 2006 ) . c. junghans , m. bachmann , and w. janke , j. chem .
phys . * 128 * , 085103 ( 2008 ) . c. junghans , m. bachmann , and w. janke , europhys .
lett . * 87 * , 40002 ( 2009 ) .
f. h. stillinger , t. head - gordon , and c. l. hirshfeld , phys .
e * 48 * , 1469 ( 1993 ) ; f. h. stillinger and t. head - gordon , phys .
e * 52 * , 2872 ( 1995 ) .
m. bachmann , h. arkin , and w. janke , phys .
e * 71 * , 031906 ( 2005 ) .
|
we discuss the hierarchy of subphase transitions in first - order - like nucleation processes for an examplified aggregation transition of heteropolymers .
we perform an analysis of the microcanonical entropy , i.e. , the density of states is considered as the central statistical system quantity since it connects system - specific entropic and energetic information in a natural and unique way .
nucleation , first - order transition , polymer , structural phases , microcanonical analysis
|
the concept of dimensions with regard to stochastic processes has been studied in various contexts . for example
, motoo and watanabe @xcite considered a class @xmath0 of martingale additive functionals of general markov processes , and proved that there exists a basis @xmath1 of @xmath0 such that every element in @xmath0 can be represented as a sum of stochastic integrals based on @xmath1 and a discontinuous part .
this is a broad extension of the study by ventcel @xcite , where the brownian motion on @xmath2 was considered .
the number of elements constituting the basis is sometimes called the _ martingale dimension _ , which coincides with the usual dimension of the underlying state space in typical cases .
some related arguments are found in the papers by kunita and watanabe @xcite , cramr @xcite , and so on .
later , as a variation , davis and varaiya @xcite introduced the concept of _ multiplicity of filtration _ in a more abstract setting , that is , on filtered probability spaces . as another variation of these studies , in the first part of this article
, we introduce the concept of _ index _ of regular dirichlet forms @xmath3 using energy measures of functions in @xmath4 .
this definition is purely analytic .
we then prove some of its properties ; among others , we describe the identification of the index with the martingale dimension of the diffusion process associated with @xmath3 when the dirichlet form is strong local .
this fact has been proved by kusuoka @xcite when the underlying space is a self - similar fractal ( although the original definition of the index is equivalent to but different from that in this article ) .
our result is regarded as a natural generalization of kusuoka s one . in the latter part of this article
, we focus on the case that the underlying space @xmath5 is a fractal set .
in such a case , it is difficult to obtain an explicit value of the index ( in other words , the martingale dimension ) of canonical dirichlet forms , unlike the case of spaces with differential structures . thus far ,
the only result that has been obtained is that dirichlet forms @xmath3 associated with regular harmonic structures on post - critically finite ( p.c.f . )
self - similar fractals with some extra assumptions have an index of one ( @xcite ) . as an application of this fact , we prove that every function @xmath6 in @xmath4 has a `` first - order derivative '' almost everywhere with respect to a minimal energy - dominant measure , and the total energy of @xmath6 ( namely , @xmath7 ) is described as the square integral of the derivative .
this representation appears as if the dirichlet form were defined on a one - dimensional object , which can be regarded as a reflection of the property that the index is one . here
, of course , derivatives in the usual sense do not exist since the underlying space @xmath5 does not have a differential structure . instead , we take a good reference function @xmath8 from @xmath4 and define the derivative of @xmath9 as the infinitesimal ratio of the fluctuation of two functions @xmath6 and @xmath8 . from another viewpoint
, we have a taylor expansion of @xmath6 with respect to @xmath8 up to the first order . the concept that the total energy of @xmath6 can be described as a square integral of a type of vector - valued gradient
was first introduced by kusuoka @xcite ( see also @xcite for a related work ) .
the contribution of this paper is the observation that only one good reference function is sufficient for the study of the infinitesimal behaviour of every function in @xmath4 and almost every point in @xmath5 when the index is one .
recently , pelander and teplyaev @xcite studied a topic similar to the one addressed in this article and proved the existence of derivatives of a type of smooth function on some self - similar fractals . in their results ,
the existence of the derivative is assured almost everywhere with respect to self - similar measures , while in this paper , the derivative is proved to exist almost everywhere with respect to some energy measures . since self - similar measures and energy measures are mutually singular in many cases ( @xcite ) , these results are not comparable . as a general theory ,
a good reference function can be taken from a certain dense subset of @xmath4 .
exactly what function we can take is not obvious .
we will provide a partial answer to this problem ; in the case of non - degenerate generalized sierpinski gaskets , we prove that every nonconstant harmonic function can be taken as a reference function . as a byproduct of this result , the energy measures of arbitrary nonconstant harmonic functions are mutually absolutely continuous .
this fact is new even for the case of a two - dimensional standard sierpinski gasket .
this article is organized as follows . in section 2
, we introduce the concept of index of regular dirichlet forms and study its properties . in section 3 , the index is characterized in terms of the martingale dimension when the dirichlet form is strong local . in section 4 ,
we review the theory of self - similar dirichlet forms on p.c.f . self - similar fractals ; we then discuss the index of such fractals and related topics . in the last section ,
we prove the existence of derivatives of functions on p.c.f .
fractals when the index is one . some properties of sierpinski gaskets are also investigated in detail .
let @xmath5 be a locally compact , separable , and metrizable space , and @xmath10 , a @xmath11-finite borel measure on @xmath5 with full support .
we denote the borel @xmath11-field of @xmath5 by @xmath12 .
let @xmath13 denote the set of all continuous real - valued functions on @xmath5 , and @xmath14 , the set of all functions in @xmath13 with compact support . for @xmath15
, @xmath16 denotes the real @xmath17-space on the measure space @xmath18 .
suppose that we are given a symmetric regular dirichlet form @xmath3 on @xmath19 .
that is , @xmath4 is a dense subspace of @xmath19 , @xmath20 is a nonnegative definite symmetric bilinear form that satisfies the following : * ( closedness ) if a sequence @xmath21 in @xmath4 satisfies @xmath22 then there exists @xmath9 such that @xmath23 . here , we define @xmath24 * ( markov property ) for any @xmath9 , @xmath25 belongs to @xmath4 and @xmath26 . *
( regularity ) the space @xmath27 is dense in @xmath4 with norm @xmath28 and dense in @xmath14 with uniform norm .
the space @xmath4 becomes a hilbert space under inner product @xmath29 .
hereafter , the topology of @xmath4 is always considered as that derived from norm @xmath30 .
we write @xmath31 for @xmath32 for simplicity . for each @xmath9
, we can define a positive finite borel measure @xmath33 on @xmath5 as follows ( @xcite ) : when @xmath6 is bounded , @xmath33 is determined by the identity @xmath34 by utilizing the inequality @xmath35 for any borel subset @xmath36 of @xmath5 and @xmath6 , @xmath37 ( cf .
* , and ( 3.2.13 ) and ( 3.2.14 ) in p. 110 ) ) , we can define @xmath33 for any @xmath9 by a limiting procedure .
then , equation still holds for any @xmath38 .
the measure @xmath33 is called the _ energy measure _ of @xmath6 . ( in the textbook @xcite and the papers by the author @xcite , the symbol @xmath39 has been used to denote the energy measure of @xmath6 . in this paper
, we use the symbol @xmath33 instead . ) for @xmath38 , the mutual energy measure @xmath40 , which is a signed borel measure on @xmath5 , is defined as @xmath41 then , @xmath42 and @xmath40 is bilinear in @xmath6 , @xmath8 ( @xcite ) .
the following inequality is also useful in the subsequent argument : for @xmath38 and a borel subset @xmath36 of @xmath5 , latexmath:[\[\label{eq : schwarz } for two @xmath11-finite ( or signed ) borel measures @xmath44 and @xmath45 on @xmath5 , we write @xmath46 if @xmath44 is absolutely continuous with respect to @xmath45 , and @xmath47 if @xmath44 and @xmath45 are mutually singular .
the measure in the following definition is a natural generalization of the ones that were defined by kusuoka @xcite .
[ def : mdem ] a @xmath11-finite borel measure @xmath48 on @xmath5 is called a _ minimal energy - dominant measure _ of @xmath3 if the following two conditions are satisfied . 1 . ( domination ) for every @xmath49 , @xmath50 .
2 . ( minimality )
if another @xmath11-finite borel measure @xmath51 on @xmath5 satisfies condition ( i ) with @xmath48 replaced by @xmath51 , then @xmath52 . by definition , two minimal energy - dominant measures are mutually absolutely continuous . from , @xmath53 for a minimal energy - dominant measure @xmath48 and @xmath38 .
as observed from proposition [ prop : dense ] below , minimal energy - dominant measures can be realized by energy measures of some functions in @xmath4 .
[ lem : basic ] let @xmath21 be a sequence of functions in @xmath4 , @xmath9 , and let @xmath48 be a @xmath11-finite borel measure on @xmath5 .
suppose that @xmath54 for every @xmath55 and @xmath56 converges to @xmath6 in @xmath4 as @xmath57 .
then , @xmath58 .
suppose @xmath59 .
then , @xmath60 for all @xmath61 . from , @xmath62 therefore , @xmath63 .
this implies the claim .
[ lem : mdem ] let @xmath21 be a sequence of functions in @xmath4 such that the linear span of @xmath64 is dense in @xmath4 . take a sequence @xmath65 of positive real numbers such that @xmath66 converges .
define @xmath67 .
then , @xmath48 is a minimal energy - dominant measure . if @xmath6 is a linear combination of @xmath68 for some @xmath69 , then @xmath58 by using . from lemma [ lem : basic ] ,
@xmath58 for all @xmath9 .
therefore , assumption ( i ) in definition [ def : mdem ] holds .
next , suppose that a @xmath11-finite borel measure @xmath51 on @xmath5 satisfies ( i ) in definition [ def : mdem ] with @xmath48 replaced by @xmath51
. then , @xmath70 for any @xmath55 .
since @xmath71 for @xmath72 , we obtain @xmath52 .
therefore , assumption ( ii ) in definition [ def : mdem ] holds .
[ lem : null ] let @xmath48 be a minimal energy - dominant measure and @xmath36 , a borel set of @xmath5 . then , @xmath59 if and only if @xmath73 for every @xmath9 .
the only if part is obvious from the definition of the minimal energy - dominant measure .
suppose that @xmath73 for every @xmath9 .
take @xmath21 and @xmath65 as given in lemma [ lem : mdem ] and @xmath74 .
then , @xmath51 is a minimal energy - dominant measure by lemma [ lem : mdem ] .
since @xmath51 and @xmath48 are mutually absolutely continuous and @xmath75 , we obtain @xmath59 . for a signed borel measure @xmath44 and a @xmath11-finite borel measure @xmath45 on @xmath5 such that @xmath46
, we denote the radon nikodym derivative of @xmath44 with respect to @xmath45 by @xmath76 ( or @xmath77 ) .
[ lem : energyineq ] let @xmath48 be a minimal energy - dominant measure and @xmath38 . 1 .
it holds that @xmath78 @xmath48-a.e .
in particular , @xmath79 2 .
let @xmath21 and @xmath80 be sequences in @xmath4 and @xmath81 and @xmath82 in @xmath4 as @xmath57 .
then , @xmath83 converges to @xmath84 in @xmath85 .
\(i ) since @xmath86 @xmath48-a.e . for all @xmath87
, we have , for @xmath48-a.e .
, for all @xmath88 , @xmath89 therefore , @xmath90 the assertions are derived from this inequality .
\(ii ) since @xmath91 , it is sufficient to consider the case that @xmath92 for all @xmath61 and @xmath93 . from , we have @xmath94 which converges to @xmath95 as @xmath57 .
let @xmath21 be a sequence in @xmath4 and @xmath48 , a minimal energy - dominant measure . for @xmath96 ,
denote the radon nikodym derivative @xmath97 by @xmath98 .
when @xmath48 is a finite measure , one of the concrete ways to construct @xmath98 is as follows .
let @xmath99 be a sequence of @xmath11-fields on @xmath5 such that @xmath100 , @xmath101 , and each @xmath102 is generated by finitely many borel subsets of @xmath5 .
for each @xmath103 , @xmath102 is provided by a partition of @xmath5 consisting of finitely many disjoint borel sets @xmath104 for some @xmath105 .
then , for each @xmath96 , the radon nikodym derivative @xmath106 of @xmath107 with respect to @xmath108 is defined by @xmath109 here , @xmath110 by convention .
we define @xmath111 from the martingale convergence theorem , @xmath112 . for each @xmath113
, we define @xmath114 then , @xmath98 is a borel measurable function on @xmath5 and it is equal to @xmath97 .
we thus have the following claim which is evident from the manner in which @xmath115 is constructed .
[ lem : z ] 1 . for every @xmath113 and @xmath116 , @xmath117 .
2 . for every @xmath118 and @xmath116 ,
the @xmath119 matrix @xmath120 is symmetric and nonnegative definite .
even if @xmath48 is an infinite measure , it is evident to see that we can define a version of @xmath121 such that the claims of lemma [ lem : z ] hold .
hereafter , we always take a version of the radon nikodym derivatives @xmath98 for given @xmath21 and @xmath48 in this manner .
denote the usual real @xmath122-space by @xmath122 , namely , @xmath123 the canonical inner product and the norm on @xmath122 will be denoted by @xmath124 and @xmath125 , respectively .
we define @xmath126 [ prop : dense ] @xmath127 is dense in @xmath4 .
it is sufficient to prove that @xmath128 @xmath48-a.e .
for @xmath6 in some dense subset of @xmath4 , where @xmath48 is an arbitrarily fixed minimal energy - dominant measure . take a c.o.n.s .
@xmath21 in @xmath4 .
define @xmath129 . from lemma [
lem : mdem ] , @xmath48 is a minimal energy - dominant measure with @xmath130 . we define @xmath131 and @xmath98 for @xmath96 as and , respectively .
then , by construction , @xmath132 for @xmath133 . in particular
, the following holds : @xmath134 fix a gaussian measure @xmath135 on @xmath122 such that the support of @xmath135 is @xmath122 and @xmath135 does not charge any proper closed subspaces of @xmath122 .
let @xmath136 and define @xmath137 here , the infinite sum given above converges in @xmath4 .
indeed , @xmath138 we denote the map @xmath139 by @xmath140 , which is a contraction operator .
then , from lemma [ lem : energyineq ] ( ii ) , @xmath141 converges to @xmath142 in @xmath85 . on the other hand , for @xmath116 , @xmath143 this implies that @xmath144 converges as @xmath145 for @xmath116 .
the limit must be @xmath146 for @xmath48-a.e.@xmath147 .
we define @xmath148 .
fix @xmath116 and take @xmath149 and @xmath150 from @xmath122 .
then , @xmath151 converges absolutely .
indeed , as in , @xmath152 from this domination , @xmath153 provides a bounded nonnegative definite and symmetric bilinear form on @xmath122 .
then , there exists a bounded nonnegative definite and symmetric operator @xmath154 on @xmath122 such that @xmath155 the kernel of @xmath154 , which is denoted by @xmath156 , is equal to @xmath157 . when @xmath158 , @xmath159 from , which implies that @xmath156 is a proper closed subspace of @xmath122
. therefore , @xmath160 for @xmath158 , in particular , for @xmath48-a.e .
@xmath147 .
now , we set @xmath161 this is a borel subset of @xmath162 .
the above observation together with the fubini theorem implies @xmath163 .
then , @xmath164 for @xmath135-a.e .
@xmath165 , where @xmath166 .
consequently , there exists some @xmath167 with @xmath168 such that for @xmath169 , @xmath170 @xmath48-a.e . , where @xmath171 .
that is , @xmath172 is a minimal energy - dominant measure for such @xmath8 .
the map @xmath173 is contractive , and @xmath174 is dense in @xmath4 .
since @xmath175 is dense in @xmath122 , @xmath176 is also dense in @xmath4 .
this completes the proof . in the proof of proposition [ prop : dense ] , it is not necessary for the measure @xmath135 to be gaussian .
it is sufficient for the proof that @xmath135 has a full support and that it does not charge any proper closed subspaces of @xmath122 .
fix a minimal energy - dominant measure @xmath48 of @xmath3 .
let @xmath177 denote the set of all nonnegative integers .
[ def : index ] 1 .
the _ pointwise index _ @xmath178 of @xmath3 is defined as a measurable function on @xmath5 taking values in @xmath179 such that the following hold : 1 . for any @xmath69 and any @xmath180 , @xmath181 2 . if another function @xmath182 satisfies ( a ) with @xmath178 replaced by @xmath182 , then @xmath183 @xmath48-a.e.@xmath147 .
the _ index _ @xmath184 of @xmath3 is defined as @xmath185 .
in other words , @xmath184 is the smallest number satisfying the following : for any @xmath69 and any @xmath180 , @xmath186 this definition is independent of the choice of a minimal energy - dominant measure @xmath48 .
the pointwise index @xmath178 is unique up to @xmath48-equivalence .
its existence is assured by the following proposition .
[ prop : rank ] let @xmath21 be a sequence of functions in @xmath4 such that the linear span of @xmath21 is dense in @xmath4 . denote the radon nikodym derivative @xmath97 by @xmath98 for @xmath96 .
for each @xmath69 and @xmath116 , define an @xmath119 matrix @xmath187 by @xmath188 then , @xmath189 is the pointwise index of @xmath3 .
it is sufficient to prove that for any @xmath190 and @xmath191 , @xmath192 for @xmath48-a.e.@xmath147 , where @xmath193 .
denote the linear span of @xmath21 by @xmath194 .
first , suppose that @xmath195 for every @xmath196 .
there exist some @xmath197 for some @xmath69 such that @xmath198 for every @xmath199 .
let @xmath200 be an @xmath201 matrix whose @xmath202-th component is @xmath203 .
then , for @xmath204 , @xmath205 which is equal to the @xmath206-th component of the matrix @xmath207
. therefore , @xmath208 for @xmath48-a.e.@xmath147 . next
, suppose that @xmath209 for every @xmath196 .
take @xmath210 from @xmath194 such that @xmath211 in @xmath4 for each @xmath196 .
then , for every @xmath196 and @xmath212 , @xmath213 converges to @xmath214 in @xmath85 as @xmath57 from lemma [ lem : energyineq ] ( ii ) . by taking a subsequence if necessary
, we may assume that this convergence is also in @xmath48-a.e . sense .
from the lower semi - continuity of @xmath215 , we have @xmath216 this completes the proof .
[ prop : zero ] the pointwise index @xmath178 is greater than @xmath95 for @xmath48-a.e.@xmath147 . in particular , unless @xmath217 for every @xmath9 , the index @xmath184 is greater than @xmath95 .
let @xmath218 .
take any @xmath9 . since the rank of the @xmath219 matrix @xmath220 is @xmath95 for @xmath48-a.e.@xmath147 in @xmath36 , @xmath221 @xmath48-a.e . on @xmath36 .
therefore , @xmath63 . from lemma [ lem :
null ] , @xmath59 .
the latter assertion arises from the fact that @xmath48 does not vanish unless @xmath217 for every @xmath9 . for @xmath222 ,
let @xmath223 and @xmath224 denote the standard inner product and the euclidean norm on @xmath225 , respectively .
[ prop : squareroot ] let @xmath222 .
let @xmath21 and @xmath98 be the same as in proposition [ prop : rank ] .
then , the index of @xmath3 is less than or equal to @xmath184 if and only if there exists a sequence @xmath226 of @xmath225-valued measurable functions on @xmath5 such that for @xmath48-a.e.@xmath147 , @xmath227 , and @xmath228 for all @xmath229 and @xmath230 , where @xmath231 and @xmath178 is the pointwise index of @xmath3 .
first , we prove the if part .
let @xmath232 denote the @xmath233 matrix whose @xmath234-th component is @xmath235 .
then , the @xmath119 matrix @xmath236 is described as @xmath237 for @xmath48-a.e.@xmath147 , and thus its rank is less than or equal to @xmath178 .
therefore , the index of @xmath3 does not exceed @xmath184 .
next , we prove the only if part .
we may assume that @xmath178 is defined as in proposition [ prop : rank ] .
let @xmath238 and @xmath239 for @xmath240 , there exist @xmath241 such that @xmath242 for every @xmath243 .
indeed , @xmath244 is decomposed as @xmath245 , where @xmath246 is an orthogonal matrix of size @xmath247 and @xmath248 is a real diagonal matrix of size @xmath247 such that @xmath249 for all @xmath61 and @xmath250 when @xmath251 .
then , by letting @xmath252 , we have @xmath253 and all the components of the @xmath61-th row of @xmath254 are zero for @xmath251 .
therefore , it suffices to define @xmath255 so that its @xmath61-th component is the @xmath234-th component of @xmath254 .
note also that @xmath256 .
now , let @xmath257 which is regarded as a closed subset of @xmath258 ( with the product topology ) . for each @xmath259 ,
let @xmath260 we will prove @xmath261 . for each @xmath69 , take @xmath262 such that @xmath263 for all @xmath264 . by the diagonalization argument , we can take an increasing sequence @xmath265 such that @xmath266 converges to some @xmath267 for all @xmath55
. then , @xmath268 belongs to @xmath269 . in the same manner
, we can prove that @xmath269 is a compact set of @xmath270 with the product topology .
moreover , if a sequence @xmath271 in @xmath272 converges to some @xmath244 and @xmath273 for each @xmath274 , then @xmath275 has a limit point in @xmath269 .
therefore , by the measurable selection theorem ( see e.g. ( * ? ? ?
* lemma 12.1.8 and theorem 12.1.10 ) ) , there exists a borel measurable map @xmath276 such that @xmath277 for all @xmath278 . for @xmath279 ,
define the map @xmath280 by @xmath281 where @xmath282 .
then , it suffices to define @xmath226 by @xmath283 when @xmath284 , the only if part of proposition [ prop : squareroot ] is proved in a simpler and more elementary manner as follows .
we define @xmath285 and @xmath286 inductively .
it holds that @xmath287 .
we define @xmath288 then , when @xmath289 for @xmath103 , @xmath290 , @xmath96 .
since @xmath291 we have @xmath292 therefore , @xmath293 for @xmath294 .
we will discuss the stability of the pointwise index . recall that a regular dirichlet form @xmath3 on @xmath19 is called _ strong local _ if @xmath295 for any @xmath38 as long as both @xmath296 and @xmath297 are compact and @xmath8 is constant on a neighborhood of @xmath296 .
here , @xmath296 is defined as the support of the measure @xmath298 on @xmath5 .
[ prop : stability ] suppose that @xmath3 and @xmath299 are both strong local regular dirichlet forms on @xmath19 .
if these are equivalent , namely , @xmath300 and there exist positive constants @xmath301 and @xmath302 such that @xmath303 then @xmath48 is also a minimal energy - dominant measure of @xmath304 , and the pointwise indices of @xmath3 and @xmath299 coincide for @xmath48-a.e .
@xmath147 .
let @xmath305 denote the energy measure of @xmath6 with respect to @xmath299 . from (
* ? ? ? * proposition 1.5.5 ( b ) ) ( or @xcite ) , @xmath306 therefore , @xmath48 is also a minimal energy - dominant measure of @xmath299 .
take any @xmath307 , @xmath69 .
we define @xmath308 and @xmath309 . for @xmath310 , denote the @xmath274-th eigenvalue of @xmath187 ( resp .
@xmath311 ) from below by @xmath312 ( resp . @xmath313 ) . from the minmax principle , @xmath314 here
, @xmath315 is the grassmann manifold consisting of all @xmath274-dimensional subspaces of @xmath316 .
note that by the separability of @xmath315 , we may replace @xmath315 and @xmath317 in the above equations by countable dense subsets .
since @xmath318 for @xmath319 , from , @xmath320 therefore , @xmath321 for @xmath48-a.e.@xmath147 .
this implies that @xmath322 for @xmath48-a.e.@xmath147 , which completes the proof .
let @xmath323 and @xmath10 be the lebesgue measure @xmath324 . define a dirichlet form @xmath3 on @xmath325 as @xmath326 where @xmath327 is the first - order @xmath328-sobolev space on @xmath2 .
then , it is evident that @xmath329 for @xmath38 , and the lebesgue measure can be taken as a minimal energy - dominant measure . from proposition
[ prop : squareroot ] , the index is less than or equal to @xmath330 . for @xmath331 ,
let @xmath332 be a @xmath333-function on @xmath334 with compact support such that @xmath335 on @xmath336 $ ] and @xmath337 on @xmath338 $ ] .
take @xmath339 , @xmath340 , @xmath341 .
then , @xmath342 and @xmath343 therefore , @xmath344 when @xmath345 .
since @xmath346 is arbitrary , the pointwise index is @xmath330 @xmath324-a.e .
and the index is @xmath330 . from proposition [ prop : stability ] ,
the same is true for the dirichlet form @xmath347 on @xmath325 defined by @xmath348 where @xmath349 is a @xmath350 matrix valued measurable function on @xmath2 such that there exist some positive numbers @xmath351 and @xmath352 satisfying @xmath353 when @xmath349 is degenerate or unbounded , the pointwise index @xmath178 should be equal to @xmath354 @xmath324-a.e .
as long as the domain of the dirichlet form contains sufficiently many functions so that the argument similar to the above one is valid . in the example above , the index can be calculated easily because the dirichlet form is given by the square integral of the gradient . otherwise , determining the index is not straightforward and it may be difficult to determine .
for instance , it is an open problem to determine the index of the canonical dirichlet forms on sierpinski carpets , which are typical infinitely ramified self - similar fractals .
in this section , we discuss the probabilistic interpretation of the index of @xmath3 . for this purpose ,
let us review the theory of additive functionals associated with @xmath3 on @xmath19 , following ( * ? ? ?
* chapter 5 ) .
the capacity @xmath355 associated with @xmath3 is defined as @xmath356 if @xmath246 is an open subset of @xmath5 , and @xmath357 for general subsets @xmath36 of @xmath5 .
a set @xmath358 with @xmath359 is called an _ exceptional set_. a statement depending on @xmath116 is said to hold for _ q.e . _ ( quasi - every ) @xmath147 if the set of @xmath147 for which the statement is not true is an exceptional set . in what follows ,
we consider only the case that @xmath3 is strong local . from the general theory of regular dirichlet forms
, we can construct a diffusion process @xmath360 on @xmath361 defined on a filtered probability space @xmath362 associated with @xmath3 .
here , @xmath363 is a one - point compactification of @xmath5 and @xmath364 is a minimum completed admissible filtration .
any numerical function @xmath6 on @xmath5 extends to @xmath361 by letting @xmath365 .
the relationship between @xmath360 and @xmath3 is explained in such a way that the operator @xmath366 $ ] produces the semigroup associated with @xmath3 , where @xmath367 denote the expectation with respect to @xmath368 .
we may assume that for each @xmath369 , there exists a shift operator @xmath370 that satisfies @xmath371 for all @xmath372 .
denote the life time of @xmath373 by @xmath374 .
a @xmath375$]-valued function @xmath376 , @xmath369 , @xmath377 , is referred to as an _ additive functional _ if the following conditions hold : * @xmath378 is @xmath379-measurable for each @xmath380 .
* there exist a set @xmath381 and an exceptional set @xmath382 such that @xmath383 for all @xmath384 and @xmath385 for all @xmath386 ; moreover , for each @xmath387 , @xmath388 is right continuous and has the left limit on @xmath389 , @xmath390 , @xmath391 for all @xmath392 , @xmath393 for @xmath394 , and @xmath395 the sets @xmath396 and @xmath247 referred to above are called a _ defining set _ and an _ exceptional set _ of the additive functional @xmath244 , respectively .
a finite ( resp .
continuous ) additive functional is defined as an additive functional such that @xmath397 ( resp .
@xmath388 is continuous ) on @xmath398 for @xmath387 .
a @xmath399$]-valued continuous additive functional is referred to as a positive continuous additive functional . from (
* theorems 5.1.3 and 5.1.4 ) , for each positive continuous additive functional @xmath244 , there exists a unique measure @xmath400 on @xmath5 ( termed the revuz measure of @xmath244 ) such that the following identity holds for any @xmath386 and nonnegative borel functions @xmath6 and @xmath401 on @xmath5 : @xmath402h(x)\,\mu(dx ) = \int_0^t \int_k e_x\left[h(x_s)\right]f(x)\,\mu_a(dx)\,ds.\nonumber \end{aligned}\ ] ] further , if two positive continuous additive functionals @xmath403 and @xmath404 have the same revuz measures , then @xmath403 and @xmath404 coincide up to the natural equivalence .
let @xmath405 be a measure on @xmath406 defined as @xmath407 .
let @xmath408 denote the integration with respect to @xmath405 .
we define the energy @xmath409 of additive functional @xmath244 as @xmath410\ ] ] if the limit exists .
let @xmath0 be the space of martingale additive functionals of @xmath360 that is defined as @xmath411<\infty$ and $ e_x[m_t]=0 $ for q.e.\ $ x\in k$}\end{array}\!\ !
\right.\right\}.\ ] ] due to the assumption that @xmath3 is strong local , any @xmath412 is in fact a continuous additive functional ( ( * ? ? ?
* lemma 5.5.1 ( ii ) ) ) .
each @xmath412 admits a positive continuous additive functional @xmath413 referred to as the quadratic variation associated with @xmath414 that satisfies @xmath415=e_x[m_t^2 ] , \ t>0\text { for q.e.\ $ x\in k$},\ ] ] and the following equation holds : @xmath416 we set @xmath417 .
then , @xmath418 is a hilbert space with inner product @xmath419 ( ( * ? ? ?
* theorem 5.2.1 ) ) . for @xmath420 , we set @xmath421 . since @xmath422 is bilinear and symmetric with respect to @xmath423 and @xmath424 is a positive measure , for any nonnegative function @xmath6 in @xmath425 , it holds that @xmath426 and @xmath427 for @xmath428 and @xmath429 , we can define the stochastic integral @xmath430 ( ( * ? ? ?
* theorem 5.6.1 ) ) , which is a unique element in @xmath418 such that @xmath431 we may write @xmath432 for @xmath430 since @xmath433 , @xmath386 , @xmath368-a.e . for q.e.@xmath116 as long as @xmath6 is a continuous function with compact support on @xmath5 ( ( * ? ? ?
* lemma 5.6.2 ) ) . from (
* lemma 5.6.2 ) , we also have @xmath434 the space @xmath435 of the continuous additive functionals of zero energy is defined as @xmath436<\infty$ for q.e.\ $ x\in k$ and $ t>0 $ } \end{array}\!\right\}\right .. \ ] ] for each @xmath437 , we have a quasi - continuous modification @xmath438 of @xmath439 in the restricted sense , that is , @xmath440 @xmath10-a.e . , and
for any @xmath441 , there exists an open subset @xmath254 of @xmath5 such that @xmath442 and @xmath443 is continuous ( ( * ? ? ? * theorem 2.1.3 ) ) .
then , the fukushima decomposition theorem ( * ? ? ?
* theorem 5.2.2 ) says that there exist unique @xmath444}\in{{\stackrel{\circ}{\smash{\cal m}\rule{0pt}{1.3ex}}}}$ ] and @xmath445}\in{{\cal n}}_c$ ] such that @xmath446}+n_t^{[u ] } , \quad t>0,\ p_x\mbox{-a.e.}\mbox { for q.e.\,}x.\ ] ] from ( * ? ? ?
* theorem 5.2.3 ) , @xmath447}{\rangle}}$ ] is equal to the energy measure @xmath448 of @xmath439 .
therefore , @xmath449},m^{[v]}{\rangle}}=\nu_{u , v},\quad u , v\in{{\cal f}}.\ ] ] we recall the following claim .
[ lem : density ] let @xmath450 be a dense subset of @xmath14 with uniform norm and @xmath451 , a dense subset of @xmath4 . then , @xmath452}\mid
u\in c_1,\ v\in { { \cal f}}_1\}$ ] is dense in @xmath453 . by using this lemma , we can prove some basic properties for revuz measures as follows .
[ lem : m ] let @xmath48 be a minimal energy - dominant measure of @xmath3 , and let @xmath454 .
then , the following hold . 1 .
@xmath455 is absolutely continuous with respect to @xmath48 , @xmath456 and @xmath457 2 . if @xmath458 and @xmath459 in @xmath453 as @xmath460 , then @xmath461 ( i ) : when @xmath462}$ ] for @xmath463 and @xmath464 , @xmath465 , which is absolutely continuous with respect to @xmath48 . from lemma [
lem : density ] , , and , @xmath466 for all @xmath428 .
in addition , @xmath467 from .
equations and follow from an argument similar to the proof of lemma [ lem : energyineq ] ( i ) .
( ii ) : this claim is proved in exactly the same manner as lemma [ lem : energyineq ] ( ii ) .
the _ af - martingale dimension _ of @xmath360 ( or of @xmath3 ) is defined as a smallest number @xmath184 in @xmath177 satisfying the following : there exists a sequence @xmath468 in @xmath418 such that every @xmath428 has a stochastic integral representation @xmath469 where @xmath470 for each @xmath471 .
if such @xmath184 does not exist , the af - martingale dimension is defined as @xmath472 .
this definition is basically consistent with the works by motoo and watanabe @xcite and davis and varaiya @xcite . from now on
, we will omit the symbol af ( an abbreviation of `` additive functionals '' ) and only write martingale dimension .
[ th : index ] the index of @xmath3 coincides with the martingale dimension of @xmath360 .
this theorem is a natural generalization of ( * ? ? ?
* theorem 6.12 ) .
the remainder of this section describes the proof of this theorem .
let @xmath473 fix a minimal energy - dominant measure @xmath48 with @xmath130 and a sequence @xmath21 in @xmath4 such that the linear span of @xmath21 is dense in @xmath4 . denote the radon nikodym derivative @xmath97 by @xmath98 for @xmath96 .
for each @xmath116 , define pre - inner product @xmath474 on @xmath475 as @xmath476 for @xmath116 , let @xmath477 be the set of all equivalent classes of @xmath475 with respect to the equivalent relation @xmath478 derived from pre - inner product @xmath474 .
that is , @xmath479 , where @xmath480 if and only if @xmath481 .
[ lem : dimension ] the dimension of @xmath477 is equal to the pointwise index @xmath178 of @xmath3 for @xmath48-a.e.@xmath147 .
fix @xmath116 and take @xmath69 .
let @xmath482 .
then , it is an elementary fact that @xmath483 where @xmath484 is defined as @xmath485 for @xmath486 and @xmath487 . by letting @xmath145
, we obtain @xmath488 for @xmath48-a.e.@xmath147 from proposition [ prop : rank ] .
let @xmath489 note that for @xmath490 , @xmath491 belongs to @xmath475 for each @xmath116 .
we define a pre - inner product @xmath492 on @xmath493 by @xmath494 for @xmath495 , we define @xmath496}\in{{\stackrel{\circ}{\smash{\cal m}\rule{0pt}{1.3ex}}}}.\end{aligned}\ ] ] we note that the sum above is in fact a finite sum .
[ lem : isometry ] the map @xmath497 preserves the products . for @xmath495 and @xmath498 , from , , and , @xmath499},m^{[f_j]}{\rangle}}(dx)\\ & = \frac12\sum_{i , j}\int_k g_i(x ) g'_j(x ) z^{i
, j}(x)\,\nu(dx)\\ & = { \langle}g , g'{\rangle}_{z}. \end{aligned}\ ] ] this completes the proof . by virtue of lemma [ lem : density ] and the fact that the linear span of @xmath21 is dense in @xmath4 , @xmath500 is dense in @xmath418 .
define @xmath501 [ lem : onb ] suppose that the index @xmath184 of @xmath3 is finite and nonzero . then , there exist @xmath502 , @xmath471 such that @xmath503 where @xmath178 is the pointwise index of @xmath3 . the proof is based on the gram schmidt orthogonalization .
let @xmath504 which is dense in @xmath477 for @xmath48-a.e.@xmath116 . here , by abuse of notation ,
an element of @xmath475 is also regarded as its equivalent class in @xmath477 .
we regard each element of @xmath505 as an @xmath475-valued constant function on @xmath5 and @xmath505 as a subset of @xmath506 . since @xmath505 is a countable set , we can write @xmath507 and @xmath508 .
we define a map @xmath509 by @xmath510 we define @xmath511 and @xmath512 inductively . by taking account of lemma
[ lem : dimension ] , for @xmath48-a.e.@xmath147 , there exists a unique @xmath513 , @xmath514 for each @xmath61 , such that @xmath515 set @xmath516 for @xmath471 . here , note that @xmath517 depends on @xmath147 .
then , @xmath518 satisfy the desired properties .
first , we prove that the martingale dimension is less than or equal to the index @xmath184 .
it suffices to assume @xmath519 .
take @xmath518 in lemma [ lem : onb ] , and define @xmath520 then , @xmath521 is a nondecreasing sequence and @xmath522 .
let @xmath523 and @xmath524 .
keeping in mind that @xmath525 , we define @xmath526}.\ ] ] note that @xmath527 in the above equation can be replaced by @xmath528 . for @xmath529
, we have @xmath530}\right)\nonumber\\ & = \frac12 \sum_{i , j=1}^\infty \int_{k_m\setminus k_l}g_i^{(k)}g_j^{(k)}\,d\mu_{{\langle}m^{[f_i]},m^{[f_j]}{\rangle}}\nonumber\\ & = \frac12\int_{k_m\setminus k_l}\left(\sum_{i , j=1}^\infty g_i^{(k)}g_j^{(k)}z^{i , j}\right)\,d\nu \qquad\mbox{(from { \rom{(\ref{eq : maf } ) } } ) } \nonumber\\ & = \frac12\int_{k_m\setminus k_l}1_{\{p(\cdot)\ge k\}}(x)\,\nu(dx ) \qquad\mbox{(from { lemma~\ref{lem : onb}})}\nonumber\\ & \to 0 \qquad\mbox{as } m > l\to\infty.\end{aligned}\ ] ] here , the infinite sums are actually finite sums . from this , @xmath531 is a cauchy sequence in @xmath453 .
we denote the limit by @xmath532 . from a calculation similar to
, we have @xmath533 and @xmath534 for @xmath535 , for every @xmath524 . by lemma
[ lem : m ] ( ii ) , we obtain @xmath536 and @xmath537 by letting @xmath538 .
now , we will prove that any @xmath428 is expressed as @xmath539 here , since @xmath540 from lemma [ lem : m ] ( i ) and , @xmath541 and @xmath542 is well - defined . to begin with ,
suppose that @xmath414 is described as @xmath543},\mbox { where $ i\in{{\bbb n}}$ and $ u$ is a bounded borel function on $ k$}\\ \mbox{such that $ u=0 $ on $ k\setminus k_l$ for some $ l\in{{\bbb n}}$}. \end{array}\ ] ] we may assume @xmath544 . in order to prove , it suffices to prove that the revuz measures of both sides coincide .
we have @xmath545 and @xmath546 since @xmath547},m_l^{(k)}{\rangle}}(dx)=1_{k_l}(x)\mu_{{\langle}m^{[f_i]},m^{(k)}{\rangle}}(dx)$ ] and @xmath548 , we also have @xmath549},m^{(k)}{\rangle}}}{d\nu}(x)\\ & = u(x)\frac{d\mu_{{\langle}m^{[f_i]},m_l^{(k)}{\rangle}}}{d\nu}(x)\\ & = \sum_{j=1}^l u(x)g_j^{(k)}(x)z^{i , j}(x)\\ & = { \langle}v(x),g^{(k)}(x){\rangle}_{z(x)},\end{aligned}\ ] ] where @xmath550 is defined as @xmath551 for @xmath552 and @xmath553 for @xmath554 . for @xmath48-a.e.@xmath116
, @xmath555 is an orthonormal basis of @xmath477 .
therefore , @xmath556 combining these equalities , we conclude that @xmath557 and holds .
the set of all linear combinations of additive functionals @xmath414 expressed as is dense in @xmath418 , which is proved in the same manner as lemma [ lem : density ] .
thus , holds for every @xmath428 by approximation .
therefore , the martingale dimension is less than or equal to @xmath184 .
next , we prove that the index is less than or equal to the martingale dimension @xmath184 .
we may assume @xmath519 .
then , there exist @xmath558 such that every @xmath428 has a representation @xmath559 .
take arbitrary finitely many functions @xmath560 from @xmath4 .
for each @xmath561 , we write @xmath562}=\sum_{k=1}^p h_k^i\bullet m^{(k)}$ ] . then , @xmath563},m^{[g_j]}{\rangle } } = \sum_{m , n=1}^p h_m^i h_n^j\cdot\mu_{{\langle}m^{(m)},m^{(n)}{\rangle}}.\ ] ] therefore , @xmath564 , where @xmath565 , and @xmath566 is a @xmath567 matrix and @xmath568 is a @xmath569 matrix provided by @xmath570 this implies that @xmath571 @xmath48-a.e.@xmath147 .
thus , the index is less than or equal to @xmath184 .
the additive functionals @xmath468 constructed in the first part of the proof satisfy @xmath572 from .
therefore , @xmath573 for @xmath574 , where @xmath575 is taken to be borel measurable , and in particular , @xmath576 this is consistent with the definition of the multiplicity of filtration in @xcite .
[ cor : one ] suppose that the index of @xmath3 is one .
for @xmath577 , the following properties are equivalent . 1 .
@xmath578 is a minimal energy - dominant measure . 2 .
for any @xmath428 , there exists @xmath579 such that @xmath580 .
in particular , for @xmath9 , @xmath581}$ ] satisfies property ( ii ) above if and only if @xmath582 . from theorem
[ th : index ] , equation , and proposition [ prop : zero ] , there exists @xmath583 such that @xmath584 and for each @xmath428 , there exists @xmath585 satisfying @xmath586 .
suppose ( i ) holds .
there exists @xmath587 such that @xmath588 . since @xmath589 , @xmath590 @xmath48-a.e . for @xmath428 ,
take @xmath585 as above and define @xmath591 .
then , @xmath592 and @xmath593 therefore , ( ii ) holds .
next , suppose ( ii ) holds .
there exists @xmath594 such that @xmath595 .
then , @xmath596 .
this implies that @xmath597 .
therefore , ( i ) holds .
in this section , we take self - similar fractals as @xmath5 . we follow @xcite to provide a framework .
let @xmath5 be a compact metrizable topological space , @xmath247 be an integer greater than one , and set @xmath598 .
further , let @xmath599 be a continuous injective map for @xmath600
. set @xmath601 . for @xmath600 , we define a shift operator @xmath602 by @xmath603 .
suppose that there exists a continuous surjective map @xmath604 such that @xmath605 for every @xmath600 .
we term @xmath606 a self - similar structure .
we also define @xmath607 , @xmath608 for @xmath609 , and denote @xmath610 by @xmath611 . for @xmath612 , we define @xmath613 and @xmath614 . by convention ,
@xmath615 is the identity map from @xmath5 to @xmath5 . for @xmath612 and @xmath616
, @xmath617 denotes @xmath618 . for @xmath619 ,
let @xmath620 be a @xmath11-field on @xmath5 generated by @xmath621 .
then , @xmath622 is a filtration on @xmath5 and the @xmath11-field generated by @xmath623 is equal to @xmath12 .
we set @xmath624 where @xmath625 is a shift operator that is defined by @xmath626 .
the set @xmath627 is referred to as the post - critical set .
we assume that @xmath5 is connected and the self - similar structure @xmath628 is _ post - critically finite _ ( p.c.f . ) , that is , @xmath627 is a finite set .
figure [ fig : pcf ] shows some typical examples of p.c.f .
self - similar sets @xmath5 , where the set of black points is @xmath629 .
the three - dimensional sierpinski gasket is realized in @xmath630 , and other sets are realized in @xmath631 .
let @xmath632 for @xmath633 and @xmath634 .
for any @xmath635 , there exists a unique element @xmath636 such that @xmath637 .
we denote @xmath638 by @xmath639_m$ ] for each @xmath633 , and define @xmath639_0=\emptyset$ ] .
the sequence @xmath640_m}\}_{m=0}^\infty$ ] is a fundamental system of neighborhoods of @xmath147 ( ( * ? ? ?
* proposition 1.3.6 ) ) . for a finite set @xmath641 ,
let @xmath642 be the space of all real - valued functions on @xmath641 .
we equip @xmath642 with an inner product @xmath643 defined by @xmath644 .
let @xmath645 be a symmetric linear operator on @xmath646 ( also considered to be a square matrix of size @xmath647 ) such that the following conditions hold : 1 .
@xmath648 is nonpositive definite , 2 .
@xmath649 if and only if @xmath439 is constant on @xmath629 , 3 .
@xmath650 for all @xmath651 .
we define @xmath652 for @xmath653 .
this is a dirichlet form on @xmath646 , where @xmath646 is identified with the @xmath328 space on @xmath629 with the counting measure ( ( * ? ? ?
* proposition 2.1.3 ) ) . for @xmath654 with @xmath655 , we define a bilinear form @xmath656 on @xmath657 as @xmath658 here , @xmath659 for @xmath660 and @xmath661 .
we refer to @xmath662 as a _ harmonic structure _ if for every @xmath663 , @xmath664 then , for @xmath665 and @xmath666 , we obtain @xmath667 .
we consider only a harmonic structure that is _ regular _ , namely , @xmath668 for all @xmath600 .
several studies have been conducted on the existence of regular harmonic structures such as @xcite .
we only remark here that all nested fractals have canonical regular harmonic structures .
nested fractals are self - similar sets that are realized in euclidean spaces and have good symmetry . for the precise definition ,
see @xcite .
all the fractals shown in figure [ fig : pcf ] except hata s tree - like set are nested fractals .
hata s tree - like set also has regular harmonic structures ; see proposition [ prop : hata ] below .
we assume that a regular harmonic structure @xmath662 is given .
let @xmath10 be a borel probability measure on @xmath5 with full support .
we can then define a strong local and regular dirichlet form @xmath3 on @xmath19 associated with @xmath662 by @xmath669 ( see the beginning of @xcite . ) for a map @xmath670 and a function @xmath6 on @xmath5 , @xmath671 denotes the pullback of @xmath6 by @xmath672 , that is , @xmath673 . the dirichlet form @xmath3 constructed above satisfies the self - similarity : @xmath674 from ( * ? ? ? * theorem 3.3.4 ) , there exists a constant @xmath675 such that @xmath676 where , in general , @xmath677 for @xmath358 . by utilizing this inequality ,
it is easy to prove that the capacity associated with @xmath3 of any nonempty subset of @xmath5 is uniformly positive ( see , e.g. , ( * ? ? ?
* proposition 4.2 ) ) .
the energy measures have the following properties . for the proof , see
e.g. ( * ? ? ? * lemma 4 ) and its proof .
[ lem : energymeas ] let @xmath9 and @xmath33 be the energy measure of @xmath6 . 1 . for @xmath665 and a borel subset @xmath36 of @xmath5 , @xmath678 2 .
the measure @xmath33 has no atoms . in particular , @xmath679 .
3 . for @xmath680 and a borel subset @xmath36 of @xmath681 , @xmath682 in addition , note that since @xmath683 and @xmath684 , @xmath685 the underlying measure @xmath10 does not play an important role with regard to the energy measures since they are independent of the choice of @xmath10 . for each @xmath686
, there exists a unique function @xmath687 such that @xmath688 and @xmath689 .
such a function @xmath401 is termed a _ harmonic function_. the space of all harmonic functions is denoted by @xmath690 . for any @xmath680 and @xmath691 , @xmath692 . by using the linear map
@xmath693 , we can identify @xmath690 with @xmath646 .
in particular , @xmath690 is a finite dimensional subspace of @xmath4 , and the norm @xmath30 on @xmath690 is equivalent to @xmath694 . for each @xmath600
, we define a linear operator @xmath695 as @xmath696 which is also considered as a square matrix of size @xmath647 .
let @xmath697 be the set of all constant functions on @xmath5 , which is a one - dimensional subspace of @xmath690 .
let @xmath698 be an arbitrary projection operator from @xmath690 to @xmath697 .
define @xmath699 , where @xmath700 is an identity operator on @xmath690 . for each @xmath600 ,
set @xmath701 . for @xmath702 , define @xmath703 and @xmath704 . by convention ,
@xmath705 and @xmath706 .
let @xmath707 be a constant function on @xmath629 with value @xmath708 .
since each @xmath709 has an eigenvalue @xmath708 with @xmath710 as eigenfunctions , @xmath711 .
then , @xmath712 .
therefore , for any @xmath680 , @xmath713 in addition , since @xmath714 and @xmath648 is symmetric , @xmath715 for @xmath619 , let @xmath716 denote the set of all functions @xmath6 in @xmath4 such that @xmath717 for all @xmath718 .
let @xmath719 .
functions in @xmath720 are referred to as _ piecewise harmonic functions_. from ( * ? ? ? * lemma 2.1 ) , @xmath720 is dense in @xmath4 .
[ lem : matrix ] let @xmath680 and @xmath721 .
then , @xmath722 it is sufficient to prove when @xmath723 . from (
* lemma 4 ( ii ) ) , equation with @xmath724 replaced by @xmath725 holds . by combining and
, we obtain . in what follows ,
we fix a minimal energy - dominant measure @xmath48 with @xmath130 .
[ prop : equivalence ] let @xmath665 .
define @xmath726 .
that is , @xmath727 then , @xmath48 and @xmath728 are mutually absolutely continuous . in particular , @xmath59 implies @xmath729 .
moreover , for any @xmath730 and @xmath718 , @xmath731 let @xmath36 be a borel set of @xmath5 .
suppose @xmath732 .
let @xmath9 and @xmath718 .
then , @xmath733 .
since @xmath734 , we have @xmath735 by lemma [ lem : energymeas ] ( iii ) . since @xmath718 is arbitrary , @xmath73 . from lemma [ lem :
null ] , @xmath59 .
therefore , @xmath736 .
next , suppose @xmath59 .
let @xmath9 and @xmath718 .
then , there exists @xmath737 such that @xmath738 . from lemma [ lem : energymeas ] ( i ) ,
@xmath739 . from lemma [ lem :
null ] , @xmath729 , that is , @xmath740 .
therefore , @xmath732 .
this implies @xmath741 . for the proof of , take @xmath742 and @xmath743
then , from lemma [ lem : energymeas ] , @xmath744_n})}{\nu'_m(k_{[x]_n } ) } = \frac{r_w^{-1}\nu_{\psi_w^ * f , \psi_w^ * g}(\psi_w^{-1}(k_{[x]_n } ) ) } { r_w^{-1}\nu(\psi_w^{-1}(k_{[x]_n } ) ) } = \frac{\nu_{\psi_w^ * f , \psi_w^ * g}(k_{[\psi_w^{-1}(x)]_{n - m } } ) } { \nu(k_{[\psi_w^{-1}(x)]_{n - m}})}.\ ] ] letting @xmath460 , we obtain . for each @xmath745 ,
define @xmath746 so that @xmath747 , @xmath748 .
let @xmath749 for @xmath750 .
[ prop : indexk ] define @xmath751 .
then , the index of @xmath3 is @xmath184 .
it suffices to prove that the index is less than or equal to @xmath184 .
take a sequence @xmath21 from @xmath720 such that the linear span of @xmath21 is dense in @xmath4 .
for @xmath69 , define an @xmath119 matrix @xmath187 , @xmath116 , by @xmath752 from proposition [ prop : rank ] , it is sufficient to show that @xmath753 take a sufficiently large @xmath633 such that @xmath754 for all @xmath755
. let @xmath718 . since @xmath756 is expressed as a linear combination of @xmath757 for every @xmath755 , it is easy to prove @xmath758 as in the first part of the proof of proposition [ prop : rank ] . from proposition [ prop :
equivalence ] , @xmath759 where @xmath728 is provided in proposition [ prop : equivalence ] .
therefore , @xmath760 .
since @xmath718 is arbitrary , we obtain . for determining the index of @xmath3 ,
the following result is the most general one available at present , which is an extension of the result described in @xcite .
[ th : main0 ] suppose that for each @xmath745 , @xmath761 is a fixed point of @xmath762 for some @xmath600 and @xmath763 is connected .
then , the index ( or equivalently , the martingale dimension ) of @xmath3 is one .
the author expects that the index is always one without the assumption of theorem [ th : main0 ] , but does not know the proof nor the counterexamples at the moment .
all nested fractals satisfy the assumption of theorem [ th : main0 ] ( see e.g. ( * ? ? ?
* theorem 1.6.2 and proposition 1.6.9 ) for the proof ) .
a typical example that does not satisfy the assumption is hata s tree - like set ( shown in the lower right - hand side of figure [ fig : pcf ] ) .
more precisely , let @xmath764 such that @xmath765 and @xmath766 .
hata s tree - like set is a unique nonempty compact subset @xmath5 of @xmath767 such that @xmath768 , where @xmath769 and @xmath770 for @xmath771 .
then , @xmath772 and @xmath773 is not a fixed point of @xmath774 nor @xmath775 .
however , even in this case , we can prove that the index is one by the following direct computation .
[ prop : hata ] let @xmath5 be hata s tree - like set and @xmath10 , a borel probability measure on @xmath5 with full support . for a dirichlet form @xmath3 on @xmath19 associated with an arbitrary regular harmonic structure @xmath662 , the index is one . from an elementary calculation , all harmonic structures
@xmath662 are provided by @xmath776 for @xmath777 , @xmath778 and @xmath779 ( cf .
* example 3.1.6 ) ) , where we identify @xmath780 with @xmath630 .
we may assume @xmath781 to prove the claim . then , the matrices @xmath782 and @xmath783 defined by are given by @xmath784 as projections on @xmath690 , take @xmath785 and @xmath786 .
then , @xmath787 therefore , @xmath788 and @xmath789 .
this implies that for @xmath718 with @xmath633 , @xmath790 unless @xmath791 . for @xmath750 ,
define @xmath792 and @xmath793 as in the statement before proposition [ prop : indexk ] . for @xmath103 and @xmath635 , let @xmath794_n})/\nu(k_{[x]_n})$ ] , where @xmath110 .
then , from lemma [ lem : matrix ] , matrix @xmath795 is equal to @xmath796_n}\nu(k_{[x]_n}))^{-1}\,{}^t\!a'_{[x]_n } d a'_{[x]_n}$ ] . its rank is less than or equal to @xmath708 unless @xmath639_n= \underbrace{11\dotsm1}_{n } $ ] .
since @xmath795 converges to @xmath797 @xmath48-a.e.@xmath147 as @xmath460 by the martingale convergence theorem , and @xmath798_n= \underbrace{11\dotsm1}_{n}\}=\emptyset$ ] , we conclude that @xmath799 @xmath48-a.e.@xmath147 . from proposition
[ prop : indexk ] and proposition [ prop : zero ] , the index is one .
the index discussed in this section may be different from that defined in @xcite .
the author does not know whether these two definitions are equivalent or not .
in this section , we discuss the concept of derivatives of functions on p.c.f . self - similar fractals .
we use the same notations as those in the previous section . from , , and lemma [ lem : energymeas ] ( iii ) , we have @xmath803_n } } f(y ) = { \mathop{\rm osc}}_{y\in k}(\psi^*_{[x]_n}f)(y ) \le
c_5\sqrt{{{\cal e}}(\psi^*_{[x]_n}f ) } = c_5\sqrt{\frac{1}{2}r_{[x]_n}\nu_f(k_{[x]_n})}. { \hskip1000pt minus 1fil \llap{\proofbox}}\ ] ] for @xmath801 , define a map @xmath804 so that @xmath805 is a unique function in @xmath806 satisfying @xmath807 on @xmath808 .
note that @xmath809 for all @xmath274 and @xmath810 as @xmath460 .
we write @xmath811 for @xmath812 . for each @xmath819 , from
, @xmath820 let @xmath821 be the conditional expectation of @xmath822 with respect to @xmath172 for given @xmath823 , that is , @xmath824_{n(k)}})}\int_{k_{[x]_{n(k)}}}\left|1-\frac{d\nu_{g_{n(k)}}}{d\nu_g}\right|\,d\nu_g,\quad
x\in k\setminus v_*,\ ] ] where the right - hand side is defined as @xmath95 if @xmath825_{n(k)}})=0 $ ] . then , @xmath826 from , @xmath827 therefore , for @xmath172-a.e.@xmath147 , @xmath828 , in particular , @xmath829 .
since @xmath830_{n(k)}})}\left|\int_{k_{[x]_{n(k)}}}\left(1-\frac{d\nu_{g_{n(k)}}}{d\nu_g}\right)\,d\nu_g\right| = \left|1-\frac{\nu_{g_{n(k)}}(k_{[x]_{n(k)}})}{\nu_{g}(k_{[x]_{n(k)}})}\right|,\ ] ] we obtain the claim .
[ lem : lower2 ] there exists a constant @xmath831 satisfying the following : for @xmath172-a.e.@xmath147 , there exists @xmath832 such that @xmath802_{n(k)}}}g(y ) \ge c_7 \sqrt{r_{[x]_{n(k)}}\nu_g(k_{[x]_{n(k ) } } ) } , \quad k\ge k_0(x).\ ] ] moreover , for @xmath172-a.e.@xmath147 , @xmath833_{n}}}g(y)>0 $ ] for every @xmath801 . from lemma [ lem : lower1 ] , for @xmath172-a.e.@xmath147 , there exists @xmath832 such that @xmath834_{n(k)}})\ge\frac12\nu_g(k_{[x]_{n(k ) } } ) , \quad k\ge k_0(x).\ ] ] let @xmath835 .
since @xmath836 is a finite dimensional space and both of the maps @xmath837 and @xmath838 provide norms on @xmath836 , there exists a constant @xmath839 depending only on @xmath3 such that for every @xmath840 , @xmath841 this inequality also holds for @xmath691 .
since @xmath842_{n(k)}}^ * g_{n(k)}$ ] coincides with @xmath843_{n(k)}}^ * g)\in{{\cal h}}$ ] , we have @xmath844_{n(k)}}\nu_{g_{n(k)}}(k_{[x]_{n(k ) } } ) } & = \sqrt{\nu_{h(\psi_{[x]_{n(k)}}^ * g)}(k ) } \qquad\mbox{(from { lemma~\ref{lem : energymeas}}~(iii))}\nonumber\\ & = \sqrt{2{{\cal e}}({h(\psi_{[x]_{n(k)}}^ * g ) } ) } \qquad \mbox{(from { \rom{(\ref{eq : total})}})}\nonumber\\ & \le \sqrt{2}c_8{\mathop{\rm osc}}_{y\in v_0}h(\psi_{[x]_{n(k)}}^ * g)(y)\nonumber\\ & = \sqrt{2}c_8{\mathop{\rm osc}}_{y\in \psi_{[x]_{n(k ) } } ( v_0)}g(y)\nonumber\\ & \le \sqrt{2}c_8{\mathop{\rm osc}}_{y\in k_{[x]_{n(k)}}}g(y).\end{aligned}\ ] ] combining and , we obtain the first claim . to prove the second claim , let @xmath845_{n}})=0 \mbox { for some } n\in{{\bbb z}}_+\}$ ] . since @xmath640_{n}}\}_{n=0}^\infty$ ] is fundamental system of neighborhoods of each @xmath635 , @xmath36 is nothing but @xmath846 .
therefore , @xmath847 .
this and the first claim imply the second claim .
[ th : main ] suppose that the index of @xmath3 is one .
let @xmath813 .
then , for every @xmath9 , for @xmath172-a.e.@xmath147 , there exists a unique real number @xmath848 such that @xmath849 where @xmath850 is negligible in the following sense:@xmath851_{n(k)}}}|r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n(k ) } } } g(y)}=0,\ ] ] where the sequence @xmath816 is taken so that holds .
in particular , @xmath852_{n}}}|r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n } } } g(y)}=0.\ ] ] moreover , it holds that @xmath853 and the map @xmath854 is a bounded linear map .
if @xmath8 belongs to @xmath720 , can be replaced by the stronger claim @xmath855_n}}|r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_n } } g(y)}=0.\ ] ] take an arbitrary c.o.n.s .
@xmath21 of @xmath4 , and define @xmath856 for @xmath96 . from proposition
[ prop : squareroot ] and the assumption that the index is one , there exists a sequence @xmath226 of real - valued functions on @xmath5 such that @xmath857 @xmath172-a.e.@xmath147 . for @xmath9 , @xmath6
is uniquely expressed as @xmath858 for some @xmath859 .
define a map @xmath860 by @xmath861 for @xmath862 .
this is well - defined ; indeed , for @xmath863 and @xmath864 for some @xmath819 , @xmath865 and @xmath866 which imply that @xmath867 is not only well - defined as a bounded linear operator from @xmath4 to @xmath868 but it also satisfies the relation @xmath869 for every @xmath870 . in particular , @xmath871 @xmath172-a.e .
we write @xmath872 for @xmath873 .
take @xmath9 and define @xmath874
. we will show that we can take @xmath875 as @xmath848 . for @xmath116 , define @xmath876 then , @xmath877 belongs to @xmath4 , and we have @xmath878_n})}{\nu_g(k_{[x]_n } ) } & = \frac{\nu_{f(\cdot)-a(x)g(\cdot)}(k_{[x]_n})}{\nu_g(k_{[x]_n})}\nonumber\\ & = \frac{\nu_f(k_{[x]_n})-2a(x)\nu_{f , g}(k_{[x]_n})+a(x)^2\nu_g(k_{[x]_n})}{\nu_g(k_{[x]_n})}\nonumber\\ & \stackrel{n\to\infty}{{\longrightarrow } } \frac{d\nu_f}{d\nu_g}(x)-2a(x)\frac{d\nu_{f , g}}{d\nu_g}(x)+a(x)^2\nonumber\\ & \qquad\quad \mbox{(for $ \nu_g$-a.e.\,$x$ , from the martingale convergence theorem)}\nonumber\\ & = { \gamma}_f(x)^2 - 2a(x){\gamma}_f(x){\gamma}_g(x)+a(x)^2\qquad\mbox{(from { \rom{(\ref{eq : gamma})}})}\nonumber\\ & = 0 \qquad \mbox{for } \nu_g\mbox{-a.e.\,}x.\end{aligned}\ ] ] take a sequence @xmath816 such that holds . from @xmath879 , lemma [ lem : upper ] and lemma [ lem : lower2 ] , for @xmath172-a.e.@xmath147 , for sufficiently large @xmath880 ,
@xmath881_{n(k)}}}|r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n(k ) } } } g(y ) } \le\frac{{\mathop{\rm osc}}_{y\in k_{[x]_{n(k)}}}r_x(y)}{{\mathop{\rm osc}}_{y\in k_{[x]_{n(k ) } } } g(y ) } \le \frac{c_6}{c_7}\sqrt{\frac{\nu_{r_x}(k_{[x]_{n(k)}})}{\nu_g(k_{[x]_{n(k)}})}}.\ ] ] combining and , we obtain . since @xmath882
we obtain .
next , we prove the uniqueness of @xmath848 .
in fact , the weaker property does imply the uniqueness .
suppose two functions @xmath883 , @xmath884 and remainder terms @xmath885 , @xmath886 satisfy and with @xmath848 and @xmath850 replaced by @xmath883 , @xmath884 and @xmath885 , @xmath886 , respectively .
we can take an increasing sequence @xmath887 such that holds with @xmath888 replaced by @xmath889 , @xmath890_{n'(2k)}}}|\widehat r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n'(2k ) } } } g(y)}=0 , \quad\mbox{and}\quad \lim_{k\to\infty}\frac{\sup_{y\in k_{[x]_{n'(2k+1)}}}|\widetilde r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n'(2k+1 ) } } } g(y)}=0.\ ] ] there exists @xmath848 for @xmath172-a.e.@xmath147 such that and hold with @xmath891 replaced by @xmath892 in .
we have @xmath893 from lemma [ lem : lower2 ] , for @xmath172-a.e.@xmath147 , @xmath833_{n}}}g(y)>0 $ ] for every @xmath274 .
take such @xmath147 from @xmath894 . for each @xmath801 , choose @xmath895_n}$ ] such that @xmath896_n}}g(y)$ ] . from , for every @xmath819 ,
@xmath897_{n'(2k)}}}|r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n'(2k ) } } } g(y ) } + \frac{2\sup_{y\in k_{[x]_{n'(2k)}}}|\widehat r_x(y)|}{{\mathop{\rm osc}}_{y\in k_{[x]_{n'(2k ) } } } g(y)}.\end{aligned}\ ] ] by letting @xmath898 , we have @xmath899 for @xmath172-a.e.@xmath147 . in the same way
, we can prove that @xmath900 for @xmath172-a.e.@xmath147 .
therefore , the uniqueness holds . 1 .
pelander and teplyaev obtained a result similar to theorem [ th : main ] ( ( * ? ? ?
* theorem 4.8 ) ) ; however , they proved the a.e .
existence of derivatives with respect to some self - similar measures . since self - similar measures and energy measures are mutually singular in many cases ( @xcite ) , their result and ours are not comparable .
2 . even when @xmath5 is not a self - similar fractal , theorem [ th : main ] should hold with suitable modification .
moreover , when the index @xmath184 is greater than one , it is expected that there exist suitable reference functions @xmath904 such that any function @xmath6 in @xmath4 has a taylor expansion with respect to @xmath904 up to the first order .
we leave such generalization for future studies . in theorem
[ th : main ] , the reference function @xmath8 can be taken from a dense subset @xmath127 of @xmath4 ; however , exactly what type of function can be chosen remains unresolved .
we will provide a partial solution to this problem . take @xmath330-dimensional level-@xmath212 sierpinski gaskets as @xmath5 for @xmath905 and @xmath906 ( see figure [ fig : sg ] ) .
set @xmath629 consists of @xmath907 points denoted as @xmath908 .
we may assume that the index set @xmath175 of contraction maps @xmath909 is a subset of @xmath910 , and for each @xmath911 , @xmath912 is a fixed point of @xmath762
. there exists a canonical harmonic structure @xmath662 since @xmath5 is a nested fractal , where @xmath645 is provided by @xmath913 for some @xmath914 and @xmath915 satisfies @xmath916 for some @xmath917 .
we may assume @xmath918 without loss of generality .
let @xmath10 be the normalized hausdorff measure on @xmath5 and @xmath3 , the dirichlet form on @xmath19 associated with @xmath662 .
we further assume that @xmath3 ( or @xmath662 ) is _ nondegenerate _ , namely , for every @xmath600 , the map @xmath695 defined in is bijective .
in other words , there are no nonconstant harmonic functions that are constant on some small cells .
the author expects that this assumption is always true , although it has not yet been proved in general .
the following cases are nondegenerate : in cases ( 1 ) and ( 3 ) , the matrices @xmath709 are explicitly calculated and are proved to be invertible . in case ( 2 ) , the author checked the nondegeneracy by numerical computation with the aid of computers .
[ th : sg ] under the nondegeneracy condition , every nonconstant harmonic function @xmath401 belongs to @xmath127 , that is , its energy measure @xmath924 is a minimal energy - dominant measure . in particular , for any nonconstant harmonic functions @xmath925 and @xmath926 , @xmath927 and @xmath928 are mutually absolutely continuous .
let @xmath665 and @xmath933 .
for any @xmath718 , @xmath934 , and the linear span of @xmath935 is equal to @xmath690 .
therefore , @xmath936 is expressed as a linear combination of @xmath937 .
this implies that there exists a constant @xmath938 such that @xmath939 . for any borel subset @xmath36 of @xmath681 , from lemma [ lem : energymeas ] ( iii ) , @xmath940 therefore , @xmath33 on @xmath681 is absolutely continuous with respect to @xmath48 on @xmath681 .
since @xmath718 is arbitrary , we obtain @xmath50 .
any @xmath691 can be expressed uniquely as @xmath945 for some @xmath946 .
then , @xmath947 since @xmath948 is not a zero matrix for @xmath48-a.e.@xmath147 from the definition of @xmath48 , @xmath949 is a proper subspace of @xmath950 .
therefore , when @xmath951 is the lebesgue measure on @xmath950 , for @xmath48-a.e.@xmath147 , @xmath952 for @xmath951-a.e.@xmath953 . from the fubini theorem , for @xmath951-a.e.@xmath953 , @xmath952 for @xmath48-a.e.@xmath147 . in particular , for such @xmath946 , @xmath954 satisfies @xmath955 @xmath48-a.e.@xmath147 .
therefore , @xmath172 is a minimal energy - dominant measure .
[ lem : singularity ] suppose that two sequences @xmath956 and @xmath957 in @xmath4 converge @xmath439 and @xmath958 in @xmath4 , respectively . if there exist a sequence of borel sets @xmath959 of @xmath5 such that @xmath960 and @xmath961 , then @xmath962 .
in particular , if @xmath963 for every @xmath274 , then @xmath962 .
take a sequence @xmath964 such that @xmath965 and @xmath966 for every @xmath633 .
set @xmath967 then , for any @xmath819 and @xmath968 , @xmath969 therefore , @xmath970 for every @xmath819 , which implies @xmath971 . on the other hand , for @xmath633 , @xmath972 then , for @xmath819 , @xmath973 therefore , @xmath974 .
this concludes that @xmath962 .
recall that @xmath707 is a constant function on @xmath629 with value @xmath708 .
let @xmath982 and @xmath983 be the orthogonal projection of @xmath646 onto @xmath984 .
we define @xmath985 which is a compact subset of @xmath4 that does not include any constant functions .
we define a map @xmath986 by @xmath987 let @xmath745 . from (
* theorem a.1.2 ) ( essentially , from the perron
frobenius theorem ) , @xmath988 has simple eigenvalues @xmath708 and @xmath989 , and the absolute values of any other eigenvalues are less than @xmath989 .
let @xmath990 be the column vector @xmath991 , that is , @xmath992 then , from ( * ? ? ?
* lemma 5 ) , @xmath990 is an eigenvector of @xmath993 with respect to the eigenvalue @xmath989 .
let @xmath994 be an eigenvector of @xmath988 with respect to the eigenvalue @xmath989 .
then , @xmath995 should be @xmath95 , and other components should be the same by symmetry .
therefore , @xmath994 can be taken as @xmath996 then , @xmath997 and the following holds .
take @xmath1002 , which is possible by lemma [ lem : existence ] .
we may assume @xmath1003 .
let @xmath401 be an arbitrary nonconstant function in @xmath690 .
define @xmath1004 .
note that @xmath1005 . in order to prove that @xmath924 is also a minimal energy - dominant measure
, it suffices to show that @xmath847 .
we will derive a contradiction by assuming @xmath1006 .
for @xmath801 , let @xmath1007 be the conditional expectation of @xmath1008 with respect to @xmath172 given @xmath102 , that is , @xmath1009_n}\cap b)}{\nu_g(k_{[x]_n})}\quad\mbox{for $ x\in k\setminus v_*$}.\ ] ] then , from the martingale convergence theorem , @xmath1007 converges to @xmath1008 @xmath172-a.e .
in particular , there exists @xmath1010 such that @xmath1011 converges to @xmath708 as @xmath460 . for @xmath801 , define @xmath1012_n}^ * g$ ] , @xmath1013_n}^ * h$ ] , and @xmath1014_n}^{-1}(b)$ ] . from lemma [ lem : energymeas ] , @xmath1015 as @xmath460 and @xmath1016_n}\nu_h(b\cap k_{[y]_n})=0 $ ] .
note that @xmath814 and @xmath1017 are both nonconstant harmonic functions for every @xmath274 from the assumption of the nondegeneracy of @xmath3 .
we define @xmath1018 and @xmath1019 for @xmath801 . then ,
@xmath1020 take a sequence @xmath1021 such that @xmath1022 and @xmath1023 converge to some @xmath1024 and @xmath1025 in @xmath4 , respectively .
then , @xmath1026 , and @xmath1027 from lemma [ lem : singularity ] .
since @xmath649 if and only if @xmath439 is constant , and @xmath1028 is not constant , there exists @xmath911 such that @xmath1029 . for @xmath801 , define @xmath1030 and @xmath1031 .
there exists a sequence @xmath816 such that each @xmath1032 and @xmath1033 converges in @xmath4 as @xmath898 .
the limits will be denoted by @xmath1034 and @xmath1035 , respectively .
both limits belong to @xmath1036 . from lemma [ lem :
convergent ] , @xmath1037 .
since @xmath1038 for each @xmath801 from lemma [ lem : singular ] , @xmath1039 from lemma [ lem : singularity ] .
now , @xmath1040 for every @xmath1041 , which is confirmed by and .
let us assume that @xmath1042 for some @xmath1043 . for @xmath801
, define @xmath1044 and @xmath1045 . then , from lemma [ lem : convergent ] , both @xmath1046 and @xmath1047 converge to @xmath1048 in @xmath4 . since @xmath1049 for all @xmath274 from lemma [ lem : singular ] , @xmath1050 from lemma [ lem : singularity ] . since @xmath33 is not a null measure ,
this is a contradiction , which implies that @xmath1051 for all @xmath1043 . but
this means that @xmath1052 and @xmath1053 is constant , which contradicts the fact that @xmath1054 .
consequently , we conclude that @xmath847 and complete the proof . 1 . for most p.c.f .
self - similar fractals except sierpinski gaskets ( for example , all fractals shown in figure [ fig : pcf ] except the first three ) , dirichlet forms associated with harmonic structures are _ not _ nondegenerate .
that is , there exist nonconstant harmonic functions that are constant on some small cells , which means that the energy measures of such functions are not minimal energy - dominant measures .
therefore , in order to obtain the claim in theorem [ th : sg ] , nondegeneracy is a necessary assumption .
thus , considering only sierpinski gaskets is not as restrictive as it may appear .
2 . the key point of the proof above
is the fact that @xmath1055 for every @xmath750 .
thus , the assertion of theorem [ th : sg ] is still true when we perturb the harmonic structure @xmath662 slightly .
3 . with regard to the radon nikodym derivative @xmath1056 ,
some numerical computation has been carried out in @xcite .
99 , ` stochastic processes as curves in hilbert space ' , _ theory probab .
_ 9 ( 1964 ) 169179 . , ` the multiplicity of an increasing family of @xmath1057-fields ' , _ ann .
_ 2 ( 1974 ) 958963 . , _
dirichlet forms and symmetric markov processes _ , de gruyter studies in mathematics 19 ( walter de gruyter , berlin , 1994 ) . , ` self - similar energies on p.c.f .
self - similar fractals ' , _ j. london math .
_ ( 2 ) 74 ( 2006 ) 93112 . , ` on singularity of energy measures on self - similar sets ' , _ probab .
theory related fields _ 132 ( 2005 ) 265290 . , ` martingale dimensions for fractals ' , _ ann . probab . _ 36 ( 2008 ) , 971991 . , ` on singularity of energy measures on self - similar sets ii ' , _ bull .
london math .
_ 38 ( 2006 ) 10191032 .
, _ analysis on fractals _ , cambridge tracts in mathematics 143 ( cambridge university press , cambridge , 2001 ) .
, ` on square integrable martingales ' , _ nagoya math . j. _ 30 ( 1967 ) 209245 .
, ` dirichlet forms on fractals and products of random matrices ' , _ publ . res .
inst . math .
_ 25 ( 1989 ) 659680 . ,
` lecture on diffusion processes on nested fractals ' , in _ statistical mechanics and fractals _
, 3998 , lecture notes in math .
1567 , springer - verlag , berlin , 1993 .
, ` measures associes une forme de dirichlet .
applications ' , _ bull .
france _ 106 ( 1978 ) 61112 . , _ brownian motion on nested fractals _ , mem .
83 ( 1990 ) , no . 420 .
, ` composite media and asymptotic dirichlet forms ' , _ j. funct . anal . _ 123 ( 1994 ) 368421 . , ` on a class of additive functionals of markov processes ' , _ j. math .
kyoto univ .
_ 4 ( 1964/1965 ) 429469 . , ` existence of eigenforms on fractals with three vertices ' , _ proc .
royal soc .
edinburgh _ 137 a ( 2007 ) 10731080 . , ` products of random matrices and derivatives on p.c.f .
fractals ' , _ j. funct .
anal . _ 254 ( 2008 ) 11881216 . , ` local behavior of smooth functions for the energy laplacian on the sierpinski gasket ' , preprint , 2008 .
, _ multidimensional diffusion processes _ , grundlehren der mathematischen wissenschaften 233 ( springer - verlag , berlin , 1979 ) . , ` gradients on fractals ' , _ j. funct . anal . _ 174 ( 2000 ) 128154 . , ` additive functionals on a multi - dimensional wiener process ' , _ dokl . akad .
nauk sssr _ 139 ( 1961 ) 1316 .
|
we introduce the concept of index for regular dirichlet forms by means of energy measures , and discuss its properties . in particular , it is proved that the index of strong local regular dirichlet forms is identical with the martingale dimension of the associated diffusion processes . as an application ,
a class of self - similar fractals is taken up as an underlying space .
we prove that first - order derivatives can be defined for functions in the domain of the dirichlet forms and their total energies are represented as the square integrals of the derivatives .
|
under irradiation of microwaves dramatic new effects occur in two - dimensional electron systems ( 2des ) in gaas - algaas heterostructures previously used to study quantum hall effects ( qhe).@xcite as first reported by zudov _
et al._,@xcite and also by ye _
et al._,@xcite in the millimeterwave ( mw ) frequency range ( @xmath0 ghz ) and in a small magnetic field ( @xmath1 t ) , at low temperatures ( @xmath2 k ) a new type of oscillations arises in the magnetoresistance @xmath3 of a high - mobility 2des .
these oscillations are periodic in @xmath4 and can occur in a @xmath5 weaker than the onset of shubinikov - de haas oscillations ( sdh ) .
characteristically , their period is controlled by the ratio of the microwave frequency to the electron cyclotron frequency , @xmath6 , where @xmath7 , @xmath8 is the effective mass of the conduction band electrons in gaas . in particular ,
the oscillations can be associated with cyclotron resonance ( cr , @xmath9 ) and its harmonics ( @xmath10 ) . although the photoconductivity ( pc ) effect was well - known in the literature@xcite such findings came as a surprise .
not only the observation of higher - order cr ( hence oscillations rather than a single peak ) , but also its large amplitude , was completely unexpected .
moreover , later it was found that in samples of even higher mobility , the oscillation minima approach zero .
recently , mani _
et al._@xcite and zudov _
et al._@xcite reported the observation of new `` zero - resistance states '' ( zrs ) associated with such minima in very clean samples . at low temperatures , zrs are characterized by an exponentially vanishing diagonal resistance and an essentially classical hall resistance .
such a new effect in a 2des induced by microwaves is a subject of much current experimental@xcite and theoretical work.@xcite some of the theoretical ideas can be traced back to earlier studies of negative photoconductivity phenomena in semiconductors.@xcite these models , as well as others , remain to be tested experimentally .
the observation of zrs is complemented by an observation of zero - conductance states under similar experimental conditions , but in corbino samples.@xcite
photoconductivity related to cr of a 2des was first reported by maan _
et al._@xcite in the far - infrared frequency range and in a strong magnetic field . in the mw frequency range ,
distinct magnetoresistance signals due to electron spin resonance ( dobers _ et al._@xcite ) or magnetoplasmon resonance ( vasiliaou _ et al._@xcite ) were observed in 2des in gaas - algaas heterostructures . in these earlier studies using samples of electron mobility @xmath11 @xmath12/vs or less
, the pc signal exhibits a single peak in the @xmath3 , and its amplitude is typically less than 1% .
the mw - induced giant amplitude magnetorsistance oscillations were discovered@xcite using a high - mobility ( @xmath13 @xmath12/vs ) 2des . in a small magnetic field
the @xmath3 shows periodic ( in @xmath4 ) oscillations , with the @xmath5 positions of major maxima and minima in the oscillatory structure roughly conforming to@xcite @xmath14 phenomenologically such oscillations resemble sdh oscillations except that their period relates to @xmath15 rather than to @xmath16 , the landau level filling factor .
most remarkably , the @xmath3 at the minima was found to reduce from its dark value , indicating of negative pc .
this was a crucial step leading to the discovery of zrs ; increasing sample mobility favors oscillation amplitude , and , therefore , the @xmath3 value at the minima becomes progressively smaller under similar experimental conditions .
ga@xmath17as heterojunction , having a 2des of density @xmath18 @xmath19 and mobility @xmath20 @xmath12/vs .
, width=302 ] vanishingly small @xmath3 at the minima was initially observed in a gaas / al@xmath21ga@xmath17as heterojunction 2des approaching a mobility @xmath22 @xmath12/vs.@xcite the data of yang _
_ are shown in fig .
as marked by arrows , in this sample the strongest minimum in the @xmath3 evolves into zrs as the mw frequency is increased ( hence , the minimum moves towards higher @xmath5 ) . our current data from an ultraclean 2des@xcite are shown in fig .
the sample was cleaved from a al@xmath23ga@xmath24as / gaas / al@xmath23ga@xmath24as qw ( width 30 nm ) , with a mobility @xmath25 @xmath12/vs and a density @xmath26 @xmath19 , respectively . with the irradiation of mw ( @xmath27 ghz and the power incident on the sample surface @xmath28w )
, the @xmath3 shows a series of oscillations and the novel zero - resistance states at the first four minima . shows sharp oscillations and a series of zero - resistance states ;
hall resistance @xmath29 shows an essentially classical , linear dependence on b. , width=377 ] the width of the zrs encompasses a large range of the filling factor @xmath16 .
for example , the first zrs spans from @xmath30 to @xmath31 and similar width of @xmath32 is observed for other zrs minima in this sample .
these data indicate that the transport under mws is controlled by @xmath15 rather than by @xmath16 . while the @xmath3 here remarkably resembles that in qhe , the hall resistance , @xmath29 , shows no sign of forming plateaus . in the very - high mobility 2des ,
the major peaks of mw - induced oscillations ( which are relevant to zrs regime ) can still be well described by eq . 1 for @xmath33 , but their shapes are largely asymmetric .
higher - order peaks are better described by a @xmath5-dependent phase shift with respect to cr harmonics ( @xmath34 ) , which saturates at @xmath35.@xcite peaks and zrs minima , showing a roughly linear dependence of the peak values on inverse temperature , and a thermally activated resistance at the minima .
, width=377 ] one of the most interesting and puzzling results in the zrs regime is the temperature dependence .
the @xmath4 oscillations are shown in fig . 3 for different temperatures but under the same mw irradiation .
below a temperature @xmath36 k both the maxima and the minima exhibit strong temperature dependence . using standard arrhenius plot we present the @xmath37 of the first four minima on a logarithmic scale , versus @xmath38 .
all four minima conform to a general expression , @xmath39 , over at least one decade in the @xmath3 .
unusually large @xmath40 is found for these minima .
for example , @xmath41 k for the first minimum , exceeds both the mw photon energy ( @xmath42 k ) and the landau level spacing ( @xmath43 k ) by about an order of magnitude . as also shown in fig .
3 , the @xmath3 peaks are roughly linear with @xmath38 . using corbino samples ,
we have observed zero - conductance states ( zcs ) under similar experimental conditions.@xcite typical data from a high mobility ( @xmath44 @xmath12/vs ) 2deg corbino ring are shown in fig .
temperature dependence , as well as i v measurements at zcs , shows that 2des behaves like a macroscopic insulator in this regime.@xcite
remarkably , we have observed a series of new zrs associated with fractional @xmath15 , such as and @xmath45 ; the associated minima develop into zrs as the microwave power is increased ( attenuation of the microwave power is marked next to the @xmath3 traces ) .
, width=245 ] @xmath46 and @xmath47 , in an ultraclean 2deg sample ( similar to the sample used in fig .
2 ) exposed to mw of lower frequency ( @xmath48 ghz ) .
figure 5 shows the development of such zrs with increasing mw power @xmath49 .
at all @xmath49 values the previously reported `` integer '' zrs are observed . beyond the @xmath50 peak ,
a @xmath51 peak and associated minimum ( around @xmath52 t ) are already visible at lower power ( attenuation @xmath53 db ) . with increasing @xmath49 , this minimum develops into a new wide zrs .
the @xmath46 structure has first been noticed in ref . and can also be seen in fig . 1 ; the feature becomes stronger in samples with higher mobility.@xcite dorozhkin@xcite and willett _
et al._@xcite have also reported features associated with @xmath54 . in this experiment
we also observe fractional zrs forming about @xmath45 .
another important observation is the ( overall ) suppression of the @xmath3 at @xmath55 . from the @xmath5 position we interpret these new zrs as resulting from two - photon processes , where two photons of frequency @xmath56 participate in the transition between landau levels separated by @xmath57 and @xmath58 , respectively@xcite .
this observation may suggest the importance of virtual processes in the microwave - induced oscillations and zrs . in the higher frequency region ( @xmath59 ghz ) , we observed a strong modulation of sdh oscillations by the presence of mw .
this phenomenon has been reported in ref . for samples of a moderately - high mobility . the data in fig .
6 shows similar results from an ultra - clean 2des .
in addition to the first node , higher - order nodes were observed in this sample .
it is interesting to investigate the photoconductivity oscillations and zrs in a high - mobility 2d hole system ( 2dhs ) primarily for the following reasons : 1 ) comparing with 2d electron system , 2dhs has different band structure parameters such as larger effective mass and significant nonparabolicity of the dispersion ; 2 ) 2dhs possesses strong spin - orbit interactions . our preliminary result for a 2dhs
is shown in fig .
the 2dhs is from a ( 001 ) gaas / al@xmath60ga@xmath61as qw grown by mbe .
the qw has a well width of 15 nm ; holes are provided by a c delta - doping layer situated 50 nm away from the well .
a 100-@xmath62m - wide hall bar , on which measurements were performed , is defined by lithography and is contacted by in / zn alloy . under microwave irradiation .
, width=377 ] to simultaneously detect photoresistivity , @xmath63 , and resistivity , @xmath64 , a double - modulation technique was employed .
the main feature of @xmath65 , as shown for mw frequency of 26.5 ghz , is a positive peak at @xmath66 kg and a negative valley ( a suppression of @xmath64 ) around @xmath67 kg , both having relatively small magnitude .
the 2.5 kg peak could be associated with the cr of the 2d holes which have an effective mass @xmath68 ( measured in separate transmission experiments in a similar @xmath5 range ; for measurement techniques , see ref . ) .
the nature of the 6 kg valley is not yet clear . from the mobility of the holes , @xmath69 @xmath12/vs and a mass value of 0.3 m@xmath70
, we estimate a transport scattering time of @xmath71 ps .
this value is a factor of two smaller than that of the 2des ( where @xmath72 ps ) in which the giant amplitude oscillations were originally observed,@xcite a fact that may be considered in explaining the lack of oscillations in this 2dhs .
however , the band structure of holes must be taken into account to understand these data .
oscillatory and vanishing magnetoresistance can be induced by millimeterwave irradiation in high mobility two - dimensional electronic systems traditionally used for studies of the quantum hall effect . depending on the sample geometry , zero resistance states ( in hall samples ) or zero conductance states ( in corbino samples ) are observed .
some new aspects of the observations are reported here which may help to distinguish between competing theoretical models of the phenomenon . ,
s. das sarma and a. pinczuk ( wiley , new york , 1996 ) .
zudov , r.r .
simmons , and j.l .
reno , _ phys .
* b 64 * , 201311(r ) ( 2001 ) ; see also cond - mat/9711149 ( 1997 ) .
engel , d.c .
tsui , j.a .
simmons , j.r .
wendt , g.a .
vawter , and j.l .
reno , _ appl .
lett . _ * 79 * , 2193 ( 2001 ) .
maan , th .
englert , and d.c .
tsui , _ appl .
_ * 40 * , 609 ( 1982 ) .
m. dobers , k. v. klitzing , and g. weimann , _ phys .
rev . _ * b38 * , 5453 ( 1988 ) .
e. vasiliadou , g. muller , d. heitmann , d. weiss , k. v. klitzing , h. nickel , w. schlapp , and r. losch , _ phys .
* b 48 * , 17145 ( 1993 ) .
mani , j.h .
smet , k. von klitzing , v. narayanamurti , w.b .
johnson , and v. umansky , _ nature _ * 420 * , 646 ( 2002 ) ; _ phys .
lett . _ * 92 * , 146801 ( 2004 ) ; _ phys .
_ b 69 , 161306 ( 2004 ) .
zudov , r.r .
pfeiffer , and k.w .
west , _ phys .
lett . _ * 90 * , 046807 ( 2003 ) .
yang , m.a .
zudov , t.a .
knuuttila , r.r .
pfeiffer , and k.w .
west , _ phys .
_ * 91 * , 096803 ( 2003 ) .
dorozhkin , _ jetp lett . _ * 77 * , 577 ( 2003 ) .
r. g. willett , l. n. pfeiffer , and k. w. west , _ phys .
lett . _ * 93 * , 026804 ( 2004 ) . a detailed analysis of the oscillation phase can be found in m. a. zudov , _ phys .
_ * b 69 * , 041304(r ) ( 2004 ) .
a. e. kovalev , s. a. zvyagin , c. r. bowers , j. l. reno , and j. a. simmons , _ solid state commun .
_ * 130 * , 379 ( 2004 ) .
s. a. studenikin , m. potemski , a. sachrajda , m. hilke , l. n. pfeiffer , and k. w. west , cond - mat/0404411 .
durst , s. sachdev , n. read , and s.m .
girvin , _ phys .
lett . _ * 91 * , 086803 ( 2003 ) .
anderson and w.f .
brinkman , cond - mat/0302129 .
andreev , i.l .
aleiner , and a.j .
millis , _ phys .
* 91 * , 056803 ( 2003 ) .
i. a. dmitriev , m. g. vavilov , i. l. aleiner , a. d. mirlin , and d. g. polyakov , cond - mat/0310668 . i. a. dmitriev , a. d. mirlin , and d. g. polyakov , _ phys
lett . _ * 91 * , 226802 ( 2003 ) . m. g. vavilov and i. l. aleiner , _ phys . rev . _
* b 69 * , 035303 ( 2004 ) .
see , e.g. , references in r. r. du _ et al .
_ , _ physica _ * e22 * , 7 ( 2004 ) .
ryzhii , _ sov .
solid state _ * 11 * , 2078 ( 1970 ) ; v.i .
ryzhii , r.a .
suris , and b.s .
shchamkhalova , _ sov .
semiconductors _ * 20 * , 1289 ( 1986 ) . c. l. yang , m. a. zudov , r. r. du , l. n. pfeiffer and k. w. west , 2000 , unpublished .
b. e. cole , f. m. peeters , a. ardavan , s. o. hill , j. singleton , w. batty , j. m. chamberlain , a. polisskii , m. henini , and t. cheng , _ j. phys . : condens
. matter _ * 9 * , 3163 ( 1997 ) .
|
giant - amplitude oscillations in dc magnetoresistance of a high - mobility two - dimensional electron system can be induced by millimeterwave irradiations , leading to zero - resistance states at the oscillation minima . following a brief overview of the now well - known phenomenon ,
this paper reports on aspects of more recent experiments on the subject .
these are : new zero - resistance states associated with multi - photon processes ; suppression of shubnikov - de haas oscillations by high - frequency microwaves ; and microwave photoconductivity of a high - mobility two - dimensional hole system .
|
there is compelling evidence for the existence of dark matter in the universe . already in 1933 , @xcite had collected enough data to postulate the presence of more mass than what could be inferred from visible galaxies . over the past 80 years , additional observations ranging from the rotation curves of spiral galaxies @xcite to large - scale structure @xcite seem to point in the same direction .
one outstanding prediction of cosmological simulations using cold dark matter particles is that the milky way halo should be heavily populated with thousands of smaller dark matter subhalos as a result of the hierarchical assembly process @xcite . dark matter
subhalos would include any dark matter configuration , from those hosting the largest known dwarf spheroidal galaxies in the milky way to the lightest predicted dark matter substructures with masses around @xmath6 @xcite . conceptually in this scenario , the bulk of the subhalo population is made up by small - scale dark matter substructures with limited or null star formation , which would be almost impossible to detect in existing optical surveys .
should dark matter subhalos without major star formation episodes exist in large numbers , one of the only ways to detect them might be by tracking gamma rays from dark matter annihilation .
the all - sky coverage and unprecedented sensitivity of the large area telescope ( lat ) , on board nasa s _ fermi _ satellite , is enabling the most effective search for potential dark matter subhalos to date .
the obvious place to look for subhalo candidates is among the 3033 sources detected and characterized in the third _ fermi_lat catalog ( 3fgl ) @xcite , in particular , among the 1010 sources listed as unassociated with counterparts of known gamma - ray - producing source classes .
it is plausible that after years of observations , _ fermi _ has already detected dark matter subhalos .
the challenge now is to locate them @xcite .
previous attempts to pinpoint dark matter subhalos have systematically searched for unassociated sources with spectra that are consistent with dark matter annihilation @xcite .
machine - learning classification targeting dark matter subhalos has also been performed using @xmath7-means clustering in the first _ fermi _ lat catalog ( 1fgl ) @xcite and random forest in the second _ fermi _ lat catalog ( 2fgl ) @xcite . here
we present an application of machine - learning classifiers that aims to chart all galactic sources outside the plane in the 3fgl . by specifically isolating 3fgl objects at high latitude
we hope to reduce the search space for dark matter subhalos and unknown source classes as much as possible .
the paper is organized as follows . in section 2
we introduce the spectral prescription for locating galactic objects outside the plane in the 3fgl . in section 3
we introduce machine - learning classifiers .
section 4 covers training sample and variable selection . in sections 5 and 6
we discuss cross validation and prediction results . in section 7 , we discuss possible interpretations of the results including a visual search for ultra - faint dwarf galaxy and globular cluster counterparts , as well as direct comparison with statistics from pulsar population synthesis models .
we also place constraints on the annihilation cross section by comparing the number of potential subhalos to predictions from cosmological simulations .
finally , in section 8 we present our conclusions and outlook .
at high galactic latitude , the totality of known galactic gamma - ray emitters correspond to pulsars or globular clusters hosting msps @xcite .
two additional undiscovered gamma - ray - producing source classes have been postulated to exist : dark matter subhalos @xcite and dwarf galaxies @xcite .
dark matter annihilation is expected to dominate the gamma - ray signal from both known galactic dwarf galaxies and from dark matter subhalos .
if two dark matter particles annihilate through typical standard model channels , the decay and hadronization of these particles would create a gamma - ray spectrum that extends up to the rest mass of the dark matter particle with a sharp cutoff @xcite .
the gamma - ray spectrum is given by @xmath8 where @xmath9 is the thermally averaged annihilation cross section , @xmath10 is the dark matter particle mass , and @xmath11 is the gamma - ray yield per annihilation and depends on the particle physics model under consideration @xcite .
the second part of the equation , or the so - called astrophysical factor @xmath12 , corresponds to the integration of the dark matter density squared @xmath13 along the line of sight @xmath14 over a solid angle @xmath15 .
the nearest and/or most massive dark matter subhalos would be easiest to detect even if they host no stars . figure 1 shows the predicted gamma - ray spectrum for the annihilation of 30 gev dark matter particles into bottom quark ( @xmath5 ) pairs
. the spectral shape from dark matter annihilation would be quite recognizable .
unfortunately , a dark matter annihilation spectrum around a few gev would not be unique .
most known gamma - ray pulsars display nearly identical spectra with sharp exponential cutoffs @xcite . to illustrate this point ,
we have overlaid in figure 1 a typical msp spectrum drawn from the second _ fermi _ lat pulsar catalog ( 2fpc ) @xcite .
this pulsar - dark matter spectral degeneracy has driven a lively debate about the origin of excess gev emission at the galactic center @xcite ; however , the spectral degeneracy permeates to other astrophysical settings beyond the galactic center . indeed ,
because of this gamma - ray degeneracy , one can not expect to distinguish spectrally between pulsars , low - luminosity globular clusters hosting msps , ultra - faint dwarf galaxies , and dark matter subhalos .
by contrast , none of the known extragalactic sources have exponentially curved lat spectra , except for the large magellanic cloud ( lmc ) pulsar @xcite .
variability is also a far more likely characteristic of extragalactic objects than of galactic objects at high latitude .
one can turn the spectral degeneracy between some galactic classes and the marked contrast with extragalactic objects into assets by using known pulsar and globular cluster gamma - ray spectra to help hunt down undiscovered dark matter subhalos and ultra - faint dwarf galaxies in the 3fgl with machine - learning classifiers .
we describe such search strategy next .
by machine learning classifier , we denote any algorithm with the ability to perform accurate predictions , after having trained on the properties of a well - known training set of data . for most of this work
, we will concentrate on supervised machine learning classifiers or sets of models that can input a list of variables of a data set and output a prediction model that best describes the relationship between the variables and known classes . for an overview of popular machine - learning classifiers ,
we refer the reader to @xcite . among the plethora of available machine - learning classifiers ,
we have settled on two highly accurate variants of classification trees , i.e. random forest and xgboost @xcite .
random forest is an ensemble learning method that grows a large forest of randomized classification trees and aggregates their predictions made from a list of input variables @xcite .
individual classification trees are constructed by randomly sampling @xmath7 variables from @xmath16 input variables at each node @xcite .
the forest selects a random sample with replacement from the original training set using bagging ( bootstrap aggregating ) .
any data left out of the bootstrap sample helps to measure the internal accuracy directly in the form of an out - of - bag estimates . to classify an object , each tree in the forest issues a prediction . the predictions from all trees for the same object
are then collected and a class is determined through a majority decision . aside from being easy to use , random forest provides outstanding performance and the ability to track proximity matrices .
we adopted the _
randomforest _ set of routines , which implements the original random forest for classification and regression to the r language @xcite .
the final model is based on 1000 trees , with a total of @xmath17 variables sampled at each split , where @xmath18 is the final number of variables .
the extreme gradient boosting xgboost is a modified version of gradient boosting @xcite .
the fundamental difference with random forest is that xgboost uses boosting to reweigh the training set sequentially @xcite .
in contrast to bagging , boosting uses all instances at each repetition but issues a weight for each instance in the training set .
these evolving weights adjust the learners to focus on different instances at each pass .
one of the key problems in tree classifiers is how to find the best split at each node . to expedite this decision
, xgboost finds the best solution over all the possible splits according to percentiles of variable distribution @xcite . in all experiments ,
we boost trees with a learning rate @xmath19 and a maximum number of iterations set to 5 .
in order to construct a base training set for the galactic population outside the plane , we use 143 3fgl pulsars identified by pulsations ( psrs ) , 15 3fgl globular clusters ( glcs ) , and 24 3fgl pulsars with no pulsation seen in lat yet ( psrs ) .
the extragalactic set includes 1745 sources from all agn classes in the 3fgl .
it contains 3 non - blazar active galaxies ( agn ) , 573 blazar candidates of uncertain type ( bcu , bcu ) , 660 bl lacs ( bll , bll ) , 484 flat - spectrum radio quasars ( fsrq , fsrq ) , 5 narrow - line seyferts 1 ( nlsy1,nlsy1 ) , 15 radio galaxies ( rdg , rdg ) , 3 soft spectrum radio quasars ( ssrq ) , 1 compact steep spectrum quasar ( css ) , and 1 seyfert galaxy ( sey ) .
the entire dataset is divided into a training set ( a random sample of 2/3 of the total ) and a testing set ( remaining 1/3 of the total ) .
variable selection is essential to machine - learning classifiers .
starting from a set of parameters or variables the algorithm must be able to classify an object into one of a set of distinct classes .
for any classification problem , there are certain variables that best capture a specific class of objects .
the advantage of using machine - learning classifiers is that the algorithms can explore the entire variable space at once .
initially we started with a total of 35 3fgl variables , excluding positional , uncertainty and descriptive variables . in view of the inherent difficulties in sampling gamma - ray spectra for any given object
, we augmented this original variable set with eight additional derived variables defined by hardness ratios ` hr_{ij}`= @xmath20 and flux ratios ` fr_{ij } ` = ` flux_{i}`/ ` flux_j ` between consecutive @xmath21 bands for 0.10.3 gev ( band 1 ) , 0.31 gev ( band 2 ) , 13 gev ( band 3 ) , 310 gev ( band 4 ) , and 10100 gev ( band 5 ) .
figure 2 shows the out - of - bag error as a function of the number of variables used .
the out - of - bag estimates tend to asymptote to a minimum at @xmath22 .
the value shown for @xmath23 corresponds to the optimal scenario when using only ` signif_curve ` and ` variability_index ` as variables . for each classifier
, we also quantify feature importance to pinpoint the features that best discriminate between classes .
table 1 ranks variable importance ( from most to least important ) in terms of the improvement achieved from each variable or @xmath24 for all splits and trees for the xgboost model . for comparison , table 2
ranks the overall percentage decrease in accuracy rate averaged over all trees or @xmath25 measured by random forest @xcite .
as can be seen , the variable rankings are not identical but there is commonality on the top variables .
the differences are an indication of the distinct paths the classifiers follow to achieve a successful prediction . after exploring different variable combinations , we find that the variables that most clearly differentiate galactic and extragalactic populations include ` signif_curve ` , ` variability_index ` , and ` spectral_index ` .
we also include flux ratios and hardness ratios for five energy bands in our final models .
cc + ` signif_curve ` & 0.47 + ` variability_index ` & 0.26 + ` fr_{45 } ` & 0.10 + ` spectral_index ` & 0.08 + ` fr_{12 } ` & 0.03 + ` fr_{23 } ` & 0.03 + ` pivot_energy ` & 0.02 + ` fr_{34 } ` & 0.01 + cc + ` variability_index ` & 0.58 + ` fr_{45 } ` & 0.40 + ` signif_curve ` & 0.40 + ` pivot_energy ` & 0.33 + ` hr_{45 } ` & 0.30 + ` spectral_index ` & 0.26 + ` hr_{23 } ` & 0.24 + ` fr_{23 } ` & 0.22 +
during initial tests , we quickly found out that random forest suffered from the fact that the training set was highly imbalanced , i.e. 90% extragalactic versus 10% galactic trainees .
as constructed , random forest tends to build classification trees dominated by the majority class . to alleviate this problem , we artificially increased the training set using over - sampling as implemented in the synthetic minority over - sampling technique ( @xmath26 ) @xcite
. @xmath26 creates a much larger synthetic minority class by replicating the existing sample using five nearest neighbors . with the @xmath26 approach we were able to achieve a test set error rate of 2.8% for random forest .
class imbalanced was not a problem for xgboost and the training was done with the original training and testing samples , reaching a test set error rate of 3.3% .
figure 3 shows the receiver operating characteristic ( roc ) curves for both classifiers @xcite . at various threshold values
, we plot the true positive rate ( the fraction of correctly identified ) versus the false positive rate ( the fraction of candidates incorrectly classified ) . as shown , random forest slightly outperforms xgboost , but only after taking into account the class imbalance .
a perfect classifier would have a true positive rate equal to one .
a complementary measure of accuracy can be accomplished by applying random forest and xgboost retroactively to unassociated sources listed in the 1fgl and 2fgl . since a number of _ fermi _ discoveries
have been made since those early catalogs were first released , we can now directly measure how many high - latitude galactic candidates selected using machine - learning classifiers have been confirmed either as pulsars or globular clusters .
out of 22 galactic candidates above @xmath27 significance picked by the classifiers in the 1fgl , 18 have been confirmed as pulsars to date .
this implies that the accuracy achieved by the classifiers exceeds 80% at that significance level . in figure 4
we show the number of 3fgl galactic candidates at @xmath28 as a function of flux . for comparison , 1fgl and 2fgl candidates using the same machine - learning classifiers are also shown .
we can clearly see the tremendous progress in pulsar discoveries since the release of the 1fgl based on 11 months of _ fermi _ lat data .
next we applied the classifier models to the entire 3fgl unassociated sample . in order to focus on sources outside the galactic plane , we excluded unassociated 3fgl sources within 5@xmath29 of the plane .
this cut leaves us with a starting list of 675 unassociated objects from a total of 1010 .
since we are only interested in high - quality predictions , we further impose the condition that in order to retain a galactic candidate both classifiers must agree ( probability @xmath30 that the object is galactic in both classifiers ) . finally , we only report predictions for 3fgl objects with a detection significance @xmath31 . based on these criteria ,
we find a total of 34 high - latitude galactic candidates with energy flux @xmath32 erg @xmath2 s@xmath3 between 100 mev and 100 gev at @xmath0 .
table 3 lists these candidates . to check for consistency , we compared our list of candidates with results from spectral fitting of 3fgl unassociated sources @xcite , as well as with pulsar predictions using a combination of random forest and logistic regression @xcite .
the list of galactic candidates at high galactic latitude is in good agreement with both of these works .
armed with these 34 potential high - latitude galactic objects , we can now place the results in context of known and hypothesized gamma - ray source classes .
one possibility is that some of these newly discovered objects are either low - luminosity globular clusters @xcite or ultra - faint dwarf galaxies . in order to examine this possibility
we have visually inspected digitized sky survey ( dss ) and sloan digital sky survey ( sdss ) images @xcite .
we confined our visual inspection to the area enclosed within the 3fgl 95% uncertainty ellipses of the 34 candidates .
our search mainly focused on visually diffuse and extended objects in the optical , but none were found at least 1 magnitude above the dss optical limit .
it is important to note that the visual approach is limited to objects that can easily standout in the images but it is severely hampered when trying to reach ultra - faint dwarf galaxies as the ones detected by sdss @xcite and des @xcite . for six optical fields with reliable multi - band sdss photometry ( 3fgl j0318.1 + 0252 , 3fgl j1120.6 + 0713 , 3fgl j1225.9 + 2953 , 3fgl j1625.10021 , 3fgl j2103.71113 , 3fgl j2212.5 + 0703 ) , we systematically searched for unusual concentrations of rr lyrae .
rr lyrae stars are excellent distance indicators , as well as superb tracers of stellar substructures away from the galactic plane @xcite . in order to pinpoint rr lyrae star candidates within the 3fgl 95% uncertainty ellipses , we adopted a simple sdss color cut scheme based on equations 14 derived by @xcite .
we find no anomalous excess of rr lyrae stars in any of the 6 fields with significance greater than @xmath33 .
it is quite possible that fainter stellar structures with very few members could be present below the optical detection limits @xcite . given the ultra - fast pace of pulsar discoveries with _ fermi _ , it is not surprising that at least five of the galactic candidates identified here have been already confirmed as pulsars since the 3fgl release ( see table 3 ) . in the 3fgl
there are 89 pulsars listed at @xmath0 with energy flux larger than @xmath34 erg @xmath2 s@xmath3 between 100 mev and 100 gev . of the 89 known high - latitude gamma - ray pulsars ,
77 are msps with spin period @xmath35 ( s ) and spin - down rate @xmath36 satisfying log@xmath37 + 19.5 + 2.5 log@xmath38 p @xmath39 0 @xcite .
an addition of 34 pulsars would bring the total number of _ fermi _ lat detected pulsars at high latitude to 123 . in order to compare this number to theoretical predictions from pulsar simulations
, we performed a new population synthesis of msps @xcite in a similar fashion as in the study of @xcite .
however , we have used improved pulsar spin - down formalism , empirical gamma - ray luminosity , and beam geometries of the high - energy emission models of the high - altitude slot gap ( two pole caustic ) @xcite ( tpc ) , of the outer gap @xcite ( og ) , and of the pair - starved polar cap @xcite ( pspc ) .
in addition to the radio beam geometry used in @xcite , we implemented a radio aligned model ( altpc ) in which the radio emission has the same geometry as the high - energy tpc model .
the characteristics of thirteen radio surveys provided the detection thresholds for simulated radio pulsars , and the _ fermi _ point source threshold map in the 2fpc was scaled for various observing periods to determine the detection of simulated gamma - ray pulsars .
the models predict that _ fermi _ lat should detect anywhere from 100 to 126 msps in four years with an energy flux larger than @xmath40 erg @xmath2 s@xmath3 at @xmath0 ( table 4 ) . assuming that 85% of the 34 galactic candidates are msps , as is observed in the 3fgl catalog at high galactic latitude , then the total number of 3fgl msps at @xmath0 would be @xmath41 ( see table 4 ) .
therefore , the simulated pulsar population is in excellent agreement with the projected number of 3fgl msp detections .
in fact , depending on the pulsar emission model there is some wiggle room ( @xmath42420 ) for additional 3fgl msp discoveries at high galactic latitude .
below @xmath27 and above @xmath43 significance , the machine - learning classifiers find an additional 33 galactic candidates that could easily cover this difference .
the locations of the 34 high - significance galactic candidates and known 3fgl pulsars at @xmath0 are shown in figure 5 . taking advantage of the predictions in @xcite , we can also directly compare the number of galactic candidates at @xmath28 to the expected number of nearby subhalos detectable by _ fermi _ lat after four years of observations . because our classifiers rely on the 3fgl variables , we are only sensitive to dark matter annihilation in the 100 mev300 gev energy range .
it is important to note that no significantly exponentially curved candidates at energies @xmath44 gev have been found in any of the hard _
fermi_-lat source catalogs @xcite . as in the original approach outlined in @xcite , the subhalo predictions in @xcite
rely on the distribution of hundreds of thousands of subhalos that were simulated in 6 ultra - highly resolved milky - way sized halos as part of the aquarius project @xcite .
@xcite calculated the number of detectable aquarius subhalos as a function of _ fermi _ lat flux and galactic latitude .
recently , @xcite used results of the via lactea ii simulation @xcite scaled to the planck 2015 cosmological parameters and found a slightly smaller number of detectable subhalos , but still consistent with the aquarius results considering the range of assumptions involved . for our comparison of high - latitude candidates with dark matter predictions ,
we removed recently detected pulsars from table 3 .
this brings down the number of @xmath28 subhalo candidates from 17 to 14 .
figure 6 shows the upper limits on @xmath9 for dark matter masses @xmath10 between 30 gev and 10 tev annihilating into a bottom quark ( @xmath5 ) final states based on the detection of 14 potential subhalo candidates at @xmath28 .
note that these limits would only be slightly better if one includes the role of halo substructure to boost the subhalo annihilation flux @xcite .
more precisely , following @xcite , we estimate at most a factor @xmath45 10% stronger limits when boosting the annihilation signal in the mass range @xmath46@xmath47 @xmath48 . for comparison
, we also show constraints if eventually no dark matter subhalos turn up in the 3fgl from the aquarius @xcite and via lactea ii simulations @xcite respectively .
while the current limit is not constraining enough to rule out the canonical thermal relic cross section at energies below 100 gev @xcite , the curve is starting to approach competitive values . indeed
, further associations of some of these candidates with conventional astrophysical sources may only lead to more stringent limits . as noted by @xcite
, there are significant uncertainties in the number of predicted dark matter subhalos that could shift these dark matter limits by a factor of a few .
the point remains that _ fermi _ lat might have detected several dark matter subhalos by now , even under the most pessimistic assumptions .
the absence of an overwhelming number of subhalo candidates provides complementary verification of more robust annihilating dark matter limits inferred from dwarf galaxies @xcite and the galactic center @xcite .
we find that the set of variables provided in the _ fermi _ lat catalogs have the ability to effectively predict gamma - ray source classes in the 3fgl dataset .
after careful examination of various galactic demographics , we find that the 34 additional high - latitude galactic candidates predicted using machine - learning classifiers can be accommodated by existing pulsar population synthesis models without the need to introduce undiscovered globular clusters , dark matter subhalos , or gamma - ray emitting ultra - faint dwarf galaxies . on the other hand , if these objects were produced by annihilating dark matter , the upper limits on the annihilation cross section are starting to approach values at or below the canonical thermal cross section for energies @xmath49 gev
. the discovery of radio and gamma - ray pulsations will be crucial to address the spectral degeneracy between dark matter annihilation and pulsar emission .
however , blind searches will face greater obstacles in noisy msps and fainter gamma - ray sources as _ fermi _ continues operations .
table 4 shows projected discoveries of msps for 10 years of _ fermi _ lat data taking .
the most promising follow - up strategy to break these degeneracies will rest on our ability to detect pulsations going from the brightest to the faintest galactic candidates .
some of these searches for the most elusive gamma - ray pulsars are being conducted by the distributed volunteer computing sources , @xmath50 @xcite .
new discoveries will require even larger computing resources and new search strategies .
optical , ultraviolet and x - ray searches for binary objects with temporal variability could also enhance the chances for finding millisecond pulsars @xcite .
incidentally , the addition of new msps will also bring us closer to the detection of nanohertz gravitational waves based on pulsar - timing arrays @xcite .
should additional high - latitude galactic candidates be confirmed as pulsars , new swaths of annihilation cross sections will be disfavored by direct comparison with statistics from cosmological numerical simulations of milky way - like galaxies .
therefore , subhalo searches represent a powerful complementary method to existing probes of dark matter annihilation .
clearly , there ought to be dedicated multiwavelength campaigns to map the error ellipses of high - latitude galactic candidates for which no radio / gamma - ray pulsations are found .
finally , the improvements in position and photon flux afforded by pass 8 analysis @xcite should further enhance machine - learning predictions in the future _ fermi _
lat fourth source catalog ( 4fgl ) .
this research was supported by a senior appointment to the nasa postdoctoral program at the goddard space flight center , administered by universities space research association through a contract with nasa .
masc is a wenner - gren fellow and acknowledges the support of the wenner - gren foundations to develop his research .
we thank seth digel for helpful suggestions .
3fgl j0212.1 + 5320 & 134.93 & -7.65 & 1.00 & 1.00 + 3fgl j0238.0 + 5237 & 138.85 & -6.92 & 0.94 & 0.85 & + 3fgl j0312.10921 & 191.51 & -52.36 & 0.95 & 0.94 & + 3fgl j0318.1 + 0252 & 178.45 & -43.64 & 1.00 & 1.00 & + 3fgl j0336.1 + 7500 & 133.11 & 15.50 & 0.98 & 1.00 & + 3fgl j0523.32528 & 228.20 & -29.83 & 0.89 & 1.00 & + 3fgl j0545.6 + 6019 & 152.4964 & 15.7493 & 0.67 & 0.89 & + 3fgl j0758.61451 & 233.9599 & 7.5619 & 0.90 & 0.83 & + 3fgl j0802.35610 & 269.9308 & -13.1755 & 0.72 & 0.90 & + 3fgl j0838.82829 & 250.6050 & 7.8008 & 0.97 & 1.00 & + 3fgl j0933.96232 & 282.2351 & -7.8937 & 1.00 & 1.00 & + 3fgl j0953.71510 & 251.9380 & 29.6055 & 0.96 & 0.98 & + 3fgl j0954.83948 & 269.8445 & 11.4575 & 0.61 & 0.52 & + 3fgl j1035.76720 & 290.3908 & -7.8284 & 1.00 & 1.00 & psr @xcite + 3fgl j1119.92204 & 276.4696 & 36.0588 & 0.97 & 1.00 & + 3fgl j1120.6 + 0713 & 251.5322 & 60.6852 & 0.85 & 0.73 & + 3fgl j1225.9 + 2953 & 185.1521 & 83.7648 & 0.94 & 1.00 & + 3fgl j1539.23324 & 338.7592 & 17.5342 & 0.90 & 1.00 & + 3fgl j1544.61125 & 356.2111 & 32.9844 & 0.63 & 0.73 & psr @xcite + 3fgl j1557.04225 & 335.6413 & 8.3622 & 0.50 & 0.57 & + 3fgl j1624.24041 & 340.5718 & 6.1421 & 1.00 & 1.00 & psr ( @xmath50 ) + 3fgl j1625.10021 & 13.8808 & 31.8378 & 1.00 & 1.00 & + 3fgl j1653.60158 & 16.6181 & 24.9246 & 0.98 & 1.00 & + 3fgl j1702.85656 & 332.3978 & -9.2447 & 0.62 & 0.84 & + 3fgl j1744.17619 & 317.1046 & -22.4711 & 0.99 & 1.00 & psr @xcite + 3fgl j1753.64447 & 347.0854 & -9.4164 & 0.98 & 0.99 & + 3fgl j1946.45403 & 343.8883 & -29.5630 & 1.00 & 0.99 & psr @xcite + 3fgl j2039.65618 & 341.2312 & -37.1551 & 0.98 & 1.00 & + 3fgl j2103.71113 & 37.8579 & -34.4231 & 0.93 & 0.87 & + 3fgl j2112.53044 & 14.8984 & -42.4487 & 0.99 & 0.99 & + 3fgl j2117.6 + 3725 & 82.7982 & -8.2737 & 0.81 & 0.90 & + 3fgl j2133.06433 & 328.7390 & -41.2683 & 0.93 & 1.00 & + 3fgl j2212.5 + 0703 & 68.7429 & -38.5650 & 0.92 & 0.99 & + 3fgl j2233.1 + 6542 & 109.3427 & 6.5614 & 0.74 & 0.95 & +
|
nearly 1/3 of the sources listed in the third _ fermi _ large area telescope ( lat ) catalog ( 3fgl ) remain unassociated .
it is possible that predicted and even unanticipated gamma - ray source classes are present in these data waiting to be discovered . taking advantage of the excellent spectral capabilities achieved by the _ fermi _
lat , we use machine learning classifiers ( random forest and xgboost ) to pinpoint potentially novel source classes in the unassociated 3fgl sample outside the galactic plane . here
we report a total of 34 high - confidence galactic candidates at @xmath0 .
the currently favored standard astrophysical interpretations for these objects are pulsars or low - luminosity globular clusters hosting millisecond pulsars ( msps ) . yet , these objects could also be interpreted as dark matter annihilation taking place in ultra - faint dwarf galaxies or dark matter subhalos .
unfortunately , _ fermi _
lat spectra are not sufficient to break degeneracies between the different scenarios .
careful visual inspection of archival optical images reveals no obvious evidence for low - luminosity globular clusters or ultra - faint dwarf galaxies inside the 95% error ellipses .
if these are pulsars , this would bring the total number of msps at @xmath0 to 106 , down to an energy flux @xmath1 erg @xmath2 s@xmath3 between 100 mev and 100 gev .
we find this number to be in excellent agreement with predictions from a new population synthesis of msps that predicts 100126 high - latitude 3fgl msps depending on the choice of high - energy emission model .
if , however , these are dark matter substructures , we can place upper limits on the number of galactic subhalos surviving today and on dark matter annihilation cross sections .
these limits are beginning to approach the canonical thermal relic cross section for dark matter particle masses below @xmath4 gev in the bottom quark ( @xmath5 ) annihilation channel .
|
a transition - edge sensor ( tes ) is a thin superconducting film that can be used as a sensitive thermometer when voltage biased within the normal metal - superconducting transition region and read out with superconducting squid sensors @xcite .
tes based devices are used as extremely sensitive bolometers and calorimeters to detect radiation in a wide energy range from gamma - rays to sub - millimeter radiation @xcite , and typically the thermal conductance to the bath is controlled by mounting the tes on a thin insulating sin membrane .
although the performance of these detectors is already excellent , the most sensitive tes devices have not yet reached the theoretical limits in energy resolution .
this is mostly due to excess noise that has been shown to be present in many devices @xcite .
several candidates for the noise sources have been proposed , such as thermal fluctuations within the tes @xcite , fluctuations in the cooper - pair density @xcite or phase - slips @xcite , but a definitive answer is still missing .
however , before resorting to more exotic noise sources to explain the data , one should be sure the known noise mechanisms are fully understood .
these are the thermal fluctuations of the electrical degrees of freedom , or johnson noise , and the thermal fluctuations of the energy degrees of freedom , usually called thermal fluctuation noise or phonon noise @xcite .
both of these noise mechanisms are unavoidable in bolometric tes detectors ; moreover , their magnitude depends on the details of the device in question .
it is therefore of utmost importance to characterize the detector accurately , both electrically and thermally , before one can understand the origin of noise .
a useful tool in characterizing tes detectors is the measurement of their frequency - dependent complex electrical impedance @xcite .
importantly , within the transition this impedance depends not only on the electrical , but also on the thermal circuit of the device , through the electrothermal feedback effect @xcite . with the help of this technique ,
a more accurate picture of the electrical and thermal properties of tes sensors has emerged : first of all , it was realized that the dependence of the detector resistance on current , and not only on temperature , is critical for the detector response , as well @xcite .
this was later shown to influence the johnson noise directly @xcite , and some of the excess noise could then be explained as non - equilibrium johnson noise .
in addition , it has become clear that for many detectors , the simplest thermal circuit of one heat capacity connected to heat bath through one thermal conductance is not adequate @xcite .
a more complex thermal circuit then adds new components to the thermal fluctuation part of the noise spectrum @xcite . here
, we present a study of the noise and complex impedance of several different designs of tes devices . many of the devices are based on the so - called corbino - geometry tes ( cortes)@xcite , where current spreads out radially from a central contact with radius @xmath0 into the outer contact at @xmath1 , instead of flowing linearly [ fig .
1 ( a ) ] .
although it is a bit more complicated to fabricate , it offers advantages in modelling , as the superconducting and normal regions separate due to the non - uniform current density profile .
this means that we can determine the phase boundary radius @xmath2 from the measured resistance @xmath3 , which depends logarithmically on @xmath4 in corbino geometry : @xmath5 where @xmath6 is the normal state resistance of the device .
once we know the size of the n and s phases , their expected theoretical heat capacities can be calculated as well .
this level of theoretical description is not possible in more common square shaped tes devices .
cortes has also shown an excess noise component , if compared to the simplest thermal model .
this was originally explained by the fluctuations of the n - s boundary , or fluctuation superconductivity noise ( fsn ) @xcite .
however , here we show that most of the `` excess '' noise is simply internal thermal fluctuation noise originating from a more complex thermal circuit within the device . by analyzing the data thoroughly , we suggest that most of the `` excess '' noise is generated by the thermal decoupling of the spatially separated superconducting and normal regions of the devices
. data from typical square shaped devices also supports this picture : due to non - uniformities of real devices and/or the lateral proximity effect @xcite , phase separation can still take place , although not in such a controllable manner as in cortes devices . and @xmath7 are the radii of the inner and outer superconducting contacts and @xmath2 is the phase boundary radius .
( b ) a schematic side view of a cortes with an absorber .
the arrows indicate the path of the bias current .
[ cortes ] ]
the simplest thermal model of a tes device is shown in fig .
[ block](a ) , consisting of a single heat capacity connected to the heat bath .
the complex impedance @xmath8 of a device with this one - body thermal circuit always traces a semicircle in the complex plane as a function of frequency @xcite .
thus , it is experimentally straightforward to determine whether a more complex thermal circuit is required by studying the shape of the @xmath8 curve .
whenever additional thermal blocks are added to the system , @xmath8 develops bulges outwards from the ideal case ( see fig .
[ mvsitfn ] ( a ) ) . the shape and size of the new features in @xmath8
depend on the heat capacities @xmath9 of the added bodies and the thermal conductance links @xmath10 between them , in such a way that the features grow in size with both @xmath9 and @xmath11 . ]
[ mvsitfn ] ( a ) shows a typical measured impedance curve of a cortes device , together with a one- , two- and three - body theoretical model fits . although simpler one and two - body models seem to fit some device designs by other groups @xcite , for our devices a three - body model is the simplest model that fits the data well , as is clear from fig . [ mvsitfn ] ( a ) .
recently , simple three - body models have also been used by other groups for more accurate modelling of their devices @xcite . ] in this work , we have therefore used the thermal model shown in fig .
[ block](b ) , which is the simplest three - body model that is physically justifiable for our devices .
in addition to the tes film with heat capacity @xmath12 where the bias power @xmath13 is dissipated , there is a hanging thermal body with heat capacity @xmath14 , and an intermediate heat capacity @xmath15 between the tes and the heat bath .
@xmath16 could be associated with the supporting sin membrane , and @xmath14 could represent the absorber on top of a tes film [ fig . 1 ( b ) ] ,
but in general it could be some other part of the device also .
most devices studied here do not actually have absorbers .
we call this the ih ( intermediate + hanging ) model , and use analytical expressions for the impedance and noise of a tes with this thermal circuit presented in @xcite .
their derivation and full theoretical discussion will be published elsewhere @xcite .
we simply note that with each new thermal block , a new thermal noise source is also introduced to the system , arising from fluctuations in energy between the bodies through the thermal link .
we call the noise due to any additional block internal thermal fluctuation noise ( itfn ) to distinguish it from the thermal noise between the system and the heat bath ( phonon noise )
. the intermediate body is characterized by its temperature @xmath17 and heat capacity @xmath15 .
we emphasize that we do not make the simplifying assumptions that @xmath18 or that @xmath19 , when the tes is biased .
associated with the intermediate block , we thus have four different values of dynamic thermal conductance , two for each physical link connected to @xmath15 , evaluated at each temperature end of the link @xcite .
the parameters @xmath20 and @xmath21 can be calculated from a careful analysis of the measured i - v characteristics , as described in ref @xcite .
in our fitting procedure we use @xmath17 , @xmath15 and @xmath22 as the free parameters . the other unknown conductances
@xmath23 , @xmath24 and @xmath25 are then calculated using the free parameters and the known values of @xmath26 and @xmath27 , as @xcite @xmath28 to simplify the model and to minimize fitting parameters , it is also assumed that the links on both sides of the intermediate block have the same thermal exponent @xmath29 , i.e. @xmath30 and @xmath31 .
this is physically reasonable if all the conductances are dominated by the phononic transport properties of the sin membrane .
later , we will show that the intermediate body fitting parameters are nearly constant throughout the transition , and only change when @xmath32 is changed .
also , @xmath15 appears to depend on both the area and thickness of the membrane .
this behavior is thus consistent with the assumption that the intermediate body represents the sin membrane phonons . furthermore , @xmath33 is usually always slightly below @xmath21 , and not close to the bath temperature .
in contrast to the intermediate body , the second added heat capacity @xmath14 in our model is hanging [ fig .
[ block ] ( b ) ] , which means that no steady state power flows through it , and the average values of @xmath26 and @xmath34 are equal .
this simplifies the description in comparison to the intermediate body , as only one thermal conductance @xmath35 is required .
the joule power of the tes bias current is all assumed to be dissipated inside @xmath12 . in real x - ray and @xmath36-ray devices with thick absorbers on top of the tes film @xcite
, @xmath14 could well be the absorber , in which case @xmath35 describes the thermal conduction within the tes and the absorber . in devices without an absorber ( most samples in this work ) , the idea is to model electronic degrees of freedom , thus @xmath35 models the thermal conduction within the tes film .
the hanging block is naturally still a simplification , in reality there can be also a thermal coupling from @xmath37 directly to the bath or to @xmath38 .
however , the effect of the missing link , which would have a magnitude of the order of @xmath39 , is believed to be minor for the devices studied here , where @xmath35 is much larger than @xmath39 , as will be seen later .
any effects of temperature gradients within the tes film are also not captured in this simple model .
numerical estimates of the gradients within a cortes device @xcite have shown that the isothermal simplification is a reasonable assumption .
for the cortes devices with phase separated n and s regions , one may at this point wonder , if the blocks @xmath12 and @xmath37 could actually directly represent the normal and superconducting regions of the tes film , respectively .
the great benefit of the cortes geometry is that we can a priori calculate the values of @xmath12 and @xmath37 as a function of bias , which is not possible for the more standard square shaped devices . as we later show
, the fitted values of the heat capacities and their behavior as a function of bias point do indeed follow a trend predicted by this interpretation .
@xmath35 is thus still the electronic thermal conductance , but now it can be affected by the location of the ns boundary , for example .
understanding of how @xmath35 should behave as a function of the bias ( phase boundary location ) is quite sketchy at the moment , and it needs to be studied further theoretically .
the hanging body @xmath37 produces internal thermal fluctuation noise ( itfn ) that appears in current noise spectra at the high - frequency side of the effective thermal time constant , just like johnson noise @xcite , see fig .
[ mvsitfn ] ( b ) .
in addition , when @xmath40 is large enough , the roll - off for the itfn noise occurs at very high frequencies , even above the electrical cut - off frequency of the read - out circuit .
thus , the itfn noise produced by the hanging body can look very similar to the johnson noise .
quite often in the past , excess noise in tes devices has been quantified by the parameter @xmath41 @xcite , which is a measure of excess johnson noise .
thus , one has to be very careful about drawing conclusions based on just the noise data : if the thermal circuit is not adequately characterized , one can misinterpret itfn noise as excess johnson noise . in our devices the time constant for itfn roll - off is indeed near the electric cut - off of our read - out circuit , so that fitting the spectra either with the three - body itfn model , or two - body + excess johnson noise ( @xmath41 parameter ) give almost identical results , as illustrated in fig .
[ mvsitfn ] ( b ) .
the only difference seen is that with the itfn model , the roll - off has a steeper slope , as it is a higher order roll - off produced by the combination of the thermal and electrical circuits . however , the point is that in many cases the impedance can only be fitted with the three block model [ see fig .
[ mvsitfn ] ( a ) ] , which supports the choice of itfn noise over other noise sources .
the experiments were performed in a compact homemade dilution refrigerator with a base temperature @xmath42 mk .
the tes devices were voltage biased by a shunt resistor ( 8.9 m@xmath43 ) at the sample stage , and a nist two - stage squid @xcite mounted on the 1 k stage ( equivalent input current noise density @xmath44 pa/@xmath45 and geometric input inductance of 300 nh ) served as the current amplifier for the readout .
the squid was operated in the flux locked loop , with dedicated room temperature electronics designed by sron .
more details on the setup can be found in @xcite .
all studied tes devices were first characterized by measuring their resistance vs. temperature transitions with a four - probe lock - in measurement , to find the normal state resistance @xmath46 and critical temperature @xmath47 .
next , each detector was connected to the squid readout , and a series of current - voltage ( i - v ) curves was measured at several different bath temperatures .
the thermal exponent @xmath29 was extracted from the @xmath48 dependence of the i - v data , as explained in ref . @xcite .
then , using the known @xmath29 , the tes temperature @xmath26 and dynamic thermal conductance @xmath27 at any point in the transition were determined from the i - v curves @xcite .
we also calculated the transition steepness parameter @xmath49 , which is a dimensionless measure of the sensitivity of the tes @xcite .
finally , a set of noise and complex impedance measurements were performed at a desired @xmath32 and at various bias points within the transition .
the tes impedance was extracted from the measured circuit impedance by dividing out the effect of the read - out circuit transfer function , as described in ref
. @xcite . from the tes
impedance curves , the low and high frequency limits @xmath50 and @xmath51 can be estimated , which can then be used to calculate the important transition parameters @xmath52 and @xmath53 as @xmath54 where @xmath55 and @xmath56 is the tes resistance . additionally , @xmath57/(2+\beta)$ ] from the impedance data should give @xcite the same value as @xmath58 , so that we can check for consistency between the impedance and i - v measurements . notice how @xmath59 depends both on @xmath60 and @xmath61 .
we have performed impedance measurements between 4 hz and 100 khz both by the white noise method @xcite , where a white noise excitation is used and the broadband response is measured , and by the lock - in method where a sinusoidal excitation is used at certain frequency points @xcite .
both methods can in general be used , however , we have seen that the sine - wave method is more reliable , as the total heating power generated by the excitation is typically less in the sine - wave method .
data with heating problems could be identified by the consistency check mentioned above : with excess heating @xmath62 did not agree with @xmath58 , indicating a heating induced shift in the bias point .
the measured impedance and noise data are finally fitted _ simultaneously _ to the ih model equations .
we emphasize that the fitting is done by eye , and free fit parameters are varied manually , as high - dimensional non - linear least - squares fitting would be demanding to implement . in some cases we found that the three - body equations still did not explain all the observed noise , even if the impedance fit was good . in those cases
we have quantified the remaining truly excess noise as excess johnson noise using the m - parameter .
note also that although here we only show the complex plane plots of @xmath8 , we also make sure that the real and imaginary parts fit separately as a function of frequency . included in the fits , but not shown in the plots because of their small values ,
are the johnson noise of the known shunt resistor , and the equivalent white input noise of the squid .
we have also used the equation for the ( lowest order ) non - equilibrium tes johnson noise @xcite @xmath63 in the analysis for all the devices .
even though @xmath8 is measured only up to 100 khz , the theory curves are always calculated up to 2 mhz , in order to see the high - frequency differences between the fits , as demonstrated in section [ sec : slice1 ] .
in this section we present the measured data and fit results from several different detectors .
table [ testable ] lists some of the key parameters for each device studied .
most of them have radially spreading current distribution as in the fully circulary symmetric cortes device shown in fig .
[ cortes ] .
however , the device labelled stes is divided into four equal size parallel slices ( see inset , fig . [ stesparam ] ) , and the detectors labelled slice 1 and slice 2 are individual slices ( fig . [ 111fits ] ) .
also , data from one traditional square - shaped device is also presented .
all measurements were done at a regulated bath temperature of 60 mk unless stated otherwise .
the detectors were not shielded against external magnetic fields , nor was earth s field compensated for .
.[testable ] device parameters for the tess measured in this work . [ cols=">,^,^,^,^",options="header " , ] the data we have measured on a full cortes turned out to be quite difficult to fit accurately even with the three - body ih model , as seen from the examples shown in fig .
[ otesdata ] .
the general trends are , nevertheless quite well reproduced , including the very strong deviations of @xmath8 from the simplest circular shape and all the trends in the noise spectra .
typical values for the most important fit parameters were @xmath64 pj / k , @xmath65 pj / k , @xmath66 pj / k and @xmath67 nw / k throughout the transition for the bare devices without an absorber .
[ otesdata ] also shows the data after the deposition of a 2 @xmath68 m thick bi absorber on top , as shown schematically in fig .
[ cortes](b ) . in that case
, all other parameters stayed about the same , except for @xmath37 , which , quite reasonably , doubled to @xmath69 pj / k .
we should note that the obtained @xmath70 values are reasonable if compared to estimates from the detector size , except for @xmath38 which is surprisingly large .
this could be because of the 120 nm thick alox insulator layer separating the nb bias lines [ [ cortes](b ) ] in the cortes device , creating unwanted thermal links and heat capacity inside , as shown in ref .
@xcite for a different device .
-parameter was used .
[ otesdata ] ] very accurate determination of how the addition of the absorber affected the detector characteristics is unfortunately complicated , because the absorber processing changed also the transition properties , by lowering the @xmath71 from 98 mk to 85 mk , and by broadening the transition . for the bias
values shown in fig .
[ otesdata ] , @xmath72 decreased from @xmath73 to @xmath74 at @xmath75 , whereas at @xmath76 it stayed constant at @xmath77 .
these changes influence the detector responsivity , which means that noise and @xmath8 are affected not only through the increase of @xmath37 but also directly .
the large difference of noise at @xmath75 can be attributed mostly to the change in @xmath60 , whereas at @xmath78 the lowering of noise is consistent with the lower @xmath79 of the device after the addition of the absorber .
we have also measured a set of data from another absorberless cortes device , shown in fig .
[ stesfits ] .
it was otherwise nominally identical to the full device discussed above , but was cut into four parallel slices ( dubbed stes ) , with an sem image shown in the inset of fig .
[ stesparam](b ) . in this case
, the impedance fits to the ih model look nearly perfect , and the noise data agrees also quite well without the need of an empirical @xmath41 parameter , except at the very highest frequencies near the cut - off ( reason unknown at the moment ) . achieving better fits than in the full cortes case was easier , most likely because the stes device has a lower responsivity ( lower maximum @xmath80 ) . in fig .
[ stesparam ] we plot the obtained fit parameters of interest as a function of the bias point @xmath81 .
the heat capacities , fig .
[ stesparam ] ( a ) , do not vary strongly , and their values are generally consistent with the cortes results .
@xmath82 pj / k is again high , even higher than in the cortes device . to compare the data with the suggestion that @xmath12 and @xmath37 are associated with the normal and superconducting regions , respectively
, we also show theoretical curves corresponding to the calculated heat capacities of the normal and superconducting phase regions , using the location of the phase boundary calculated from eq .
[ eq : rb ] , the known film thicknesses and sommerfeld constants @xmath83 j / k@xmath84m@xmath85 and @xmath86 j / k@xmath84m@xmath85 from literature @xcite .
in addition , the heat capacity of the s region was calculated using two different assumptions : ( a ) the jump at @xmath79 is given by simple bcs theory as @xmath87 ( dash - dotted line ) , and ( b ) the jump is slightly suppressed due to proximity effect according to ref .
we see that @xmath37 is clearly consistent with it being associated with the s region , however @xmath12 is a bit elevated compared to the simplest theory .
moreover , lower in the transition at @xmath88 , there seems to be a trend that @xmath89 increases and @xmath90 decreases .
this trend is very clear in the data , as forcing the values of @xmath37 and @xmath12 for @xmath88 to equal the values at @xmath91 produces very poor fits , as shown in figs .
[ stesfits ] and [ stesparam ] . , 0.2 , 0.3 , 0.4 , 0.5 , 0.8 ,
rest are omitted for clarity .
dashed lines show fits with alternative parameter values shown in fig .
[ stesparam ] as crosses .
[ stesfits ] ] as a function of the bias point.(a ) the heat capacities , and ( b ) the thermal conductances .
@xmath27 is fixed by the i - v data and is not a fit parameter .
theory curves show the expected dependence of the normal region ( dashed line ) and the superconducting region with two different models for the jump at @xmath79 : full jump ( dash - dotted ) and suppressed according to @xcite ( solid line ) .
crosses show alternate parameter values used for `` bad fits '' shown in fig .
[ stesfits ] .
inset in ( b ) is an sem image of the stes .
[ stesparam ] ] the usual cortes devices have an alox insulator layer between the bias leads .
furthermore , from the high values of @xmath38 obtained above , and from previous measurements in ref .
@xcite , we suspect that the alox layer could be responsible for the high value of @xmath38 .
therefore , we also studied devices where only a quarter `` slice '' of the full cortes disk is retained : this way the insulator layer is not required , as the center contact lead can come from the opposite side ( see inset fig .
[ 111fits ] ) .
however , we still retain a geometry that promotes a phase separation . here
we report on measurements on two such samples , with different au / ti thicknesses , leading to different @xmath79 and @xmath6 values ( table [ testable ] ) .
the au / ti layer thicknesses were increased compared to the full cortes devices , in order to keep @xmath46 approximately the same .
also , the thickness of the sin membrane ( 300 nm ) was chosen to be the same as in the usual cortes devices for sample 1 , whereas sample 2 had much thicker sin ( 750 nm ) .
the first slice tes has relatively low values of @xmath92 , with the result that the measured impedance data , shown in fig . [
112fits ] , lacks any striking features .
the data is therefore too easy to fit : that is , we can find several ways to fit the data ( both impedance and noise ) , and choosing the best fit is not simple based on this data alone . to illustrate this ,
we have fitted the same data in three different ways , with the resulting fitted curves almost identical up to 100 khz , but with significant differences in the obtained parameter values .
the three cases we consider are : case 1 : we force @xmath93 to the theoretically calculated value and let the other parameters vary freely . in case 2 , @xmath94 is not fixed but also free ( as in all the fits for the cortes and stes devices above ) . finally , in case 3 , we set @xmath95 , reducing the thermal model to two bodies : the tes and the intermediate .
= 0.1 , 0.2 , 0.4 , 0.5 , 0.6 , 0.8 , rest omitted for clarity .
we also show the best fit for case 1 ( solid black line ) and case 2 ( dashed red line ) .
the dotted green line in ( a ) is the impedance according to the thermal model of fig .
[ block](a ) , illustrating how much our data deviates from the simple one - body case , and it was calculated using @xmath96 and @xmath97 obtained from case 1 at @xmath98 = 0.4 .
[ 112fits ] ] from the cortes sample results above , we still expect that the three - block model is required to describe the sample physics , thus we first discuss the comparison between cases 1 and 2 , and only later comment on the case 3 fits .
[ 112fits ] shows a comparison of cases 1 and 2 with the data .
it is clear that both cases fit the data well , and deviate from each other only at high frequencies close to 100 khz , with case 1 making a more pronounced kink in the calculated impedance and slightly weaker roll - off for the noise . .
case 1 is shown by filled symbols , and case 2 with open symbols . ( a )
squares : @xmath99 , triangles : @xmath100 and circles : @xmath89 .
lines are the theoretical calculations as discussed in the text .
( b ) thermal conductances of interest .
( c ) @xmath101 and @xmath102 parameters calculated from the i - v ( line ) and impedance data ( points ) .
[ 112param ] ] looking at the obtained parameters plotted in fig .
[ 112param ] , we find case 1 quite interesting . with @xmath93 forced to the theoretical values , @xmath103 follows the calculated theoretical value of the reduced bcs heat capacity @xcite , supporting again the proposed picture of thermal decoupling between the n and s phases .
as expected , due to the lack of alox , @xmath104 pj / k is much reduced compared to the cortes and stes samples , even though the sin membrane dimensions are the same .
case 2 reproduces all the trends , but now @xmath12 is higher than the theory ( as in the stes device ) and @xmath103 lower , but with their sum approximately constant between the two cases .
we also note , how for both cases @xmath103 grows fast below @xmath105 , similar to the stes device . on the other hand , comparing the two cases in terms of @xmath40 fit values produces a striking difference : the trends are quite clear , but opposite for the two cases .
thus , it is hard to draw solid conclusions , except that one can see that the highest values obtained @xmath106 nw / k are consistent with the increased tes bilayer film thickness , if compared to the stes results .
the trend in case 1 of decreasing @xmath40 with decreasing @xmath3 is of course consistent with the fact that thermal conductance starts to decrease in the superconducting phase , and that the size of the s phase increases when going down in bias . in cases 1 and 2
we did not need to use the @xmath41-parameter ( excess noise ) in the upper part of the transition , but had to include it in the lower part , in case 1 below 30 % bias and in case 2 already at 50 % .
this is in contrast to the cortes and stes fits , where @xmath41-parameter was not used at all ( those fits were not as accurate , though ) .
the fitted values of m are shown in fig .
[ mbetactot](a ) . for this sample
, we also found that in a narrow bias range @xmath107 = 0.3 to 0.35 there was an enhancement of the mid - frequency ( 1 khz - 30 khz ) noise that could be explained neither by the thermal circuit nor an @xmath41-noise component .
curiously , this bias range corresponds to the sudden drop in the @xmath60 parameters , as can be seen in fig . [ 112param](c ) ) .
( a ) , the dashed lines show the calculated one - block impedances using @xmath97 of case 1 ( short dash ) and case 3 ( long dash ) . [ 112cabs0 ] ] to study case 3 , we show in fig .
[ 112cabs0 ] representative impedance fits at two bias points ( solid lines ) , with case 1 fits included for comparison . again
, if we only look at frequencies up to the measurement maximum , we do not see any difference in the fits .
the difference in the noise fits ( not shown ) is also limited to a small variation in the roll - off . in case 3
we naturally do not have the high frequency itfn component present because @xmath103 is missing , and thus a large part of the noise at high freqencies has to be accounted for by excess noise with @xmath41 parameter , at all bias points .
the effect of the analysis case on the obtained @xmath41-parameter values vs the bias point is shown in fig .
[ mbetactot ] ( a ) .
in addition , the chosen fitting case has a big effect also on @xmath108 , as it is proportional to @xmath51 [ fig .
[ mbetactot ] ( b ) ] .
the plots remind us that many parameters strongly depend on the chosen thermal model and how it is interpreted .
if we do not know with certainty which of the possible thermal models is correct , we should not jump to conclusions about the nature of the excess noise based on the dependence of @xmath41 parameter on bias , magnetic field etc .
in other words , the @xmath8 data may look like it fits a simpler one- or two - block model , but the underlying true model could still be different .
for example , if one is developing a detector where @xmath108 is an important parameter , mistakes can be made in tes design if the decisions are based on @xmath108 from a wrong fit . for the case of this particular device , we believe that cases 1 and 2 are more accurate because of the evidence from the other samples studied here , where two - block model fits are impossible to the impedance data .
and ( c ) the sum of @xmath93 and @xmath103 between different fits cases .
symbols : values from fits .
line : bcs heat capacity with the bilayer correction , dashed line , full bcs theory .
[ mbetactot ] ] another interesting point to notice is that even though we have three different sets of values for @xmath93 and @xmath103 , the sums @xmath109 are essentially constant between the different fitting cases , as shown in fig .
[ mbetactot](c ) .
the values are also consistent with the modified bcs theory @xcite , except at the lowest bias points , where the rapid increase mentioned before is clearly seen . at this point
we wish to speculate about the observations at low bias points . a possible scenario for the effects seen
( increasing @xmath70 , increasing @xmath41-noise ) could well be related to vortex physics .
if vortices are not pinned , they could contribute to heat capacity and also generate excess noise by the so called phase - slip shot noise mechanism @xcite .
the sudden drop in @xmath110 is also correlated with the onset of the need for @xmath41- parameter in case 1 .
the lower @xmath60 would then correspond to extra resistance in the transition , caused by the vortex motion . in this light
we can perhaps also understand why the full cortes devices do not require @xmath41 to fit their noise , as the extra nb top layer over the tes film may pin vortices more strongly , and prevent the generation of excess phase slip noise .
another option is that some of the excess @xmath41-noise could be generated by the fsn mechanism @xcite .
however , fsn theory in its current state can not explain the increased heat capacity or lowered @xmath60 .
the geometrical design of slice tes sample 2 was identical to sample 1 , however the sin and tes layer thickness are different ( table [ testable ] ) , resulting also to a higher @xmath79 .
in contrast to sample 1 , the second slice tes featured a more complicated transition with very large peaks in @xmath58 as shown in the inset of fig .
[ 111fits](b ) .
the high values and large variation in @xmath58 produces complicated impedance features and high noise levels .
sample 2 was also measured at four different bath temperatures : 140 mk , 110 mk , 85 mk and 60 mk .
[ 111fits ] shows the data for a few representative bias points measured at a bath temperature of 60 mk .
the fits were done with the full ih model using case 2 , keeping @xmath12 free .
now case 3 is again out of the question due to the complex shapes of the @xmath8-data .
case 1 was tried but did not produce as good fits as case 2 .
insets : ( a ) sem image of the tes , ( b ) total @xmath101 calculated from i - v .
[ 111fits ] ] again , the main features of both @xmath8 and noise data are captured with the ih model . however , at bias points where @xmath60 is very large ( for example @xmath111 ) it was more difficult to find good fits , and determination of @xmath51 is more uncertain .
the inevitable result of this is more scatter in the fitted parameters . in fig .
[ 111param ] we have plotted all the results from all four bath temperature runs in one graph , as @xmath12 , @xmath103 and @xmath40 should not depend on @xmath32 at all according to the thermal model ( @xmath38 depends on @xmath48 through its influence on @xmath112 , thus we show only @xmath113 60 mk fits ) . comparing to sample 1 , we see that @xmath12 and @xmath37 have the same trends , and only slightly higher values , consistent with the increase of @xmath79 . on the other hand
, @xmath38 is three times larger .
this again supports the picture that @xmath38 originates from the sin membrane , as it is 2.5 times thicker for sample 2 .
@xmath40 is of the same order of magnitude , and does not follow a monotonous trend throughout the whole transition .
most of the noise data did not require any additional @xmath41-noise , all the fits shown in fig .
[ 111fits ] are without the @xmath41 parameter .
this , we speculate , could be because of the much higher @xmath60 compared to sample 1 : the themal noise grows so much that is swamps any possible additional noise sources .
, except @xmath38 is shown only at @xmath113 60 mk .
( a ) squares : @xmath99 , circles : @xmath114 and triangles : @xmath100 .
dashed black and solid red lines are theoretical calculations for @xmath99 and @xmath89 , respectively , whereas the dash - dotted line is their sum .
the reduced bcs value @xcite was used for @xmath103 .
[ 111param ] ] the data for @xmath115 was not shown because it depends on the bath temperature .
for comparison with sample 1 , at 60 mk sample 2 had @xmath116 = 1.1 nw / k and @xmath117 = 0.97
nw / k at 0.5 @xmath118 . as can be seen for bias point @xmath119 in fig .
[ 111fits ] , the measured noise sometimes develops a peak near the high - frequency roll - off , when the tes is biased low in the transition .
this indicates electrothermal oscillations and it is predicted by our thermal model , as shown by the fit .
it arises due to a resonant - like interaction between the electrical and thermal circuits : when @xmath103 starts to decouple from a tes that has a very small heat capacity and large enough @xmath101 , the tes response becomes oscillatory .
the increased responsivity means that all noise sources develop a peak , including the johnson noise , for example .
finally , we show also results for a simple square tes . the measured detector was a bare @xmath120 tes with nb bias lines and no extra features added ( more sample parameter details in table [ testable ] ) .
it has a smoothly changing @xmath121 with a fairly low maximum value of 50 . in fig .
[ ttesfits ] we plot the measured and fitted impedance and noise data .
just as for all the other geometries , our square tes clearly has a large high frequency noise component .
again , the shapes of the impedance curves shows that the model without a hanging block ( case 3 discussed in section [ sec : slice1 ] ) will not work here either .
the ih model again produces a very good fit for both @xmath8 and noise , which gives us confidence that the noise in the square tes can be explained by the same mechanism as in the other geometries .
note that no @xmath41 parameter was needed in these fits . ] fig .
[ ttesparam ] shows the relevant parameters from the fits .
we again observe a decreasing @xmath12 and an increasing @xmath37 , consistent with our model of thermal decoupling of the n and s regions , and comparable values to the corbino - like samples . to calculate
some estimate for the n and s phase heat capacities , we take a crude model of a linear transition , where the ratio of the volumes is given directly by @xmath98 , so for example in the middle of the transition we would have exactly half of the device in the superconducting state .
this model produces a correct order of magnitude for @xmath12 and @xmath37 , but naturally no real agreement , although the parameter values do seem to change linearly with @xmath3 , as our overly simple model predicts .
the values of thermal conductances are again consistent with the corbino devices , and @xmath35 does not show any simple dependence on @xmath3 . .
lines show the simple estimates for @xmath12 ( dashed ) and @xmath37 ( solid ) discussed in the text .
[ ttesparam ] ]
we have fabricated several different geometrical designs of tes devices , and studied their impedance and noise properties .
all data are consistent with the picture that a three block thermal circuit is required to accurately describe the data . in all devices ,
the fitted heat capacity values follow trends which suggest that one of the extra heat capacity blocks arises from the insulating films such as the sin membrane and , if present , the alox layer used in some devices .
the other two blocks seem to represent the normal and superconducting regions of the tes film itself , with the observation that the finite value of thermal conductance within the tes film ( or a possible thermal boundary resistance between the n and s phases ) , in combination with the large value of the heat capacity in the superconducting phase near @xmath79 , leads to fairly large observable features in the impedance and noise data at high frequencies . in most cases
, this internal thermal fluctuation noise component explains all of the `` excess noise '' in the devices .
however , in some devices indications of additional noise sources remained , especially low in the transition . in our ti / au devices the resistivity of the tes film is such that @xmath40 is `` low enough '' . by this
we mean that the resulting cut - off frequency of the itfn noise due to the hanging block is near the electrical cut - off of the readout , so that we can observe its effect in the measured data . our data , therefore , agrees with the previous suggestion that a large itfn noise exists in ti / au tess @xcite .
in contrast , the lower resistivity mo / au or mo / cu devices usually have a lower excess noise level that is white voltage noise , in other words noise that looks like excess johnson noise @xcite .
our model could perhaps explain part of this difference between ti based and mo based devices through the higher thermal conductance @xmath35 of the mo / au and mo / cu tes films .
if @xmath35 grows , the itfn noise level will fall , and the cut - off is pushed to a higher frequency .
thus if the itfn roll - off happens much after the readout cut - off , itfn noise will look like excess johnson noise in the frequency range of the measurement .
more measurements on mo devices are required to test this hypothesis .
the results reported here indicate that in an optimal tes design , the internal thermal conductances within the metallic parts of the device ( tes film , absorber ) should be maximized .
as the actual phase separation between the superconducting and normal regions may be impossible to prevent , the effect of the superconducting heat capacity is minimized if the heat capacity of the normal part of the tes is larger .
we note that the leading designs in terms of reported energy resolution @xcite employ added normal metal features , and thus we speculate that the reduction of their noise could originate , at least partly , due to the increased normal region heat capacity .
we acknowledge helpful discussions about data analysis with m. lindeman , j. van der kuur , j. ullom and d. swetz .
this work was supported by the finnish funding agency for technology and innovation tekes and eu through the regional funds , and the finnish academy project no .
128532 . m. p. would like to thank the national graduate school in materials physics for funding .
m. e. eckart , j. s. adams , s. r. bandler , r. p. brekosky , j. a. chervenak , f. m. finkbeiner , r. l. kelley , c. a. kilbourne , f. s. porter , j. e. sadleir , and s. j. smith , aip conf .
proc . * 1185 * , 430 ( 2009 ) .
n. iyomoto , s. r. bandler , r. p. brekosky , a .-
brown , j. a. chervenak , f. m. finkbeiner , r. l. kelley , c. a. kilbourne , f. s. porter , j. e. sadleir , and s. j. smith , appl .
* 92 * , 013508 ( 2008 ) .
i. j. maasilta and k. m. kinnunen , aip conf . proc . * 1185 * , 38 ( 2009 ) .
note that there are errors in equations ( 3 ) and ( 7 ) of this reference : the corrected versions should read @xmath122/(2+\beta)$ ] and @xmath123^{-1}$ ] .
|
we have studied the origin of excess noise in superconducting transition - edge sensors ( tes ) with several different detector designs .
we show that most of the observed noise and complex impedance features can be explained by a thermal model consisting of three bodies .
we suggest that one of the thermal blocks and the corresponding thermal fluctuation noise arises due to the high - frequency thermal decoupling of the normal and superconducting phase regions inside the tes film .
our results are also consistent with the prediction that in thin bilayer proximitized superconductors , the jump in heat capacity at the critical temperature is smaller than the universal bcs theory result .
|
big bang nucleosynthesis ( bbn ) remains one of the most successful probes of early times in the hot big bang cosmology .
standard bbn ( sbbn ) assumes the standard model proposition of only three massless light neutrinos , leaving the primordial element abundances characterized by only one parameter , the baryon - to - photon ratio @xmath9 .
now , the evidence for neutrino mass from the super - kamiokande atmospheric neutrino observations is overwhelming @xcite .
in addition , the solar neutrino deficit and the lsnd experiment @xcite are hinting at the presence of a fourth light neutrino that would necessarily be `` sterile '' due to the @xmath10 decay width @xcite .
the role of neutrino masses and their mixing bring another aspect to the physical evolution of the early universe , when neutrinos played a large role .
the effects of neutrinos on big bang nucleosynthesis can be in part parameterized by the effective number of neutrinos @xmath11 ( _ i.e. _ the expansion rate ) , and in the weak physics which determine the neutron - to - proton ratio .
in many works @xmath11 is regarded as the sole determinant of neutrino effects in bbn .
however , we find that the effect of neutrino mixing on bbn is not well described by this one parameter .
characterizing the effects of neutrino mixing on bbn by @xmath11 can become overly complicated , since the parameter has been used to describe several distinct physical effects , including lepton number asymmetry and change in energy density .
another feature we find is that matter - enhanced mixing of neutrinos in the early universe can alter @xmath5 spectra to be non - thermal , which uniquely affects the nucleon weak rates by lifting fermi blocking of neutron decay by neutrinos and suppression of neutrino capture on neutrons .
the observation of the abundances of primordial elements has greatly increased in precision within the past few years @xcite . in sbbn , each observed primordial abundance corresponds to a value of the baryon - to - photon ratio @xmath12 , where @xmath13 is the fractional contribution of baryon rest mass to the closure density and @xmath14 is the hubble parameter in units of @xmath15 .
the theory is apparently successful since the inferred primordial abundances correspond within errors to a single value of @xmath9 .
sbbn took a new turn in 1996 when it was found that the deuterium abundance at high redshifts may be significantly lower than that inferred through the cosmic helium abundance @xcite .
the advent of high - resolution spectroscopy of quasars allowed the measurement of deuterium abundance in high - redshift clouds between us and distant quasars .
the `` early days '' of these high - redshift spectral deuterium measurements were filled with controversy due to a disparity in the inferred deuterium to hydrogen ( d / h ) ratio between two observational groups working with different quasars .
more recently , the number of quasars with low d / h has grown .
the deuterium abundance has converged to @xmath1610% precision : @xmath17 .
the @xmath9 value inferred from this observation is @xmath18 .
the prediction for the corresponding primordial @xmath19he mass fraction is @xmath20 .
based on stellar absorbtion features in metal - poor hii regions , the observed helium abundance has been argued by olive , steigman , and skillman ( oss ) @xcite to be @xmath21 .
the corresponding baryon - to - photon ratio is @xmath22 . the discrepancy between the central values of @xmath23 predicted by the burles and tytler deuterium observations and @xmath24 has led to much speculation on non - standard bbn scenarios which could reconcile the two measurements by decreasing the @xmath23 predicted by bbn .
however , the systematic and statistical uncertainty in the @xmath23 and @xmath24 puts them well within their @xmath25 errors .
furthermore , izotov and thuan @xcite exclude a few potentially tainted hii regions used in oss from their analysis and arrive at a significantly higher @xmath26 . the systematic uncertainty claimed in observed @xmath0
may also be underestimated @xcite .
it has been suggested that this dubious @xmath0 and d / h discrepancy with standard bbn could be hinting at the presence of sterile neutrino mixing in the early universe @xcite .
however , we find that the overall effect of active - sterile mixing in the early universe would only increase @xmath0 and thus increase the possible disparity .
active - sterile mixing can produce an electron lepton number asymmetry @xmath27 before and during bbn @xcite . if the sign of @xmath27 is the same throughout the universe , the helium abundance would change to be more favorable ( @xmath28 ) or less favorable ( @xmath29 ) with observation . taking into account the effects of energy density , lepton number generation and alteration of neutrino spectra
, we find that @xmath30 can be large in the positive direction , but limited in the negative direction @xcite .
in particular , we find that matter - enhanced transformations can leave non - thermal @xmath5 spectra which significantly affect the weak rates through the alleviation of fermi - blocking and suppression of neutrino capture on neutrons @xcite .
furthermore , causality limits the size of the region with a certain sign of @xmath27 to be within the size of the horizon during these early times ( @xmath31 at weak freeze - out ) @xcite .
the sign of the asymmetry will vary between regions , and the net effect of transformation on @xmath0 will then be the average of both the @xmath32 and @xmath33 bbn yields , which leaves @xmath29 .
two possible schemes exist for producing a nonzero @xmath27 involving matter - enhanced ( msw ) transformation of neutrinos with sterile neutrinos .
these two schemes differ greatly in the energy distribution of the asymmetry in the spectra of the @xmath34 the asymmetry .
the resonant transformation of one neutrino flavor to another depends on the energy of the neutrino . in the case of bbn ,
the active neutrinos all start off as thermal , and sterile neutrinos are not present .
as the universe expands , decreases in density and cools , the position of the msw resonance moves in energy space from lower to higher neutrino energies .
the energy of the resonance is @xcite @xmath35 where @xmath36 is the neutrino energy normalized by the ambient temperature @xmath37 .
the asymmetry @xmath38 is generated as the resonance energy moves up the neutrino spectrum . as the universe cools , the resonance energy increases .
this produces a distortion of the neutrino or anti - neutrino spectrum .
there are two plausible cases for producing an asymmetry @xmath27 that would affect the production of primordial helium through the weak rates : * * direct two neutrino mixing : * an asymmetry in @xmath27 is created directly through a @xmath39 or @xmath40 resonance . in this case , the resonance starts at low temperatures for low neutrino energies , when neutrino - scattering processes are very slow at re - thermalization .
this case leaves the @xmath5 or @xmath41 spectrum distorted , with the spectral distortion from a thermal fermi - dirac form residing in the low energy portion of the spectrum .
the position of the distortion cut - off moves through the spectrum before and during bbn . in calculating the effects of this scenario on bbn
, we included the evolving non - thermal nature of the neutrino spectrum . *
* indirect three neutrino mixing : * an asymmetry in @xmath42 or @xmath43 created by a @xmath44 resonance is later transferred into the @xmath27 through a @xmath45 resonance ( @xmath32 ) .
this will happen for the anti - neutrino flavors for the @xmath33 case . in this scenario ,
the resonant transitions occur at higher temperatures when neutrino - scattering is more efficient .
both the @xmath46 and @xmath5 spectra re - thermalize , but since the temperature is below weak decoupling , the asymmetry in neutrino lepton numbers remain .
the effect of this scenario on bbn can be described by suppression or enhancement over the entire @xmath34 spectra .
bbn can be approximated as the freeze - out of nuclides from nuclear statistical equilibrium in an `` expanding box . ''
nuclear reactions freeze - out , or fail to convert one nuclide to another when their rate falls below the expansion rate of the box ( the hubble expansion ) . as different reactions rates freeze - out at different temperatures , the abundances of various nuclides get altered and then become fixed .
the nuclide produced in greatest abundance ( other than hydrogen ) is @xmath19he .
the reactions that affect the @xmath19he abundance ( @xmath0 ) the most are the weak nucleon interconversion reactions @xmath47 these reactions all depend heavily on the density of the electron - type neutrinos and their energy distributions @xcite .
the two neutrino transformation scenarios for altering the @xmath5 spectrum described above affect @xmath0 through these rates and through the effect of increased energy density on the expansion rate . in this situation ,
either the populations of the @xmath5 or @xmath41 energy spectra are cutoff at lower energies .
the reaction rates that alter the overall @xmath49 rates the most because of the spectral cutoff are @xmath50^{-1 } [ 1+e^{-e_e / k t_\nu}]^{-1}}\ ] ] @xmath51^{-1 } [ 1+e^{-e_e / k t_\nu}]^{-1}}\ ] ] ( the notation in these expressions follows that in ref .
the rate in eq .
[ rate1 ] is modified with the @xmath41 spectra , and the rate in eq .
[ rate2 ] is modified with the @xmath5 spectra . for a spectral cutoff for @xmath41 , corresponding to @xmath52 transformations , the fermi - blocking term @xmath53^{-1}$ ] in neutron decay ( eq .
[ rate1 ] ) becomes unity at low @xmath41 energies .
the rate integrand is then greatly enhanced when low energy neutrinos would otherwise block the process ( see figure [ intnueb ] ) . in the case of a spectral cutoff for @xmath5 ,
corresponding to @xmath54 transformations , the spectral term @xmath55^{-1}$ ] and thus the rate integrand go to zero for low energy neutrinos ( see fig [ intnue ] ) .
5 truein 0 cm 5 truein 0 cm the resultant change in @xmath0 for this `` two neutrino '' case is shown in figure [ dycut ] . for @xmath56 transformations , @xmath32 , the neutron decay rate in eq .
[ rate1 ] is enhanced , fewer neutrons are present , and @xmath0 is decreased . for @xmath39 transformations , @xmath57 , the @xmath58 rate in eq .
[ rate2 ] is suppressed , more neutrons are present , and @xmath0 is increased .
the effect of the rate in eq .
[ rate2 ] is much greater than that of the rate in eq .
[ rate1 ] on the neutron - to - proton ratio and @xmath0 .
this causes a larger magnitude change in the neutron - to - proton ratio for the @xmath33 case , and thus a larger magnitude change in @xmath0 in the positive direction than in the negative .
the sign of the asymmetry @xmath27 is randomly determined @xcite .
because of causality , different horizons will have an asymmetry of random sign @xcite , fifty - percent positive and fifty - percent negative .
the effect of the asymmetry on @xmath0 will then be the average of these two cases , which is also plotted in figure [ dycut ] .
3.5 truein 0 cm in the case that an asymmetry is produced in @xmath34 indirectly through an initial transformation between @xmath59 or @xmath60 , the neutrinos can re - thermalize through self - scattering at higher temperatures .
the number densities remain asymmetric , but the asymmetry is spread uniformly over the entire energy spectra . the effect on the weak rates is not as drastic as the above case , but is still present . in this case , the increase in energy density due to the thermalization of sterile neutrinos also affects @xmath0 through the increase in the expansion rate .
thus , for @xmath32 , the suppression of @xmath0 is counteracted by the increased expansion rate . the increase in energy density always causes a positive change in @xmath0 , while the change caused by the weak nucleon rates depends on the sign of @xmath27 , thus the overall resulting change in @xmath0 is greater in the positive direction than in the negative ( see figure [ dyoverall ] ) . for the @xmath32 case , the negative change @xmath30 has a minimum at @xmath61 since at this point the positive effect of energy density begins to dominate the negative effect of the modification of the weak rates .
3.5 truein 0 cm
the characterization of the effects of neutrino transformations on bbn through a single parameter @xmath62 hides varied physical effects of neutrino mass and mixing that are non - trivially related .
these effects include an increase in energy density and modification in the nucleon weak rates .
the modification of these rates is through either an asymmetry that is distributed over the entire spectrum , or an asymmetry confined to low energy @xmath34 .
we find that the distribution of the @xmath27 in the spectra is important to the weak rates , the neutron - to - proton ratio , and the production of @xmath19he .
in addition , the change in the predicted @xmath0 is always greater in the positive direction than in the negative , thus when averaged over causal horizons , @xmath30 is always positive .
the changes @xmath63 are comparable to the uncertainty in current @xmath0 measurements , thus resonant neutrino mixing can measurably affect the bbn predictions .
it is important to note that there may remain non - trivial portions of parameter space in non - standard bbn including , but not limited to , neutrino mixing , spatial variations in @xmath9 , and massive decaying particles that fit within the observed uncertainties of the primordial abundances .
|
we investigate the effects of matter - enhanced ( msw ) transformation of neutrinos in the early universe on the primordial helium yield ( @xmath0 ) .
we find that @xmath0 is affected much more by the msw - induced alterations in the neutrino energy spectra than by the associated change in expansion rate .
specifically , the absence due to transformation of low energy electron neutrinos can significantly affect neutron - proton weak interconversion rates through the lifting of fermi - blocking of neutron decay at low energies and through halting low - energy neutrino capture on neutrons .
we find that the change of @xmath0 within a causal horizon is @xmath1 for @xmath2 in the case of @xmath3-@xmath4-@xmath5 mixing , with the lower limit at @xmath6 , and @xmath7 for @xmath8 in the case of @xmath5-@xmath4 mixing .
|
to obtain the estimates of the fisher information , for the post - selected two - photon input state as plotted in fig . 3 of the main text ,
the probability functions for each of the outcomes were first estimated . theoretically , the probability functions are : @xmath74 , \\ p\left(|\delta| = 0|\theta \right ) & = \frac{1}{4}\left [ 3 - \mathcal{i}^\prime + ( 1 + \mathcal{i}^\prime ) \cos(2 \theta ) \right ] , \end{split}\ ] ] where @xmath75
. the detection efficiency of each outcome @xmath76 will affect the observed statistics so we correct for this in order to estimate the probability functions which would be observed with ideal detection efficiency .
these functions are fitted to our data by allowing the fourier coefficients ; @xmath77 to vary .
we fit probability functions with positive and negative @xmath78 separately .
maximum - likelihood estimation is used to find these coefficients .
firstly , estimates of ideal detection rates are obtained by dividing observed detection rates by the intrinsic detection inefficiency given by the combinatorics of a multiplexed detection system @xcite . to account for the unequal count rates across each interference fringe we fit the functions in eq .
( [ two_probs ] ) to normalized count rates , that is the counts of a specific outcome divided by the total number of counts at this position .
the likelihood function to be maximized is : @xmath79 where @xmath80 are the observed number of counts and @xmath81 is a vector of such over @xmath1 , @xmath82 is an estimate for the expected number of observed counts of @xmath78 type given @xmath83 , @xmath84 is a maximum likelihood estimate for the total number of events an ideal detector would record at this value of @xmath1 .
the log likelihood function was maximized numerically .
many random initial guesses of @xmath83 are followed by gradient ascent to try and find the global likelihood maximum . to find the statistical error associated with our analysis , this curve - fitting procedure
was conducted many times with counts which has been simulated base on the probability functions fitted to the real data .
the standard error of these repeats was taken to be the statistical error .
the probability functions for the four - photon interference fringes were granted an additional fourier term as they should contain oscillations with double the frequency over the two - photon state .
these coefficients do not have an expression in terms of @xmath26 .
the same procedure was carried out for the post - selected four - photon input state to perform curve fitting and error analysis .
[ [ b - predicted - f - vs - mathcaliprime - for - two - photon - fringes . ] ] b : predicted @xmath7 vs @xmath85 for two - photon fringes . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
the dominant source of background noise in the two - photon interference fringes we observed is accidental coincidences between detected photons which did not arise from a spdc event .
we measured the accidental coincidences by recording the coincidences after delays had been added to the output signals of our avalanche photodiodes such that any pairs from an spdc event would no longer be recorded as a coincidence .
this leaves only the rate of coincidences caused by uncorrelated photons .
we measured this rate to be @xmath86hz after correcting for non - uniform detection efficiency .
the two - photon count rate when the detectors signals were matched was @xmath87hz which results in a probability of a count being background of @xmath88% .
these rates were found by integrating over two minutes with the phase of the hwp @xmath89 and the intensity of the laser the same as it was when the two - photon interference fringes were recored .
the errors in average count rates were found assuming poisson statistics and have been propagated to the error in background probability by standard error propagation techniques . rather than correct for background noise
, we include this effect into our analysis such that the ideal phase dependent two - photon probabilities , @xmath90 , are modified to : @xmath91 where @xmath92% is the probability of observing a count due to background noise . using these probabilities we find the largest fisher information to be at a different point on the interference fringe , for the two - photon fringes the position with highest phase sensitivity is given by : @xmath93 { \frac{\zeta^2 - 2\zeta}{{\mathcal{i}^\prime}^2(\zeta-1)^2 - 1 } } \ , \right).\ ] ]
the resulting fisher information can be found by evaluating eq .
( 3 ) , within the main text , at this point .
this gives an algebraic expression for optimal fisher information : @xmath94}.\ ] ] inserting @xmath95 into eq .
( [ optimal_fi ] ) gives the theoretical line displayed in fig .
3 of the main text .
[ [ c - predicted - f - vs - mathcaliprime - for - four - photon - fringes . ] ] c : predicted @xmath7 vs @xmath85 for four - photon fringes . ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
to predict the fisher information of the four - photon state we follow a procedure similar to that of ref .
we experimentally determine properties of the four - photon state and then use this , in combination with background noise rates , to predict the fisher information . as in appendix
b we measured the rate of four - fold coincidences , when the coincidence time bins were matched , to be @xmath96 hz and when the time bins were mismatched to be @xmath97 hz .
the probability of a four - fold coincidence being background noise is therefore @xmath98 . as stated in the main text , the expression for the four - photon state when @xmath55 is @xmath99 where we have used abbreviated notation @xmath100 $ ] since there is only one set of spectral functions @xmath48 . passing this state through a hwp at @xmath101
applies a beamsplitter like transformation : @xmath102 , @xmath103 to the state : @xmath104 |\mathbf{0}\rangle \\ = & \ , \frac{1}{\sqrt{2 + 2\lambda_4}}\ ! \left[\
! \sum_i \lambda_i^2 \!\left(\frac{\sqrt{3}}{\sqrt{2 } } |4\rangle_{1,i } \!-\ ! |2\rangle_{1,i}|2\rangle_{2,i } \!+\ ! \frac{\sqrt{3}}{\sqrt{2 } } |4\rangle_{2,i } \right)\!+ \sum_{i > j } \lambda_i \lambda_j \big ( |2\rangle_{1,i}|2\rangle_{1,j } \!+\ ! |2\rangle_{1,i}|2\rangle_{1,j } \!-\ !
|2\rangle_{1,i}|2\rangle_{1,j } \big)\!\right ] \\
\end{split}\ ] ] yielding @xmath105 and hence @xmath106/[2 - 2p(|\delta| = 4)]$ ] . using the four - photon interference fringe
when @xmath50 , from the fitted functions , we find that @xmath107 once the background noise has been accounted for . therefore @xmath108 . then using this value for @xmath54 , and the level of background noise , we could evaluate an expected fisher information across @xmath109 when @xmath50 ( @xmath70 ) and for large @xmath44 ( @xmath56 ) to have maximums of @xmath110 and @xmath111 .
errors on these values have been calculated by finding the expected value for fisher information using the extremal errors for the background noise .
the points are plotted as green circles in fig . 3 of the main text and the error bars are smaller than the points .
as can be observed in fig . 3 these values suitably agree with the experimentally obtained values for fisher information when @xmath70 and when @xmath56 .
|
quantum metrology promises high - precision measurements beyond the capability of any classical techniques , and has the potential to be integral to investigative techniques
. however , all sensors must tolerate imperfections if they are to be practical .
here we show that photons with perfectly overlapped modes , which are therefore fully indistinguishable , are not required for quantum - enhanced measurement , and that partially - distinguishable photons do not have to be engineered to mitigate the adverse effects of distinguishability .
we quantify the effect of distinguishability on quantum metrology experiments , and report results of an experiment to verify that two- and four - photon states containing partially - distinguishable photons can achieve quantum - enhanced sensitivity with low - visibility quantum interference .
this demonstrates that sources producing photons with mixed spectral states can be readily utilized for quantum metrology . in an ideal scenario ,
the use of non - classical states of @xmath0 photons to measure an optical phase @xmath1 will enable the scaling of precision to be increased beyond the shot - noise limit ( snl)@xmath2to the fundamental heisenberg limit@xmath3 @xcite .
however , a real sensor will operate in non - ideal conditions with non - ideal parameters .
effects such as photon loss and phase diffusion have been shown to remove much of the advantage offered by non - classical techniques , so that the snl can only be beaten by a constant factor @xcite , and this motives further study of imperfections in quantum metrology @xcite .
proposals for photonic quantum metrology typically exploit quantum interference of photons which have exactly the same parameters and are therefore perfectly indistinguishable .
achieving this indistinguishability is a major technical challenge in practice , in particular for immature and developing photon source technology .
here we study the effect on precision measurements using non - classical probe states that contain partially - distinguishable photons , which leads to degraded quantum interference .
we find that despite high levels of distinguishability , it is still possible to achieve a quantum advantage in interferometry if this is the only imperfection and provided there is non - zero indistinguishability , heisenberg scaling is still achieved .
this classifies the effect of distinguishability as separate to those of optical loss and phase diffusion .
we have performed a proof - of - principle experiment to observe the quantum advantage that can be achieved with 2- and 4-photon probe states with varied distinguishability .
quantum metrology promises an advantage where high precision is needed whilst minimising probe intensity to avoid damaging the system under investigation @xcite . to become useful outside of specialized settings , developments in quantum metrology
must address the optical needs of samples being investigated .
creating states which possess desired optical properties with highly - indistinguishable photons is a technological challenge and has been achieved only at a handful of wavelengths @xcite .
however , systems which could use only partially - distinguishable photons would have a greatly - reduced technological challenge associated with creating the photon source .
such systems could dramatically increase the early application of photon sources in spectral regions where highly - indistinguishable photons have not yet been demonstrated @xcite .
furthermore , schemes based on low - visibility quantum interference could potentially benefit from reduced loss and increased brightness by removing spectral filters .
interference between two indistinguishable photons , known as hong - ou - mandel ( hom ) interference , is now the standard experiment used to quantify the indistinguishability of photons .
this is important since hom interference is the origin of supra - classical performance central to many proposed quantum - optical technologies and is integral to quantum metrology schemes , linear - optical quantum computers , quantum communication and the first loophole - free bell - inequality violation @xcite .
the importance of hom interference has recently inspired many works on photon distinguishability , present in an input state , and its effect on non - universal quantum computers @xcite . in a metrological setting , studies have focused on the effect of photons becoming distinguishable within an interferometer @xcite which removes heisenberg limited scaling .
photonic devices have been reported with fidelities of @xmath4 which indicates that distinguishability already present before the state enters the interferometer can be the dominant effect @xcite . in ref .
@xcite a novel approach was presented which utilised two photons with a carefully - engineered additional degree of freedom , and measurement of this additional degree of freedom mitigated the effects of distinguishability on metrological schemes . in this letter
we investigate the effect of distinguishability in a situation wherein photons are measured with detectors that resolve only the path the photons are in , and not information about additional degrees of freedom which the photons have .
positive results which apply to this scenario would show that interferometers and detectors do not need to be modified in order to operate using partially - distinguishabile photons .
indeed , the phase sensitivity will be shown here to degrade gently with increasing distinguishability , and precision scaling with increasing photon number remains proportional to the heisenberg limit .
we experimentally observed this degradation by controlling distinguishability between pairs of photons and pairs of bi - photons via a temporal delay .
these photons were found to exhibit supra - classical phase sensitivity despite their distinguishability .
we use the fisher information to quantify the sensitivity of interference fringes to small changes of the unknown phase , @xmath1 , due to its relation with achievable precision in the seminal cramr - rao bound : @xmath5 , where @xmath6 is the variance of an unbiased estimator of @xmath1 and @xmath7 is the fisher information , a function of the probabilities associated with different measurement outcomes @xcite .
this bound can be saturated asymptotically by a large number of measurements , and therefore @xmath7 well characterises the achievable precision of a scheme .
we consider photon - number - counting measurements which are described by the set of projectors @xmath8 such that @xmath9 projects onto the subspace containing all states with @xmath10 photons in one path and @xmath11 in the other . , enters an mach - zehnder interferometer before being measured with photon number - counting measurements .
the overlap of the temporal modes governs how much quantum interference occurs at the beamsplitters . ] to explore the role of photon distinguishability in metrological schemes , we consider distinguishability introduced by mode - mismatch .
an example of a scheme using a probe state with partially - distinguishable photons is displayed in fig .
[ main_scheme ] . here
the probe state is a @xmath12-photon dual - fock state @xmath13 where @xmath14 and @xmath15 are boson creation operators for modes @xmath16 and @xmath17 .
if the temporal modes @xmath18 and @xmath19 are mismatched @xmath20 , then imperfect quantum interference occurs .
output statistics for number - counting measurements can be calculated by changing the basis of the probe state such that it is expressed in terms of modes with well - defined interference relations @xcite .
re - expressing the temporal mode @xmath19 as a linear combination of @xmath18 and an orthogonal temporal mode , @xmath21 , the input state becomes : @xmath22 where @xmath23 is the overlap of the functions @xmath18 and @xmath19 and serves as a measure of indistinguishability @xcite . for @xmath24 the photons
will be uncorrelated as quantum interference will not occur .
conversely for @xmath25 the photons will undergo maximal quantum interference .
@xmath26 is a simple parameterisation of distinguishability which transitions between quantum and classical measurement statistics .
an interferometer acting on both temporal modes performs the operation @xmath27 with @xmath28 generating an orthogonal transformation . after transforming the two - photon input state ,
the probabilities for different detection outcomes are : @xmath29 this allows us to calculate the phase estimation capabilities of this state by computing the fisher information : @xmath30 for comparison , the snl for this setup is @xmath31 , therefore if the photons have any indistinguishability ( @xmath32 ) then the shot - noise limit can be surpassed around @xmath33 .
additionally , sensitivity scales quadratically with increasing photon number i.e. proportional to the fundamental heisenberg limit .
if @xmath1 is not near zero then an adaptively controlled phase can be used to counteract @xmath1 producing the sensitivity obtained near @xmath33 @xcite .
this point on the interference fringe has been highlighted as a loss - resistant part , retaining the most phase sensitivity in the presence of balanced loss across the arms of the interferometer @xcite . in ref .
@xcite , an alternative approach has been presented to mitigate the effects of distinguishability by using spatially - engineered photons and detectors which resolve the positions of each photon .
this approach requires increasing distinguishability in order to mitigate its adverse effects . to verify the usefulness of states with partially - distinguishable photons
, we performed an experiment using two- and four - photon states post - selected from a type - i spontaneous parametric down - conversion ( spdc ) process .
states were post - selected by recording only detection events with the desired number of photons .
this allowed us to isolate the effects of distinguishability since the effect of optical loss is also nullified by this procedure .
the experimental setup is shown in fig .
[ setup ] .
differently from the example above , the two - photon state that type - i spdc produces is spectrally entangled @xmath34 .
therefore we use a generalised measure of indistinguishability , or exchange symmetry , @xmath35 which reduces to @xmath36 for unentangled photons
i.e. @xmath37 @xcite .
@xmath26 can be measured experimentally by observing the coincidence probability after a 50:50 beamsplitter , @xmath38 , as in the hong - ou - mandel experiment @xcite : @xmath39 and takes the place of @xmath40 in eq .
( [ fisher ] ) when @xmath41 . for a two - photon state created by two single - photon sources separability
is guaranteed ensuring @xmath26 is positive and supra - classical precision will be achieved .
we note that regardless of the spectral structure of such a two - photon state , be it pure or mixed , a single exchange - symmetry parameter , as given by @xmath42 , will completely determine the output statistics after a linear network @xcite .
therefore the two - photon state from type - i spdc will give results indicative of all other photon sources with the same @xmath42 .
the fisher information of the four - photon data is not determined by @xmath43 ; however , for some values of @xmath44 , it is determined by the four - photon equivalent of a hom dip .
the post - selected four - photon state entering an interferometer may be expressed as @xmath45\,\hat{a}_{2}^\dagger[g_i]\right)^2|\mathbf{0}\rangle$ ] with @xmath46 and @xmath47 , and with two sets of orthonormal spectral functions @xmath48 and @xmath49 @xcite .
when @xmath50 , due to the symmetry of type - i spdc under exchange of paths , the two sets of functions are the same , with @xmath51 @xcite .
whereas , when @xmath44 is large @xmath52 .
when @xmath50 , the value of @xmath53 $ ] dictates how often there is genuine four - photon interference and how often there are simply two pairs of mutually indistinguishable photons . by determining @xmath54
the fisher information obtained when @xmath55 ( @xmath56 ) and when @xmath44 is large @xmath57 can be predicted as described in appendix c. we use a half - wave plate ( hwp ) as a polarisation interferometer as displayed in fig .
[ setup]a .
the indistinguishability parameter @xmath58 is controlled by delaying one polarisation mode by a distance @xmath44 with a translation stage prior to the interferometer .
our experiment uses an 80fs pump pulse , and momentum conservation dictates that the down - conversion process results in a theoretical indistinguishability of @xcite : @xmath59 , where @xmath60 is the gamma function @xcite and @xmath61 is a constant dependent on the properties of the crystal and pump laser used for down - conversion .
we obtained @xmath58 experimentally by fitting @xmath62 , to normalised coincidence rates shown in fig .
[ setup]a , by allowing @xmath63 and @xmath61 to vary . at various different values of @xmath44
the phase of the interferometer was scanned over the range @xmath64 obtaining interference fringes with different levels of quantum interference shown in fig .
[ setup]c(1 - 4 ) .
the fisher information was obtained from fitted curves of @xmath65 to normalised interference fringes over @xmath1 .
see appendix a for the fitting procedures . .
values achieved with two ( four ) photons are shown in blue ( orange ) .
non - linear relation between @xmath66 and @xmath67 is due to background noise and is modeled in appendix b. green points are estimates of @xmath68 we expect for the four - photon state when @xmath69 and when @xmath70 as described in appendix c. error bars arise from the statistical uncertainty in the counts and are found by monte - carlo simulation of experimental data followed by the same analysis as the real data which provides a spread of estimates of the fisher information for each fringe . ] the results for fisher information are shown in fig .
[ final_plot ] for both the two- and four - photon input states .
the two - photon data demonstrates an approximately - linear degradation of the quantum - enhanced sensitivity in line with the theoretical prediction above .
the slight deviation from the linear degradation is due to background noise , which we do not correct for , as modeled in appendix b. the effect of increased background noise is to shift the point of highest sensitivity further away from @xmath33 in addition to lowering the sensitivity of the fringes .
for the four - photon input state , since @xmath26 by itself does not infer a value for the fisher information , we use an experimentally - determined estimate of @xmath71 to estimate the fisher information we expect around @xmath55 , and for a large @xmath44 , as is shown by green points in fig .
[ final_plot ] ( appendix c describes how we estimate @xmath54 which is similar to the method of ref .
@xcite ) .
fig .
[ final_plot ] shows the sensitivity of the four - photon fringes decays from the predicted value when @xmath69 approximately down to the predicted value for @xmath70 as expected . as we can not measure the function @xmath72 we can only predict the value of @xmath7 in the extremal cases of small and large @xmath44 .
we note that our four - photon input states give a higher precision that previously reported experiments using similar states @xcite . in order for
supra - classical values of fisher information to be extracted contrast must remain high around @xmath73 .
this is apparent in the fringes we obtained experimentally as shown in fig .
[ setup]c .
we have shown that the effects of photon distinguishabiltiy have a distinct nature from errors considered previously , as heisenberg - limited scaling remains with partially - distinguishable photons .
surprisingly , any amount of quantum interference can grant a quantum advantage in the absence of any other imperfections .
the experiment we performed allowed us to see the gentle degradation of precision in agreement with our theoretical prediction .
we conclude that sources of highly - indistinguishable photons are not necessary to gain a quantum advantage in metrology .
our results reduce technical difficulties associated with making photon sources useful for metrology . using novel photon sources such as integrated sources @xcite , lattice defect sources @xcite , atomic sources @xcite , heralded sources @xcite using fast switching , or quantum memories @xcite and quantum dots @xcite may enable quantum enhanced measurements to be performed in new spectral regions .
* acknowledgments : * the authors would like to thank j. d. a. meinecke , t. ono and d. bonneau for helpful discussions .
this work was supported by epsrc , erc , picque , bboi , us army research office ( aro ) grant no .
w911nf-14 - 1 - 0133 , u.s .
air force office of scientific research ( afosr ) and the centre for nanoscience and quantum information ( nsqi ) .
j.l.ob . acknowledges a royal society wolfson merit award and a royal academy of engineering chair in emerging technologies .
j.c.f.m . and
j.l.ob acknowledge fellowship support from the engineering and physical sciences research council ( epsrc , uk ) .
j.s.c . acknowledges support from the university of bristol .
47ifxundefined [ 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 * * , ( ) link:\doibase 10.1103/physreva.91.063842 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevx.5.041015 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , vol .
( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( )
|
since the large hadron collider ( lhc ) started its run , many experimental results have come out that have provided an opportunity to carry out further investigation of the phenomenology of qcd - based effective theories .
nonrelativistic qcd ( nrqcd ) is one of the most successful effective theories describing quarkonium production and decays @xcite . under the nrqcd framework ,
the cross section is factorized into the summation of the products of the short - distance coefficient ( sdc ) , which is independent of the quarkonium state and can be calculated perturbatively , and the long - distance matrix element ( ldme ) , which only depends on the quarkonium state and requires the fit of experimental data to extract its value .
the cross section for the process of heavy quarkonium @xmath2 production or decay can be expressed as d(h)=_ndf_no^h(n)[eqn : nrqcd ] where @xmath3 is the sdc for the @xmath4 state @xmath5 , and @xmath6 is the ldme of state @xmath5 for quarkonium @xmath2 .
nrqcd succeeded in many processes where the color - singlet ( cs ) model @xcite failed ; however , it still faces many challenges .
@xmath7 polarization at hadron colliders is among the most puzzling questions that nrqcd encounters .
references @xcite investigated the polarization of directly produced @xmath7 , while ref . @xcite provided polarization results for prompt @xmath7 hadroproduction , which is the first next - to - leading - order ( nlo ) result comparable with experiment .
the three letters employed three sets of ldmes , which were obtained from different fit strategies .
all of them can describe @xmath7 production , yet none of them can explain all the polarization measurements .
in addition , the universality of the ldmes is another challenge .
reference @xcite reconciled experimental data of @xmath7 production at the tevatron and hera;however , their ldmes resulted in unphysical cross sections when employed to @xmath7 associated with a photon production at hadron colliders @xcite .
another interesting example is the transverse momentum ( @xmath8 ) integrated cross sections for the @xmath7 hadroproduction .
the theoretical results at qcd nlo obtained with collinear factorization @xcite overshoot the experimental data .
having resummed @xmath9 ( where @xmath10 denotes the bjorken-@xmath10 ) and considered the all - twist contributions in the dense side , ma and venugopalan @xcite remedied the discrepancy .
however , they @xcite did not include the @xmath0 and @xmath11 feeddown contributions .
whether the inclusion of these parts will ruin the conclusions requires further investigation . actually , ref .
@xcite has already studied the @xmath0 production processes in which the @xmath8 of the @xmath0 completely comes from the initial states . exploiting the unintegrated gluon distribution , the authors announced that the color - octet ( co ) ldme for @xmath0 is almost an order of magnitude smaller than the one obtained through the collinear factorization calculations .
reference @xcite , however , found that the final - state gluon emission processes actually dominate the @xmath0 hadroproduction in the midrapidity region .
for the above reasons , testing nrqcd is still an important work .
p - wave quarkonia productions provide an excellent laboratory to test nrqcd .
above all , at lo in @xmath12 ( the typical relative velocity of quark and antiquark in quarkonium ) , only one ldme is to be obtained from experiment ; one does not suffer from the entanglement of too many free parameters in the fit of experimental data , in contrast to the @xmath7 case @xcite .
further , at nlo , the @xmath8 distribution of both @xmath13}$ ] and @xmath14}$ ] channels behaves as @xmath15 in the large @xmath8 region .
higher order corrections can not exceed this behavior ; thus , one can expect nlo predictions to give a good precision in this phase space region .
in contrast to the @xmath16}$ ] case , the significance of nnlo correction is still in the mist .
finally , feeddown from higher excited states to p - wave quarkonia , say , @xmath0 , @xmath17 , @xmath18 , or @xmath19 , can almost be neglected [ e.g.,@xmath20 at lhcb @xcite ] .
notice the advantages stated above : we say the case of @xmath0 is clean " ; it is much easier to make definite conclusions in this case than in the @xmath7 case .
on the other hand , the study of @xmath0 production is not only important itself for phenomenological concerns , but also provides an opportunity for precise study of @xmath7 phenomenology ( e.g.,ref .
@xcite indicates that @xmath0 feeddown contribution is essential to the study of @xmath7 polarization ) .
many theoretical works on @xmath0 production have been published .
the authors of refs .
@xcite obtained ldme for @xmath0 production at lo and employed them to the prediction of prompt @xmath7 hadroproduction and/or polarization .
et al_. @xcite presented the first nlo study of @xmath0 hadroproduction and gave a favorable choice of the ldme for @xmath0 production .
et al_. @xcite calculated @xmath0 production associated with a @xmath21 pair at hadron colliders .
et al_. @xcite provided a detailed study on the polarization of hadroproduced @xmath0 and @xmath0-generated @xmath7 ; at the same time , they compared the theoretical prediction with some of the recent experimental data @xcite .
et al_. @xcite calculated @xmath0 hadroproduction at lo in @xmath22 and extracted both the cs and co ldmes from the fit of the experimental data , where their cs ldme is several times larger than the value obtained by the potential model and higher order terms in @xmath23 contribute significantly .
this paper is devoted to the theoretical predictions of @xmath0 hadroproduction , for one thing , as an alternative test of nrqcd .
recently , a number of experiment results have come out from the lhc collaborations , among which are many measurements on the @xmath0 yield and the ratio @xmath1 .
this paper will answer the question whether a single ldme can explain all the experimental data .
for another thing , the popular values of the ldmes for @xmath0 production , both at lo and nlo , were all given before these experimental results were published ; they are out of date . this paper will provide a detailed analysis on the determination of the ldmes and reasonable values of it at both lo and nlo . finally , as suggested in ref .
@xcite , we also observe the nrqcd scale dependence of the cross sections to determine whether nlo prediction stands up .
this paper is organized as follows .
section ii gives a brief introduction to the nrqcd framework for @xmath0 hadroproduction calculation . in sec .
iii , we present the parameter choices in our numerical computation and an analysis to see whether the nrqcd scale dependence is severe . in sec .
iv , we present the results of the fit and the values of the ldmes at both lo and nlo and investigate the universality of the ldmes in detail .
section vi is a concluding remark .
this section provides quite a brief review of the nrqcd formulas for the calculation of @xmath0 production .
we do not discuss in detail how the equations are derived .
interested readers can refer to some relative references , e.g. @xcite . for @xmath0 production at lo in @xmath12 , eq.([eqn :
nrqcd ] ) can be written as d(_cj)=df_^3p_j^[1]o^_cj(^3p_j^[1])+df_^3s_1^[8](2j+1)o^_c0(^3s_1^[8])[eqn : s ] .
the value of cs ldme can be evaluated through @xcite o^_cj(^3p_j^[1])=(2j+1)|r_p(0)|^2 , where @xmath24 is the derivative of the wave function of the related quarkonium with respect to the radius at the origin .
the calculation of @xmath25}}$ ] has been described in detail in many of our previous papers;see , e.g .
, @xcite . to evaluate @xmath26}}$ ] , we notice that it is independent of the long - distance asymptotic states and replace @xmath27 in eq.([eqn : s ] ) with a @xmath21 state @xmath13}$ ] , so that we obtain d(^3p_j^[1])=df_^3p_j^[1]o^^3p_j^[1](^3p_j^[1])+df_^3s_1^[8]o^^3p_j^[1](^3s_1^[8])[eqn : fac ] , where we have used the relation @xmath28}}(^3s_1^{[8]})\rangle=(2j+1)\langle o^{^3p_0^{[1]}}(^3s_1^{[8]})\rangle$ ] .
we should keep in mind that eq.([eqn : fac ] ) is to extract cs sdc , which is expanded in @xmath22 . as a result , the quantities @xmath28}}(^3p_j^{[1]})\rangle$ ] and @xmath28}}(^{3}s_{1}^{[8]})\rangle$ ] should also be evaluated perturbatively , and the value of @xmath22 in them should be in accordance with that in the sdcs .
the left- and right - hand side of eq.([eqn : fac ] ) should keep those terms up to the same order as in the perturbative expansion .
the evaluation of @xmath29}}$ ] follows the ordinary procedure : writing the squared amplitudes through reading feynman diagrams and multiplying it by the flux density and the phase space unit
. both @xmath29}}$ ] and @xmath28}}(^3s_1^{[8]})\rangle$ ] are infrared ( ir ) divergent , and the ir divergences from the two quantities cancel each other .
we adopt the two - cutoff phase space slicing method @xcite and find that the cross section excluding the terms ( denoted as @xmath30 ) corresponding to the squared real - correction diagrams , in which a gluon connects the quarkonium ( as displayed in fig .
[ fig : gc ] ) , integrated over the gluon soft region is free of divergence .
we denote this finite part of the cross section as @xmath31 . neglecting the finite terms proportional to the size of the small region
, @xmath30 can be expressed as d^s(^3p_j^[1])=-u_^s d f_^3s_1^[8]o^^3p_j^[1](^3p_j^[1 ] ) , [ eqn : scs ] where @xmath32 is 3 for su(3 ) gauge field and u_^s=+ln()+ln()-_e- [ eqn : us ] , with @xmath33 and @xmath34 being the energy and absolute value of momentum of @xmath0 , respectively , @xmath35 the euler s constant , and @xmath36 the scale to complement the dimension .
@xmath37 is an arbitrary positive number small enough to provide the soft approximation with sufficient accuracy . up to the order maintained in our calculation , the transition rate of @xmath21 state @xmath14}$ ] into @xmath13}$ ]
can be calculated in the dimensional regularization scheme as o^^3p_j^[1](^3s_1^[8])^nlo =- u^c _ o^^3p_j^[1](^3p_j^[1])^lo , [ eqn : ldm ] where @xmath38 is defined as u^c_|__=-_e-+ln ( ) [ eqn : ucmu ] and u^c_|_=-_e++ln ( ) [ eqn : ucms ] in the @xmath39-cutoff ( in which @xmath39 is the upper bound of the integrated gluon energy ) and @xmath40 renormalization scheme , respectively .
@xmath39 is a scale rising from the renormalization of the ldme .
substituting eqs.([eqn : ldm ] ) and ( [ eqn : scs ] ) into eq.([eqn : fac ] ) , we can solve the sdc for @xmath13}$ ] as df_^3p_j^[1]^nlo = df^f_^3p_j^[1]-u_d f_^3s_1^[8]^lo , [ eqn : soft ] where u_=u_^s - u_^c , the expressions of which for the @xmath39-cutoff and @xmath40 renormalization scheme are u_|__=ln()+ln()-2 + 2ln(2 ) and u_|_=ln()+ln ( ) , respectively . both of the terms on the right - hand side of eq.([eqn : soft ] )
are finite .
now , all the short - distance coefficients are ir divergence free;then the components for calculating the cross section for @xmath0 hadroproduction are well defined . substituting eq.([eqn : soft ] ) into eq.([eqn : s ] )
, we obtain the complete expression of the cross section for @xmath0 production : d^nlo(_cj)=d^f(_cj ) -u_o^_cj(^3p_j^[1])df_^3s_1^[8]^lo + ( 2j+1)o^_c0(^3s_1^[8])df_^3s_1^[8]^nlo .
[ eqn : com ]
to calculate @xmath41})$ ] and @xmath42})$ ] , we apply our feynman diagram calculation package ( fdc ) @xcite to generate the entire needed fortran source .
before we present the numerical results , we should comment on the obtaining the co ldme . focusing on the last two terms in the right - hand side of eq.([eqn : com ] ) , one can notice that , if @xmath39 varies its value , in order to fit the cross section @xmath43 to the experimental data , the ldme in the last term should change accordingly , which is to say , the dependence on @xmath39 is partly absorbed into the co ldme . if we proceed with our calculation to infinite order in @xmath22 , the @xmath39 dependence can be totally absorbed into the co ldme .
consequently , this scale actually can be any positive value holding the convergence of @xmath22 expansion .
if our results significantly depend on @xmath39 , the dropped terms in higher orders must contribute significantly , and the calculation up to this order does not reach a sufficient accuracy . up to nlo , the condition of @xmath39 independence requires d f^lo_^3s_1^[8]df^nlo_^3s_1^[8 ] , [ eqn : cond ] as well as that the proportional ratio should be universal for all the processes .
we define a quantity r=/ ( ) , [ eqn : defr ] to determine whether the @xmath39 dependence is severe .
if @xmath44 is constrained in a small range throughout the whole @xmath8 region for all the processes , we know for sure the dependence on @xmath39 can be absorbed into the ldme and vice versa . in this paper
, we provide the values of the co ldme for different renormalization schemes and @xmath39 choices . in the numerical calculation , we have the following common choices of parameters : @xmath45 @xcite for both lo and nlo calculation , and @xmath46 .
the soft cutoff @xmath37 independence is checked in the calculation and @xmath47 is used .
since the energy scale of most of the phase space region exceeds b - quark mass , @xmath48 is used .
we employ cteq6 m @xcite as the parton distribution function ( pdf ) and two - loop @xmath22 running for up - to - nlo calculation , and cteq6l1 @xcite and one - loop @xmath22 running for lo .
the renormalization and factorization scales are chosen as @xmath49 . in our fit , we exclude the @xmath50 data points .
this is because , for one thing , relativistic correction contributes a part proportional to the qcd lo sdc , when @xmath8 is larger than about 7gev @xcite ; this part can be complemented by modifying the values of the ldmes , while below 7gev , this is not true . for another thing , the @xmath9 terms might ruin the perturbative expansion in this region @xcite . to extract the co ldme , we employ all the existing data on @xmath0 hadroproduction except for those in ref .
@xcite , which measured the fraction of the @xmath7 hadroproduction cross section through the @xmath0 feeddown to the prompt one .
reference @xcite provided the prompt @xmath7 hadroproduction cross section with the same center - of - mass energy and rapidity range .
however , the @xmath8 s of the two sets of data do not coincide.thus , we can not extract the exact central values and error bars of the @xmath0 hadroproduction cross sections and consequently we can not use these data directly . for this reason , we give up using them in our fit .
there are six sets of data involved in our analysis .
all of them are listed in table [ table : data ] . except for the data mentioned above
, we also noticed another set of data ( denoted as e3a ) , which was published in ref .
since the data in e3 are the updated version of those in e3a , we do not use e3a for fit;however , we plot them in the figures for reference . .
all the sets of data used in our fit . [ cols="^,^,^,^,^,^,^",options="header " , ] in refs .
@xcite ( corresponding to e1@xmath51e4 ) , the values of @xmath8 are given for the @xmath7 generated from @xmath0 feeddown , in accordance with which we should do the so - called @xmath8 shift as @xmath52 .
here we choose @xcite @xmath53 , @xmath54 , @xmath55,and @xmath56 , which are different from ref .
@xcite , where @xmath57 and @xmath58 are @xmath59 and @xmath60 , respectively .
the branching ratios @xcite are @xmath61 , @xmath62 , and @xmath63 for @xmath64 to @xmath7 , respectively .
before we carry out the fit , we shall first investigate whether @xmath44 defined in eq.([eqn : defr ] ) is universally a constant to hold the @xmath39 independence .
e2 and e3 are in the same experimental condition , as are e5 and e6 ; accordingly , there are four conditions to present . for these experimental conditions ,
the values of @xmath44 are presented in fig.[fig : ratio ] .
we can see that , except for e2 ( as well as e3 ) , for all three conditions , @xmath44 is almost a constant , ranging from about 104 to 118 , as @xmath8 varies from @xmath65 to @xmath66 .
as we expected in the introductory section that nlo results should provide a sufficiently precise prediction , the @xmath39 dependence can not be severe . for e2 ,
the situation is a little worse ( @xmath44 ranges from 132 to 105 ) , yet , not so bad to ruin the results .
we can now expect that , for e1 and e4@xmath51e6 , the theoretical prediction should be in good agreement with the experiment , and if we carry out the fit by employing the four sets of data individually , we should obtain almost the same results of the values of the ldme . for e2 and e3
, the theoretical prediction might agree with the experiment qualitatively .
@xmath44 is not always a constant up to nlo precision . as an example , for @xmath17 hadroproduction , @xmath44 varies significantly for different experimental conditions and phase space regions .
interested readers can refer to ref .
@xcite , in which the detailed results for @xmath17 hadroproduction are presented .
we present the values of the co ldme extracted from the fit of each set of the experimental data at both lo and nlo , and see if they correspond with one another . in the rest of this paper , we abbreviate @xmath67})\rangle$ ] as @xmath68 .
for lo calculation , & & o_e1^lo=0.240.13 , o_e2^lo=1.260.03 , + & & o_e3^lo=0.190.06 , o_e4^lo=0.130.05 , [ eqn : ldmelo ] + & & o_e5^lo=1.220.07 , o_e6^lo=0.670.07 , and the @xmath69 are 6.4 , 0.0078 , 1.1 , 0.69 , 0.11 , and 0.43 , respectively .
for nlo calculation , as @xmath70 in @xmath39-cutoff renormalization scheme , & & o_e1^nlo=1.970.17 , o_e2^nlo=2.340.06 , + & & o_e3^nlo=2.280.06 , o_e4^nlo=2.000.07 , [ eqn : ldmenlo ] + & & o_e5^nlo=2.030.05 , o_e6^nlo=2.040.06 , and the @xmath69 are 2.8 , 0.034 , 0.18 , 0.17 , 0.068 , and 0.35 , respectively . to begin with , we can see that , for all the conditions except for e2 , the @xmath71 for nlo is smaller than that for lo .
moreover , for e1 and e4@xmath51e6 , the obtained values of the co ldme for nlo are almost the same ( @xmath72 ranges from 1.97 to 2.04 ) . as we analyzed in the previous section
, we do not expect theoretical prediction for e2 and e3 to agree with the experiment in high precision ; however , even for e2 and e3 , the obtained values of the co ldme are very close to those for e1 and e4@xmath51e6 . by contrast
, for the values of @xmath72 obtained for the lo range from 0.13 to 1.26 , the largest is about ten times the smallest , and there is no common value for any group of the sets of data ; the distribution of the values is dispersive .
we can conclude that , no universal value exists for lo ldme , since up to lo , the precision is not sufficient to describe all the experiments .
we can also see that the ldme given in ref .
@xcite is too large .
one might make wrong conclusions if using that value .
the ldme given in ref .
@xcite is much larger than the upper bound of the series of the values presented above .
the large value of the ldme might lead to overestimation of the absorption effect ( as well as other nuclear matter effects ) .
now we carry out a global fit , using all the experimental data in e1@xmath51e6 for lo , and e1 and e4@xmath51e6 for nlo , and obtain ^lo=0.310.09 , o__^nlo=2.010.04 .
[ eqn : ldmedefault ] the @xmath69 are 2.4 and 0.47 for lo and nlo , respectively .
the consideration is that we trust the precision of the nlo results for e1 and e4@xmath51e6 .
however , for e2 and e3 , the situation is not clear .
thus , we fit e1 and e4@xmath51e6 to obtain nlo co ldme as a default value to present our results , and we employ the ldme to see whether it can explain the experiment e2 and e3 . theoretical predictions for the @xmath0 production cross section and the ratio @xmath1 are listed in figs.[fig : pt ] and [ fig : r ] , respectively .
references @xcite only provide results for @xmath7 @xmath8 , while ref .
@xcite provides results for both @xmath7 and @xmath0 @xmath8 .
since our calculations are carried out at @xmath0 @xmath8 , for this reason , the distributions for e5 and e6 illustrated in fig.[fig : pt ] are with respect to @xmath0 @xmath8 . since the uncertainty of the ldme at nlo is small
, we do not draw the band rising from this uncertainty . however , the uncertainty of the ldme for lo is quite large , but we do not bother with this matter here
. we will provide a more reasonable band for lo uncertainty later .
we can see from figs.[fig : pt ] and [ fig : r ] that nlo results are in very good agreement with all the experiments , while lo results can not agree with most of the experimental data . as we mentioned above , for e2 and e3
, the nlo calculation might not be able to provide sufficiently precise results , since as displayed in fig.[fig : ratio ] that @xmath39 dependence is severe for this experimental condition .
this fact might arise from the large rapidity ( denoted as @xmath73 ) , since large @xmath73 eventually introduces two scales @xmath74 ( the energy of @xmath0 ) and @xmath75 . when @xmath76 , @xmath77 , which might ruin the perturbative expansion .
one might resume these terms to achieve well - converged results .
since lo results obtained from the default choice of the ldme can not provide good predictions , we shall give a range of the ldme to cover all the experimental data .
here we choose the range between the upper and lower bound of the values given in eq.([eqn : ldmelo ] ) : @xmath78 .
we can see in fig.[fig : lo ] that the large band can cover most of the experimental data , and the upper bound overestimates the significance of @xmath0 feeddown contributions for some of the conditions . still , we are not sure whether they are able to explain new experiments ; however , a band presented for the range given above might cover the experimental data in the sense of statistics . and we know for sure that , a single value of the co ldme can not give reasonable predictions at lo . at the end of this section , we present the values of the co ldme at nlo for different choices of @xmath39 . for the same reason , we exclude e2 and e3 data .
the ldmes are listed as follows : _
m_c^nlo=2.250.04 , o_m_c/2^nlo=1.680.04 , o__qcd^nlo=0.700.04 .
[ eqn : ldmescale ] here we have used @xmath40 renormalization scheme ( in the calculation of nlo correction to the co ldme ) .
the @xmath69 are 0.48 , 0.46 , and 0.42 , respectively .
fitting experimental data at different @xmath39 s is actually an alternative procedure of solving the ldme running equation , which can be obtained from the renormalization of the ldme as _
o^_cj(^3p_j^[1])[eqn : running ] as fig.[fig : ratio ] shows , @xmath44 is almost a constant for the experimental conditions in the fit , so we can expect that the ldmes listed above would be consistent with eq.([eqn : running ] ) . from fig.[fig : ratio ] and eq.([eqn : defr ] ) , the typical value of @xmath22 is about 0.09 ( at large @xmath8 ) . employing this value , we find that the ldmes in eq.([eqn : ldmescale ] ) satisfy eq.([eqn : running ] ) .
we also present here the ldme obtained with the inclusion of e2 and e3 , and see whether the results change much : _ _ ^nlo=2.090.04 , o_m_c^nlo=2.350.04 , + o_m_c/2^nlo=1.770.04 , o__qcd^nlo=0.770.04 , [ eqn : ldmescalecom ] where the subscript @xmath39 denotes the @xmath39-cutoff renormalization scheme as well as @xmath70 , and the subscripts @xmath79 , @xmath80 and @xmath81 refer to the @xmath40 renormalization scheme , with @xmath39 being the corresponding values .
the @xmath69 are 0.51 , 0.52 , 0.50 , and 0.46 , respectively .
the difference between the ldmes fitted by including and excluding e2 and e3 ranges from @xmath82 to @xmath83 .
the difference increases as @xmath39 gets smaller , which is caused by the different behavior of @xmath44 for the four experimental conditions . including e2 and e3 enhances the @xmath69 slightly , which is to say that theoretical prediction can fit e2 and e3 equally as well as it fits e1 and e4@xmath51e6 .
figure[fig : nlo ] presents the comparison of theoretical predictions for the eight ldmes to the experimental data . actually , all the eight ldmes result in good agreement with the experiment .
for e1 and e4@xmath51e6 , the bands hold small as @xmath8 varies , while for e2 and e3 , the bands get very large in high @xmath8 regions , which is to say for e1 , e4@xmath51e6 , and the small @xmath8 region in e2 and e3 , the @xmath39 dependence can be absorbed into the ldmes , while in the large @xmath8 regions in e2 and e3 , the problem of @xmath39 dependence becomes severe .
the values of @xmath67})\rangle$ ] at lo obtained in this paper , which range from 0.00013gev@xmath84 to 0.00126gev@xmath84 , are smaller than those in ref .
@xcite ( ( [email protected])gev@xmath84 ) , ref .
@xcite ( ( [email protected])gev@xmath84 ) , and ref .
@xcite ( ( [email protected])gev@xmath84 ) , which employed the data obtained through extrapolation carried out in ref .
the nlo ldmes are slightly different from those obtained in refs .
@xcite , which is due to the different parameter choices between our paper and those cited above . with the same parameter choice
, we can obtain exactly the same ldmes with refs .
in this paper , we calculated @xmath0 production cross sections and the ratio @xmath1 at hadron colliders , and compared the theoretical predictions with the experiment .
we presented a detailed analysis on the co ldmes and found that , at lo , there does not exist any universal value of the co ldme to explain all the experiments , while at nlo , the co ldme obtained from a global fit is able to explain all the experimental data . at lo
, we obtained the value of @xmath72 ranging from 0.13 to 1.26 when fitting individual experiments e1@xmath51e6 .
the upper and lower bounds of @xmath72 result in quite a large band , which can cover most of the experimental data ; however , the upper bound overestimates the significance of @xmath0 feeddown contributions for some of the experimental conditions . as for the nlo case , we carried out a global fit by using eight schemes .
each of them agree well with the experimental data .
we also investigated the @xmath39 dependence of the results and found that , for e1 , e4@xmath51e6 , and the small @xmath8 region in e2 and e3 , the dependence on @xmath39 can be absorbed into the ldmes , while for the large @xmath8 region in e2 and e3 , the problem of @xmath39 dependence is relatively severe .
one needs to resum the large log terms rising from large rapidity to achieve better results .
our work provides a strong support of the nrqcd effective theory . , * * , ( ) .
, , , , , * * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , * * , ( ) .
, , , * * , ( ) .
( ) , * * , ( ) .
( ) , * * , ( ) .
( ) , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , * * , ( )
, * * , ( ) . , , , , * * , ( ) .
( ) , * * , ( ) .
( ) , * * , ( ) .
( ) , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , , , , , * * , ( ) . , , , * * , ( ) . , * * , ( ) . , * * , ( ) . , * * , ( ) . , , , , , . , ( ) . , , , , * * , ( ) . , , , . , ( ) .
( ) , * * , ( ) .
|
we carry out a global analysis of the experimental data on the @xmath0 production cross section and the ratio @xmath1 at the lhc and the tevatron . the related long - distance matrix elements ( ldmes ) at both leading order ( lo ) and next - to - leading order ( nlo ) in the qcd coupling constant are renewed .
we also present the transverse momentum distribution of the @xmath0 production cross section and the ratio @xmath1 for several experimental conditions and find that nlo predictions agree with all sets of experimental data .
by contrast , at lo , one can not explain all the data with a unique value of the color - octet ldme .
a brief analysis of the nonrelativistic qcd scale dependence of the cross sections shows that , for the conditions we are concerned with in this paper , the dependence can be almost totally absorbed into the ldme .
|
[ secintro ] of all the topological defects @xcite , which may have been formed during phase transitions in the early universe , cosmic strings @xcite are those which have attracted most attention from a cosmological point of view .
they were introduced into cosmology some 20 years ago by kibble @xcite , zeldovich @xcite and vilenkin @xcite , and were considered for a long time as possible sources for density perturbations and hence for structure formation in the universe .
indeed , the latest data from the boomerang and maxima experiments @xcite disagree with the predictions @xcite for the cosmic microwave background anisotropies based on topological defect models ( see also ref .
this seems to point to the conclusion that if cosmic strings were formed in the early universe they could not have been responsible for structure formation
. however , cosmic strings are still cosmologically relevant and enjoy wide interest in cosmology . the most common field - theoretical model , which is used in order to describe the generation of cosmic strings during a phase transition , is the abelian higgs model .
this model is defined by the action : @xmath2 where @xmath3 is the ricci scalar , @xmath4 the abelian field strength , @xmath5 is a complex scalar field with vacuum expectation value @xmath6 and @xmath7 is the usual gauge covariant derivative .
we use units in which @xmath8 and a `` mostly minus '' metric .
the simplest form of a cosmic string is just a direct generalization of the nielsen - olesen flux tube @xcite , which takes into account its gravitational effect .
because of the cylindrical symmetry , we use a line element of the form : @xmath9 and the usual nielsen - olesen ansatz for the + 1 flux unit : @xmath10 since the flux tube is very concentrated along the symmetry axis , it is quite natural to expect that it will generate a spacetime geometry with the asymptotic behavior of the general static and cylindrically - symmetric vacuum solution of einstein equations @xcite , the so - called kasner solution : @xmath11 where @xmath12 sets the length scale while @xmath13 represents the asymptotic structure , as will be discussed below .
the parameters @xmath14 must satisfy the kasner conditions : @xmath15 more information about the solutions can be obtained by inspection of the full system of field equations with the appropriate energy - momentum tensor , which we will not write here ( see however section 4 or refs.@xcite ) . here
we will only give a condensed summary .
for cylindrical symmetry , the components of the energy - momentum tensor of the flux tube @xmath16 have the property of @xmath17 .
this means that the solution will have a symmetry under boosts along the string axis , i.e. , @xmath18 .
the kasner conditions ( [ kasner2 ] ) then leave only two options , which are indeed realized as solutions of the full einstein - higgs system @xcite .
the standard conic cosmic string solution @xcite is characterized by an asymptotic behavior given by ( [ kasner1 ] ) with : @xmath19 which is evidently locally flat . in this case , the parameter @xmath13 represents a conic angular deficit @xcite , which is also related to the mass distribution of the source .
in addition to the cosmic string solutions , there exists a second possibility : @xmath20 which is the same behavior as that of the melvin solution @xcite .
eq.([mw ] ) is therefore referred to as the melvin branch .
note however that the magnetic field in this solution inherits the exponential decrease ( with @xmath21 ) of the original nielsen - olesen flux tube .
therefore , it is much more concentrated than in the original melvin solution where it decreases only according to a power law .
the main difference between the two branches lies in the tolman mass , which is zero for the conic cosmic string solutions , but non - zero for the melvin solutions @xcite .
moreover , the central magnetic field is generally larger for the melvin solutions .
the string - like solutions of the abelian higgs model are the simplest and most studied ones ( see e.g. @xcite-@xcite , @xcite-@xcite ) .
it is , however , probable that when high energy corrections are taken into account , gravity is not purely tensorial .
a minimal modification suggested by string theory @xcite is the introduction of a scalar degree of freedom , the dilaton @xmath22 , turning gravity into a scalar - tensor theory in the spirit of jordan - brans - dicke ( jbd ) theory .
studies of cosmic string solutions in the framework of jbd theory and its extensions already exist in the literature @xcite , and the typical characteristics of the string - like solutions are found to be quite different from those of the gauge strings of pure tensorial gravity .
the main difference is the absence of asymptotically conic solutions due to the long range effect of the massless jbd field . unlike the brans - dicke field , which is postulated not to couple to matter ,
the dilaton may do so and may further change the situation .
this is represented by the following action @xcite , which is written in the so called `` jbd / string frame @xcite '' : @xmath23 the parameter @xmath24 serves as a general coupling constant between the dilaton and matter fields .
newton s constant is added explicitly in order to keep track of the dimensionality of the various fields .
this makes the action in the `` einstein / pauli frame @xcite '' contain the standard einstein - hilbert lagrangian : @xmath25 it is related to the former by the metric redefinition : @xmath26 where @xmath27 is the metric tensor used in ( [ higgsdilaction1 ] ) and @xmath28 is used in ( [ higgsdilaction2 ] ) . from now on we will use the einstein frame for a clearer physical picture .
this system was studied by gregory and santos @xcite .
their analysis was done to first order in the parameter @xmath29 and is therefore valid only for weak gravitational fields .
they concentrated in the one branch which contains for @xmath30 asymptotically conic solutions .
however , since for @xmath31 no asymptotically conic solutions exist , there is no natural criterium to prefer one branch over the other , even if one is `` more flat '' than the other .
moreover , the weak - field approximation breaks down both near the core of the string and far away from the string thus calling for a further investigation of the system . in this paper
, we expand the study of cosmic strings in dilaton gravity in two directions .
+ 1)we analyze both branches and do not limit ourselves to weak gravitational fields .
+ 2)we modify the dilaton coupling to matter fields in such a way that it will be uniform and the original einstein - higgs system ( [ higgsaction ] ) is recovered in a certain limit .
[ secasympt ] the field equations for the einstein - higgs - dilaton action ( [ higgsdilaction2 ] ) are straightforward to obtain and we give them already for flux tubes with cylindrical symmetry .
their structure becomes more transparent if we use dimensionless quantities . as a length scale
we use @xmath32 ( the correlation length `` in the superconductivity terminology ) .
we therefore change to the dimensionless length coordinate @xmath33 and we use the metric component @xmath34 which we still denote by @xmath35 .
we also introduce the ' ' bogomolnyi parameter " @xmath36 in addition to @xmath29 , which has been already defined above . in terms of these new quantities
we get , for fixed @xmath24 , a two parameter system of five coupled non - linear ordinary differential equations ( the prime denotes @xmath37 ) : @xmath38 @xmath39 @xmath40 @xmath41 @xmath42 the sixth function , the metric component @xmath43 , turns out to be equal to @xmath44 due to the high symmetry .
we have also to keep in mind the existence of the constraint which comes from the ( @xmath45 ) einstein equation , and gets the following form : @xmath46 the field equations are supplemented by the following boundary conditions , that should be satisfied by the scalar and gauge fields : @xmath47 these are just the usual flux tube boundary conditions borrowed from the flat space version .
moreover , regularity of the geometry on the symmetry axis @xmath48 will be guaranteed by the `` initial conditions '' : @xmath49 the presence of the dilaton requires additional conditions , which we impose on the dilaton field on the axis : @xmath50 the condition @xmath51 follows directly from the equations of motion and the constraint ( [ constraint ] ) , using also eqs.([boundarycond])-([initcond ] ) . as for the condition @xmath52 , notice that the equations of motion ( 2.1)-(2.6 ) are invariant under the following transformation : @xmath53 where @xmath54 is a constant
thus , we can trivially get the solution for @xmath55 from the solution for @xmath52 by simple rescaling .
since we are looking for string - like solutions to the system , we may easily get the asymptotic behavior of the metric components from the assumption that the right - hand sides of the einstein equations vanish exponentially fast , as in the case of pure tensorial gravity .
consequently , the system reduces to a simple set of equations , which are easily integrated to power law solutions : @xmath56 the 3 parameters @xmath57 , @xmath58 and @xmath59 are subjected to two conditions , which may be viewed as a generalization of the kasner conditions , eq.([kasner2 ] ) : @xmath60 these two conditions leave a one - parameter family of solutions , which may be visualized by the ellipse in the @xmath61 ( or equivalently @xmath62 ) plane ( fig.[ellipsea0 ] ) : @xmath63 the 2 branches of the einstein - higgs system ( the cosmic string branch and the melvin branch ) are represented in this picture by the top ( @xmath64 ) and bottom ( @xmath65 ) points of the ellipse , respectively .
these points correspond to an asymptotically constant dilaton ( @xmath66 ) , and as we will show later , only the first is realized as an actual solution for the special value of the coupling constant @xmath30 .
the parameters @xmath57 , @xmath58 and @xmath59 are related to the matter distribution through the following three quantities : the tolman mass ( per unit length ) @xmath67 , or rather its dimensionless representative @xmath68 : @xmath69 and two others : @xmath70 which is related to the string angular deficit , and : @xmath71 which may be interpreted as a dilaton charge .
note that @xmath72 is manifestly positive definite and @xmath73 is positive definite for @xmath74 .
the tolman mass is also non - negative since it is proportional to the power @xmath57 which is non - negative due to ( [ dilatonkasner ] ) .
we further notice that , as in the non - dilatonic case @xcite , @xmath72 may be expressed in terms of the mass parameter @xmath75 and the magnetic field on the axis , which we represent by the dimensionless parameter @xmath76 : @xmath77 the parameters @xmath57 , @xmath58 and @xmath59 are easily found to be expressible in terms of @xmath75 , @xmath73 and @xmath72 or as : @xmath78 the first of the two relations ( [ dilatonkasner ] ) is satisfied identically ; the second translates in terms of @xmath75 , @xmath73 and @xmath79 to : @xmath80
[ secsols ] regarding the complexity and non - linearity of the system , it is quite evident that a detailed structure of the solutions can be obtained only numerically .
we have performed a numerical analysis of the system and found that the solutions are paired , as in the non - dilatonic case , and have the characteristics that will be described below .
the solutions have been found numerically by first discretizing the radial coordinate and then applying a relaxation procedure to the set of non - linear and coupled algebraic equations , @xmath81 , in the set of scalar functions evaluated at the grid points , @xmath82 .
if the initial guess @xmath83 is good enough then the iteration obtained by @xmath84 , where @xmath85 is found by solving the linear equation : @xmath86 ( gauss elimination with scaling and pivoting will do ) , will converge towards a solution for the discretized variables .
we constructed a starting guess satisfying the boundary and initial conditions and still having several free parameters .
changing gradually these parameters , we eventually stumbled into the region of convergence and a solution was found .
having obtained one solution of this type , the rest could easily be generated by moving around in the parameter space in sufficiently small steps .
representative plots of the solutions are given in figs.[soltna2a-1g0.4 ] , [ soltna2a - s3g0.4 ] and [ soltna2a0g0.4 ] .
the simplest picture emerges in the case @xmath30 .
the two branches of solutions for this paticular value of @xmath24 are particularly interesting . for fixed @xmath87 ,
both solutions exist up to some critical @xmath88 ; see fig.5 .
one is an ordinary cosmic string , as in einstein gravity with the additional feature of a non - constant dilaton .
this solution has @xmath89 , @xmath90 , @xmath66 independent of @xmath87 and @xmath88 .
the metric is very similar to the usual gauge string , but there are small deviations , due to the presence of the dilaton .
it becomes identical to the metric of the einstein gravity gauge string only in the bogomolnyi limit ( @xmath91 ) , where eqs.([eqf])-([constraint ] ) reduce to : @xmath92 with @xmath93 and @xmath94 , i.e. first order differential equations for @xmath35 , @xmath95 and @xmath96 .
a typical solution of this kind is depicted in fig.[soltna2a-1g0.4](a ) . the second branch in the case @xmath30 has for all values of @xmath87 and @xmath88 below the critical curve ( fig.5 ) , @xmath97 and @xmath98 .
accordingly , @xmath99 . this solution is quite different from its non - dilatonic analog , the main difference being the asymptotic spatial geometry which is now cylindrical . at the critical curve ( fig.5 ) ,
the two types of solutions become identical .
asymptotically the corresponding spatial geometry is conic with deficit angle @xmath100 .
the critical curve itself can be approximated by a power - law : @xmath101 where @xmath102 and @xmath103 .
this is quite , although not exactly , similar to the result obtained in tensorial gravity @xcite .
the @xmath87 and @xmath88 - independence of @xmath57 , @xmath58 and @xmath59 for @xmath30 is however the only case of such behavior .
the generic one is such that for a given value of @xmath24 , the powers of the asymptotic form of the metric components and the dilaton change with @xmath87 and @xmath88 .
put differently , we have now a situation where the values of @xmath57 , @xmath58 and @xmath59 depend upon the three parameters @xmath24 , @xmath87 and @xmath88 . therefore , the universal asymptotic form of the metric tensor , which corresponds to the two extremal points on the ellipse , is now replaced by a motion along the ellipse whose specific details depend on the value of @xmath24 .
additional values of @xmath24 were studied in detail and here we give few further results for two more special representative points , so in all there are : @xmath104 these values were selected for the following reasons : concentrating on the einstein - dilaton - maxwell system , @xmath30 corresponds to certain compactified superstring theories @xcite , @xmath105 corresponds to jordan - brans - dicke theory and @xmath106 corresponds to ordinary kaluza - klein theory . in both cases
@xmath107 , @xmath105 the same two - branch picture emerges .
one branch always contains the upper point of the ellipse ( @xmath64 ) at the limit @xmath108 , such that for small @xmath88 the `` almost asymptotically conic '' solutions of gregory and santos @xcite are obtained . for large @xmath88 the deviations from asymptotically conic geometry are more pronounced . at the same time
a gravitational newtonian potential appears , so this kind of string exerts force on non - relativistic test particles far outside its core .
the other branch exhibits also an @xmath87 and @xmath88 dependence of the powers @xmath57 , @xmath58 and @xmath59 , but has its @xmath108 limit at @xmath24-dependent points .
this @xmath24-dependence can be understood if we note that the @xmath108 limit corresponds to two types of asymptotic behavior .
one is the asymptotically conic cosmic string behavior and the other is that of the dilatonic melvin universe @xcite , which in our coordinate system corresponds to an asymptotic behavior with the following powers : @xmath109 the situation therefore is the following .
first we fix @xmath24 .
one branch ( the `` upper '' or `` cosmic string '' branch ) has the point @xmath64 , i.e. , @xmath110 , @xmath90 as a starting point for @xmath108 , and as @xmath87 and @xmath88 change , the powers @xmath58 and @xmath59 move along the ellipse while @xmath57 changes accordingly ( see [ dilatonkasner ] ) . the other branch ( the `` lower '' or `` dilatonic melvin '' branch ) starts , for @xmath108 , with the above values eq.([dilatonmelvin ] ) for the powers @xmath57 , @xmath58 and @xmath59 and the motion along the ellipse is done in the opposite direction so the two branches approach each other and `` meet '' at a point in between . in the meeting point , the solutions not only have the same asymptotic behavior , but they become actually identical .
there are no asymptotically conic solutions except for @xmath30 .
actually , this property can be understood analytically . for this
, we note that a useful relation is obtained by adding the equation for the dilaton , ( [ eqphi ] ) to the equation for the metric component @xmath111 , ( [ eqn ] ) and integrating from @xmath112 to @xmath113 , using the boundary conditions ( on the axis ) : @xmath114 where we denote by @xmath115 the integration variable . in the case
@xmath30 , this equation is easily integrated again and gives in this case : @xmath116 it follows that the dilaton charge is minus twice the tolman mass , @xmath117 .
this imposes an additional condition on the powers of the asymptotic metric components for any @xmath87 , @xmath88 which selects two points on the ellipse .
the first is : @xmath118 which is evidently locally flat .
it is also easy to prove the converse , i.e. , that there are no asymptotically conic solutions for @xmath31 . in order to do it ,
we start again with eq.([inteqnphi ] ) and notice also that for asymptotically conic solutions we need @xmath89 and @xmath90 , which yield @xmath66 .
thus , @xmath119 and @xmath120 vanish asymptotically faster than @xmath121 and the left - hand side of eq.([inteqnphi ] ) vanishes asymptotically .
but the right hand side of this equation is an integral of a sum of 4 positive - definite terms and it can vanish only if @xmath30 .
it also follows that in this asymptotically conic solutions @xmath122 , while the central magnetic field is given by : @xmath123 where @xmath124 is the deficit angle .
there is as usual a second point on the ellipse namely : @xmath125 as indicated above , this represents the same asymptotic behavior as the dilatonic melvin solution ( with @xmath30 ) .
this solution has non - vanishing dilaton charge and tolman mass .
the central magnetic field fulfills the equation : @xmath126
[ secunif ] the model analyzed so far has a dilaton which couples ( in the einstein / pauli frame ) with different strengths to various matter fields . as a consequence
, we do not return to the pure tensorial ( einstein ) gravity in the limit of vanishing dilaton coupling .
the only case of identical solutions of the two systems is the cosmic string solutions with @xmath30 and @xmath127 .
it seems therefore natural to consider a uniform coupling of the matter fields to the dilaton .
this uniform coupling will be described by the following action : @xmath128 the corresponding field equations for the cylindrically symmetric case will be a simplified version of ( [ eqf])-([constraint ] ) : @xmath129 @xmath130 @xmath131 @xmath132 @xmath133 with the constraint : @xmath134 the asymptotic behavior of the metric components and the dilaton field are obtained as before and the results are identical to ( [ asymptvac])-([elipse ] ) .
analysis of the new system shows that the solutions do not change much except in the case @xmath30 , where an asymptotically conic solution does not exist any more , see fig.6 .
actually , there are no asymptotically conic solutions in this system for any combination of parameters and in this respect the solutions are more similar to the string - like solutions in jbd theory .
it is quite simple to show that the only case where asymptotically conic solutions exist , in the case of this uniform coupling , is @xmath105 . the way to do it is to obtain an equation analogous to eq.([inteqnphi ] ) and to see that it can be satisfied by asymptotically constant @xmath111 and @xmath22 only for @xmath105 .
this is just the einstein - higgs system with a massless real scalar .
we have analyzed in detail , using both numerical and analytic methods , the cosmic string solutions in dilaton - gravity .
contrary to previous works , we did not limit ourselves to one branch in the weak - field approximation .
we found it most natural to work in the einstein frame , but all our results are trivially transformed into the string frame .
we have shown that the solutions generally come in pairs .
that is , any point in parameter space gives rise to two solutions with completely different asymptotic behaviors for the dilaton and for the geometry .
generally , neither of the two branches describe asymptotically flat solutions .
this should be contrasted with the non - dilatonic case , where one branch always describes asymptotically flat solutions . only for a very special coupling to the dilaton , which is related to certain compactified superstring theories ,
do asymptotically flat solutions appear .
our aim in this paper was to consider cosmic strings in the simplest dilaton - gravity models parametrized by a phenomenological coupling constant @xmath24 .
in the special case of fundamental string theory , which is related to our models for @xmath30 , there is an infinity of curvature corrections ( controlled by the reciprocal string tension @xmath135 ) and an infinity of string loop corrections ( controlled by the string coupling which is proportional to @xmath136 .
however , since our solutions for @xmath137 are everywhere non - singular and with finite dilaton , we expect that they will be only slightly modified in the full ( yet unknown ) quantum string theory .
the asymptotic behaviors of the solutions in the two branches seemed at first to depend in an extremely complicated way on the parameters of the theory .
however , a universal picture emerged , as explained in section 3 .
these results were obtained by coupling the dilaton `` uniformly '' to matter in the string frame . for comparison ,
we re - analyzed the whole problem using a uniform coupling in the einstein frame which may seem to be more natural . among other things , the correct tensorial ( einstein )
gravity solutions now come out in the limit of vanishing dilaton coupling .
but otherwise the solutions in the two models are quite similar .
a. vilenkin and e.p.s .
shellard , _ cosmic strings and other topological defects " _ ( cambridge univ . press ,
cambridge , 1994 ) .
a. vilenkin , phys . rep .
* 121 * , 263 ( 1985 ) .
m. hindmarsh and t.w.b .
kibble , rep .
. phys . * 58 * , 477 ( 1995 ) .
kibble , j. phys .
* a9 * , 1387 ( 1976 ) . ya .
b. zeldovich , mon . not .
. soc . * 192 * , 663 ( 1980 ) .
a. vilenkin , phys . rev . *
d23 * , 852 ( 1981 ) .
a.e . lange _
et al . _ ,
phys . rev . * d63 * , 042001 ( 2001 ) .
a.t . lee _
et al . _ , astro - ph/0104459 .
a.h . jaffe _
et al . _ ,
lett . * 86 * , 3475 ( 2001 ) .
r. durrer , a. gangui and m. sakellariadou , phys .
lett . * 76 * , 579 ( 1996 ) .
a.e . lewin and a. albrecht , phys . rev . * d64 * , 023514 ( 2001 ) .
r. durrer , m. kunz and a. melchiorri , astro - ph/0110348 .
nielsen and p. olesen , nucl .
phys . * b61 * , 45 ( 1973 ) .
d. kramer , h. stephani , e. herlt and m. maccallum , _ exact solutions of einstein s field equations " _
( cambridge univ . press , cambridge , england 1980 ) .
d. garfinkle , phys .
* d32 * , 1323 ( 1985 ) .
m. christensen , a.l .
larsen and y. verbin , phys . rev . * d60 * , 125012 - 1 ( 1999 ) .
l. marder , proc .
london a * 252 * , 45 ( 1959 ) .
bonnor , j. phys .
a * 12 * , 847 ( 1979 ) .
melvin , phys . lett . * 8 * , 65 ( 1964 ) .
b. linet , phys .
a * 124 * , 240 ( 1987 ) .
p. laguna - castillo and r.a .
matzner , phys .
* d36 * , 3663 ( 1987 ) .
d. garfinkle and p. laguna , phys .
* d39 * , 1552 ( 1989 ) .
laguna and d. garfinkle , phys . rev .
* d40 * , 1011 ( 1989 ) .
green , j.h .
schwarz and e. witten , _ superstrings theory " _
( cambridge univ . press ,
cambridge , 1987 ) . c. gundlach and m. ortiz , phys
. rev . * d42 * , 2521 ( 1990 ) .
l.o . pimentel and a.n .
morales , rev .
. fis . * 36 * , s199 ( 1990 ) .
guimaraes , class .
quantum grav .
* 14 * , 435 ( 1997 ) .
r. gregory and c. santos , phys .
* d56 * , 1194 ( 1997 ) . c. santos , class .
quantum grav .
* 18 * , 1835 ( 2001 ) .
v. faraoni , e. gunzig and p. nardone , fund .
* 20 * , 121 ( 1999 ) .
a. sen , nucl .
b388 * , 457 ( 1992 ) .
f. dowker , j. gauntlett , d. kastor and j. traschen , phys . rev . *
d49 * , 2909 ( 1994 ) . , @xmath138 , @xmath139 .
( a ) cosmic string branch .
( b ) dilatonic melvin branch .
, title="fig:",width=294 ] + ( a ) + , @xmath138 , @xmath139 . ( a ) cosmic string branch .
( b ) dilatonic melvin branch .
, title="fig:",width=294 ] + ( b ) , @xmath138 , @xmath139 .
( a ) cosmic string branch . ( b ) dilatonic melvin branch . , title="fig:",width=294 ] + ( a ) + , @xmath138 , @xmath139 .
( a ) cosmic string branch .
( b ) dilatonic melvin branch .
, title="fig:",width=294 ] + ( b ) , @xmath138 , @xmath139 .
( a ) cosmic string branch .
( b ) dilatonic melvin branch . , title="fig:",width=294 ] + ( a ) + , @xmath138 , @xmath139 .
( a ) cosmic string branch .
( b ) dilatonic melvin branch .
, title="fig:",width=294 ] + ( b ) , @xmath138 , @xmath139 .
( a ) cosmic string branch . ( b ) dilatonic melvin branch . , title="fig:",width=294 ] + ( a ) + , @xmath138 , @xmath139 .
( a ) cosmic string branch .
( b ) dilatonic melvin branch .
, title="fig:",width=294 ] + ( b )
|
the static string - like solutions of the abelian higgs model coupled to dilaton gravity are analyzed and compared to the non - dilatonic case . except for a special coupling between the higgs lagrangian and the dilaton , the solutions are flux tubes that generate a non - asymptotically flat geometry .
any point in parameter space corresponds to two branches of solutions with two different asymptotic behaviors . unlike the non - dilatonic case , where one branch is always asymptotically conic , in the present case the asymptotic behavior changes continuously along each branch .
+ @xmath0 _ department of natural sciences , the open university of israel , _ _ p.o.b .
39328 , tel aviv 61392 , israel _
0.4 cm @xmath1 _ department of physics , university of odense , _
_ campusvej 55 , 5230 odense m , denmark _ 1.1 cm _ pacs : 11.27.+d , 04.20.jb , 98.80.cq_
|
the standard cosmology was lack of a mechanism to produce cosmological perturbations . in order to compensate for this flaw in the standard model , there are currently two main paradigms for structure formation inflation @xcite and topological defects @xcite . while the beauty and simplicity of the former appears to have enticed more adherents and studies , the latter has proved computationally much more challenging to make robust predictions with which to confront observations @xcite .
these two paradigms are fundamentally different in the way they generate cosmological perturbations .
the standard adiabatic inflation produces primordial perturbations on all scales of cosmological interest via quantum fluctuations and the causal constraint during inflation , and these perturbations grow over time in an uncorrelated manner . as a consequence
, the perturbations today can be thought of as simply transfered from the initial irregularities that inflation set up , and this transfer function can be easily obtained in the linear theory and thus well understood in the literature . on the other hand ,
topological defects are the byproducts of the spontaneous symmetry - breaking phase transition in the early universe , and hence carry energy that was carved out of the originally homogeneous background energy of the universe . therefore due to causality
, defects induce perturbations only on sub - horizon scales , via gravitational interactions while evolving .
this mechanism that prevents the growth of super - horizon perturbations is called the ` compensation mechanism ' . in addition , due to the certain topology of the defect network , the resulting perturbations are correlated and thus non - gaussian , in contrast to the standard adiabatic inflationary perturbations .
it then follows that to compute the perturbations in models with defects , we need to know the evolution history of defects for the entire dynamic range during which the cosmological perturbations of our interest were seeded .
this is what makes the computation of defect - induced perturbations so difficult . in the literature
the power spectra of this kind of models have been investigated using the full einstein - boltzmann equations .
however , the study of the phase information of these perturbations still remains difficult because of the limited computation power .
although there have been some detailed treatments for theories with causal seeds @xcite , we shall in this paper present a simpler formalism , which is an approximation to the full einstein - boltzmann equations , to provide not only a physically transparent way for understanding the evolution of density perturbations in models with source , but also a computationally economical scheme to investigate the phase information of the resulting cosmological perturbations .
this formalism is parallel to those presented in ref .
@xcite and ref .
@xcite , but we give some modifications to incorporate the inclusion of the cosmological constant and a more detailed treatment for the effect of baryon - photon coupling / decoupling .
we also note that part of the solutions in ref .
@xcite are incorrect due to the incorrect initial conditions and the incorrect assumptions about the form of the subsequent perturbations induced by the source ( see text later ) .
we shall correct these mistakes and further provide a complete and explicit set of analytic solutions for the matter density perturbations . with an accurate extrapolation scheme ,
these solutions become also valid for models with any reasonably chosen background cosmology . the formalism and its solutions to be developed here
will be completely general and thus suitable for any models with or without causal / acausal source .
the structure of this paper is as follows . in section [ synchronous_gauge_perturbation_theory ] , under the fluid approximation
, we first derive in the synchronous gauge the full perturbation equations with source terms , in flat cosmologies with a cosmological constant @xmath0 .
this is done by considering the stress - energy conservation of the fluids ( [ conservation - rad - mat ] ) and the source ( [ conservation - seeds ] ) , and the linearly perturbed einstein equations ( [ perturbed - einstein ] ) .
the fluid components considered here are cold dark matter ( cdm ) , baryons ( b ) , and photons ( @xmath3 ) , and we employ the baryon - photon tight - coupling approximation to derive the perturbation equations before the last - scattering epoch . in this context , we also investigate the role of the so - called stress - energy pseudotensor ( [ conservation - pseudo ] ) . the initial conditions of these perturbation equations are discussed ( [ initial - conditions ] ) , and we use the approximation of instantaneous decoupling to deal with the decoupling of photons and baryons at the epoch of last scattering ( [ instantaneous - decoupling ] ) .
we then numerically justify the accuracy of this formalism in the context of standard cdm models , by comparing its results with those of the full einstein - boltzmann solver @xcite ( [ verification - stdcdm ] ) . within reasonable ranges of cosmological parameters
, our approach provides satisfactory precision at greatly reduced numerical cost . in section
[ solutions - of - matter - perturbations ] , we derive the matter perturbation solutions of the equations presented in section [ synchronous_gauge_perturbation_theory ] .
the perturbations of radiation and matter are first divided into two parts : the initial and the subsequent perturbations . with some change of variables ,
these equations are then ready to be solved by the green - function technique ( [ evolution - equations - and - decomposition - of - perturbation ] ) . with this technique
, we find the exact solutions on scales much larger or much smaller than the horizon size , namely the super - horizon or the sub - horizon solutions respectively ( [ super - horizon - and - sub - horizon - modes ] ) .
some degeneracy among the green functions for the matter perturbation solutions is then found and used to reduce their effective number ( [ degeneracy - of - the - green - functions ] ) . with this great simplification , solutions on intermediate scales
are then easily obtained by an accurate interpolation scheme based on the well - known standard cdm transfer function ( [ solutions - on - intermediate - scales ] ) .
we also discuss the effect of baryons ( [ the - effect - of - baryons ] ) .
a simple and accurate extrapolation scheme is then introduced to obtain solutions in the @xmath4 or @xmath2 cosmologies ( [ k - neq-0-or - l - neq-0-models ] ) , where @xmath5 is the curvature of the universe ( see appendix [ background_cosmologies ] ) .
all our green - function solutions are numerically verified to high accuracy . in section [ important - properties ] , we use our green - function solutions to investigate some important properties of cosmological matter density perturbations .
we first demonstrate the relation between our solutions and the standard cdm transfer function ( [ the - standard - cdm - model ] ) .
we also prove that in models with causal source , the resulting matter perturbations today are independent of the way the source energy is initially compensated into the background contents of the universe ( [ independence - of - the - initial - conditions ] ) .
finally we use our green - function solutions to study the compensation mechanism and the scale on which it operates ( [ compensation - and - total - matter - perturbations ] ) .
we find that this compensation scale is determined not only by the dynamics of the universe , but also by the properties of the source near the horizon scale .
once the detailed features of the source near the horizon scale are known , this compensation scale can be accurately located using our green functions .
a summary and conclusion is given in section [ matpert - conclusion ] . in appendix [ background_cosmologies ] , we define the convention of some notations used in this paper , and present for reference the solutions for the dynamics of various background cosmologies , including the consideration of non - zero curvature and a cosmological constant .
in this section , we derive the linear evolution equations for cosmological perturbations . to calculate the density and metric perturbations , we model the contents of the universe as perfect fluids : radiation ( photons and neutrinos ) and pressureless matter ( cdm and baryons ) .
we shall use the photon - baryon tight - coupling approximation until the epoch of last scattering , at which we assume instantaneous decoupling , also taking into account the effect of silk damping due to the photon diffusion .
after the decoupling , the baryonic perturbations originating from the perturbations of the photon - baryon coupled fluid are then merged linearly into the cdm content . in scenarios with causal seeds ,
the radiation and matter fields are assumed to be initially uniform , and then perturbed by the causal seeds after they are formed .
the radiation , matter , and causal seeds are assumed to interact only through gravity , meaning that their stress - energy tensors are separately covariantly conserved .
we shall work in the synchronous gauge , in which the perturbations @xmath6 to the spacetime metric @xmath7 obey the constraint @xmath8 . throughout this paper ,
we use a signature @xmath9 for the spacetime metric , and units in which @xmath10 .
thus the perturbed flat friedmann - robertson - walker ( frw ) metric is given by @xmath11.\ ] ] we shall work in the linear theory , requiring @xmath12 .
greek alphabet will denote the spacetime indices ( e.g. @xmath13 ) , and mid - alphabet latin letters the spatial indices ( e.g. @xmath14 ) . although the synchronous gauge is sometimes criticised in the literature due to its residual gauge freedom , it is still well suited to models in which the universe evolves from being perfectly homogeneous and isotropic . in such models ,
all perturbation variables can be initially set to zero ( before the causal seeds are generated ) , and this is normally referred to as the ` initially unperturbed synchronous gauge ' ( iusg ) @xcite .
it possesses no residual gauge freedom .
thus the einstein equations are completely causal in iusg , with the values of all perturbation variables at a given spacetime point being completely determined by initial conditions within the past light cone of the point .
one example of such models is the cosmic defect models , which have been of most interest in the study of models with causal seeds . in section [ conservation - rad - mat ] , we derive in the iusg the conservation equations of radiation and matter fields . in section [ conservation - seeds ] , we consider the conservation of source stress energy . in section [ perturbed - einstein ] , we derive the linearly perturbed einstein equations .
then , in section [ conservation - pseudo ] , we employ the concept of stress - energy pseudotensor to investigate the internal energy transfer among various fields . in section [ instantaneous - decoupling ] , we describe the approximation of instantaneous decoupling of photons and baryons at the epoch of last scattering . in section [ verification - stdcdm ] , we numerically verify the accuracy of our formalism for the standard cdm model , in comparison with the results from cmbfast @xcite , a fast einstein - boltzmann solver .
the contents of the universe are considered as perfect fluids , whose energy - momentum tensors have the form @xmath15 here @xmath16 , @xmath17 , and @xmath18 are the density , pressure , and four - velocity of the @xmath19th fluid respectively . in the homogeneous background , we have @xmath20 , which implies that @xmath21 to first order .
we thus define the velocity perturbation as @xmath22 , i.e. , @xmath23 .
the equation of state and the sound speed are defined respectively as @xmath24 consequently , the covariant conservation of stress energy for each fluid @xmath25 gives @xcite @xmath26 where @xmath27 , @xmath28 is the spatial trace of @xmath6 , and we have decomposed the velocities as @xmath29 , with @xmath30 and @xmath31 . in the regime of photon - baryon tight coupling , we have only two main fluids : the cdm component and the tightly - coupled photon - baryon fluid . they will be denoted as @xmath32 respectively , and discussed separately as follows .
note that we have ignored the neutrinos in the radiation .
we first consider the cdm fluid , i.e. @xmath33 . with @xmath34 for pressureless matter ,
the equations of stress - energy conservation ( [ conserve-1])([conserve-3 ] ) become @xmath35 as we can see , any perturbations in the cdm velocity will decay as @xmath36 .
thus we can simply choose @xmath37 in the iusg .
once @xmath37 , it will remain so as there is no linear gravitational source .
as a consequence , the cdm obeys a single nontrivial conservation law resulting from equation ( [ conserve - c0 ] ) @xmath38 where the second equation results from the initial condition @xmath39 , as required by the iusg . for the tightly - coupled photon - baryon ( @xmath40 ) fluid
, we have @xmath41 thus we can define @xmath42 where the second result comes from the fact that @xmath43 and @xmath44
. definitions ( [ eqn - of - state ] ) then give @xmath45 with these results , the equations of stress - energy conservation for the @xmath40 fluid can be obtained from equations ( [ conserve-1])([conserve-3 ] ) : @xmath46 in cosmological applications , such as cmb anisotropies , we are more interested in the photon perturbations rather than the perturbations in the @xmath47 fluid . therefore by using equations ( [ v - p - rho - gammab ] ) and ( [ r - gammab ] ) , we can extract the photon component from the above equations to yield @xcite @xmath48 where we have ignored neutrinos in the radiation so as to replace the subscript @xmath3 with @xmath49 .
the velocity can then be eliminated to yield a single second - order equation : @xmath50 we note that although the photon velocities are missing in this equation , they can be recovered at any given moment using equation ( [ conserve - r1 ] ) .
an alternative presentation of equations ( [ conserve - r1 ] ) and ( [ conserve - r ] ) is via the entropy perturbation @xmath51 .
it is defined as the fluctuation in the number of photons per dark matter particle @xmath52 thus equations ( [ conserve - r1 ] ) and ( [ conserve - r ] ) can be rewritten as @xmath53 as we shall see , @xmath54 can only have a white noise power spectrum on super - horizon scales . from equation ( [ conserve - r2 ] ) , this implies a @xmath55 power spectrum in @xmath56 on these scales .
adding the fact that the entropy fluctuation @xmath51 starts from zero on super - horizon scales due to the fixed number of dark matter particles per photon , it then follows from equation ( [ entropy1 ] ) that both @xmath51 and @xmath57 have a @xmath58 fall off outside the horizon .
therefore in numerical simulations , as long as the initial horizon size is smaller than the scales of our interest , we can simply set @xmath59 as part of the initial condition .
the causal source we shall consider is weak , so it will appear only as first - order terms in the perturbed einstein equations .
thus in the linear theory we are considering here , they can be treated as being stiff , meaning that their evolution depends only on their own self - interactions and the background dynamics of the universe , but not on their self - gravity or on the weak gravitational field of the inhomogeneities they produce .
this assumption will enable us to separate the calculation of their dynamics from that of the inhomogeneities they induce , allowing us to evolve them as if they are in a completely homogeneous background . since the source is stiff , its energy - momentum tensor @xmath60
need only be locally covariantly conserved with respect to the background : @xmath61 where @xmath62 .
another important aspect of cosmic structure formation with causal seeds like cosmic defects is the fact that the sources , formed at very early times , will ultimately create under - densities in the initially homogeneous background , out of which they are carved .
this is a direct result of energy conservation in the universe , and is normally termed ` compensation ' .
we shall discuss this issue in more detail later .
at first we have ten einstein equations @xmath63 or equivalently , @xmath64 where @xmath65 is the ricci tensor , @xmath66 is the gravitational constant , @xmath67 , @xmath0 is the cosmological constant , @xmath68 is the einstein tensor , and @xmath69 is the scalar curvature . linearly perturbing the above equations , we obtain @xmath70 or equivalently , @xmath71 where @xmath72 a closed set of the ten linearly perturbed einstein equations are then @xmath73 \tilde{h}_{ij}\right\ } & = & \ddot{\tilde{h}}_{ij}+2{\frac{\dot{a}}{a}}\dot{\tilde{h}}_{ij } -\nabla^2 { \tilde{h}}_{ij}-{\frac{1}{3}}h_{,ij } + { \frac{1}{9}}\delta_{ij}\nabla^2 h \nonumber\\ & & + \tilde{h}_{ik , kj } + \tilde{h}_{jk , ki}-{\frac{2}{3}}\delta_{ij}\tilde{h}_{kl , kl } = 16\pi g\tilde{\theta}_{ij } + a^2 \lambda \tilde{h}_{ij}\ ; , \label{deinstein2}\\ 2\delta g_{00 } = \tilde{h}_{ij , ij}-{\frac{2}{3}}\nabla^2h+2{\frac{\dot{a}}{a}}\dot{h } & = & 6\left({\frac{\dot{a}}{a}}\right)^2 \sum_n \omega_n\delta_n + 16\pi g\theta_{00}\ ; , \label{deinstein3}\\ 2\delta g_{0i } = \dot{\tilde{h}}_{ij , j}-{\frac{2}{3}}\dot{h}_{,i } & = & -6\left({\frac{\dot{a}}{a}}\right)^2 { \sum_n } ' ( 1+\kappa_{\scriptscriptstyle n})\omega_n v^i_{\scriptscriptstyle n } + 16\pi g\theta_{0i}\ ; , \label{deinstein4}\end{aligned}\ ] ] where the traceless parts are defined by @xmath74 , and similarly for @xmath75 and @xmath76 .
the prime over the sum in equation ( [ deinstein4 ] ) indicates the sum over all fluids except cdm .
we note from the above results that in the iusg the cosmological constant @xmath0 does not appear as extra terms in the perturbation equations except in ( [ deinstein2 ] ) , the ` @xmath77 ' component . within the photon - baryon tight - coupling regime ,
the above perturbation equations simplify as : @xmath78 + 8\pi g \theta_+ \ ; , \label{einstein1}\\ \ddot{\tilde{h}}_{ij}+2{\frac{\dot{a}}{a}}\dot{\tilde{h}}_{ij } -\nabla^2 { \tilde{h}}_{ij}-{\frac{1}{3}}h_{,ij } + { \frac{1}{9}}\delta_{ij}\nabla^2 h + \tilde{h}_{ik , kj } + \tilde{h}_{jk , ki}-{\frac{2}{3}}\delta_{ij}\tilde{h}_{kl , kl } = 16\pi g\tilde{\theta}_{ij } + a^2 \lambda \tilde{h}_{ij}\ ; , \label{einstein2}\\ \tilde{h}_{ij , ij}-{\frac{2}{3}}\nabla^2h+2{\frac{\dot{a}}{a}}\dot{h } = 6\left({\frac{\dot{a}}{a}}\right)^2 \left[\omega_{\rm c}\delta_{\rm c}+(1+r)\omega_{\rm r}\delta_{\rm r}\right ] + 16\pi g\theta_{00}\ ; , \label{einstein3}\\ \dot{\tilde{h}}_{ij , j}-{\frac{2}{3}}\dot{h}_{,i } = -8\left({\frac{\dot{a}}{a}}\right)^2 ( 1+r)\omega_{\rm r } v_{\rm r}^i+16\pi g\theta_{0i}\;. \label{einstein4}\end{aligned}\ ] ] we note that if the source obeys the covariant conservation equations ( [ conserve - seed1 ] ) and ( [ conserve - seed2 ] ) , then equations ( [ einstein3 ] ) and ( [ einstein4 ] ) are preserved by equations ( [ einstein1 ] ) . in the standard cdm model where
the source is absent , equation ( [ einstein1 ] ) can be greatly simplified on super - horizon scales ( @xmath79 ) in the radiation or matter era : @xmath80 since @xmath81 by definition , we know @xmath82 deep in the radiation era .
thus the above equations both have a growing mode @xmath83 .
this result has an important implication for numerical simulations of structure formation with causal sources . in this case ,
if numerical errors appear as white noise on super - horizon modes @xmath84 , then they will have a growing behavior @xmath85 .
for the horizon crossing mode @xmath86 , this becomes @xmath87 @xcite .
this means that although energy conservation together with causality should forbid the growth of perturbations on super - horizon scales , any numerical errors seeded from early times would induce a spurious growing mode on these scales . to overcome this problem
, one needs to perfectly compensate the source energy in the initially homogeneous background . in the following section
, we shall discuss one of the methods that can achieve this .
the concept of the stress - energy pseudotensor in an expanding universe was first remarked in this context by veeraraghavan and stebbins @xcite , and further investigated by pen , spergel and turok @xcite . to introduce this concept ,
we start from a perturbed minkowski space @xmath88 , where the bianchi identity @xmath89 leads to an ordinary conservation law @xmath90 at linear order in @xmath91 .
adding the fact that the einstein equations give @xmath92 where @xmath93 is the sum of non - linear terms in @xmath91 , we see that the right - hand side of this equation provides an ordinarily conserved tensor , the stress - energy pseudotensor . the generalization of this result to an frw model is straightforward , with only the corrections due to the expansion of the universe .
moving all these corrections ( derivatives of the scale factor ) to the right - hand side of the einstein equations while keeping only the linear terms in @xmath94 , we obtain a pseudo - stress - energy tensor @xmath95 : @xmath96 -\frac{1}{8 \pi g}\frac{\dot{a}}{a}\dot{h}+\theta_{00 } \ ; , \label{tau00}\\ \tau_{0i } & = & -\frac{1}{2\pi g}\left({\frac{\dot{a}}{a}}\right)^2 ( 1+r)\omega_{\rm r } v_{\rm r}^i+\theta_{0i } \ ; , \label{tau0i}\\ \tau_{ij } & = & \delta_{ij } \frac{1}{8 \pi g}\left({\frac{\dot{a}}{a}}\right)^2 \omega_{\rm r}\delta_{\rm r } -\frac{1}{8 \pi g } \frac{\dot{a}}{a}(\dot{\tilde{h}}_{ij}-\frac{2}{3}\dot{h}\delta_{ij } ) + \theta_{ij}\;. \label{tauij}\end{aligned}\ ] ] this tensor obeys an ordinary conservation law @xmath97 according to the einstein equations , or equivalently @xmath98 this is not a fundamentally new conservation law , but it describes the interchange of energy and momentum among the different components in the universe , i.e. the radiation , matter , and the source in our case .
this description appears to be physically more transparent than the original einstein equations .
another advantage of invoking this formalism is that it is easier for numerical simulations to specify the initial conditions and to maintain proper compensation on super - horizon scales . as we shall explain later
, @xmath99 can only have a white - noise power spectrum on super - horizon scales . thus integrating equations ( [ conserve - tau1 ] ) and ( [ conserve - tau2 ] ) over time shows that @xmath100 has a @xmath58 power spectrum and that @xmath101 has a @xmath55 power spectrum .
therefore , as long as the horizon size at the beginning of the simulation is smaller than the scales of our interest , we can set @xmath102 as the initial condition , allowing for perturbations to grow only inside the horizon and for @xmath100 to fall off as @xmath58 outside the horizon . for simulations of structure formation with causal source ,
a check of @xmath103 on super - horizon modes will tell us whether or not the compensation is well obeyed . to make use of the pseudo - stress - energy tensor formalism in the study of cosmological perturbations ,
we combine the conservation equation for radiation ( [ entropy2 ] ) , the definition of pseudoenergy ( [ tau00 ] ) , and one of the alternative einstein equations using the pseudo - stress - energy tensor ( [ conserve - tau1 ] ) , to yield a convenient closed set of equations : @xmath104 \delta_{\rm c}+ 2(1+r)\omega_{{\rm r}}s \right\ } , \label{density - eqn2}\\ \dot{\tau}_{00 } & = & \theta_{0i , i}+ \frac{1}{2\pi g}\left(\frac{\dot{a}}{a}\right)^2 ( 1+r)\omega_{{\rm r}}\dot{s}\;. \label{density - eqn3}\end{aligned}\ ] ] here we have used equations ( [ conserve - c ] ) , ( [ entropy ] ) , ( [ entropy1 ] ) and ( [ tau0i ] ) to eliminate @xmath105 , @xmath54 , @xmath106 and @xmath101 respectively . by analogy to the results in ref .
@xcite , here we have built both the pseudoenergy @xmath100 and the entropy fluctuation @xmath51 into the above formalism .
as required by the iusg , all perturbation variables are zero before any mechanism of structure formation starts to act on the initially homogeneous and isotropic universe . in causal models
, causality also requires that local physical processes can never induce correlated perturbations on scales much larger than the horizon .
therefore , when the initial irregularities of the universe are first formed ( e.g. via the formation of cosmic defects , or the presence of inflation ) , the spatial part of @xmath60 , @xmath107 and @xmath6 can only have white - noise power spectra on super - horizon modes their spatial perturbations being uncorrelated on scales larger than the horizon size @xcite .
the same applies to @xmath108 and therefore @xmath109 .
it then follows from equations ( [ conserve - r2 ] ) , ( [ conserve - seed2 ] ) and ( [ conserve - tau2 ] ) respectively that the power spectra of @xmath56 , @xmath110 and @xmath101 all fall off as @xmath55 outside the horizon . from equations ( [ entropy1 ] ) and ( [ conserve - tau1 ] ) , we also have the spectra of @xmath51 , @xmath57 and @xmath100 proportional to @xmath58 on these scales , as previously discussed . as a summary , we have for super - horizon modes @xmath111 that @xmath112 where ` @xmath113 ' means the power spectrum is proportional to @xmath114 .
since the production time of the initial irregularities is normally so early that the horizon size @xmath115 at that time is much smaller than the cosmological scales @xmath116 of our interest ( i.e. @xmath117 ) , the above conditions can be regarded as general initial conditions for all scales of cosmological interest .
if we require @xmath118 in our analysis , we can simply choose @xmath119 as the initial conditions , because their power spectra all decay as either @xmath55 or @xmath58 outside the horizon .
with such a choice , we can see from equation ( [ tau00 ] ) that there is still freedom for the choice of @xmath120 , @xmath54 and @xmath105 into which to compensate @xmath121 when @xmath121 was first formed .
nevertheless , as we shall analytically prove later , no matter how @xmath121 was compensated into the background contents of the universe when the causal source was first formed , the resulting matter density perturbations today would be the same .
we note that this was first numerically observed in ref .
@xcite , and here we shall provide a thorough interpretation to it using our analytical solutions to be obtained later . we also note that none of the above arguments will hold if the initial perturbations are seeded in an acausal way , which is nevertheless not of our current interest .
one thing we have not included in our formalism is the treatment at and after the decoupling epoch @xmath122 . before this epoch , photons and baryons
are assumed to be tightly coupled , forming a single @xmath3b fluid . at the decoupling epoch @xmath122 , baryons and photons
are assumed to be instantaneously decoupled from each other , so that @xmath123 and @xmath124 evolve separately afterwards .
a numerical fit to the redshift of the decoupling epoch is @xcite @xmath125 , \label{z - d}\\ b_1 & = & 0.313(\omega_{\rm m0}h^2)^{-0.419 } \left[1 + 0.607(\omega_{\rm m0}h^2)^{0.674}\right],\\ b_2 & = & 0.238(\omega_{\rm m0}h^2)^{0.223}.\end{aligned}\ ] ] although this is the result for the decoupling epoch of baryons and there is another fit for that of photons , these two epochs the recombination of baryons and the last scattering of photons coincide approximately in the absence of subsequent reionization @xcite .
in addition , the photons and baryons are not in fact perfectly coupled , and this leads to the diffusion damping of photons and silk damping of baryons @xcite during the decoupling epoch .
to model these effects , we apply damping envelopes to both @xmath123 and @xmath124 at the decoupling epoch @xmath126 , i.e.@xmath127 where the tilde indicates the fourier transform of a quantity and @xmath128 is the wave number .
the photon diffusion damping envelope can be approximated by the form @xcite @xmath129 where @xmath130 \right\}^{p1/p2}f_1\;,\\ m_{\gamma } & = & 1.46(\omega_{\rm m0}h^2)^{0.0303 } \left(1 + 0.128\arctan\left\{\ln\left[(32.8\omega_{\rm b0 } h^2)^{-0.643}\right]\right\}\right),\\ p_1 & = & 0.29,\\ p_2 & = & 2.38 ( \omega_{\rm m0}h^2)^{0.184},\\ f_1 & = & 0.293(\omega_{\rm m0}h^2)^{0.545 } \left[1+(25.1\omega_{\rm m0}h^2)^{-0.648}\right],\\ f_2 & = & 0.524(\omega_{\rm m0}h^2)^{0.505 } \left[1+(10.5\omega_{\rm m0}h^2)^{-0.564}\right].\end{aligned}\ ] ] silk damping for the baryons can likewise be approximated as @xcite @xmath131 where @xmath132 in some scenarios with causal sources , the damping envelopes ( [ d_gamma ] ) and ( [ d_b ] ) may depart from the form of exponential fall - off here to a power - law decay towards smaller scales .
this is due to the survival of perturbations which are actively seeded during the decoupling process .
for example , in models with cosmic strings , the departure appears on scales smaller than of order a few arc - minutes ( i.e. the multipole index @xmath133 ) @xcite .
certainly this is beyond the scale range of our interest .
moreover , since the decoupling process is relatively a short instant in the entire evolution history of the perturbations , the contribution from these survived small - scale perturbations should be relatively small .
adding the fact that we expect the post - decoupling contribution in the perturbations seeded by defects to have a power - law fall - off on small scales due to a certain topology of the source @xcite , the small - scale power in the final perturbations is likely to be dominated by this post - decoupling contribution , rather than the primary perturbations ( those seeded before and during the decoupling , whose power spectrum first exponentially decays and then turns to a power - law fall - off ) .
therefore , on the scales of our interest , the damping approximation employed here should be still appropriate for models with cosmic defects .
now we consider the evolution of @xmath123 and @xmath124 after the decoupling epoch @xmath126 . from the energy conservation law ( [ conserve-1])([conserve-3 ] ) , we have for the baryon perturbations @xmath134 this implies @xmath135 , meaning that the evolution of @xmath124 and @xmath120 will soon converge to the same behavior .
we also know that matter perturbations grow as @xmath136 in the matter era so that @xmath137 $ ] is relatively small when compared with either @xmath138 or @xmath139 . as a consequence , in the calculation of @xmath138 and @xmath139 to linear order , it is appropriate to combine @xmath124 and @xmath120 at the decoupling epoch @xmath126 as @xmath140 and the same for their time derivatives
. then we have only two fluids after the decoupling : the photon fluid ( @xmath141 ) and the matter fluid , which is linearly combined from the cdm and baryon fluids ( @xmath142 ) .
eventually we can take the matter perturbations at the present epoch to be @xmath143 . to sum up , we first evolve the cdm and @xmath3b perturbations up to the decoupling epoch @xmath126 given by ( [ z - d ] ) , noting that our formalism extracts the photon component @xmath3 from the @xmath3b fluid .
we then apply damping envelopes to @xmath144 and @xmath145 , as illustrated by equation ( [ damping - effects ] ) , to account for the photon diffusion and silk damping .
@xmath146 is then obtained by linearly combining @xmath145 and @xmath147 , as shown in equation ( [ delta - m - decoupling ] ) .
finally we carry on the evolution of @xmath148 and @xmath149 from the epoch @xmath126 to the present , using our previous perturbation equations with @xmath150 and the subscript ` c ' replaced by ` m ' . to verify our scheme for evolving cosmological perturbations ,
we first calculate the cdm transfer function in the context of the adiabatic inflationary cdm model : @xmath151 where @xmath152 is the present conformal time . to this end , we employ equations ( [ density - eqn1 ] ) , ( [ density - eqn2 ] ) and ( [ density - eqn3 ] ) in the absence of the source terms , and the approximation of instantaneous decoupling described above .
we start the evolution in the deep radiation era when @xmath153 , @xmath154 , and @xmath155 for a given mode @xmath128 . in this case
, one choice of the initial conditions is @xmath156 figure [ fig - transtc ] shows our results for the cdm transfer functions @xmath157 at the present epoch in different cosmologies , together with the results obtained from cmbfast @xcite .
+ it is clear that they agree very well .
the discrepancy of the two reaches its maximum of about 5% at the scale @xmath158mpc@xmath159 in the open model with @xmath160 and @xmath161 .
we have also checked our results against those in ref .
@xcite , and they are in agreement again within a @xmath162% error . in addition , from the bottom curves in figure [ fig - transtc ] , we notice the oscillations resulting from the photon - baryon coupling before @xmath122 in cosmologies with high baryon fractions @xmath163 .
next , we calculate the radiation transfer function at the decoupling epoch , since the radiation perturbations at this epoch will appear as the intrinsic cmb anisotropies .
we define this transfer function as : @xmath164 where we normalize the radiation perturbations at @xmath122 to both the amplitude of the super - horizon cdm perturbations today and the initial cdm power spectrum , as we did for @xmath157 ( see eq .
[ [ cdm - transfer - function ] ] ) . this definition will enable us to verify not only the scale dependence of the evolution of perturbations , but also their normalizations .
figure [ fig - transtr ] shows our results , again as a comparison with the results from cmbfast .
+ we see that although the scale dependence of our results is slightly different from that of the cmbfast results , the overall normalisation appears to be quite accurate .
the sideway shift of the oscillatory peaks in our results when compared with the peaks from cmbfast has a maximum of about 5% in the flat model with @xmath165 and @xmath161 .
this discrepancy results naturally from the instantaneous - decoupling approximation in our formalism . as a result , despite the small inaccuracy , our formalism provides a much more numerically efficient way than the full einstein - boltzmann scheme in calculating the density perturbations .
we first consider density perturbations about a flat frw model with a cosmological constant @xmath0 , which are causally sourced by an evolving source field with the energy - momentum tensor @xmath166 .
as seen in the previous section , with the photon - baryon tight coupling approximation in the synchronous gauge , the linear evolution equations of the radiation and cdm perturbations can be given by equations ( [ density - eqn1 ] ) , ( [ density - eqn2 ] ) and ( [ density - eqn3 ] ) , which are derived from equations ( [ entropy2 ] ) , ( [ tau00 ] ) and ( [ conserve - tau1 ] ) .
this set of equations has the advantage in controling the initial condition for numerical simulations , as well as understanding the law of stress - energy conservation . for analytic simplicity
, however , we shall drop the use of @xmath100 in this section , and employ equations ( [ conserve - r ] ) and ( [ einstein1 ] ) to form an alternative set of evolution equations for density perturbations : @xmath167 = 4 \pi g \theta_+
. \label{delta - two}\end{aligned}\ ] ] we note again that the cosmological constant @xmath0 affects only the background dynamics ( i.e. , the evolution of the scale factor @xmath168 ) , but does not contribute extra terms in the above perturbation equations . after the decoupling epoch @xmath122 , the treatment is essentially the same as that introduced in section [ instantaneous - decoupling ] .
we have numerical verified in the context of the adiabatic inflationary cdm model that the set of equations ( [ delta - one ] ) and ( [ delta - two ] ) and the set of equations ( [ density - eqn1 ] ) , ( [ density - eqn2 ] ) and ( [ density - eqn3 ] ) indeed give identical transfer functions of density perturbations , with a numerical discrepancy of less than @xmath169% .
assuming that the causal source was formed at some initial time @xmath115 and then evolved to the current time @xmath170 , it proves useful to split the source - seeded linear perturbations into initial ( i ) and subsequent ( s ) parts @xcite : @xmath171 the initial perturbations @xmath172 originate from the source configuration at @xmath115 , while the subsequent perturbations @xmath173 are actively and cumulatively seeded by the later evolution of the source at each @xmath174 , where @xmath175 .
this is equivalent to having the initial conditions @xmath176 because the source induces isocurvature perturbations , @xmath177 must compensate @xmath178 on comoving scales @xmath179 to prevent acausal perturbation growth on super - horizon scales .
one of the aims of this paper is to show analytically how this compensation mechanism can be achieved .
now we can solve the system of equations ( [ delta - one ] ) and ( [ delta - two ] ) by employing the integral equation with green functions : @xmath180 where @xmath181 .
the easiest method of obtaining the green - function solutions is to go to fourier space and solve the resulting homogeneous system of ordinary differential equations with appropriate initial conditions .
since the green functions depend only on the modulus of @xmath181 , it follows that their fourier amplitudes must depend only on the modulus of @xmath182 .
thus we have @xmath183 , \label{delta_i_n } \\ \widetilde{\delta}^{\rm s}_{n}({\bf k},\eta ) & = & 4 \pi g \int_{\eta_{\rm i}}^{\eta } \widetilde{\cal g}^{n{\rm s } } ( k;\eta,\hat{\eta } ) \widetilde\theta_{+}({\bf k},\hat{\eta } ) \ , d\hat{\eta}\ , .
\label{delta_s_n}\end{aligned}\ ] ] we notice that equation ( [ delta_s_n ] ) is different from the form in ref .
@xcite , where the authors identified our @xmath184 as @xmath185 .
this identification is incorrect , because @xmath184 and @xmath185 have different initial conditions , as we shall see . for simplicity , we assume no baryons and therefore set @xmath150 for now , and shall relax this constraint later . with the change of variable @xmath186 where @xmath187 ( leading to @xmath188 ) , and with the formalism ( [ delta_i_n ] ) and ( [ delta_s_n ] ) , we can rewrite equations ( [ delta - one ] ) and ( [ delta - two ] ) in fourier space as @xmath189{\widetilde{\cal g}}^{\rm c } = 0\ , \label{g_one}\end{aligned}\ ] ] where a prime represents a derivative with respect to @xmath190 , @xmath191 , @xmath192 or @xmath193 , and @xmath194 , @xmath195 or @xmath196 . according to equations ( [ delta_i_n ] ) and ( [ delta_s_n ] ) ,
the initial conditions ( [ inic_deltai ] ) and ( [ inic_deltas ] ) now become : @xmath197 with all the other green functions and their time derivatives vanishing .
there are three things we should notice here .
first , it is required that @xmath198 for @xmath199 , and that @xmath200 for @xmath201 .
second , the green functions @xmath202 only describe the time dependence of the homogeneous version of equations ( [ delta - one ] ) and ( [ delta - two ] ) , while the green functions @xmath184 are , by the conventional definition , the true green functions used to solve the inhomogeneous equations ( [ delta - one ] ) and ( [ delta - two ] ) . finally ,
since there are only four variables in equations ( [ g_two ] ) and ( [ g_one ] ) ( i.e. @xmath203 , @xmath204 , @xmath205 and @xmath206 ) , there must exist some dependence among the five sets of green functions ( i.e. @xmath207 , @xmath208 , @xmath209 , @xmath210 and @xmath211 ) .
this dependence can be observed from the initial conditions ( [ gnnini ] ) and ( [ gnsini ] ) , which yield @xmath212 in ref .
@xcite , the authors ignored the fact that @xmath213 in the initial condition ( [ gnsini ] ) .
this ignorance led to the absence of the second term in equation ( [ gcs-2 ] ) ( and thus the identification of @xmath214 ) , and consequently the incorrect solutions of green functions in their final results .
based on equations ( [ g_two ] ) and ( [ g_one ] ) with the initial conditions ( [ gnnini ] ) and ( [ gnsini ] ) , in the following subsections we shall analytically derive a complete set of green - function solutions for the matter perturbations , which will then be numerically verified . under the limit @xmath215 or @xmath216 , the ratio @xmath217 will approach a constant ( see below ) , so that equation ( [ g_one ] ) becomes the associated legendre equation , with solutions composed of the associated legendre functions @xmath218 and @xmath219 , where @xmath220 .
we shall use subscripts @xmath221 and @xmath222 to denote solutions in the limits @xmath215 and @xmath216 respectively . for simplicity
, we shall denote both @xmath174 and @xmath115 as @xmath174 in the following solutions .
1 . @xmath215 : when the wavelengths are much smaller than the horizon size , the radiation oscillates many times per expansion time and its effect is therefore negligible . by setting @xmath223 , equation ( [ g_one ] ) can be solved as @xmath224 where @xmath225 and @xmath226 are functions of @xmath174 .
this gives the sub - horizon solutions .
@xmath79 : when the wavelengths are much longer than the horizon size , we have @xmath227 as the consequence of zero entropy ( see eqs . [ [ entropy ] ] and [ [ initial - condition ] ] ) , giving @xmath228 . thus equations ( [ g_two ] ) and ( [ g_one ] ) yield @xmath229 where @xmath230 and @xmath231 are constants , and @xmath232 and @xmath233 are functions of @xmath174 , all determined by the initial conditions .
these are the super - horizon solutions . combined with the initial conditions ( [ gnnini ] ) and ( [ gnsini ] ) , equations ( [ gcninf ] ) and ( [ gc0 ] ) can be solved to yield the following results . for clarity
, we shall denote @xmath234 in @xmath184 and @xmath235 in @xmath202 both as @xmath236 : @xmath237 \!-\!6(y - w)(3wy+1 ) \right\ } , \label{gcsinf}\\ \widetilde{\cal g}^{\rm cs}_0 & = & \frac { 2(y^6w - w^6y - 5y^4w + 5w^4y + 15y^2w - 15yw^2 + 5w - 5y ) } { 5a ( y^2 - 1)^2 ( w^2 - 1)}\ , , \label{gcs0}\\ \widetilde{\cal g}^{\rm cc}_{1\infty } & = & \frac{1}{2}(3y^2 - 1)(3w^2 - 2)-\frac{9}{2}wy(w^2 - 1 ) \nonumber\\ & & + \frac{3}{4}w(w^2 - 1)(3y^2 - 1 ) \log\left[\frac{(y+1)(w-1)}{(y-1)(w+1)}\right ] , \label{gcc1inf}\\ \widetilde{\cal g}^{\rm cc}_{10}&= & \frac { 2yw^5 -20yw^3 + 20y^2w^2 + 20w^2 -30yw -15y^4 -5 + 3y^6 + 25y^2 } { 5(y^2 - 1)^2(w^2 - 1 ) } \ , , \label{gcc10}\\ \widetilde{\cal g}^{\rm cc}_{2\infty}&= & \widetilde{\cal g}^{\rm cs}_{\infty}\ , , \label{gcc2inf}\\ \widetilde{\cal g}^{\rm cc}_{20}&= & \widetilde{\cal g}^{\rm cs}_0-\frac{4}{3}\widetilde{\cal g}^{\rm cr}_{20}\ , , \label{gcc20}\\ \widetilde{\cal g}^{\rm cr}_{1\infty}&= & 0\ , , \label{gcr1inf}\\ \widetilde{\cal g}^{\rm cr}_{10}&= & \frac { 3(y^6 - 5y^2w^4 + 4yw^5 + 10y^2w^2 - 5y^4 - 5w^4 + 10y^2 - 20yw+ 10w^2 ) } { 5(y^2 - 1)^2(w^2 - 1)^2}\ , , \label{gcr10}\\ \widetilde{\cal g}^{\rm cr}_{2\infty}&= & 0\ , , \label{gcr2inf}\\ \widetilde{\cal g}^{\rm cr}_{20 } & = & \frac { -3}{10 a } \left [ \frac { ( y^2 + 4y + 5)(y - 1)^2}{(y + 1)^2 } \log\left(\frac { y - 1}{w - 1}\right ) + \frac { ( y^2 - 4y + 5)(y + 1)^2}{(y - 1)^2 } \log\left(\frac { w + 1}{y + 1}\right ) \right . \nonumber \\ + & \!\!2 & \!\ ! \left .
( w - y ) \frac { ( 4yw^3 -6y^2w^2 -10w^2 + y^5w -4y^2w + 7yw + 6y^2 - 4y^3 + 5 + y^4 ) } { ( w^2 - 1)(y^2 - 1)^2 } \right ] . \label{gcr20}\end{aligned}\ ] ] we note that equations ( [ gcc2inf ] ) , ( [ gcc20 ] ) and ( [ gcr2inf ] ) result directly from the initial conditions ( [ gnnini ] ) and ( [ gnsini ] ) .
they are consistent with equation ( [ gcs-2 ] ) . despite the complicated forms presented here , all these green functions have simple asymptotic behaviors in the radiation- or matter - dominated regimes .
since we are more interested in the matter perturbations today and we know from equation ( [ g_one ] ) that @xmath238 when @xmath239 , we can design a ` source transfer function ' as @xmath240 note that this is different from the definition of the standard cdm transfer function ( [ cdm - transfer - function ] ) . equations ( [ gcsinf])([gcr20 ] ) then lead to the source transfer functions : @xmath241 , \label{tcsinf}\\ \widetilde{t}^{\rm cs}_0 & = & \frac { 2w}{5a ( w^2 - 1)}\ , , \label{tcs0}\\ \widetilde{t}^{\rm cc}_{1\infty } & = & \frac{3}{2}(3w^2 - 2 ) + \frac{9}{4}w(w^2 - 1 ) \log\left(\frac{w-1}{w+1}\right ) , \label{tcc1inf}\\ \widetilde{t}^{\rm cc}_{10}&= & \frac { 3}{5(w^2 - 1 ) } \ , , \label{tcc10}\\ \widetilde{t}^{\rm cc}_{2\infty}&= & \widetilde{t}^{\rm cs}_{\infty}\ , , \label{tcc2inf}\\ \widetilde{t}^{\rm cc}_{20}&= & \widetilde{t}^{\rm cs}_0-\frac{4}{3}\widetilde{t}^{\rm cr}_{20 } = -\frac { 2}{5 a } \left [ \frac { w}{w^2 - 1 } -\log\left(\frac { w + 1}{w - 1}\right ) \right ] \ , , \label{tcc20}\\ \widetilde{t}^{\rm cr}_{1\infty}&= & 0\ , , \label{tcr1inf}\\ \widetilde{t}^{\rm cr}_{10}&= & \frac { 3}{5(w^2 - 1)^2}\ , , \label{tcr10}\\ \widetilde{t}^{\rm cr}_{2\infty}&= & 0\ , , \label{tcr2inf}\\ \widetilde{t}^{\rm cr}_{20 } & = & \frac { 3}{10 a } \left [ \frac { 2 w}{w^2 - 1 } -\log\left(\frac { w + 1}{w - 1}\right ) \right ] . \label{tcr20}\end{aligned}\ ] ] we plot these source transfer functions in figures [ fig - tcs ] , + [ fig - tcc ] , + and [ fig - tcr ]
. + they are now only functions of the initial time , but not of the final time . in the context of topological defects ,
the defect source was formed at @xmath242
. therefore it would be also interesting to investigate the asymptotic behaviors of the source transfer functions with very early initial times . for @xmath242 , equations ( [ tcsinf])([tcr20 ] )
become : @xmath243 where the subscript ( i ) denotes the condition @xmath242 .
these asymptotic behaviors can be clearly seen in figures [ fig - tcs ] , [ fig - tcc ] and [ fig - tcr ] .
we note that on sub - horizon scales , @xmath244 has a maximum at @xmath245 as seen in figure [ fig - tcs ] .
adding the fact that cosmic defects seed matter perturbations only on sub - horizon modes due to the compensation mechanism , it follows that the defect - induced matter perturbations are seeded mainly during the radiation - matter transition era .
this is a generically different mechanism from inflationary models , in which matter perturbations are seeded during inflation in the deep radiation era when all the modes are well outside the horizon .
nevertheless , the defect and inflationary models both provide scale - invariant perturbations at horizon crossing , and these perturbations evolve similarly after horizon crossing .
+ we note the the asymptotic behaviors indicated in equations ( [ tc3-propto ] ) and ( [ tc4-propto ] ) can be clearly seen in figures [ fig - tcs ] and [ fig - tc3 ] .
therefore , the original ten green functions for solving @xmath262 and @xmath149 have now been reduced to four functions : two for @xmath262 ( @xmath263 and @xmath264 ) , and two for @xmath149 ( @xmath265 and @xmath266 ) .
we shall concentrate only on the solutions of @xmath262 , while leaving those of @xmath149 elsewhere @xcite . to calculate
@xmath267 we need @xmath268 using equation ( [ delta_s_n ] ) ; to calculate @xmath269 we need @xmath252 and @xmath264 using equation ( [ delta_i_n-2 ] ) . in solving @xmath269 , we note that @xmath264 transfers the initial perturbations of both matter and radiation @xmath270 $ ] to today , while @xmath268 transfers the initial perturbations of their time derivatives @xmath271 $ ] to the present . with @xmath264 and @xmath272 ( @xmath273 ) as the two basis green functions
, we can now work out the solutions on intermediate scales , using results derived in previous sections . in the matter era , @xmath274 on all scales so from equation ( [ g_one ] ) we know that @xmath275 when @xmath276 .
this can be clearly seen from figures [ fig - tcs ] and [ fig - tc3 ] . in the radiation era , the perturbations @xmath270 $ ] or @xmath271 $ ] that were seeded well before the horizon crossing will evolve in the same way as in the standard cdm model due to the same zero entropy fluctuation initial condition . therefore the solution interpolating between @xmath277 and @xmath278 for @xmath242 will be the standard cdm transfer function . thus we can write down a fit of the solution for the full gamut of @xmath128 and @xmath115 as @xmath279 t(k ) i(k;\eta_{\rm i } ) , \quad i=3,4 \,,\ ] ] where @xmath280^{-1 } , \label{tk-2}\\ i(k ; \eta_{\rm i } ) & = & \frac{1 + 30\eta_{\rm i } } { 1 + 30\eta_{\rm i}(1+\frac{k\eta_{\rm i}}{2\pi } ) } , \label{iketa}\end{aligned}\ ] ] and @xmath128 is in units of @xmath281 ( see equation ( [ eta_eq ] ) ) .
here @xmath282 is the standard cdm transfer function without baryons ( modified from ref .
@xcite ; see eq .
[ [ cdm - transfer - function ] ] for the definition of @xmath157 ) , and @xmath283 is a small correction near the horizon crossing to make the analytic solutions ( [ gc ] ) fit the numerical results . for a given mode which is initially outside the horizon , the background contents of the universe compensate the defect source until horizon crossing .
therefore the detailed behavior of these green functions near the horizon scale will affect the so - called compensation scale , beyond which no perturbations can grow .
this means that the correction function @xmath283 in equation ( [ gc ] ) actually plays an important role in getting the compensation scale right , and we shall discuss this further in section [ compensation - and - total - matter - perturbations ] .
we have verified numerically for both @xmath264 and @xmath272 that the fit ( [ gc ] ) is accurate within a @xmath284 error for any @xmath128 and @xmath115 ( note that the initial conditions of @xmath264 and @xmath272 in the numerical verifications can be obtained from eqs . [
[ gnnini ] ] , [ [ gnsini ] ] and [ [ g_n_34 ] ] ) . figure [ fig - gc34 ] shows the numerical solutions of @xmath264 and @xmath272 ( = @xmath268 ) within a chosen domain of @xmath285 .
+ + it confirms the asymptotic behaviors indicated by equations in ( [ tc30 ] ) ( see also eqs .
[ [ tcsinf ] ] , [ [ tcs0 ] ] and [ [ tcc1inf ] ] ) , and plotted in figures [ fig - tcs ] and [ fig - tc3 ] . the asymptotic behaviors shown by equations ( [ tc3-propto ] ) and ( [ tc4-propto ] ) can be also marginally observed from figure [ fig - gc34 ] .
schematically , we can divide the @xmath286-plane into three regions for the solutions of @xmath287 ( @xmath288 ) . as shown in figure [ fig - k - eta ] , these three domains are : region i ( @xmath289 ) , region ii ( @xmath290 and @xmath291 ) , and region iii ( @xmath292 ) . + in region i , the solution of @xmath287 is @xmath293 because the horizon crossing happens after @xmath294 , after which @xmath295 as argued before . in region
ii , the solution is @xmath296 because all modes in this region are inside the horizon all the time .
we notice that @xmath293 merges with @xmath296 at the boundary of regions i and ii , where @xmath297 . in region
iii , the solution along the @xmath128 direction is in the same form as the standard cdm transfer function .
this is because modes with larger @xmath128 cross the horizon earlier , so that their perturbations are suppressed after the horizon crossing for longer until @xmath294 .
in addition , the solution along the @xmath115 direction in region iii is in the same form as @xmath293 .
this is because modes in this region are initially on super - horizon scales , and a given mode with different initial time @xmath115 will experience the same amount of suppression resulting from the period between the horizon crossing and @xmath294 .
therefore , regions i , ii and iii illustrate the intrinsic property of the solution ( [ gc ] ) .
there is one important issue which we have not discussed
the effect of baryons .
prior to the photon - baryon decoupling at @xmath122 , the cdm and baryons are dynamically independent . in this era , the photon - baryon fluid propagates as acoustic waves with a sound speed given by equation ( [ c_gammab ] ) , preventing baryons from collapsing on small scales .
therefore there exists a sound horizon at the decoupling epoch @xmath298 ( hereafter simply the sound horizon ) which is the distance such waves can travel prior to @xmath122 , and which is the largest scale at which the baryons can affect the evolution of density perturbations .
it has been shown that inside the sound horizon @xmath298 , not only are the cdm perturbations seeded before @xmath122 suppressed due to the presence of baryons ( e.g. @xcite ) , but also the baryons themselves have an exponentially decaying power due to the silk damping @xcite ( see also eq . [ [ damping - effects ] ] ) , with acoustic oscillations due to the velocity overshoot @xcite .
after the decoupling , baryons evolve in the same way as the cdm does , so the matter perturbations today can be obtained by linearly combining the cdm and baryonic fluctuations at @xmath122 ( see section [ instantaneous - decoupling ] and eq .
[ [ delta - m - decoupling ] ] ) , and then evolving them to today .
it follows that the baryonic effects tend to suppress the matter perturbations seeded before the decoupling epoch ( @xmath299 , see the horizontal dashed line in figure [ fig - k - eta ] ) and on scales inside the sound horizon ( i.e. for @xmath300 , see the vertical dashed line in figure [ fig - k - eta ] ) . the perturbations seeded after @xmath122 or on scales @xmath301 will not be affected by the baryons . with this argument , we can impose a suppression factor on our current solution ( [ gc ] ) to account for the effect of baryons , i.e. the solution with the inclusion of baryons can be written as @xmath302 where @xmath303 accounts for the baryonic suppression : @xmath304 where @xmath305 is the usual standard cdm transfer function with the baryonic dependence .
one accurate fit of @xmath305 is provided in ref .
we note that the ratio @xmath306 is unity outside the sound horizon ( @xmath307 ) , and is less than unity inside the sound horizon .
referring to figure [ fig - k - eta ] , equation ( [ b_k ] ) means that the value of @xmath303 is less than unity in the region to the right and above the dashed lines , and is unity otherwise .
we also note that in the low-@xmath308 models , the sound horizon can be smaller than the radiation - matter equality horizon , i.e. , it is possible that @xmath309 @xcite .
in addition , there is a transition era ( @xmath310 ) which is not included in equation ( [ b_k ] ) .
this is because in this era the baryonic effects do not fully operate as in the regime @xmath311 so that a good fit is not trivial to obtain .
we have numerically verified equation ( [ b_k ] ) , although an accurate fit to the missing era @xmath310 has yet to be found .
the solutions we have obtained so far have assumed @xmath312 . for @xmath4 or @xmath2 ,
the growing behavior of the cdm perturbations departs from that of a flat @xmath1 model only at very late times in the matter era ( see later for a more detailed argument ) .
this allows us to apply a universal suppression factor on @xmath313 to account for the effects of curvature or @xmath0 : @xmath314 where @xmath128 is in units of @xmath315 , and @xmath316 is given by @xcite @xmath317}\ .\ ] ] in equation ( [ gcbol ] ) , the leading factor @xmath318 results from the fact that the ratio of scale factors @xmath319 is proportional to @xmath318 and that the green function @xmath320 is proportional to this ratio .
the factor @xmath321 accounts for the suppression of the linear growth of density perturbations in a @xmath4 or @xmath0-universe relative to an @xmath322 and @xmath323 universe @xcite ( also verified in ref .
the reason for @xmath128 to have the unit @xmath315 in equation ( [ gcbol ] ) is that the horizon size at radiation - matter equality @xmath294 is proportional to @xmath324 ( see eq .
[ [ eta_eq ] ] in appendix [ background_cosmologies ] ) . for @xmath4 or @xmath2 , the extrapolation scheme ( [ gcbol ] )
will be inaccurate when @xmath174 is close to @xmath152 , i.e. , when the background dynamics at @xmath174 significantly departs from that of a flat @xmath1 model .
nevertheless , this extrapolation scheme is still appropriate for most models with active source for two reasons .
first , in the context of cosmic defects , the power of matter perturbations on the scales of our interest ( @xmath325@xmath326 ) is mainly seeded around @xmath294 ( see figure [ scalefactor ] and the discussion after eq .
[ [ tc0-i ] ] ) .
at this time , the curvature or @xmath0 effects are negligible .
second , at late times when the curvature or @xmath0 effects become important , these scales of our interest are already well inside the horizon so that any curvature terms in the perturbation equations can be neglected . therefore , the only required change in the perturbation equations to account for the effects of curvature or @xmath0 is simply to incorporate the correct background dynamics , and this involves only modifications in @xmath327 , @xmath328 and @xmath329 , whose solutions are given in appendix [ background_cosmologies ] .
as can be seen in figure [ scalefactor ] , the presence of curvature or a cosmological constant affects the background dynamics only at late times .
more precisely , we verify that for @xmath330 and @xmath331 , the largest observable scale for matter perturbations @xmath332 corresponds to the horizon sizes at @xmath333 respectively , whereas in these models the curvature or cosmological - constant domination occurs only at a much later epoch when @xmath334 . at these moments ( @xmath333 )
, the scale factor in the @xmath4 or @xmath2 models departs from that in the flat @xmath1 model only by less than one percent .
indeed , we have numerically verified that the extrapolation scheme ( [ gcbol ] ) is accurate within a 5% error for @xmath335 and @xmath336 in @xmath0-models , for @xmath337 and @xmath338 in open @xmath1 models , and for @xmath339 and @xmath340 in closed @xmath1 models .
these ranges of cosmological parameters have apparently covered the values of our interest .
with the green - function solutions we have found , we can now analytically investigate some important aspects about the growth of cosmological matter perturbations .
first we investigate the relationship between our green functions and the standard cdm transfer function , and thereby to justify the use of the standard cdm transfer function in the analytic solution ( [ gc ] ) . in the standard cdm model ,
there are no subsequent perturbations , so we have @xmath341 . as discussed in equations ( [ rad - superh ] ) and ( [ mat - superh ] ) , we also know that the cdm perturbations have a growing mode @xmath342 on super - horizon scales ( @xmath79 ) for @xmath343 or @xmath344 . for the super - horizon modes in the radiation era and all modes in the matter era ,
this allows us to write @xmath345 where @xmath346 is the coefficient of the growing mode in the radiation era ( @xmath347r : @xmath344 and @xmath79 ) or in the matter era ( @xmath347 m : @xmath343 ) . thus using our green - function solutions ( [ delta_i_n-2 ] ) and ( [ gc ] ) with the initial conditions @xmath59 and @xmath348 as required by the adiabatic inflationary model
, we can derive the standard cdm transfer function as @xmath349 \nonumber\\ & = & \frac{1}{4\eta_{\rm eq}^2 } \left [ a^2\eta_{\rm i}^2\widetilde{t}^{\rm c}_{30\rm(i)}+ 2a^2\eta_{\rm i } \widetilde{t}^{\rm c}_{40\rm(i ) } \right]t(k ) = \frac{2}{5}t(k)\ , , \label{transteta}\end{aligned}\ ] ] where we have used @xmath350 and equations ( [ aflat ] ) , ( [ t_limit ] ) , ( [ tc3-propto ] ) and ( [ tc4-propto ] ) , and the last expression was obtained based on the formalism ( [ gc ] ) .
first , we note that the two terms involving @xmath351 and @xmath352 are equal , meaning that the two sets of initial perturbations @xmath270 $ ] and @xmath271 $ ] contribute equally to the present matter perturbations .
second , the @xmath353 here is nothing but the standard cdm transfer function which we have defined earlier .
third , the coefficient @xmath354 in the final result of equation ( [ transteta ] ) is well known ( e.g. @xcite ) , and here we obtained it using our green - function solutions .
this coefficient can be also obtained by first knowing from equation ( [ density - eqn3 ] ) that @xmath100 is a constant on super - horizon scales ( @xmath111 ) , and then using its definition ( [ tau00 ] ) and equation ( [ a_j ] ) to compare its expressions for @xmath347r , m. one will find @xmath355 , which implies @xmath356 for @xmath111 .
thus the above derivation and result not only illustrate the relation between our green functions and the the standard cdm transfer function @xmath353 , but also justify the use of @xmath353 in our formalism ( [ gc ] ) .
one important problem for structure formation with causal seeds is to investigate how the source energy was compensated into the radiation and matter background when the seeds were formed at @xmath115 . from the result ( [ tau00-propto ] ) we know that the power spectrum of the pseudo energy @xmath357 should decay as @xmath58 on super - horizon modes .
as argued in equation ( [ initial - condition ] ) , we can thus take @xmath358 as part of the initial conditions provided that the scales of interest are well outside the horizon initially . for similar reasons we can take @xmath359 , where @xmath360 .
in addition , from equation ( [ tau00 ] ) without baryons , we have @xmath361 since @xmath362 is required at @xmath115 , it follows that for a given @xmath363 , one can have different ways of compensating it into between @xmath108 and @xmath364 .
it is thus vital to check the dependence of the resulting @xmath365 on the way we compensate @xmath363 into the background initially .
consider the following two extreme cases , both satisfying @xmath366 on super - horizon scales at @xmath115 : 1 .
[ case1 ] @xmath367 , @xmath368_{\rm i}$ ] : using equation ( [ delta_i_n-2 ] ) , the normalized resulting initial perturbations can be calculated as @xmath369_{\rm i } } = \widetilde{\cal g}^{\rm c}_4 = \widetilde{\cal g}^{\rm cs}\,.\ ] ] 2 .
[ case2 ] @xmath370_{\rm i}$ ] , @xmath371 : similarly we have @xmath372_{\rm i } } = \widetilde{\cal g}^{\rm c}_{3 } \frac{2w(w^2 - 1)}{a(3w^2 + 1 ) } \,.\ ] ] to see the difference in @xmath373 today resulting from these two cases , one can calculate @xmath374 where we have used equations ( [ tcs0 ] ) , ( [ tc30 ] ) and ( [ gc ] ) .
this implies that no matter how the source @xmath363 is compensated into the background when it was formed ( i.e. with any portions between @xmath108 and @xmath364 initially ) , it results in the same @xmath373 today on scales which were outside the horizon at @xmath115 .
we note that this independence of the initial conditions was first numerically observed in ref .
@xcite , and here we have provided an analytic proof . with a complete set of green functions for both initial and subsequent perturbations , we can now investigate the resulting total cdm perturbations and therefore the compensation mechanism in models with active source .
having seen the independence of the resulting @xmath375 on the way the source energy is initially compensated into various background components , we can invoke equation ( [ delta1 ] ) for @xmath376 , and equation ( [ delta_s_n ] ) for @xmath377 to obtain @xmath378 .
for a given mode at which @xmath379 initially , we have : @xmath380 \label{delta_comp1}\\ & = & \frac{8\pi g a_0}{5a^2 a_{\rm eq } } \left [ - t(k ) \widetilde\theta_{00}({\bf k } , \eta_{\rm i } ) + \int_{\eta_{\rm i}}^{\eta_0 } { t'}(k ; \hat{\eta } ) \frac{\dot{a}(\hat{\eta})}{a(\hat{\eta } ) } \widetilde\theta_+({\bf k } , \hat{\eta } ) d\hat{\eta } \right ] \label{delta_comp2}\\ & = & \frac{8\pi g a_0}{5a^2 a_{\rm eq } } \left\ { - t(k ) { \widetilde\theta_{00}({\bf k } , \eta_0 ) } + \right .
\nonumber\\ & & \left . \int_{\eta_{\rm i}}^{\eta_0 } \left [ { t'}(k ; \hat{\eta } ) \frac{\dot{a}(\hat{\eta})}{a(\hat{\eta } ) } \widetilde\theta_+({\bf k } , \hat{\eta } ) + t(k ) \dot{\widetilde\theta}_{00}({\bf k } , \hat{\eta } ) \right ] d\hat{\eta } \right\ } , \label{delta_comp3}\end{aligned}\ ] ] where @xmath381 and @xmath382 is given by ( [ gc ] ) .
the function @xmath383 is plotted in figure [ fig - tk - prime ] .
+ here we notice that the quantities inside the outer most brackets in equations ( [ delta_comp2 ] ) and ( [ delta_comp3 ] ) are equivalent to nothing but the coefficient of the growing mode in cdm perturbations . using equation ( [ delta_comp2 ] )
, one can obtain the resulting perturbations @xmath384 by knowing the initial @xmath385 and integrating the evolution history of @xmath386 .
hence this expression is convenient for numerical purposes .
in addition , we see that the first term in equation ( [ delta_comp2 ] ) comes simply from the initial source energy , serving with an opposite sign to account for energy conservation .
this is the so - called compensation . on the other hand , the second term results from the subsequent evolution of @xmath386 , which actively creates the cdm density perturbations on sub - horizon scales ( see later ) .
this term also provides a way for defects to create non - gaussianity . alternatively , equation ( [ delta_comp3 ] ) provides a both physically and mathematically transparent way of interpreting how the perturbations are seeded by the source .
first consider the integral term for a given mode @xmath128 .
when the mode is well outside the horizon , i.e. @xmath387 , @xmath388 equals @xmath353 by definition .
hence the two terms inside the inner brackets reduce to @xmath389 due to source stress - energy conservation ( [ conserve - seed1 ] ) . since the power spectrum of @xmath390 falls off as @xmath58 outside the horizon ( see eq . [ [ theta0i - propto ] ] ) , we expect the quantity inside the brackets to be negligible until the given mode approaches horizon crossing . near horizon
crossing , @xmath390 is no longer small , and @xmath388 starts departing from @xmath353 ( i.e. @xmath391 constant @xmath392 , see figure [ fig - tk - prime ] ) , so the two terms inside the inner brackets begin to contribute to the integral .
this also explains why the correction function @xmath283 in equation ( [ gc ] ) is important in affecting the compensation scale .
after horizon crossing , the significance of the two terms inside the inner brackets then depends on the subhorizon behaviours of their power spectra . as for the first term in equation ( [ delta_comp3 ] )
, we see that for a superhorizon mode today , the integral in ( [ delta_comp3 ] ) is negligible as argued above so that only the first term contributes .
it serves to give the opposite sign to the source energy so as to account for energy conservation on superhorizon scales today , and thus for the compensation at the present epoch @xmath152 . on the other hand ,
if a given mode is well inside the horizon today , then the first term will be negligible provided that the source energy @xmath393 has a power - law fall - off inside the horizon , as it does for cosmic strings .
therefore in calculating cdm perturbations on scales of our interest , which are well inside the horizon today , the first term in equation ( [ delta_comp3 ] ) is negligible , so that it will not affect our compensation argument observed from the integral .
this argument can be further strengthened by deriving the pseudo - energy today . from the definition of @xmath100 ( [ tau00 - 2 ] ) and the final result of equation ( [ delta_comp3 ] ) , one obtains @xmath394 d\hat{\eta}\ , .
\label{tau00_0}\end{aligned}\ ] ] from this result , one can clearly see that for super - horizon modes , @xmath353 is unity by definition so that only the integral survives .
we have also seen from an earlier argument that on super - horizon scales , the quantity inside the square brackets is nothing but the @xmath390 , which has a @xmath58 fall - off power spectrum ( see eq .
[ [ theta0i - propto ] ] ) .
it follows immediately from equation ( [ tau00_0 ] ) that the pseudo - energy today , @xmath395 , has a @xmath58-decay power spectrum outside the horizon .
this result confirms the super - horizon behavior of @xmath100 presented in equation ( [ tau00-propto ] ) . on the other hand
, although @xmath396 is approximately unity for sub - horizon modes , the usual sub - horizon power - law decay in @xmath393 ( as in the case of cosmic strings ) will still make the first term in equation ( [ tau00_0 ] ) negligible inside the horizon .
thus we can see explicitly in a neat mathematical form how compensation acts on a given length - scale . from this analysis
we can also see that the compensation scale is determined not only by the functions @xmath388 and @xmath353 , but also by the properties of the source near the horizon scale .
once the detailed behavior of the source near the horizon scale is known , we can accurately locate the compensation scale using equation ( [ delta_comp3 ] ) or ( [ tau00_0 ] ) .
we note that this result is different from the claim in ref .
@xcite , where multi - fluid compensation back - reaction effects were studied to show that the compensation scale arises naturally and uniquely from an algebraic identity in the perturbation analysis . ref .
@xcite also investigated the compensation scale , and found constraints on the generation of super - causal - horizon energy perturbations from a smooth initial state , under a simple physical scheme .
the compensation wavenumber was found to be constrained with @xmath397 due to causality , depending on the behavior of the causal events .
this result is not inconsistent with our finding above , where we further provide a quantitative way to locate the compensation scale for any given specific model .
in this paper we present a formalism which can be used to study the evolution of cosmological perturbations in the presence of causal seeds . in this formalism
we invoked the fluid approximation in the synchronous gauge to model the contents of the universe , and assumed photon - baryon tight coupling until the last - scattering epoch to account for the baryonic effects .
the approximation of instantaneous decoupling of photons and baryons was then employed at the last - scattering epoch . in particular
, we demonstrated the accuracy of our formalism in the context of the standard cdm model , by comparing our results of density perturbations with those calculated from cmbfast .
we then derived the analytic solutions of matter density perturbations in a flat @xmath1 cosmology .
the errors in ref .
@xcite were corrected to yield a complete set of green - function solutions for the super - horizon and sub - horizon modes ( eqs .
[ [ deltan ] ] , [ [ delta_i_n ] ] , [ [ delta_s_n ] ] , [ [ gcsinf]][[tc0-i ] ] ) . the degeneracy among these green functions was then found by comparing their initial conditions and employing the zero - entropy initial condition ( eqs .
[ [ gcs-2 ] ] , [ [ delta_i_n-2 ] ] ) .
this effectively reduces the number of the green functions needed in the perturbation solutions ( eqs . [ [ deltan ] ] , [ [ delta_s_n ] ] , [ [ delta_i_n-2]][[tc4-propto ] ] ) . with this great simplification , the solutions on intermediate scales
were then easily found by the use of the standard cdm transfer function ( eq . [ [ gc ] ] ) .
this complete set of solutions were numerically verified to high accuracy .
the baryonic effects were also considered ( eq . [ [ gcb ] ] ) .
we then extrapolated these green - function solutions to @xmath4 or @xmath2 models ( eq . [ [ gcbol ] ] ) , with numerical justifications to high accuracy . using these green - function solutions
, we investigated several important aspects of structure formation with causal source .
we first demonstrated the relation between our green functions and the standard cdm transfer function ( eq . [ [ transteta ] ] ) .
second we proved that the resulting matter perturbations today is independent of the way the source was initially compensated into the background contents of the universe ( eq . [ [ d12 ] ] ) . with our green - function solutions and the use of the pseudo - stress - energy tensor
, we finally addressed the compensation mechanism in a mathematically and physically explicit way ( eqs .
[ [ delta_comp2 ] ] , [ [ delta_comp3 ] ] , [ [ tau00_0 ] ] ) .
in particular , the compensation scale was shown to be dependent not only on the dynamics of the universe , but also on the properties of the source near the horizon scale . once given the detailed behavior of the source near the horizon scale
, the compensation scale can be accurately located using our green functions ( eq . [ [ tau00_0 ] ] ) .
although in the literature , there have been detailed treatments of theories with causal seeds , the formalism and its analytic solutions presented here will provide not only a physically transparent way for understanding the evolution of matter perturbations , but also a computationally economical scheme which is particularly pertinent when one needs to investigate the phase information of the resulting cosmological perturbations .
following the same line of development , we have been also working on the analytic solutions for radiation perturbations @xcite , which will be useful in computing the full - sky cmb anisotropies seeded by topological defects .
finally , we note that although we have been concentrating on investigating the perturbations with causal source , our green - function solutions are completely general and therefore can be also applied to the study of models with acausal source .
* acknowledgments* we thank andrew liddle , paul shellard , radek stompor , and neil turok for useful conversations .
we acknowledge support from nsf kdi grant ( 9872979 ) and nasa ltsa grant ( nag5 - 6552 ) .
a. h. guth , phys .
* d23 * , 347 ( 1981 ) for a review see a. vilenkin & e. p. s. shellard , _ ` cosmic strings and other topological defects ' _ , cambridge university press , cambridge ( 1994 ) p. p. avelino , e. p.
s. shellard , j. h. p. wu & b. allen , * 81 * , 2008 ( 1998 ) p. p. avelino , e. p. s. shellard , j. h. p. wu & b. allen , * 507 * , l101 ( 1998 ) j. h. p. wu , p. p. avelino , e. p. s. shellard & b. allen , _ ` cosmic strings , loops , and linear growth of matter perturbations ' _ ( see astro - ph/9812156 ) u. l. pen , u. seljak & n. turok , * 79 * , 1615 ( 1997 ) a. albrecht , r. a. battye & j. robinson , * 79 * , 4736 ( 1997 ) r. a. battye , a. albrecht & j. robinson , * 80 * , 4847 ( 1998 ) c. contaldi , m. hindmarsh & j. magueijo , * 82 * , 679 ( 1999 ) p. p. avelino , r. r. caldwell & c. j. a. p. martins , phys .
rev . * d56 * , 4568 ( 1997 ) b. allen , r. r. caldwell , s. dodelson , l. knox , e. p. s. shellard & a. stebbins , * 79 * , 2624 ( 1997 ) r. durrer , m. kunz & a. melchiorri , phys .
* d59 * , 123005 ( 1999 ) u. seljak & m. zaldarriaga , * 469 * , 437 ( 1996 ) w. hu , u. seljak , m. white & m. zaldarriaga , phys . rev . * d57 * , 3290 ( 1998 ) w. hu & m. white , phys
. rev .
* d56 * , 596 ( 1997 ) s. veeraraghavan & a. stebbins , * 365 * , 37 ( 1990 ) u. pen , d. spergel & n. turok , phys .
* d49 * , 692 ( 1994 ) we thank n. turok for providing the idea of spliting the photon and baryon components here .
d. j. eisenstein & w. hu , * 496 * , 605 ( 1998 ) w. hu & n. sugiyama , * 444 * , 489 ( 1995 ) w. hu & n. sugiyama , * 471 * , 542 ( 1996 ) j. silk , * 151 * , 459 ( 1968 ) r. a. battye , phys . rev .
* d55 * , 7361 ( 1997 ) j. r. bond & g. efstathiou , * 285 * , l45 ( 1984 ) j. h. p.
wu , in preparation .
r. sunyaev & ya .
b. zeldovich , ap&ss , * 7 * , 3 ( 1970 ) w. press & e. vishniac , * 236 * , 323 ( 1980 ) s. m. carroll , w. h. press & e. l. turner , annu .
astrophys . * 30 * , 499 ( 1992 ) d. j. eisenstein , _ ` an analytic expression for the growth function in a flat universe with a cosmological constant ' _ ( see astro - ph/9709054 ) c. cheung & j. magueijo , phys .
* d56 * , 1982 ( 1997 ) j. robinson & b. d. wandelt , phys .
* d53 * , 618 ( 1996 ) a. a. penzias & r. w. wilson , * 142 * , 419 ( 1965 ) k. c. jacobs , * 215 * , 1156 ( 1967 ) j. m. cohen , * 216 * , 249 ( 1967 ) a. d. chernin , * 220 * , 250 ( 1968 ) c. b. g. mcintosh , mon . not .
r. astron . soc .
* 140 * , 461 ( 1968 ) e. w. kolb & m. s. turner , _ ` the early universe ' _ , addison - wesley , redwood city , california ( 1990 )
with the discovery of the cmbr in 1964 @xcite , the universe is believed to be mainly composed of not only matter but also radiation . after the discovery , several authors worked out the solutions in some frw models with both radiation and matter @xcite . in this appendix
, we aim to derive the general solution of frw models , in the presence of both curvature and a cosmological constant .
we assume that the universe is homogeneous and isotropic , and is filled with two fluids , radiation and dark matter , whose stress - energy tensors are also homogeneous and isotropic on average .
we shall ignore the overall contribution of the stress energy from causal seeds like defect fields , because in general they are much smaller than the total energy density of radiation and matter .
thus in a frw universe with only radiation and matter components that evolve independently and adiabatically , the scale factor @xmath327 is determined by the unperturbed einstein equation , or equivalently the friedmann equation : @xmath398 where a dot represents a derivative with respect to the conformal time @xmath170 , @xmath5 is the curvature , @xmath399 is the matter energy density , @xmath0 is the cosmological constant , and we have normalized @xmath400 .
if we define @xmath401 where @xmath402 is the hubble parameter , then we have from ( [ fried ] ) that @xmath403 and @xmath404 we also notice that @xmath405 .
we define @xmath406 where we note that @xmath407 due to @xmath408 and @xmath409 according to the current observational results .
thus we can rewrite equation ( [ fried ] ) as @xmath410 where @xmath411 equation ( [ eta_a ] ) can then be numerically evaluated with certain choices of @xmath308 , @xmath412 and @xmath413 . assuming three species of neutrinos and using @xmath414gev@xmath415 @xcite and the fact that at @xmath294 both the curvature and the cosmological constant terms are negligible in ( [ fried ] )
, we obtain @xmath416 where @xmath294 is in the units measured today . in certain cases , ( [ eta_a ] )
can be exactly solved : 1 . @xmath417 ( i.e. @xmath418 ) : @xmath419 which give @xmath420 .
2 . @xmath421 ( i.e. @xmath422 ) : @xmath423 \ , , \label{aopen } \\
t(\eta ) & = & \frac{1}{\bar{a}b}\left [ \cosh{(\bar{a}\sqrt{b}\eta ) } + \frac{1}{2\sqrt{b}}\sinh{(\bar{a}\sqrt{b}\eta ) } - \frac{\bar{a}\eta}{2 } - 1 \right ] \ , . \label{topen } \end{aligned}\ ] ] 3 .
@xmath424 ( i.e. @xmath425 ) : @xmath426 \ , , \label{aclose}\\ t(\eta ) & = & \frac{1}{\bar{a}b}\left [ \cos{(\bar{a}\sqrt{-b}\eta ) } + \frac{1}{2\sqrt{-b}}\sin{(\bar{a}\sqrt{-b}\eta ) } - \frac{\bar{a}\eta}{2 } -1 \right ] \ , .
\label{tclose } \end{aligned}\ ] ] we notice that at early times equations ( [ aopen][topen ] ) and ( [ aclose][tclose ] ) reduce to equations ( [ aflat][tflat ] ) . at late times equations ( [ aflat][tflat ] ) , ( [ aopen][topen ] ) and ( [ aclose][tclose ] )
give the asymptotic forms @xmath427 or @xmath428 , & k>0,\ ; \lambda=0 .
\end{array } \right.\ ] ] figure [ scalefactor ] shows some examples of these solutions .
as we can see , the destinies of universes in different cosmologies diverge , although all have identical features around or before the radiation - matter equality @xmath429 .
this converging behavior at early times helps simplify the calculation of cosmological perturbations with causal source , since we know that this kind of perturbations are mainly contributed from the radiation - matter transition era .
|
we investigate matter density perturbations in models of structure formation with or without causal / acausal source . under the fluid approximation in the linear theory ,
we first derive full perturbation equations in flat space with a cosmological constant @xmath0 .
we then use green - function technique to obtain analytic solutions for matter perturbations in a flat @xmath1 model .
some incorrect solutions in the literature are corrected here . a simple yet accurate extrapolation scheme is then proposed to obtain solutions in curved or @xmath2 cosmologies .
some general features of these solutions are revealed . in particular
, we analytically prove that the resulting matter density perturbations are independent of the way the causal source was compensated into the background contents of the universe when it was first formed .
we also use our green - function solutions to investigate the compensation mechanism for perturbations with causal seeds , and yield a mathematically and physically explicit form in interpreting it .
we found that the compensation scale depends not only on the dynamics of the universe , but also on the properties of the seeds near the horizon scale .
it can be accurately located by employing our green functions .
.2 in
|
the top quark condensation is an attractive idea that explains the electroweak symmetry breaking in the absence of fundamental scalar bosons , and gives an understanding of that the top quark mass is of the order of the electroweak scale @xcite .
bardeen , hill and lindner ( bhl ) have proposed an interesting scenario for the top condensation @xcite . in their model ,
the renormalization group ( rg ) equations play an important role , and the composite nature of the higgs particle is reflected in the boundary conditions for the coupling constants of the standard model , namely , the compositeness conditions of bhl .
the usual standard model is considered to well describe physical phenomena in the low energies where the higgs particle can be regarded as elementary particle . on the other hand , if the higgs particle is composed of some elementary particle , the higgs - yukawa sector of the model will be useless in some high - energy region because the lagrangian written in terms of the local higgs field must be useless to describe the inside of the higgs particle . in the scenario of bhl ,
the higgs - yukawa sector becomes useless at a high - energy scale @xmath0 and above owing to the divergence of the yukawa coupling constant for the top quark at the scale .
besides , bhl have also supposed that the higgs field becomes a non - propagating auxiliary field at @xmath0 , and so the higgs - yukawa system is equivalent to a local four - fermi system at the scale .
these two phenomena can be considered to indicate the compositeness of the higgs particle .
using one of the compositeness conditions of bhl , one can predict the top quark mass depending on the value of @xmath0 .
unfortunately , the predicted masses are larger that the experimental value as for @xmath1 ( @xmath2 gev is planck scale ) , and several improvements and extensions of bhl s model have been studied @xcite .
our attempt to improve the model is very simple : we merely remove the condition that leads to the predictions contradicted by the experiment from the compositeness conditions .
we consider what occurs in this case , and reconsider the condition for the higgs particle to be the composite particle . in refs .
@xcite , it has been shown that if we use the @xmath3 functions of the mass independent renormalization scheme in bhl s model , we can not properly describe the spontaneous symmetry breaking . this is the same in our scenario for the composite higgs particle ( see appendix a ) .
ideally , we should use wilson s renormalization scheme @xcite as discussed in refs .
@xcite , however , the treatment of the rg equation is not so easy in this scheme .
actually , bhl have used the @xmath3 functions of the mass independent renormalization scheme in the continuous theory instead of ones of wilsonian s .
the validity of this substitution has been discussed in ref .
@xcite in leading order of the @xmath4 expansion , where @xmath5 is the number of colors .
there , it was concluded that this substitution is valid as an approximation except for the @xmath3 function for the mass parameter . in this paper , we will also define the renormalization transformation in our cutoff model with having the wilson renormalization approach in mind , and give the @xmath3 functions for the quartic ( @xmath6 ) and the quadratic ( @xmath7 ) coupling constants of the higgs self - interactions . to give them , we will use the perturbative expansion with respect to all the coupling constants including the yukawa coupling constant for the top quark , which does not diverge at @xmath0 in our scenario .
we will see that our results coincide with the statement of ref .
@xcite . on the basis of this observation , disregarded the problem of the gauge invariance ( see appendix a ) , we use the @xmath3 functions of the @xmath8 scheme for all the dimensionless coupling constants in this paper except for a few cases . in the next section ,
we review the compositeness conditions of bhl . in 3 , we attempt to remove the condition that leads to the quark mass predictions contradicted by the experiment .
the condition for the higgs particle to be the composite particle is reconsidered in this section . in 4 , predictions for the higgs boson mass are given .
section 5 is devoted to summary and discussions . in appendix
a , we define the renormalization transformation and give the @xmath3 functions for @xmath6 and @xmath7 with having wilson s renormalization approach in mind .
we have wilson s renormalization approach in mind .
the effective lagrangian density of the standard model at an energy scale @xmath9 will be @xmath10 where @xmath11 denotes the usual gauge invariant lagrangian for the gauge and the fermion fields , @xmath12 is the higgs doublet , @xmath13 is the gauge covariant derivative and @xmath14 denotes the higher dimensional terms .
the yukawa interaction terms are @xmath15 where @xmath16 denote the generations , @xmath17 , @xmath18 are the left - handed quark doublets in which the up ( down ) type components are mass eigen states , @xmath19 are the right - handed up ( down ) type quark singlets , @xmath20 are the left - handed lepton doublets and @xmath21 are the right - handed lepton singlets .
now suppose that the running coupling constants behave as @xmath22 near @xmath23 , where @xmath24 denotes the coupling constant of the interaction term that is proportional to the power of @xmath25 of @xmath12 , @xmath26 is a finite constant , and the small factor @xmath27 vanishes at a high - energy scale @xmath0 : @xmath28 specifically , we write as @xmath29 where @xmath30 , @xmath31 and @xmath32 are finite constants ( @xmath33 ) .
in this case , we have @xmath34 near @xmath23 , where @xmath35 , @xmath36 is given by replacing @xmath37 and @xmath12 in eq .
( [ yukawa ] ) with @xmath38 and @xmath39 , respectively , and @xmath40 is given by replacing @xmath24 and @xmath12 in @xmath14 with @xmath26 and @xmath39 , respectively .
further , if @xmath41 the lagrangian reduces to @xmath42 for @xmath43 , where we have considered that @xmath44 and @xmath45 vanish or become negligible compared to @xmath46 . through the auxiliary field method
, we can show that the system of eq .
( [ reduce ] ) is equivalent to one of the four - fermi interaction supplemented with gauge interaction , @xmath47 therefore , we can describe the higgs particle by alternative lagrangian below @xmath0 : one is the higgs - yukawa lagrangian of the standard model , and another is the lagrangian for the fermions that has the local four - fermi interaction term with other higher dimensional terms generated through the renormalization transformation from @xmath0 . on the other hand , at @xmath0 and above , we can not use the higgs - yukawa lagrangian as follows . from eqs .
( [ compositeness0 - 1 ] ) and ( [ compositeness0 - 2 ] ) , we have @xmath48 which are the compositeness conditions for the running coupling constants proposed by bhl .
the condition ( [ composite condition1 ] ) means that the higgs - yukawa sector of the standard model becomes useless at @xmath0 and above . here
, we arrange the phenomena arising from the compositeness conditions of bhl . 1 .
the higgs - yukawa system of the standard model becomes equivalent to the local four - fermi interaction system at @xmath0 .
the higgs - yukawa sector of the standard model becomes useless at @xmath0 and above because the yukawa coupling constant for the top quark diverges at the scale .
the two phenomena can be considered to indicate the compositeness of the higgs particle . using the condition ( [ composite condition1 ] )
, we can predict the top quark mass depending on @xmath0 . we can solve . ]
@xmath49 using the condition as a boundary condition , where @xmath50 and the 1-loop @xmath3 function for @xmath51 is @xmath52 the gauge coupling constants @xmath53 obtained from 1-loop @xmath3 functions are @xmath54 all the @xmath3 functions used above are of the @xmath8 scheme in the continuous theory . because @xmath55 have been determined thanks to the condition ( [ composite condition1 ] ) , the top quark mass is given by @xmath56 where @xmath57 is the vacuum expectation value of the higgs field .
unfortunately , the results of the top quark mass given in this way are larger than the experimental value , @xmath58 gev .
actually , if we determine @xmath59 by solving eq .
( [ rge - at ] ) with the boundary condition obtained from the experimental value of the top quark mass , the condition ( [ composite condition1 ] ) does not hold as for @xmath60 .
for example , we give the running of @xmath51 for @xmath61 gev and @xmath62 gev in fig .
[ at ] . to this calculation
we have converted the pole mass @xmath62 gev into the @xmath8 mass @xmath63 gev using @xmath64 according to ref .
@xcite .
= 100 mm = 70 mm
we attempt to remove the condition ( [ composite condition1 ] ) from the compositeness conditions of bhl . in this case the remaining condition ( [ composite condition2 ] ) becomes @xmath65 even in the case , we can also path integrate out the higgs field at @xmath0 , and obtain an action having a non - local four - fermi interaction term , @xmath66 where @xmath67 and the repeated indices are summed and @xmath68 means the integration with respect to the repeated arguments .
we have defined : @xmath69 and @xmath70 where @xmath71 and @xmath72 and @xmath73 are the @xmath74 and @xmath75 gauge fields , respectively .
@xmath76 can be expressed as @xmath77 when the series converges , where @xmath78 when we give eq .
( [ non - local ] ) , we have also taken account of the yukawa coupling constants for the other fermions than the top quark . therefore , we can describe the higgs particle by using alternative actions at @xmath0 and below , one is the higgs - yukawa action of the standard model , and another is the action for the fermions that has the non - local four - fermi interaction term with other higher dimensional terms generated through the renormalization transformation from @xmath0 .
we note that the equation of motion for the higgs field at @xmath0 yields @xmath79 in our case .
now , what does occur above @xmath0 ? in the last section we saw that the condition ( [ composite condition1 ] ) means the breaking down of the higgs - yukawa system at @xmath0 due to the divergence of @xmath80 . in our case
the system does not break down at the scale in this sense because we have removed the condition ( [ composite condition1 ] ) , however , the system is breaking down above @xmath0 from the viewpoint of the vacuum stability .
let us determine the running of @xmath6 .
we can solve @xmath81 using eq .
( [ composite condition2 - 1 ] ) as a boundary condition , where @xmath82 is the 1-loop @xmath3 function for @xmath83 of the @xmath8 scheme ; we will confirm the validity of the use in appendix a. as an example , we give the running of @xmath83 for @xmath61 gev and @xmath62 gev ( @xmath84 gev ) in fig .
[ ah ] .
= 100 mm = 70 mm from the figure we can see that @xmath6 becomes negative above @xmath0 , and the higgs potential becomes unbounded from below .
this means that the states corresponding to the local minimum of the potential are unstable and the true vacuum state of the higgs - yukawa system can not be defined .
after all , when we remove the condition ( [ composite condition1 ] ) from bhl s compositeness conditions , the two phenomena occur : 1 .
the higgs - yukawa system of the standard model becomes equivalent to the non - local four - fermi interaction system at @xmath0 .
the higgs - yukawa sector of the standard model becomes useless above @xmath0 because the vacuum state can not be defined above the scale .
we consider that these indicate the compositeness of the higgs particle . here , we give two comments .
( 1 ) in bhl s scenario , the 1-loop @xmath3 functions obtained in the perturbation theory did not permit an extrapolation all the way to @xmath0 because @xmath51 diverges at @xmath0 .
in contrast to this , the perturbative expansion with respect to coupling constants is reliable near @xmath0 in our scenario because all the coupling constants are within the perturbative region near @xmath0 .
( 2 ) we have discussed the vacuum stability based on the `` tree potential '' of the higgs field assuming that higher dimensional terms in the effective potential of the higgs field are negligible compared to the `` tree potential '' for @xmath85 , where @xmath86 is the physical higgs field .
the verification of this assumption is not so easy in wilson s renormalization scheme .
we give a desired form for the effective lagrangian to make the assumption valid in appendix b. the removed condition ( [ composite condition1 ] ) was derived from the condition ( [ condition for z ] ) that coincides with another version of the compositeness condition discussed in many contexts @xcite .
if one identifies @xmath27 in the condition ( [ condition for z ] ) with the wave function renormalization constant of the higgs field , the condition also implies the compositeness of the higgs particle from the viewpoint of the following : the 1-particle pole part in the 2-point function of the higgs field vanishes at @xmath0 . in our scenario ,
the pole will not vanish at @xmath0
. then does our scenario have some significance for the compositeness of the higgs particle bearing comparison with that the condition ( [ condition for z ] ) possesses ?
let us consider this point from now .
one may consider that four - fermi interactions are low - energy effective interactions of some fundamental interaction .
for the top - condensation models , this idea is studied in the so - called topcolor model @xcite , in which the local four - fermi interaction is rewritten into a form of the effective current - current interaction through fierz transformation .
let us rewrite our non - local four - fermi interaction into such a form : @xmath87 where @xmath88 and the latin indices @xmath89 distinguish the components of the right - hand sides of eqs .
( [ l ] ) and ( [ r ] ) including the difference of the colors of the quarks , @xmath90 s are defined by @xmath91 with @xmath92 .
this current - current like interaction will be effectively produced by exchanges of heavy vector bosons represented by bi - local fields .
first , we consider @xmath93 where @xmath94 s are the bi - local vector fields that are @xmath75 singlets , @xmath95 is a parameter having mass dimension 2 . the interaction eq .
( [ new interaction ] ) is invariant under @xmath96 gauge transformations if the bi - local vector fields transform as @xmath97 where @xmath98 and @xmath99 are the matrices of the @xmath74 and the @xmath100 gauge transformations acting on @xmath101 , respectively .
we introduce the fourier transformation for the bi - local vector fields : @xmath102 where @xmath103 and @xmath104 , and so the bi - local fields are expressed by an infinite number of the local fields labeld by @xmath105 .
we suppose that @xmath106 in the interaction picture , where @xmath107 and @xmath108 represent @xmath109 and @xmath110 , respectively .
then , we have @xmath111 for @xmath112 , and the interaction @xmath113 will induce @xmath114 of eq . ( [ current ] ) in this case . on the other hand , we can rewrite @xmath154 of eq .
( [ reduced2 ] ) into @xmath156 through fierz transformation , which becomes the original expression of @xmath157 of eq .
( [ non - local term ] ) for @xmath153 .
the form of eq .
( [ another form of b-4f ] ) suggests that @xmath158 is equivalent to some action written in terms of scalar fields .
we note @xmath159\nonumber \\ & & = itrln\left[k^{-1~ab : cd}_{ij : kl}(x_{1},x_{2}:y_{1},y_{2})\right],\label{path integral}\end{aligned}\ ] ] where @xmath160 s are the bi - local scalar doublets , and @xmath161 which satisfies @xmath162 . because the integrand of the path integral of the left - hand sides of eq .
( [ path integral ] ) is equal to @xmath163},\ ] ] where @xmath164,\end{aligned}\ ] ] we have @xmath165}=s_{\mbox{{\scriptsize b-4f}}}=\frac{1}{i}ln\int { \cal d}a{\cal d}a^{\ast}{\cal d}c{\cal d}c^{\ast}\exp{\left[is_{\mbox{{\scriptsize bv}}}\right]},~\ ] ] where @xmath166.\label{chi}\ ] ] thus , @xmath155 is equivalent to the action for the bi - local scalar fields @xmath167
. it will be interest to see what @xmath167 becomes for @xmath153 .
when we set @xmath168 to @xmath169 , i.e. , to be independent of @xmath170 , @xmath171 and @xmath172 become the kinetic term and the yukawa interaction terms of the higgs - yukawa sector of the standard model , respectively .
we have given an example of the action for the physics above @xmath0 , @xmath155 or @xmath167 .
the actions are equivalent to the action for fermions having @xmath154 that becomes the higgs - yukawa sector of the standard model at @xmath0 in the limit @xmath153 becomes equivalent to the sector in this limit , where @xmath173 is of eq .
( [ det ] ) .
so , the action above @xmath0 is @xmath174 or @xmath175 . ] .
it is important to note that @xmath155 or @xmath167 is not exactly equivalent to the action for the fermions having @xmath114 but induces it effectively .
if they are exactly equivalent , then we can describe the physics above @xmath0 using the higgs - yukawa sector of the standard model alternatively , because @xmath176 is equivalent to the sector .
then the vacuum of the system will be ill defined above @xmath0 .
it should be emphasized that @xmath155 could not be rewritten into the action for the single elementary local scalar field but the bi - local scalar fields .
in general , although we can not rewrite an action into another one through the auxiliary field method , the two actions might be equivalent to each other in the sense that the vertex functions obtained from both actions are identical with each other limit . ] .
however , there will exist no action for a single elementary local scalar field that is equivalent to @xmath155 , because it has been already shown that @xmath155 is equivalent to action for the bi - local scalar fields that are represented by an infinite number of local fields like eq .
( [ infinite number of fields ] ) .
therefore , we can say that the scale @xmath0 is the critical scale above which the higgs particle can not be described by the action written in terms of the elementary local field for the higgs particle but described by one for the constituent particles of the higgs particle or some action for bi - local scalar fields . from this
, we consider that the scale @xmath0 has a qualification of the compositeness scale .
now , we predict the higgs boson mass in our scenario depending on @xmath0 . the mass is obtained from @xmath177 because @xmath178 has been determined thanks to the condition ( [ composite condition2 - 1 ] ) .
the results are shown in table i. table i. the higgs boson masses obtained from eq .
( [ higgs mass ] ) .
@xmath179 s are defined by eq .
( [ pole ] ) .
the values having the superscript @xmath180 are obtained using eqs .
( [ bm ] ) and ( [ bl ] ) .
the results for @xmath181 gev are not reliable because our @xmath3 function is not reliable for @xmath182 ( see appendix a ) .
we must take notice that @xmath183 s are not the pole masses . [
cols="^,^,^,^",options="header " , ] for @xmath184 gev , we have used the @xmath3 functions of our cutoff scheme that will be defined in appendix a , namely , eqs . ( [ bm ] ) and ( [ bl ] ) for @xmath184 gev , the results does not differ from the values in table i more than @xmath185 gev . ] . on the other hand , for @xmath186 gev
, we have used eq .
( [ beta - a - lambda ] ) approximately .
this is because a problem of fine - tuning arises for @xmath187 when we use the @xmath3 functions of our cutoff scheme ( see appendix a ) .
we will show that our @xmath3 function for @xmath83 reduces to @xmath188 under an approximation .
we will also see that our @xmath3 functions are not reliable for @xmath182 , and so the results for @xmath189 gev in table i are not reliable .
we must take notice that @xmath183 s are not the pole masses .
the pole mass @xmath190 and @xmath183 are related by @xmath191 where @xmath192 depends on @xmath190 @xcite . from the results shown in ref .
@xcite , we can see that @xmath192 decreases with increasing @xmath190 for the values of @xmath183 in table i. ] , and so the pole masses will be smaller than @xmath183 . to conclude this section , we give two comments .
( 1 ) our way used to predict the higgs boson mass is the same with one to yield the ordinary stability bound of the mass @xcite , except that we treat our model as a cutoff model . in ordinary discussion of the stability bound , the lower limits on the mass
are derived from the requirement that the higgs potential is bounded below up to a certain scale where some new physics appears .
the compositeness scale @xmath0 of our scenario corresponds to such a scale .
the value of @xmath0 is a free parameter of our model , however , we consider that @xmath0 is not so large compared to @xmath193 because if @xmath194 , we encounter a problem of the fine - tuning as we will detail in appendix a. this problem can be considered a kind of the problem of hierarchy .
the absence of such a fine - tuning may require @xmath195 @xcite .
( 2 ) the higgs boson mass is sensitive to the variation of @xmath196 relative to one of @xmath0 .
this seems to support the idea that the higgs particle is composed of the top quarks .
the compositeness conditions of bhl , eqs .
( [ composite condition1 ] ) and ( [ composite condition2 ] ) lead to the two phenomena ( i ) and ( ii ) in 3 .
these are considered to indicate the compositeness of the higgs particle .
now that the experimental value of the top quark mass contradicts the condition ( [ composite condition1 ] ) , we must give up the scenario of bhl without some improvement .
in this paper we considered what occurs when we remove eq .
( [ composite condition1 ] ) . if one removes eq .
( [ composite condition1 ] ) from bhl s compositeness conditions , the remaining condition ( [ composite condition2 ] ) becomes eq .
( [ composite condition2 - 1 ] ) .
this condition leads to the two phenomena ( i ) and ( ii ) in 3 .
we consider that these , together with each other , also indicate the compositeness of the higgs particle
. the phenomenon ( i ) may not be so surprising one . in refs .
@xcite , it is shown in the large @xmath5 limit that even in the absence of any conditions for @xmath197 there exists the generalized nambu - jona - lasinio ( njl ) system to be equivalent to the higgs - yukawa system in the sense that the vertex functions obtained in both systems are identical with each other .
if the equivalence is exact , we should say that eq .
( [ composite condition2 - 1 ] ) is the condition to guarantee the statement ( ii ) , and then the condition also determine the type of the generalized njl lagrangian that is equivalent to the higgs - yukawa lagrangian .
the condition ( [ condition for z ] ) , which arises the removed condition ( [ composite condition1 ] ) , can be considered the compositeness condition also from the viewpoint of that the 1-particle pole part in the 2-point function of the higgs field vanishes at @xmath0 . in our scenario ,
the pole will not vanish at @xmath0 , and so we can not say @xmath0 to be the scale at which the asymptotic field for the higgs particle @xmath198 vanishes .
however , the scale @xmath0 is considered to be the compositeness scale of the higgs particle from another point of view as follows . in the ordinary discussion of the vacuum stability
, it would be considered that some new physics appears before the higgs potential becomes unbounded below , and the higgs - yukawa system supplemented with some new interaction does not break down . in our scenario
we consider that the higgs - yukawa system breaks down above @xmath0 , and the physics above the scale is described by the action that does not have the higgs - yukawa sector of the standard model .
this is naturally understood when we consider the non - local four - fermi interaction eq .
( [ non - local term ] ) to be a low - energy ( of course higher than @xmath0 ) effective interaction produced by exchanges of heavy vector bosons , which are represented by the bi - local fields in our scenario because of the non - locality of the four - fermi interaction .
then even if we rewrite the action for the vector fields into an action for the scalar fields , the action will not be for a single local scalar field but for the bi - local scalar fields as discussed in 3 .
therefore , we can say that the scale @xmath0 is the critical scale above which the higgs particle can not be described by the action written in terms of the local elementary field for higgs particle but described by one for the constituent particles of the higgs particle or some action for bi - local scalar fields . contrasting with this , at @xmath0 and below
, the higgs particle can be described by alternative actions , one is the action for the constituent particles of the higgs particle , namely , the action having the non - local four - fermi term eq .
( [ non - local term ] ) with higher dimensional terms generated through the rg transformation from @xmath0 , and another is the higgs - yukawa action , which is of course written in terms of the elementary local field for the higgs particle . from this point of view , we consider that the scale @xmath0 has a qualification for the compositeness scale . now
, we note that if the equivalence between some generalized njl system and the unconstrained higgs - yukawa system discussed in refs .
@xcite is exact , we can say as follows in our criterion for the compositeness : the higgs particle described by the higgs - yukawa sector is the composite particle whenever the sector breaks down at a high - energy from any reason .
so , it would appear that the higgs particle is already fated to be the composite particle when we adopt the higgs - yukawa system to describe the low - energy physics of the higgs particle or the symmetry breaking of the gauge theory .
it remains to be shown whether the vacuum of the system of @xmath155 or @xmath167 given in 3 is well defined or not .
we will not address to this question here @xcite .
our example of the new theory above @xmath0 may be cheap a little to consider it a realistic model ; it is sufficient to explain our criterion for the compositeness of the higgs particle , though .
we may be able to construct the new theory above @xmath0 based on some gauge symmetry that requires the bi - local gauge fields @xcite .
then , the following should be required in general .
( 1 ) the vacuum is well defined above @xmath0 .
( 2 ) the masses of the bi - local vector fields become large as the energy scale approaches to @xmath0 . (
3 ) the new theory breaks down at @xmath0 and below .
detailed studies of the theory above @xmath0 remains as future works .
the author expresses his sincere thanks to prof .
s. naka for useful comments and discussions .
he is grateful to other members of his laboratory for their comments and encouragement .
he thanks prof .
h. nakano for pointing out ref .
he also thanks dr . c. itoi for discussions on the vacuum stability in wilson s renormalization approach .
* appendix *
in this appendix , we define the renormalization transformation for the action and give the @xmath3 functions of our model with having wilson s renormalization approach in mind .
our definition is about the same with one of ref .
@xcite . in ref .
@xcite , the @xmath4 expansion was used , and then one can elude the problem of the gauge invariance in the cutoff theory accidentally .
we will use the perturbative expansion with respect to coupling constants ; we can use the perturbation in our scenario as we have mentioned in 3 . in our case the problem of the gauge invariance will be serious and remains unsolved in this paper .
for this reason we give the @xmath3 functions only for @xmath6 and @xmath7 , and assume that for the other coupling constants , the @xmath3 functions of the @xmath8 scheme in the continuous theory can be used approximately .
our definition of the renormalization transformation for the action , @xmath199 , is as follows .
now , we parameterize the higgs doublet as @xmath218 where @xmath219 are the would - be n.g . bosons and @xmath86 is the physical higgs boson .
we focus on @xmath86 by setting the other fields equal to zero from now .
then eq .
( [ 1-loop ] ) can be written as @xmath220=\int d^{4}x\left[-\delta z\frac{1}{2}\phi\partial^{2}\phi-\sum_{n=2,4,\cdots}^{\infty}{\widetilde \delta}\lambda_{n}\phi^{n}-\sum_{m , n}^{\infty}{\widetilde \delta}c_{mn}h(\phi^{n},\partial^{m})\right],\ ] ] where @xmath221 and @xmath222 denotes higher derivative terms .
step ii : one can consider that the momentum cutoff has decreased through step i such as @xmath223 , where @xmath224 . here , we introduce new momentum and coordinate variables , @xmath225 and @xmath226 .
note that the cutoff for @xmath227 is restored .
using @xmath228 , we can write as @xmath229+{\widetilde \delta } s[\phi]&=&\int d^{4}x'b^{-4}\biggl[-\frac{1}{2}b^{2}(1+\delta z)\phi\partial^{'2}\phi -\sum_{n=2,4,\cdots}^{\infty}(\lambda+{\widetilde \delta}\lambda_{n})\phi^{n}\nonumber \\ & & -\sum_{m , n}^{\infty}(c_{nm}+{\widetilde \delta}c_{nm})b^{m}h(\phi^{n},\partial^{'m})\biggr].\label{step 2}\end{aligned}\ ] ] step iii : we normalize the kinetic term on the right - hand side of eq .
( [ step 2 ] ) through the re - definition @xmath230 : @xmath231+{\widetilde \delta } s[\phi ] \to s[\phi]+\delta s[\phi],\ ] ] where @xmath232=\int d^{4}x\left\{-\sum_{n=2,4,\cdots}^{\infty}\delta\lambda_{n}\phi^{n}-\sum_{m , n}^{\infty}\delta c_{mn}h(\phi^{n},\partial^{m})\right\}\label{step 3}\ ] ] and @xmath233 we define @xmath234 $ ] as the variation of the action under the infinitesimal renormalization transformation in our cutoff model .
the variation of the full action under the infinitesimal renormalization transformation will not be gauge invariant , and so the effective lagrangian density will not be gauge invariant in general .
this is the problem of the gauge invariance in our cutoff model mentioned above and 1 .
the @xmath3 functions for @xmath7 and @xmath6 are @xmath235 where we have identified @xmath236 with @xmath237 because @xmath234 $ ] is the variation with decreasing @xmath9 . in the t hooft - landau gauge
, we have @xmath238 where @xmath239 we have calculated the @xmath3 functions neglecting higher dimensional terms in the effective action on the assumption that their contributions will be @xmath240 . because the calculations are not reliable for @xmath241 , our discussion based on the @xmath3 functions is not reliable for @xmath242 . however , we consider that @xmath0 is not so large compared to @xmath193 because a fine - tuning will be required if @xmath194 as we will see in the next paragraph .
let us try to determine the running of @xmath6 and @xmath7 using eqs .
( [ bm ] ) and ( [ bl ] ) .
because @xmath243 contains @xmath7 in the expression , we must solve the differential equations for @xmath6 and @xmath7 simultaneously . to solve them numerically ,
a boundary value for @xmath7 at @xmath0 is necessary because the boundary value for @xmath6 is given at @xmath0 , i.e. , eq .
( [ composite condition2 - 1 ] ) .
practically , we solve the differential equations with tuning the value of @xmath244 so as to obtain the solutions for @xmath197 and @xmath245 that satisfy the condition for the minimum , @xmath246 where @xmath57 is of eq .
( [ v ] ) gev should be consider to be one at @xmath247 because it is determined by the experiment of the muon decay @xcite . according to ref .
@xcite , we consider @xmath248 to be equal to @xmath249 approximately . ] .
when @xmath194 , the tuning for @xmath244 becomes hard .
this is the problem of fine - tuning mentioned above and in 4 . here
, we consider the case @xmath250 . in the leading order of the @xmath251 expansion , we have @xmath252 where @xmath253 is of eq . ( [ beta - a - lambda ] ) , and @xmath254 in this case we can determine @xmath6 independently of @xmath255 as performed in 3 because @xmath253 does not contain @xmath255 .
then , we can obtain a value of @xmath256 from eq .
( [ extremal ] ) without any fine - tuning , which can be used as a boundary value to determine the running of @xmath245 .
if we use @xmath257 with eq .
( [ extremal ] ) , we can see that @xmath7 is negative in @xmath258 .
this means that if we adopt the @xmath8 scheme in our scenario for the composite higgs particle , the spontaneous symmetry breaking is not properly described as mentioned in 1 .
finally , we check the consistency of the use of @xmath250 . as figs .
[ zmm4 ] and [ zmm4r ] show , @xmath251 is sufficiently small until the blowing down in the low - energy region appears .
= 68 mm = 47 mm = 68 mm = 47 mm we can see that @xmath259 in @xmath260 .
we have discussed the vacuum stability based on the `` tree potential '' @xmath261 of the higgs field assuming that higher dimensional terms can be neglected for @xmath85 . in this appendix , we give a desired form for the effective lagrangian to make our assumption valid . in the ordinary discussion of the vacuum stability , the coleman - weinberg potential is used with the rg improvement @xcite . in our discussion , however , we can not use it because we are treating our model as a cutoff model with having wilson s renormalization approach in mind .
the effective potential @xmath262 of our model will be obtained by solving the rg equation for the effective lagrangian . from eq .
( [ step 3 ] ) , we have @xmath263 by taking @xmath264 constant , where @xmath265 is defined by @xmath266.\ ] ] using @xmath267 , we can reduce eq .
( [ rg eq-0 ] ) to @xmath268 where @xmath269 is defined by @xmath270-i{\widetilde { \rm tr}}\,{\rm ln}\left[f.t.\!\left\{\left.\frac{\delta^{2}s}{\delta f_{\alpha}(y ) \delta { \overline f_{\beta}}(z)}~\right|_{0}\right\}\right]~~~~~~~~~~\end{aligned}\ ] ] and @xmath212 means taking @xmath264 constant and @xmath271 for the other fields .
we can not determine @xmath262 only from this equation because it is necessary to know the full effective lagrangian @xmath272 to know @xmath269 .
from now , we give asymptotic solutions of the equation for @xmath273 with imposing artificial conditions on @xmath269 or the effective lagrangian .
we will consider three cases , and the last case is the desired one .
we first examine the case , @xmath274 by putting @xmath275 we have @xmath276 from eq .
( [ rge for effective potential ] ) for @xmath273 . the general solution of this equation yields @xmath277 where @xmath278 is a function .
note that the first two terms on the left - hand side of eq .
( [ trivial ] ) cancel @xmath279 . the functional form for @xmath278
is determined from the condition that @xmath280 becomes quadratic at @xmath0 , i.e. , eq .
( [ compositeness0 - 3 ] ) , and we have @xmath281 where @xmath282 is a constant having mass dimension @xmath283 .
this means that our scenario for the composite higgs particle does not hold in this case .
next , we consider the case , @xmath284 first in this case , we examine the following candidate for the effective lagrangian other higher dimensional terms that give no effect on @xmath285 and @xmath286 . ] , @xmath287 where @xmath288 is the ordinary renormalizable lagrangian of the standard model , and @xmath289 denotes higher dimensional terms that have no effect on @xmath285 .
functional forms for @xmath290 , @xmath291 and @xmath292 are determined from now .
the effective lagrangian yields @xmath293 and then eq . (
[ rge for effective potential ] ) reduces to @xmath294 for @xmath273 .
the general solution of this equation yields @xmath295 where @xmath296 and @xmath297 are constants and @xmath278 is a function . on the other hand , eq .
( [ desired-1 ] ) requires @xmath298 by comparing this with eq .
( [ result of desired-1 ] ) , we have @xmath299 where @xmath300 and @xmath301 are constants having mass dimensions @xmath302 , @xmath303 and @xmath304 , respectively ) also yields certain restrictions on the functional form for @xmath289 . under this restrictions
we can choose the functional form for @xmath289 in consistent with the assumption that it has no effect on @xmath285 . ] .
note that @xmath305 does not vanish at any finite scale , and the first term on the right - hand side of eq .
( [ first possibility ] ) dominates in @xmath262 for @xmath273 .
thus , @xmath306 is not the condition for the vacuum stability in this case .
this means that our scenario for the composite higgs particle does not hold in this case too .
finally , we consider the case that eq .
( [ ne 0 ] ) holds thanks to derivative coupling terms . as a candidate
, we examine @xmath307 where @xmath308 is the covariant derivative for the top field @xmath309 , @xmath310 and @xmath311 are certain functions of @xmath312 , and @xmath313 denotes higher dimensional terms that have no effect on @xmath285 .
we give certain restrictions on @xmath314 , @xmath315 and @xmath316 from now .
the effective lagrangian yields @xmath317 if @xmath314 and @xmath315 satisfy the conditions @xmath318 eq .
( [ rge for effective potential ] ) reduces to @xmath319 for @xmath273 .
the general solution of this equation yields @xmath320 where @xmath278 is a function . from the condition that @xmath262 becomes quadratic at @xmath0 , we have @xmath321 .
after all , if @xmath272 of eq .
( [ second candidate ] ) is realized with the conditions ( [ x12 ] ) and ( [ y12 ] ) , the `` tree potential '' is an asymptotic solution of the rg equation ( [ rge for effective potential ] ) for @xmath273 . in this case
our discussion on the vacuum stability based on the `` tree potential '' is valid , i.e. , @xmath306 is the condition for the vacuum stability . in our scenario for
the composite higgs particle , @xmath272 itself must become quadratic with respective to @xmath86 at @xmath0 .
this requirement implies that @xmath314 and @xmath315 vanish at @xmath0 , and so we obtain @xmath322 from eqs .
( [ x12 ] ) , ( [ y12 ] ) and ( [ composite condition2 - 1 ] ) . the conditions for the @xmath3 functions should be considered with taking account of the contributions from higher dimensional terms in @xmath272 to the @xmath3 functions that were neglected approximately is required . ] when we gave eqs .
( [ bm ] ) and ( [ bl ] ) .
the conditions will give some restriction on the value of @xmath0 if they can hold .
the further investigation of the conditions remains as future works .
miransky , m. tanabashi and k. yamawaki , phys .
* 221 * ( 1989 ) , 177 ; mod .
* a4 * ( 1989 ) , 1043 .
+ see also : + g. cveti , hep - ph/9702381 , + v.a .
miransky , _ dynamical symmetry breaking in quantum field theory _
( world scientific publishing co. , singapore , 1993 ) and references contained therein .
bardeen , c.t . hill and m.
lindner , phys .
* d41 * ( 1990 ) , 1647 . k. yamawaki , prog .
vol . 123 ( 1996 ) , 19 ; hep - ph/9603277 . + t.e .
clark , s.t .
love and w.a .
bardeen , phys .
* b235 * ( 1990 ) , 235 . m. bando , t. kugo , n. maekawa , n. sasakura , y. watabiki and k. suehiro , phys . lett . *
b246 * ( 1990 ) , 466 .
m. bando , t. kugo , n. maekawa and h. nakano , phys . rev . * d44 * ( 1991 ) , r2975 . + for the definition of the renormalization transformation ,
see also : + m.e .
peskin and d.v .
schroeder , _ an introduction to quantum field theory _ ( addison - wesley publishing company , 1995 ) .
wilson , rev .
mod . phys .
* 47 * ( 1975 ) , 773 .
wilson and j. kogut , phys . rep . * 12c * ( 1974 ) , 75 .
+ j. polchinski , nucl .
phys . * b231 * ( 1984 ) , 269 . m. harada , y. kikukawa , t. kugo and h. nakano , prog .
* 92 * ( 1994 ) , 1161 . c. caso et al .
j. * c3 * ( 1998 ) , 1 .
m. beneke , i. efthymiopoulos , m.l .
mangano , j. womersley ( conveners ) et al .
, report in _ 1999 cern workshop on sm physics at the jhc _ , hep - ph/0003033 .
t. eguchi , phys .
rev . * d17 * ( 1978 ) , 611 .
+ k. shizuya , phys .
* d21 * ( 1980 ) , 2327 .
+ k. akama , phys .
* 76 * ( 1996 ) , 184 .
+ k. akama and t. hattori , phys .
* b392 * ( 1997 ) , 383 .
hill , phys .
lett . * b266 * ( 1991 ) , 419
. + s.p .
martin , phys .
* d45 * ( 1992 ) , 4283 . + see also : + j.d .
wells , phys .
* d56 * ( 1997 ) , 1504 and references contained therein .
several authors have studied a non - local four - fermi interaction obtained in a qed - like theory .
see : + t. kugo , _ dynamical symmetry breaking _ , edited by k. yamawaki , ( world scientific publishing co. , singapore , 1992 ) , p. 35
and references contained therein .
t. hambye and k. reisselmann , phys .
* d55 * ( 1997 ) , 7255 and see also references contained therein .
a. sirlin and r. zucchini , nucl .
* b266 * ( 1986 ) , 389 . for reviews , see : m. sher , phys . rep .
* 179 * ( 1989 ) , 273 .
c. ford , d.r.t .
jones and p.w .
stephenson , nucl .
* b395 * ( 1993 ) , 17 ; nucl .
phys . * b387 * ( 1992 ) , 373 . + g. altarelli and g.
isidori , phys .
lett . * b337 * ( 1994 ) , 141
. + j.a .
casas , j.r .
espinosa and m. quirs , phys .
lett . * b342 * ( 1995 ) , 171 ; phys . lett .
* b382 * ( 1996 ) , 374 . + p.q
. hung and g. isidori , phys .
* b402 * ( 1997 ) , 122 . c. kolda and h. murakama , hep - ph/0003170 and see also references contained therein .
a. hasenfratz , p. hasenfratz , k. jansen , j. kuti and y. shen , nucl .
* b365 * ( 1991 ) , 79 .
+ see also : + m. suzuki , mod .
lett . * a5 * ( 1990 ) , 1205 .
k. kubota and h. terao , prog .
* 102 * ( 1999 ) , 1163 .
the vacuum structure of the system of a bi - local scale field have been studied in : + t. kugo , phys .
lett . * b76 * ( 1978 ) , 625 .
+ see also ref .
bi - local fields introduced as gauge fields have been discussed in : + s. naka , s. abe , e. umezawa and t. matsufuji , prog .
* 103 * ( 2000 ) , 411 .
h. arason , d.j .
casta@xmath323o , b. kesthelyi , s. mikaelian , e.j .
piard , p. ramond and b.d .
wright , phys .
* d46 * ( 1992 ) , 3945 .
|
we consider what occurs when we remove one of the compositeness conditions proposed by bardeen , hill and lindner that leads to predictions for the top quark mass conflicting with the experimental value . through this consideration
the condition for the higgs particle to be the composite particle is reconsidered .
we show that in this case , ( i ) the higgs - yukawa system of the standard model becomes equivalent to a non - local four - fermi system at a high - energy scale @xmath0 , ( ii ) the higgs - yukawa sector of the model becomes useless above the scale because the vacuum state can not be defined .
we regard the two phenomena as indications of the compositeness of the higgs particle .
it is suggested that the new physics above @xmath0 contains bi - local fields .
addtoresetequationsection 20.5 cm 0.2 cm nup - a-2000 - 14 + hep - ph/0007225 + eizou umezawa +
|
for @xmath0 a closed surface and @xmath2 two regularly homotopic generic immersions , we are interested in the number mod 2 of quadruple points occurring in generic regular homotopies between @xmath3 and @xmath4 .
it has been shown in [ n1 ] that this number is the same for all such regular homotopies , and so it is a function of @xmath3 and @xmath4 which we denote @xmath5 . for @xmath0 orientable and @xmath1
two regularly homotopic _ embeddings _ , we give an explicit formula for @xmath6 which depends on the following data : if @xmath7 is an embedding then @xmath8 splits @xmath9 into two pieces , one compact which will be denoted @xmath10 and the other non - compact which will be denoted @xmath11 . by restriction of range
@xmath12 induces maps @xmath13 ( @xmath14 ) and let @xmath15 be the kernel of the map induced by @xmath16 on @xmath17 .
let @xmath18 be the orientation on @xmath0 which is induced from @xmath10 to @xmath19 and then via @xmath12 to @xmath0 .
our formula for @xmath6 will be in terms of the two triplets @xmath20 and @xmath21 .
our formula will be also easily extended to finite unions of closed orientable surfaces .
for two special cases a formula for @xmath6 ( for @xmath22 embeddings ) has already been known : the case where @xmath0 is a sphere has appeared in [ mb ] and [ n1 ] , and the case where @xmath0 is a torus has appeared in [ n1 ] .
the starting point for our work will be [ n2 ] where an explicit formula has been given for @xmath23 , where @xmath24 is any generic immersion and @xmath25 is any diffeomorphism such that @xmath3 and @xmath26 are regularly homotopic .
let @xmath27 be a finite dimensional vector space over @xmath28 .
a function @xmath29 is called a _
quadratic form _
if @xmath30 satisfies : @xmath31 for all @xmath32 , where @xmath33 is a bilinear form .
the following properties follow : ( a ) @xmath34 .
( b ) @xmath35 for all @xmath36 .
( c ) @xmath37 for all @xmath38 .
@xmath30 is called _ non - degenerate _
if @xmath39 is non - degenerate , i.e. for any @xmath40 there is @xmath41 with @xmath42 . for an exposition of quadratic forms
see [ c ] . in what follows
we always assume that our vector space @xmath27 is equipped with a non - degenerate quadratic form @xmath30 .
it then follows that @xmath43 is even .
a subspace @xmath44 such that @xmath45 is called a _ totally singular _ subspace .
a pair @xmath46 of subspaces of @xmath27 will be called a _ totally singular decomposition _ ( abbreviated tsd ) of @xmath27 if @xmath47 and both @xmath48 and @xmath39 are totally singular .
it then follows that @xmath49 .
( we remark that tsds do not always exist .
they will however always exist for the quadratic forms which will arise in our geometric considerations , as seen in lemma g1 below . )
a linear map @xmath50 is called _ orthogonal _ if @xmath51 for all @xmath36 .
it then follows that @xmath52 for all @xmath38 and that @xmath53 is invertible .
the group of all orthogonal maps of @xmath27 with respect to @xmath30 will be denoted @xmath54 .
the proof of the following lemma appears in [ c ] : [ ll1 ] let @xmath55 . 1 .
if @xmath44 is a totally singular subspace of dimension @xmath56 then there exists a @xmath57 such that @xmath46 is a tsd of @xmath27 .
2 . if @xmath46 is a tsd of @xmath27 and @xmath58 is a given basis for @xmath48 then there is a basis @xmath59 for @xmath39 such that @xmath60 .
[ d1 ] if @xmath46 is a tsd of @xmath27 then a basis @xmath61 of @xmath27 will be called _ @xmath46-good _ if @xmath62 , @xmath63 and @xmath60 .
the following two lemmas follow directly from the definition of quadratic form : [ ll2 ] let @xmath46 be a tsd of @xmath27 and @xmath61 an @xmath46-good basis for @xmath27 . if @xmath64 and @xmath65 then @xmath66 and @xmath67 .
[ ll3 ] let @xmath46 and @xmath68 be two tsds of @xmath27 .
let @xmath61 be an @xmath46-good basis for @xmath27 and @xmath69 an @xmath68-good basis for @xmath27 . if @xmath50 is the linear map defined by @xmath70 , @xmath71 then @xmath72 . for @xmath72
we define @xmath73 by : @xmath74 it has been shown in [ n2 ] that @xmath75 is a ( non - trivial ) homomorphism .
[ l1 ] if @xmath46 is a tsd of @xmath27 and @xmath72 satisfies @xmath76 and @xmath77 then @xmath78 . by lemma ll1 there exists an @xmath46-good basis @xmath61 for @xmath27 .
using lemma ll2 it is easy to verify that the matrix of @xmath53 with respect to such a basis has the form : @xmath79 where @xmath80 .
it follows that @xmath78
. given two tsds @xmath46 , @xmath68 of @xmath27 then by lemmas ll1 and ll3 there exists a @xmath72 such that @xmath81 and @xmath82 .
it follows from lemma l1 that if @xmath83 are two such @xmath53s then @xmath84 . and
so the following is well defined : [ d2 ] for a pair @xmath46 , @xmath68 of tsds of @xmath27 let @xmath85 be defined by @xmath86 for some ( thus all ) @xmath72 with @xmath81 and @xmath82 .
[ d3 ] for two tsds @xmath46 , @xmath68 of @xmath27 , we will write @xmath87 if @xmath88 . since @xmath89 is a homomorphism , @xmath90 for any three tsds @xmath91 .
it follows that @xmath92 is an equivalence relation with precisely two equivalence classes and that @xmath93 whenever @xmath87 .
[ l2 ] let @xmath94 and let @xmath44 be a totally singular subspace of dimension @xmath56 . if @xmath72 satisfies @xmath95 for every @xmath96 then @xmath78 . by lemma
ll1 there is a @xmath57 such that @xmath46 is a tsd of @xmath27 and an @xmath46-good basis @xmath61 for @xmath27 . using lemma ll2
it is easy to verify that the matrix of @xmath53 with respect to such a basis has the form : @xmath97 where @xmath98 is the @xmath99 identity matrix and @xmath100 is an alternating matrix , i.e. if @xmath101 then @xmath102 and @xmath103 .
since alternating matrices have even rank , it follows that @xmath78 .
[ p1 ] let @xmath46 and @xmath68 be two tsds of @xmath27 .
if @xmath104 or @xmath105 then @xmath106 .
say @xmath104 . by lemmas ll1 and ll3 there exists a @xmath72 with @xmath95 for all @xmath107 and @xmath82 .
the conclusion follows from lemma l2 .
let @xmath108 be two subspaces of @xmath27 .
we will write @xmath109 if @xmath110 for every @xmath111 , @xmath112 .
the following is clear : [ p2 ] let @xmath113 satisfy @xmath114 and @xmath109 . 1 .
if for @xmath115 , @xmath116 is a tsd of @xmath117 ( with respect to @xmath118 which is indeed non - degenerate ) then @xmath119 is a tsd of @xmath27 .
2 . if @xmath120 is another tsd of @xmath117 and @xmath121 ( @xmath115 ) then @xmath122 .
a _ surface _ is by definition assumed connected . a finite union of surfaces
will be called a _ system of surfaces_. let @xmath123 be a system of closed surfaces and @xmath124 a generic regular homotopy .
we denote by @xmath125 the number mod 2 of quadruple points occurring in @xmath126 .
the following has been shown in [ n1 ] : [ tn1 ] let @xmath123 be a system of closed surfaces ( not necessarily orientable . ) if @xmath127 are two generic regular homotopies between the same two generic immersions , then @xmath128 .
[ qii ] let @xmath123 be a system of closed surfaces and @xmath129 two regularly homotopic generic immersions .
we define @xmath5 by @xmath130 , where @xmath126 is any generic regular homotopy between @xmath3 and @xmath4 .
this is well defined by theorem tn1 .
let @xmath0 from now on denote a closed orientable surface .
a simple closed curve in @xmath0 will be called a _
circle_. if @xmath131 is a circle in @xmath0 , the homology class of @xmath131 in @xmath132 will be denoted by @xmath133 $ ] .
any immersion @xmath134 induces a quadratic form @xmath135 whose associated bilinear form @xmath33 is the algebraic intersection form @xmath136 of @xmath132 , as follows : for @xmath137 let @xmath138 be an annulus bounded by circles @xmath139 with @xmath133=x$ ] , let @xmath140 be an embedding which is regularly homotopic to @xmath141 and define @xmath142 to be the @xmath28 linking number between @xmath143 and @xmath144 in @xmath9 .
one needs to verify that @xmath142 is independent of the choices being made and that @xmath145 .
this has been done in [ p ] .
also , @xmath2 are regularly homotopic iff @xmath146 .
if @xmath7 is an embedding then @xmath8 splits @xmath9 into two pieces one compact and one non - compact .
we denote the compact piece by @xmath10 and the non - compact piece by @xmath11 . by restriction of range
, @xmath12 induces maps @xmath147 , @xmath14 .
let @xmath148 be the maps induced on homology and finally let @xmath149 , @xmath14 .
[ g1 ] let @xmath150 be an embedding , then @xmath151 is a tsd of @xmath132 with respect to the quadratic form @xmath152 .
we first show that each @xmath153 is totally singular : for @xmath154 let @xmath155 be as in the definition of @xmath156 and simply take @xmath157 . since @xmath158 , @xmath159 bounds a properly embedded ( perhaps non - orientable ) surface @xmath123 in @xmath160 . since @xmath161 is disjoint from @xmath123 , the @xmath28 linking number between @xmath159 and @xmath161 in @xmath9 is 0 , and so @xmath162 .
now , the fact that @xmath163 is a consequence of the @xmath28 mayer - vietoris sequence for @xmath164 where @xmath0 is identified with @xmath165 via @xmath12 .
if @xmath1 are two regularly homotopic embeddings then @xmath166 so @xmath167 and @xmath168 are tsds of @xmath132 with respect to the same quadratic form and so @xmath169 is defined .
we spell out the actual computation involved in @xmath169 : 1 .
find a basis @xmath61 for @xmath132 such that @xmath170 , @xmath171 and @xmath172 .
2 . find a similar basis @xmath69 using @xmath173 in place of @xmath12 .
3 . let @xmath174 be the dimension of the subspace of @xmath132 spanned by : @xmath175 4 .
@xmath176 , ( an element in @xmath28 . ) [ d6 ] if @xmath150 is an embedding then we define @xmath18 to be the orientation on @xmath0 which is induced from @xmath10 to @xmath19 and then via @xmath12 to @xmath0 ( and where the orientation on @xmath10 is the restriction of the orientation of @xmath9 . )
if @xmath177 are two embeddings then we define @xmath178 to be 0 if @xmath179 and 1 if @xmath180 .
our purpose in this work is to show : [ t1 ] let @xmath56 be the genus of @xmath0 .
if @xmath177 are two regularly homotopic embeddings then : @xmath181 our starting point is the following theorem which has been proved in [ n2 ] : [ tn2 ] for any generic immersion @xmath24 and any diffeomorphism @xmath25 such that @xmath3 and @xmath26 are regularly homotopic , @xmath182 where @xmath183 is the map induced by @xmath184 on @xmath185 , @xmath56 is the genus of @xmath0 and @xmath186 is 0 or 1 according to whether @xmath184 is orientation preserving or reversing , respectively .
let @xmath7 be an embedding , let @xmath188 be a plane and assume @xmath8 intersects @xmath189 transversally in a unique circle .
let @xmath190 then @xmath131 is a separating circle in @xmath0 .
let @xmath48 be a regular neighborhood of @xmath131 in @xmath0 and let @xmath191 be the connected components of @xmath192 .
( a lower index will always be related to the splitting of @xmath9 via a plane , the assignment of 0 and 1 to the two sides being arbitrary .
an upper index on the other hand is related to the splitting of @xmath9 via the image of a closed surface , assigning 0 to the compact side and 1 to the non - compact side . )
let @xmath193 ( @xmath115 ) be the closed surface obtained by gluing a disc @xmath194 to @xmath195 .
let @xmath196 be the embedding such that @xmath197 and @xmath198 is parallel to @xmath189 .
let @xmath199 and @xmath200 denote the inclusion maps .
the induced map @xmath201 is an isomorphism and let @xmath202 be the map @xmath203 .
this follows from the fact that the inclusions @xmath205 , @xmath206 , @xmath207 and @xmath208 all induce isomorphisms on @xmath17 and the splitting of each of the above spaces via @xmath189 induces a direct sum decomposition .
we only check that the inclusion @xmath208 induces isomorphism on @xmath17 .
indeed @xmath209 is obtained from @xmath11 by gluing a 2-handle along @xmath210 , and the inclusion of @xmath210 in @xmath11 is null - homotopic . [ s2 ] let @xmath7 be an embedding .
assume @xmath8 intersects a plane @xmath188 transversally in one circle and let @xmath215 be as above . if @xmath216 ( @xmath115 ) are both equivalent to standard embeddings , then @xmath12 is equivalent to a standard embedding . changing @xmath12 by isotopy , we may assume @xmath210 is a very thin tube .
@xmath216 is equivalent to a standard embedding @xmath217 via a regular homotopy @xmath218 having no quadruple points .
we may further assume that each @xmath219 moves @xmath193 only within the corresponding half - space defined by @xmath189 , that each @xmath220 is situated at the point of @xmath221 which is closest to @xmath189 and that these two points are opposite each other with respect to @xmath189 .
we now perform both @xmath219 , letting the thin tube @xmath48 be carried along .
if we make sure the thin tube @xmath48 does not pass triple points occurring in @xmath222 and @xmath223 then the regular homotopy @xmath126 induced on @xmath0 in this way will also have no quadruple points . since @xmath210 has approached @xmath224 from @xmath225 and since @xmath226 , we also have at the end of @xmath126 that @xmath48 approaches @xmath221 from @xmath227 .
and so we may continue moving the tube @xmath48 until it is all situated in the region between @xmath228 and @xmath229 , then canceling all knotting by having the thin tube pass itself ( this involves only double lines ) until @xmath48 is embedded as a straight tube connecting @xmath228 to @xmath229 and so the final map @xmath230 thus obtained is indeed a standard embedding . by assumption @xmath231 , @xmath115 which implies that @xmath232 , @xmath115 as tsds of @xmath233 .
( note that @xmath234 preserves the corresponding quadratic forms . ) but @xmath235 and @xmath109 and so by lemma p2 and lemma sp @xmath213 .
finally , from @xmath236 it follows that @xmath214 .
[ d5 ] let @xmath211 be two embeddings .
a regular homotopy @xmath237 ( @xmath238 ) with @xmath239 , @xmath240 will be called _ k - extendible _ ( where @xmath187 is either 0 or 1 ) if there exists a regular homotopy @xmath241 ( @xmath238 ) satisfying : the proof is by induction on the genus of @xmath0 . if @xmath257 then any @xmath12 is isotopic to a standard embedding and isotopic embeddings are equivalent .
so assume @xmath0 is of positive genus and so there is a compressing disc @xmath258 for @xmath8 in @xmath9 ( i.e. @xmath259 and @xmath260 does not bound a disc in @xmath8 . )
let @xmath261 and let @xmath48 be a regular neighborhood of @xmath131 in @xmath0 .
isotoping @xmath48 along @xmath258 as before we may assume @xmath48 is embedded as a thin tube .
there are four cases to be considered according to whether @xmath258 is contained in @xmath10 or @xmath11 and whether @xmath260 separates or does not separate @xmath8 .
_ case 1 _ : @xmath262 and @xmath260 separates @xmath8 .
it then follows that @xmath258 separates @xmath10 .
if @xmath191 denote the two components of @xmath192 and @xmath196 are defined as before then it follows from the assumptions of this case that @xmath263 and @xmath264 are disjoint and the tube @xmath210 approaches each @xmath224 from its non - compact side , i.e. from @xmath225 .
move each foot of the tube @xmath210 ( see figure f2 ) along the corresponding surface @xmath224 until they are each situated at the point @xmath265 of @xmath224 having maximal @xmath266-coordinate .
in particular it follows that now @xmath210 approaches each @xmath224 from above .
we now uniformly shrink each @xmath267 towards the point @xmath265 until it is contained in a tiny ball @xmath268 attached from below to the corresponding foot of @xmath210 , arriving at a new embedding @xmath269 .
this regular homotopy is clearly 0-extendible , and since no self intersection may occur within each of @xmath270 , @xmath222 and @xmath48 , this regular homotopy has no quadruple points .
and so by corollary g3 @xmath173 is equivalent to @xmath12 .
we now continue by isotopy , deforming the thin tube @xmath271 until it is a straight tube , and rigidly carrying @xmath272 and @xmath273 along .
we finally arrive at an embedding @xmath274 for which there is a plane @xmath189 intersecting @xmath275 as in lemma s2 with our @xmath270 and @xmath222 on the two sides of @xmath189 .
since the genus of both @xmath276 and @xmath277 is smaller than that of @xmath0 then by our induction hypothesis and lemma s2 , @xmath274 is equivalent to a standard embedding .
_ case 2 _ : @xmath278 and @xmath260 separates @xmath8 .
this time either @xmath279 or @xmath280 and assume the former holds .
in this case @xmath210 approaches only @xmath281 from its non - compact side and so we push the tube and perform the uniform shrinking as above only with @xmath270 .
this is a 1-extendible regular homotopy since we are shrinking @xmath263 which is part of @xmath11 .
now , if @xmath39 is the tiny ball into which we have shrunken @xmath282 then @xmath283 supplies separating compressing discs on both sides of @xmath8 and so we are done by case 1 . _ case 3 _ : @xmath262 and @xmath260 does not separate @xmath8 . if @xmath284 and @xmath285 is induced as above ( where @xmath286 is the surface obtained from @xmath287 by gluing two discs to it ) then both feet of the tube @xmath210 approach @xmath288 from its non - compact side .
push the feet of @xmath210 until they are both situated near the same point @xmath289 in @xmath288 having maximal @xmath266 coordinate .
let @xmath189 be a horizontal plane passing slightly below @xmath289 ( so that in a neighborhood of @xmath289 it intersects @xmath0 in only one circle . )
we may pull the tube @xmath210 until it is all above @xmath189 .
we then let it pass through itself until it is unknotted .
this is a 0-extendible regular homotopy with no quadruple points , at the end of which we have an embedding intersecting @xmath189 as in lemma s2 with an embedding of a torus above the plane @xmath189 , this embedding being already standard and an embedding of a subsurface @xmath290 of @xmath0 below the plane @xmath189 , @xmath290 being of smaller genus than that of @xmath0 .
again we are done by induction and lemma s2 .
_ case 4 _ : @xmath278 and @xmath260 does not separate @xmath8 .
we may proceed as in case 3 ( this time via a 1-extendible regular homotopy ) to obtain a standard embedding of a torus connected with a tube to @xmath288 but this time the torus is contained in @xmath291 and the tube connects to @xmath288 from its compact side .
but once we have such an embedding then the little standardly embedded torus has non - separating compressing discs on both sides and so we are done by case 3 .
if @xmath7 is an embedding and @xmath25 is a diffeomorphism such that @xmath12 and @xmath292 are regularly homotopic , then @xmath293 and @xmath294 .
( recall that @xmath183 is the map induced by @xmath184 on @xmath185 and @xmath186 is 0 or 1 according to whether @xmath184 is orientation preserving or reversing . )
we are now ready to prove theorem t1 . for two regularly homotopic embeddings
@xmath1 let @xmath300 .
we need to show @xmath301 .
if @xmath302 is also in the same regular homotopy class then @xmath303 and @xmath304 .
and so if @xmath173 is equivalent to @xmath274 and @xmath305 then also @xmath301 .
and so we may replace @xmath12 with an equivalent standard embedding @xmath212 ( proposition p3 ) and similarly replace @xmath173 with an equivalent standard embedding @xmath306 .
now @xmath212 and @xmath306 have isotopic images and so after isotopy we may assume @xmath307 and so @xmath308 for some diffeomorphism @xmath25 .
by lemma g5 and theorem tn2 the proof of theorem t1 is complete .
we conclude with a remark on systems of surfaces .
if @xmath309 is a system of closed orientable surfaces , and @xmath310 is an embedding , then we can rigidly move @xmath311 one by one , until they are all contained in large disjoint balls .
when it is the turn of @xmath312 to be rigidly moved , then the union of all other components is embedded and so only double lines occur .
if @xmath313 is another embedding then we can similarly move @xmath314 into the corresponding balls .
it follows that @xmath315 and so we obtain a formula for systems of surfaces , namely : @xmath316 .
chevalley : `` the algebraic theory of spinors . ''
columbia university press 1954 . also reprinted in : c. chevalley : `` the algebraic theory of spinors and clifford algebras , collected works , vol .
springer - verlag 1997 .
|
let @xmath0 be a closed orientable surface .
we give an explicit formula for the number mod 2 of quadruple points occurring in any generic regular homotopy between any two regularly homotopic embeddings @xmath1 .
the formula is in terms of homological data extracted from the two embeddings .
[ section ] [ thm]lemma [ thm]proposition [ thm]corollary [ thm]definition [ thm]remark
|
the study of patterns in permutations is one of the most active trends of research in combinatorics .
the richness of the notion of permutation patterns is especially evident from its plentiful appearances in several very different disciplines , such as algebra , geometry , analysis , theoretical computer science , and many others .
even if it is arguably not possible to encompass all possible applications of this notion into a simple formal environment , it seems reasonable to assert that the single mathematical structure which best catches the concept of a pattern and allows us to express a great deal of results about it is the _ permutation pattern poset_. given two permutations @xmath4 , we say that @xmath5 in the permutation pattern poset whenever there is an occurrence of @xmath2 in @xmath1 as a classical pattern .
an extremely challenging open problem concerning the permutation pattern poset is the determination of its mbius function .
the problem , originally posed by wilf @xcite , received quite recently much attention , and some partial results have been achieved @xcite .
however , a complete description of the mbius function of the permutation pattern poset is not yet available .
the same problem can be formulated for the _ consecutive pattern poset _ , where , by definition , @xmath6 whenever @xmath2 appears in @xmath1 as a _ consecutive _ pattern .
this poset is much easier than the classical one and , in particular , its mbius function is now completely understood @xcite .
we recall here that the mbius function @xmath7 of a poset is an important element of the incidence algebra of that poset , i.e. a function mapping an interval of the poset into a scalar .
more specifically , it is possible to give a recursive definition of @xmath7 as follows : @xmath8 in particular , if @xmath9 , then @xmath10 . using a duality argument ,
it is possible to show that the mbius function of an interval can be equally computed from top to bottom " , according to the following formula ( for @xmath11 ) : @xmath12 both formulas for computing @xmath7 will be frequently used throughout the paper .
consecutive and classical patterns are special ( actually , extremal ) cases of the more general notion of _ vincular patterns_. an occurrence of a vincular pattern is an occurrence of that pattern in which entries are subject to certain adjacency conditions ( see next section for a precise definition ) .
vincular patterns were introduced by babson and steingrmsson @xcite ( under the name of generalized patterns ) , and constitute a vast intermediate continent between the two lands of consecutive patterns and classical patterns . since the determination of the mbius function is still open in the classical case and totally solved in the consecutive one , it is conceivable that , if one is able to define a reasonable poset structure on permutations depending on the type of vincular patterns under consideration , then the resulting class of permutation posets ( which would somehow interpolate between the consecutive and the vincular posets ) may shed light on the differences and on the analogies between the two extremal cases . in the present paper
we propose a definition of _ vincular pattern poset of type @xmath13 _ ( whose elements are vincular patterns , or dashed permutations , as defined in section [ posets ] ) , where @xmath13 is a suitable infinite matrix whose @xmath14-th row encodes the type of the pattern of length @xmath14 belonging to the poset ( by specifying which entries of the pattern have to be adjacent in the corresponding dashed permutation ) .
this is done in section [ posets ] , where we also address the problem of understanding in which cases ( that is , for which matrices
@xmath13 ) the partial order thus defined perfectly catches the notion of an @xmath13-vincular pattern ; even if we are not able to give a complete characterization of such matrices , we find some partial results , and in particular we single out a special class of matrices for which everything works in the best possible way . in section [ quasi ]
we consider a single case , probably the closest to the consecutive one , in which all the entries of a pattern are required to be adjacent , except at most the first and the second . in section [ mobius_quasi ] we address the problem of the computation of the mbius function in this poset , which we call the _ quasi - consecutive pattern poset _ , and we completely determine the value @xmath15 when @xmath2 occurs precisely once in @xmath1 . in spite of its closeness with the consecutive case
, it seems that , in the general case , the computation @xmath15 in the quasi - consecutive pattern poset is considerably more complicated .
this is underlined , for instance , by the fact ( hinted at in the last section ) that the absolute value attained by the mbius function seems to be unbounded , whereas in the consecutive pattern poset the only possible values are @xmath16 .
denote with @xmath17 the set of all finite permutations .
elements of @xmath17 are represented in one - line notation , so that @xmath18 is the permutation of length @xmath14 mapping @xmath19 to @xmath20 , for all @xmath21 .
dashed permutation _ is a permutation in which some dashes are possibly inserted between any two consecutive elements .
for instance , @xmath22 is a dashed permutation ( of length 5 ) .
the _ type _ of a dashed permutation @xmath23 of length @xmath14 is the @xmath24-vector @xmath25 ( having @xmath26 components ) such that , for all @xmath27 , @xmath28 whenever there is no dash between @xmath20 and @xmath29 , and @xmath30 whenever there is a dash between @xmath20 and @xmath29 .
for example , the above dashed permutation @xmath22 has type @xmath31 .
we remark that , in different sources , this notion of type is expressed using a different formalism , namely by recording the lengths of each interval of adjacent elements in the dashed permutation .
for instance , the above dashed permutation is said to have type @xmath32 .
let @xmath23 be a permutation of length @xmath14 and @xmath33 be a dashed permutation of length @xmath34 .
we say that @xmath23 _ contains an occurrence of the vincular pattern _
@xmath33 when there exists a subsequence @xmath35 of elements of @xmath23 which is order - isomorphic to @xmath33 and such that @xmath36 and @xmath37 appear consecutively inside @xmath23 if there is no dash between @xmath38 and @xmath39 . in this case
we also say that @xmath33 is a vincular patter of @xmath23 .
vincular patterns were introduced in @xcite ( where they are called _ generalized patterns _ ) .
let @xmath13 be an infinite lower triangular @xmath24-matrix , and denote with @xmath40 the @xmath19-th row vector of @xmath13 .
given @xmath41 , we say that @xmath42 is an _ occurrence of a vincular pattern of type @xmath13 _ ( or an _ occurrence of an @xmath13-vincular pattern _ ) in @xmath23 when , given that @xmath33 has length @xmath43 , @xmath23 contains an occurrence of @xmath33 as a vincular pattern of type ( given by the first @xmath44 entries of ) @xmath45 . in this case
, we also write @xmath46 .
if @xmath23 does not contain any occurrence of an @xmath13-vincular pattern @xmath33 , that is @xmath47 , we say that @xmath23 _ avoids _
@xmath33 _ as an @xmath13-vincular pattern_. the most studied special cases are obtained when all the entries of @xmath13 below and on the main diagonal are equal to 1 and when @xmath13 is the null matrix . in the former case
we recover the notion of classical pattern , whereas in the latter one we get the notion of consecutive pattern . in some cases
, the notion of an @xmath13-vincular pattern can be described by means of a suitable partial order .
given @xmath48 , having lengths @xmath14 and @xmath26 respectively , we say that @xmath23 _ covers _
@xmath33 whenever @xmath33 appears as an @xmath13-vincular pattern in @xmath23 . in this case
, we write @xmath49 , or simply @xmath50 when @xmath13 is clear from the context .
the transitive and reflexive closure of this covering relation is a partial order which will be called the _ a - vincular pattern containment order _ , and the resulting poset will be called the _ a - vincular pattern poset_. when @xmath2 is less than or equal to @xmath1 in the @xmath13-vincular pattern poset , we will write @xmath51 or simply @xmath52 , when no confusion is likely to arise .
it is not difficult to realize that , in the two special cases mentioned above , the resulting posets are well known .
when @xmath13 is the lower triangular matrix having all 1 s below and on the main diagonal we obtain the classical pattern poset , whereas when @xmath13 is the null matrix we get the consecutive pattern poset .
we observe that , in these two cases , the partial order relation @xmath53 coincides with the binary relation @xmath54 , that is @xmath55 if and only if @xmath56 .
however this is not always true , for the relation @xmath54 is not transitive in general . in this direction
, it would be interesting to characterize all matrices @xmath13 for which the partial order relation ( rather than the covering one ) is directly defined in terms of occurrences of @xmath13-vincular patterns .
more precisely , it would be nice to have a characterization of those matrices @xmath13 such that @xmath57 if and only if @xmath56 .
such matrices @xmath13 are those for which the structure of the @xmath13-pattern poset perfectly describes the notion of an @xmath13-vincular pattern .
unfortunately , we have not been able to find such a characterization yet , so we leave it as our first open problem .
we have however some partial results , which we are going to illustrate below .
the first thing we observe is that in general , in the above conjectured equivalence , @xmath56 does not imply @xmath55 , nor does @xmath55 imply @xmath58 , as shown by the following two examples .
_ examples .
let @xmath13 be the ( lower triangular ) matrix all of whose elements are 0 , except for row 3 , which is @xmath59 . if @xmath60 and @xmath61 , then @xmath56 ( there is precisely one occurrence , in the subsequence 3456 ) , however @xmath62 ( any @xmath13-vincular pattern of @xmath1 of length 5 has to be consecutive in @xmath1 ) . notice that this example can be easily generalized to a matrix @xmath13 whose @xmath14-th row has only one 1 and all successive rows are identically 0 .
if @xmath13 is such that @xmath63 and @xmath64 , then it is immediate to verify that @xmath65 and @xmath66 ; thus we have @xmath67 , but @xmath68 , since the unique occurrence is of type @xmath69 .
if we restrict to the class of infinite lower triangular matrices having constant columns , we can show that one of the previous implications holds . observe
that , in this case , we can completely describe @xmath13 using the ( infinite ) vector @xmath70 whose @xmath19-th component @xmath71 is the unique nonzero value appearing in column @xmath19 of @xmath13 .
in particular , we modify the notations accordingly , by writing @xmath72 and @xmath73 in place of @xmath58 and @xmath55 , respectively . if @xmath72 , then @xmath73 . _
_ proof.__fix an occurrence of @xmath2 in @xmath1 as an @xmath70-vincular pattern .
let @xmath23 be the smallest consecutive pattern of @xmath1 containing such an occurrence .
then @xmath74 : this follows from the more general fact that , given any matrix @xmath13 , for any consecutive pattern @xmath75 of a permutation @xmath33 , we have @xmath76 ( starting from @xmath33 , repeatedly remove either the first or the last element until getting to @xmath75 ) . to conclude the proof we now need to find a ( descending ) chain of coverings from @xmath23 to @xmath2 in the @xmath70-vincular pattern poset .
to this aim , partition the elements of the fixed occurrence of @xmath2 in @xmath23 into blocks of consecutive elements of @xmath23 of maximal length , and suppose that such lengths are @xmath77 .
this implies that the vector @xmath70 certainly has a 1 in positions @xmath78 .
therefore we can remove the leftmost element of @xmath23 not belonging to the selected occurrence of @xmath2 , which is the @xmath79-th element of @xmath23 , thus obtaining a permutation covered by @xmath23 in the @xmath70-vincular pattern poset .
we can repeat this operation until the first two blocks of @xmath2 become adjacent , thus obtaining a chain of coverings in the @xmath70-vincular pattern poset from @xmath23 down to a permutation @xmath80 .
now observe that , starting from @xmath80 , we can certainly remove its leftmost element not belonging to the selected occurrence of @xmath2 to obtain a covering , since it is the @xmath81-th element of @xmath23 ( and @xmath70 has a 1 in position @xmath82 ) .
this argument can be repeated , by simply observing that , each time we remove the leftmost element of a permutation of the chain not belonging to the highlighted occurrence of @xmath2 , we create a new covering in the @xmath70-vincular pattern poset .
we can thus complete the chain of coverings from @xmath1 to @xmath23 to a chain of covering from @xmath1 to @xmath2 , which is what we need to conclude that @xmath83.@xmath84 however , also in this special case , the reverse implication does not hold , as the following example clarifies . _ _
example.__let @xmath85 ( that is , the vector whose unique nonzero component is the second one ) . if @xmath86 and @xmath87 , then @xmath88 , and so @xmath73 ; on the other hand , @xmath89 , since in the unique occurrence of @xmath2 in @xmath1 the first two elements are not consecutive .
the reason why the above counterexample holds is essentially that in @xmath70 there is a 1 preceded by a 0 . if this does not happen , we are able to prove the following proposition .
suppose that the vector @xmath70 has the first @xmath43 components equal to 1 and all the remaining components equal to 0 .
then @xmath90 if and only if @xmath73 . _
_ proof.__we only need to prove that @xmath73 implies that @xmath72 .
observe that , if the length of @xmath2 is @xmath91 , then the thesis becomes trivial .
so , from now on , we will tacitly assume that @xmath2 is sufficiently long .
we proceed by induction on the difference between the length of @xmath1 and the length of @xmath2 .
if @xmath2 and @xmath1 have the same length , then clearly @xmath92 . moreover , if the length of @xmath1 is one more than the length of @xmath2 , then the definitions of @xmath93 and @xmath94 coincide .
now suppose that the assertion holds for pairs of permutations whose lengths differ by less than @xmath14 , and let @xmath2 and @xmath1 be permutations whose lengths differ by exactly @xmath14 . since @xmath73 , there is a chain of coverings @xmath95 .
by inductive hypothesis , @xmath96 .
due to our assumptions on @xmath70 , this means that there is an occurrence of @xmath97 in @xmath1 all of whose elements must appear consecutively except at most the first @xmath98 . since @xmath99 , @xmath2 is obtained from @xmath97 by removing either one of the first @xmath98 elements or the last one . in all cases ,
it is easy to check that the occurrence of @xmath2 in @xmath1 thus resulting from the selected occurrence of @xmath100 in @xmath1 is made of consecutive elements of @xmath1 , except at most the first @xmath98 .
this is equivalent to say that @xmath72 , as desired.@xmath84
in the rest of the paper we will deal with a special case , which turns out to be particularly manageable , due to its closeness to the consecutive case .
let @xmath13 be the infinite lower triangular @xmath24-matrix for which an entry is equal to 1 if and only if it belongs to the first column .
so the first lines of @xmath13 are as follows : @xmath101 since @xmath13 has constant columns , according to the notations introduced in the previous section , it corresponds to the infinite vector @xmath102 having 1 only in the first position .
therefore an @xmath70-vincular pattern @xmath2 is interpreted as a dashed permutation having only one dash , which is placed between the first two elements of @xmath2 . due to the last result of the previous section , in this special case
, we have that @xmath103 if and only if @xmath72 .
in other words , we can assert that @xmath73 when there is an occurrence of @xmath2 in @xmath1 whose first two elements are possibly not consecutive , whereas all the remaining elements has to appear consecutively in @xmath1 .
for this reason , we will say that @xmath2 is a _ quasi - consecutive pattern _ of @xmath1 , and the resulting poset will be called the _ quasi - consecutive pattern poset_. as an example , the permutation @xmath104 contains the quasi - consecutive pattern @xmath105 , since the entries @xmath106 show an occurrence of the vincular pattern @xmath107 .
when a permutation @xmath1 does not contain the quasi - consecutive pattern @xmath2 , we say that @xmath1 _ avoids _ the quasi - consecutive pattern @xmath2 .
for instance , the preceding permutation @xmath104 avoids the quasi - consecutive pattern @xmath108 .
observe that other instances of vincular patterns with
ground permutation " @xmath108 appear in @xmath104 ( for example , there are occurrences of both @xmath109 and @xmath110 ) .
to have an idea of how intervals look like in the quasi - consecutive pattern poset , we refer the reader to subsequent figures .
the quasi - consecutive pattern poset has many similarities with the consecutive pattern poset studied in @xcite .
clearly it has a slightly more complicated structure , which would be interesting to investigate in detail .
a first structural property of such a poset is recorded in the next proposition . for any @xmath111
, @xmath1 covers at most three permutations , which are obtained by removing either the first , the second or the last entry of @xmath1 ( and suitably rearranging the remaining ones ) . _ _
proof.__if @xmath1 has length @xmath14 , a permutation @xmath33 of length @xmath26 covered by @xmath1 can appear as an occurrence either consecutive or not .
in the former case , @xmath33 has to appear either as a prefix or a suffix of @xmath1 ; thus @xmath33 is obtained by removing either the first or the last entry of @xmath1 . in the latter case ,
the only possibility is that the first two entries of @xmath33 are not consecutive in @xmath1 ; thus @xmath33 is obtained from @xmath1 by removing the second entry.@xmath84 the property proved in the above proposition gives an important feature of the quasi - consecutive poset , which will be very useful in the next section .
in this section , which contains the main results of our paper , we address the problem of the computation of the mbius function of the quasi - consecutive pattern poset .
specifically , given @xmath112 , with @xmath113 , we want to compute @xmath114 . we will completely solve the case of a single occurrence , which turns out to be not a trivial one , and leave the general case as an open problem . throughout this section
, we will frequently write permutations with a dash between the first entry and the second entry .
this is done in order to emphasize the fact that they have to be interpreted as specific vincular patterns in the quasi - consecutive pattern poset . moreover , as we already started to do in the previous paragraph , the partial order relation of the quasi - consecutive pattern poset will be denoted simply by @xmath115 ( instead of @xmath94 ) , as no other partial orders will appear from now on .
we start by giving some partial results , depending on whether @xmath1 covers precisely @xmath19 permutations , for @xmath116 .
recall that a _ monotone permutation _ is a permutation which is either increasing or decreasing ( so that , for any @xmath14 , the monotone permutations are simply @xmath117 and @xmath118 ) .
if @xmath1 covers precisely one permutation , then @xmath119 $ ] is a chain , for any @xmath0 . _
proof.__if @xmath1 covers precisely one permutation , then the same permutation is obtained by removing any of the three allowed elements of @xmath1 . in particular , if removing the first or the last entry of @xmath1 results in the same permutation , necessarily @xmath1 has to be monotone .
so @xmath2 has to be monotone too , and in this case @xmath119 $ ] is clearly a chain.@xmath84 thus , if @xmath1 covers precisely one permutation , then @xmath120 when @xmath92 , @xmath121 when @xmath1 covers @xmath2 , and @xmath122 in all the other cases .
let @xmath123 .
if @xmath1 covers precisely two permutations , then there are two distinct possibilities : * @xmath124 and @xmath125 are consecutive integers ; * either @xmath126 or @xmath127 . _
proof.__if @xmath1 covers precisely two permutations , then there exist two elements in the set @xmath128 such that removing either of them from @xmath1 gives the same permutation .
it can not happen that removing either @xmath124 or @xmath129 yields the same permutation , since otherwise @xmath1 would be monotone , and so it would cover only one permutation , which is not the case .
if removing either @xmath124 or @xmath125 yields the same permutation , then each of the remaining entries of @xmath1 has to be either bigger or smaller than both @xmath124 and @xmath125 .
thus @xmath124 and @xmath125 must be consecutive integers .
finally , if removing either @xmath125 or @xmath129 yields the same permutation , then the subpermutation determined by the elements @xmath130 is monotone and @xmath124 must be either smaller or bigger than such elements .
since @xmath1 itself can not be monotone , then @xmath124 is either 1 or @xmath14 , and so either @xmath126 or @xmath131.@xmath84 observe that , if @xmath1 covers precisely two permutations and @xmath124 and @xmath125 are not consecutive integers , then we can determine the structure of the interval @xmath3}}$ ] , and so its mbius function . indeed
, suppose that any occurrence of @xmath2 in @xmath1 does not involve @xmath124
. then necessarily @xmath2 is monotone .
thus a permutation in @xmath3}}$ ] is uniquely determined by a ( possibly empty ) set of entries to remove from the end of @xmath1 and by choosing if @xmath124 has to be removed or not . for these reasons ,
it is not difficult to realize that @xmath3}}\simeq \mathbf{2}\times \mathbf{n}$ ] , where @xmath132 denotes the chain of the first @xmath19 positive integers . therefore , in this situation , the computation of the mbius function can be easily carried out : @xmath133 notice that @xmath134 does not fit into the previous case , nevertheless using a completely analogous argument we get to the same formula for @xmath15 . on the other hand , if there is an occurrence of @xmath2 which involves @xmath124 , then necessarily every occurrence of @xmath2 involves @xmath124 . in this case
, it is easily seen that either @xmath2 is not monotone or @xmath2 has length 2 .
thus @xmath3}}$ ] is a finite chain , whose mbius function is trivial to compute .
let @xmath135 and suppose that @xmath2 occurs precisely once in @xmath1 .
we will be concerned with the computation of @xmath15 in this particular case , which we will be able to solve completely .
we will distinguish several cases , depending on the fact that the single occurrence of @xmath2 in @xmath1 does or does not involve the three elements @xmath136 and @xmath129 .
a first obvious consideration is that @xmath2 can not involve all of the three above elements , unless @xmath92 , and in this last case of course @xmath120 .
the second possibility is that two of the three elements @xmath137 are involved in the unique occurrence of @xmath2 in @xmath1 .
there are of course three distinct cases ; however in any of them the interval @xmath3}}$ ] is a chain .
for instance , if @xmath2 involves @xmath124 and @xmath129 ( and not @xmath125 ) , then @xmath2 is ( isomorphic to ) @xmath138 , for some @xmath139 , and it is immediate to see that any @xmath140}}$ ] is obtained by starting from @xmath1 and repeatedly removing the second element of the resulting permutation until we get to @xmath33 .
this is of course the only possible way to remove elements from @xmath1 and remain inside @xmath3}}$ ] ( @xmath124 and @xmath129 can not be removed , being part of the unique occurrence of @xmath2 in @xmath1 ) .
this means that the interval @xmath3}}$ ] is a chain .
the remaining cases can be dealt with in a completely analogous way .
thus , we can conclude that , in all these cases , @xmath122 , unless @xmath2 has length @xmath26 , in which case @xmath121 .
we next examine the case in which only one among @xmath136 and @xmath129 is involved in the unique occurrence of @xmath2 in @xmath1 .
there are three distinct cases to consider .
[ a1 ] suppose that @xmath124 occurs in @xmath2 ( whereas @xmath125 and @xmath129 do not ) . then @xmath122 , unless @xmath119 $ ] has rank 2 ( in which case @xmath120 ) . _
_ proof.__in this case @xmath2 is ( isomorphic to ) @xmath141 , for some @xmath142 with @xmath143 and @xmath144 . if @xmath140}}$ ] , then @xmath33 is ( isomorphic to ) @xmath145 , for suitable @xmath19 and @xmath146 , and so it is uniquely determined by two intervals of elements @xmath147 and @xmath148 to be removed from @xmath1 in a well - specified order
. therefore @xmath3}}$ ] is a _ grid _ , i.e. it is isomorphic to a product of two chains , whose lengths ( which of course depend on the values of @xmath19 and @xmath146 ) are @xmath149 ( this is due to the fact that @xmath2 does not involve @xmath125 and @xmath129 ) . thus @xmath122 , unless the two chains both have length 1 .
clearly , in the latter case @xmath120.@xmath84 suppose that @xmath125 occurs in @xmath2 ( whereas @xmath124 and @xmath129 do not ) .
then @xmath122 , unless @xmath119 $ ] has rank 2 ( in which case @xmath120 ) . _
_ proof.__the argument is completely analogous to the one used in the above proposition .
the only difference here is that the roles of @xmath124 and @xmath125 have to be swapped.@xmath84 notice that the two cases considered so far are essentially equivalent to the case of a single occurrence in the consecutive pattern poset . indeed , using the notation introduced in @xcite , it is not too difficult to realize that @xmath3}}$ ] and @xmath150 $ ] are order isomorphic .
the last case is by far the most challenging one .
[ consec_end ] suppose that @xmath129 occurs in @xmath2 ( whereas @xmath124 and @xmath125 do not )
. then @xmath122 , unless @xmath3}}$ ] has rank 2 . in this last case ,
if @xmath124 and @xmath125 are consecutive integers , then @xmath122 , otherwise @xmath120 . _
_ proof.__it is convenient to distinguish two cases , depending on whether @xmath2 appears as a consecutive pattern in @xmath1 or not .
if @xmath2 is not a consecutive pattern of @xmath1 , then it is not difficult to realize that @xmath3}}$ ] has only one atom , which can be obtained in the following way : take the unique occurrence of @xmath2 in @xmath1 and add to it the element of @xmath1 immediately to the left of its consecutive part .
observe that , in this case , if @xmath2 has length @xmath151 , then @xmath152 , and @xmath122 .
if @xmath2 is a consecutive pattern of @xmath1 , then @xmath2 is ( isomorphic to ) @xmath153 , where @xmath154 .
figure [ prop_4_5 ] shows an instance of this situation .
denote with @xmath155 the permutation obtained from @xmath1 by removing @xmath124 ( and of course suitably renaming the remaining elements ) .
similarly , denote with @xmath156 the permutation obtained from @xmath1 by removing @xmath125 and with @xmath157 the permutation obtained from @xmath1 by removing both @xmath124 and @xmath125 . finally , let @xmath158}}\ ; |\ , \rho < { \ensuremath{{\tau^{(1)}}}},\rho \nleq { \ensuremath{{\tau^{(1,2)}}}}\ } , \\ { \ensuremath{{c^{(2)}}}}&=&\ { \rho \in { \ensuremath{[\sigma,\tau]}}\ ; |\ , \rho < { \ensuremath{{\tau^{(2)}}}},\rho \nleq { \ensuremath{{\tau^{(1,2)}}}}\ } .\end{aligned}\ ] ] we start by observing that , in this case , @xmath155 and @xmath156 are the only coatoms of @xmath3}}$ ] .
so , if @xmath159 ( i.e. , @xmath124 and @xmath125 are consecutive integers ) , then clearly @xmath122 .
thus , in what follows we will suppose that @xmath160 .
we next show that @xmath161 and @xmath162 are chains .
indeed , any @xmath163 can be obtained from @xmath156 by repeatedly removing the second element of the resulting permutation , so @xmath33 is uniquely determined by the set of consecutive elements of @xmath1 which have been removed .
an analogous argument shows that also @xmath161 is a chain .
moreover , we have that @xmath164 . indeed , if we had @xmath165 , then any @xmath166 would be order isomorphic to both @xmath167 and @xmath168 , for a suitable @xmath169 .
we wish to show that , in this situation , @xmath124 and @xmath125 have to be consecutive integers . indeed
, suppose that @xmath124 and @xmath125 are not consecutive and , w.l.o.g .
, that @xmath170 .
first of all , for all @xmath19 such that @xmath171 , it can not be @xmath172 , otherwise @xmath173 and @xmath174 would not be order isomorphic .
so there should exist an element @xmath70 such that @xmath175 , which appears before @xmath176 in @xmath1 .
but then @xmath177 would be order isomorphic to @xmath33 , which is not possible , since otherwise @xmath178 .
therefore @xmath124 and @xmath125 have to be consecutive integers , which is not true ( remember that we are supposing that @xmath160 ) . we can thus conclude that @xmath164 , as desired .
our next goal is to prove that , for all @xmath179 , @xmath180 .
in fact , if @xmath181 $ ] , then @xmath182 , since otherwise we have @xmath178 , which is not true .
so , if for instance @xmath183 , then we have @xmath184 as well .
therefore @xmath185 $ ] is a chain having at least three elements ( choose @xmath186 ) , whence @xmath180
. finally , we are now in a position to prove that , for all @xmath187 , @xmath180 . indeed , if @xmath33 is covered by @xmath157 , then necessarily @xmath188=\ { \rho , { \ensuremath{{\tau^{(1,2)}}}},{\ensuremath{{\tau^{(1)}}}},{\ensuremath{{\tau^{(2)}}}},\tau \}$ ] , and it is immediate to see that @xmath180 . instead ,
if @xmath189 is not covered by @xmath157 , then @xmath188 $ ] contains the same five permutations listed above as well as a set @xmath190 of permutations less than @xmath157 ( but closer " than @xmath33 to @xmath157 ) and ( possibly ) a set @xmath191 of permutations contained in @xmath192 .
using an inductive argument ( on the distance from @xmath157 ) , we can show that , for all permutations @xmath193 , @xmath194 ; moreover , we already know ( from the previous paragraph ) that , for all @xmath195 , @xmath194 .
thus the only @xmath196 $ ] such that @xmath197 are @xmath198 , and so we can immediately conclude that @xmath180 . since of course @xmath199 , if the rank of @xmath3}}$ ] is greater than 2 , then @xmath200 and so @xmath122 .
otherwise , if @xmath124 and @xmath125 are consecutive integers , then @xmath159 , and @xmath3}}$ ] is a 3-elements chain , so that @xmath122 . if instead @xmath124 and @xmath125 are not consecutive , then @xmath3}}$ ] is the product of two 2-elements chains , whence @xmath120 . @xmath84
the last case to be considered is when none of @xmath201 and @xmath129 belongs to the unique occurrence of @xmath2 in @xmath1 .
it will be convenient to distinguish two cases , depending on whether the occurrence of @xmath2 is consecutive or not .
[ none_nonconsecutive ] if @xmath2 does not occur consecutively in @xmath1 , and does not involve @xmath136 and @xmath129 , then @xmath122 . _
_ proof.__denote with @xmath202 the permutation isomorphic to the smallest consecutive pattern in @xmath1 containing @xmath2 and with @xmath203 the permutation isomorphic to the smallest suffix of @xmath1 containing @xmath2 .
we start by observing that , for all @xmath140}}$ ] , we have @xmath204 or @xmath205 .
indeed , if @xmath206 , then the leftmost element of an occurrence of @xmath33 in @xmath1 must correspond to an entry of @xmath1 which appears on the left of @xmath2 . thus necessarily all the other elements of the occurrence of @xmath33 have to appear consecutively in @xmath1 and to contain @xmath2 , and this implies that @xmath205 .
we next give more precise information on the structure of the interval @xmath3}}$ ] ( see also figure [ particular ] ) .
* @xmath207 $ ] is a chain : this follows from the fact that @xmath208 occurs only once in @xmath209 , and the first two elements of @xmath209 belong to such an occurrence . *
@xmath210 $ ] is a chain : this is analogous to the above one , since @xmath2 occurs only once in @xmath208 and the first and last elements belong to such an occurrence . *
@xmath211 $ ] is a product of two nonempty chains : this follows from the proof of proposition [ a1 ] . in each permutation of @xmath3}}$ ] is highlighted .
] therefore we have @xmath212}\mu ( \sigma , \rho ) -\sum_{\rho \notin [ \sigma , \pi_2 ] } \mu ( \sigma , \rho ) .\ ] ] it is immediate to see that , in the last expression , the first sum is 0 . as for the second sum , we observe that each @xmath213 ( and so in particular each @xmath214 $ ] ) lies either above all the four elements of @xmath211 $ ] having nonzero mbius function , or above two of them , namely @xmath2 and one of the two atoms of @xmath3}}$ ] . in both cases , an inductive argument ( on the difference between the length of @xmath33 and the length of @xmath208 ) shows that , for all @xmath215 $ ] , @xmath216 .
from all the above considerations it then follows that @xmath122.@xmath84 [ ultima ] if @xmath2 occurs consecutively in @xmath1 , and does not involve @xmath136 and @xmath129 , then @xmath122 , unless @xmath3}}$ ] has rank 3 , the unique occurrence of @xmath2 in @xmath1 contains @xmath217 and @xmath1 covers three elements in @xmath3}}$ ] . in such a case , we have @xmath121 . _ _ proof.__denote with @xmath23 the permutation isomorphic to the smallest suffix of @xmath1 containing the unique occurrence of @xmath2 and with @xmath218 the permutation isomorphic to the prefix of @xmath1 of length @xmath26 . using an argument analogous to that of proposition [ none_nonconsecutive ] , we observe that , for every @xmath219}}$ ] , we have @xmath220 or @xmath221 . indeed , if @xmath222 , then an occurrence of @xmath33 necessarily contains both the last element of @xmath1 and the unique occurrence of @xmath2 , so that @xmath33 contains @xmath23 ( i.e. @xmath221 ) .
since @xmath23 and @xmath218 are clearly incomparable , the two intervals @xmath223 $ ] and @xmath224 $ ] constitute a partition of @xmath3}}$ ] . therefore concerning the mbius function of @xmath3}}$ ] we have :
suppose first that @xmath226 $ ] has rank at least 2 .
this situation is illustrated in figure [ prop_4_7 ] .
observe that , in this case , @xmath227 , since @xmath2 is a prefix of @xmath23 , and so @xmath226 $ ] is a chain of length at least 2 .
given @xmath228 $ ] , suppose now that @xmath229 , for all @xmath230 .
notice that @xmath33 is made by an element of @xmath1 followed by a suffix of @xmath1 containing @xmath2 .
thus , by removing the rightmost element of @xmath33 , we obtain a permutation @xmath231 contained in @xmath218 ( observe that @xmath231 is a coatom of @xmath232 $ ] ) .
moreover , the reader can immediately see that @xmath231 and @xmath23 are incomparable and that , for all @xmath233 $ ] , either @xmath234 or @xmath235 .
so we have a partition of @xmath232 $ ] and we can conclude that , for all @xmath228 $ ] , @xmath236 which is enough to assert that @xmath122 . otherwise , i.e. when @xmath226 $ ] has rank 1 , we observe that the two intervals @xmath223 $ ] and @xmath224 $ ] both fall into the scopes of proposition [ consec_end ] . indeed , there is a unique occurrence of @xmath2 and @xmath23 in @xmath218 and @xmath1 respectively , in both cases at the end of the permutation .
each permutation @xmath237 $ ] covers exactly one permutation @xmath238 $ ] ( just remove the last entry of @xmath239 , which is its only entry which can be removed in order to get a permutation lying below @xmath218 ) .
this fact will now be used to describe the structure of @xmath3}}$ ] . concerning the interval @xmath224 $ ] , we have two possibilities : either @xmath1 covers one element or two elements . in the former case ,
necessarily @xmath218 covers only one element in @xmath240 $ ] ; in the latter case , @xmath218 may cover either one or two elements .
we thus have a total of three cases , which are illustrated in figure [ prop_4_7_2 ] . in the proof of proposition [ ultima ] . ]
we first observe that , when @xmath3}}$ ] has rank 3 , the second and the third cases depicted in figure [ prop_4_7_2 ] correspond exactly to the exceptional case of the statement of the proposition , and it is immediate to observe that @xmath121 . instead , in the first case we have @xmath122 .
so from now on we suppose that the rank of @xmath3}}$ ] is at least 4 .
if we compute the mbius function of @xmath3}}$ ] from top to bottom , we can restrict ourselves to compute it in each element of @xmath223 $ ] , that is : @xmath241 it is now not too difficult to realize that , in each of the three possible cases , for any @xmath242 $ ] except that for the top three levels of the interval ( i.e. @xmath218 , its coatoms and the elements covered by such coatoms ) , the value of @xmath243 is 0 .
indeed , denote with @xmath244 the set of all permutations in @xmath188\setminus \ { \rho \}$ ] other than those whose mbius function is explicitly indicated in figure [ prop_4_7_2 ] .
then , @xmath245 we can assume inductively that each summand of the first sum is equal to 0 ; as far as the second sum is concerned , the reader is referred again to figure [ prop_4_7_2 ] to get convinced that it is equal to 0 in all cases as well .
thus , in particular , @xmath122.@xmath84 summing up all the results obtained so far in this section , we then have the following theorem , which completely solves the problem of the computation of the mbius function of the quasi - consecutive pattern poset in the case of one occurrence .
suppose that @xmath2 occurs exactly once in @xmath123 as a quasi - consecutive pattern . then @xmath122 , unless one of the following cases hold : * @xmath92 , in which case @xmath120 ; * @xmath2 is covered by @xmath1 , in which case @xmath121 .
* @xmath3}}$ ] has rank 2 and @xmath2 involves @xmath124 but not @xmath125 and @xmath129 , in which case @xmath120 ( the same holds when @xmath124 and @xmath125 are swapped ) . *
@xmath3}}$ ] has rank 2 , @xmath2 involves @xmath129 but not @xmath124 and @xmath125 , and @xmath124 and @xmath125 are not consecutive integers , in which case @xmath120 .
* @xmath3}}$ ] has rank 3 , @xmath2 involves @xmath217 but not @xmath246 and @xmath129 , @xmath2 occurs consecutively in @xmath1 and @xmath1 covers three elements in @xmath3}}$ ] , in which case @xmath121 .
the study of vincular pattern posets , which has been initiated in the present paper , is of course very far from being completed . from the point of view of vincular pattern posets in general ,
the main open problem , already stated in section [ posets ] , is perhaps that of characterizing those matrices @xmath13 for which @xmath55 if and only if @xmath56 .
concerning the main topic investigated here , namely the mbius function of the quasi - consecutive pattern poset , a lot of work has still to be done .
the case of one occurrence which we have completely solved in the previous section seems not to be really representative of the general case .
for instance , the absolute value of @xmath7 can be different from 0 and 1 ( a simple example is given by the interval @xmath247 $ ] , whose mbius function equals 2 ) , and there is computational evidence @xcite that @xmath248 is actually unbounded . despite its closeness with the consecutive case , this fact shows that the quasi - consecutive case can sometimes be very similar to the classical ( unrestricted ) case .
another conjecture suggested by @xcite is the following : if @xmath1 is the direct sum of some copies of @xmath2 , then @xmath120 .
l.f . wishes to thank einar steingrmsson for his kind hospitality at the university of strathclyde in november 2012 .
some of the results and conjectures of the present work have their roots in several stimulating discussions during that visit .
|
we introduce vincular pattern posets , then we consider in particular the quasi - consecutive pattern poset , which is defined by declaring @xmath0 whenever the permutation @xmath1 contains an occurrence of the permutation @xmath2 in which all the entries are adjacent in @xmath1 except at most the first and the second .
we investigate the mbius function of the quasi - consecutive pattern poset and we completely determine it for those intervals @xmath3}}$ ] such that @xmath2 occurs precisely once in @xmath1 .
|
the lhc has recently provided tentative evidence for a standard model higgs boson in the mass range of @xmath4 .
the hints for the higgs boson comes from several independent analyses .
first , both atlas and cms find @xmath5 excesses in the @xmath6 channel that favor the mass range of 120 to 170 gev with a production cross section equal to that of the standard model higgs with @xmath7 @xcite .
next , both cms and atlas observe too many @xmath8 events with three events clustered between 142 gev and 145 gev @xcite . finally , the tevatron s combined data in the @xmath9 channel is @xmath5 high with the excess consistent with the expected standard model production cross section for the higgs boson@xcite . while these excesses have not been officially combined , cms and atlas @xmath10-values are each @xmath11 and the excess could be greater than @xmath12 @xcite . and @xmath13 in split supersymmetry .
the recent experimental excess is consistent with @xmath14 , favoring scalar mass scales greater than @xmath15 .
potential constraints from big bang nucleosynthesis exist for gluino lifetimes greater than @xmath16s .
flavor constraints are present for scalar mass scales lower than @xmath17.,width=422 ] if this excess continues to hold , then it gives strong evidence against natural low scale supersymmetry .
natural supersymmetric standard models require new physics to raise the higgs mass and frequently involve non - perturbative couplings , elaborate model building , or the loss of gauge coupling unification @xcite . on the other hand , this range for the higgs mass is favored by fine - tuned supersymmetry breaking , most notably split supersymmetry @xcite . in split supersymmetry ,
the @xmath18 to @xmath19 mass range points at moderate scalar masses around @xmath20 for moderate to large values of @xmath13 , as shown in fig .
[ fig : higgsmass ] was used . ] .
models with high scale supersymmetry breaking and high scale r - symmetry breaking , dubbed `` supersplit supersymmetry , '' predict heavy gauginos and higgsinos and thus the low energy spectrum consists solely of the standard model @xcite .
this scenario has no direct low energy consequences but has a prediction for the standard model higgs boson mass due to the gut - scale higgs quartic coupling supersymmetric boundary condition @xcite .
the prediction from supersplit supersymmetry is @xmath21 large @xmath22 can be a consequence of a weakly broken peccei - quinn symmetry , resulting in a small @xmath23-term . the large @xmath22 value for the higgs mass in supersplit supersymmetry is @xmath24 and is relatively insensitive to supersymmetric threshold corrections and is in the range where the excesses lie @xcite .
the difference in scales between split supersymmetry and supersplit supersymmetry is due to the gaugino yukawa couplings to the higgs boson that counteract contributions from the electroweak gauge boson loops , allowing for a lower scale of supersymmetry breaking . with the higgs mass consistent with both split supersymmetry and supersplit supersymmetry , distinguishing these two scenarios
is of primary importance .
gluino signatures are the first discriminant between supersplit and split susy .
the moderately low scalar masses of split susy have phenomenological implications for gluino decays , allowing for lifetimes @xmath25 the lower end of this range gives rise to displaced vertices greater than the minimum visible impact parameter at the lhc detectors , around @xmath26 .
the upper end gives rise to r - hadrons that can be seen as champs in the muon chamber and stopped gluinos with lifetimes measurable in dedicated analyses @xcite .
measuring the lifetime of the gluino will fix the lightest squark mass , since the lifetime depends on the fourth power of the squark mass .
when combined with the requirement of the correct yield for the dark matter relic abundance , there are three regions of parameter space singled out as particularly promising thermal relic dark matter scenarios : the higgs resonance region , `` well - tempered neutralinos '' , and pure electroweakino dark matter @xcite .
these different regions have qualitatively different experimental confirmation prospect , ranging from optimistic in the case of the higgs resonance region to incredibly pessimistic for pure electroweakino dark matter .
the higgs resonance region features bino dark matter with mass @xmath27 annihilating through an @xmath2-channel higgs boson .
if gaugino mass unification is assumed , then the relatively light bino implies that the gluino is currently accessible at the lhc .
the winos are also light and can be produced in either the decays of the gluinos or by direct production at the lhc or a future linear collider .
on the other hand , the higgsinos of the higgs resonance region are typically heavy with masses above and sometimes as heavy as the remainder of this article will focus on the phenomenology of split supersymmetry and derive its predictions for the immediate and short - term future .
the organization is as follows : sec .
[ sec : gluinos ] maps out gluino phenomenology and focuses on the different discovery signatures depending on the gluino lifetime .
[ sec : splitdm ] briefly reviews dark matter scenarios in split susy before mapping out the higgs resonance region in detail due to its promising early lhc phenomenology . with the assumption of gaugino mass unification ,
the signatures of the higgs resonance region beyond the gluino are examined .
[ sec : disc ] concludes with a discussion about the prospects for measuring the properties of supersymmetry breaking .
the unmistakable signature of split supersymmetry is the quasi - stable gluino .
the relatively low higgs mass implies that the scalars are sufficiently heavy to be undetectable , but at a sufficiently low scale to allow for gluino lifetimes to be accurately measured .
assuming degenerate squarks with mass @xmath28 and kinematically unsuppressed gluino decays , @xmath29 the lifetime of the gluino is given by @xmath30 when gluinos decay into top quarks , the assumption of the decays being kinematically unsuppressed is not necessarily a good approximation .
gluino decays do not need to preserve flavor and it is possible to have single - top gluino decays @xcite .
this offers the possibility of mapping out the flavor structure of the squarks through gluino decays .
assuming that the kinematic factors not included in eq .
[ eq : gluinolifetime ] are irrelevant , the relative branching ratio of the gluino to wino versus bino is @xmath31 if the gluino decays into a wino , then the wino will subsequently decay into a bino through a weak vector boson .
these cascades have not typically been studied in long - lived gluino phenomenology , but can provide additional handles on signatures .
the most striking effect is that the decay of the wino to the bino can produce hard leptons in the final state .
folding in branching ratios into leptons from the wino decay to bino results in @xmath32 depending on whether one or two gluinos are reconstructed at a time .
[ fig : higgsmass ] shows that the gluino lifetimes predicted by a higgs mass of @xmath33 correspond to the interpolating region between a promptly decaying gluino and a long lived gluino @xmath34 the decays of long - lived gluinos can alter big bang nucleosynthesis , but the majority of the parameter space leading to @xmath35 gives gluino lifetimes @xmath36 and evades tension with early universe cosmology @xcite .
[ fig : higgsmass ] shows that @xmath37 , for and @xmath38 , is satisfied for @xmath39 . and
@xmath40 for @xmath41 and @xmath42 .
the full red contours delimit the 1@xmath43 and 2@xmath43 errors quoted by pdg on @xmath44 and @xmath40 .
the dot indicates the central values of @xmath45 and @xmath46 .
the dashed red contours delimit the 1@xmath43 and 2 @xmath43 errors for the global fit for @xmath44 from reference @xcite.,width=422 ] measuring the higgs mass enforces a tight relationship between @xmath22 and @xmath28 . at moderate to large @xmath22
, the higgs mass depends only upon @xmath47 .
the gluino lifetime depends on the fourth power of the squark mass .
therefore , small errors in the higgs mass prediction result in enormous uncertainties on the expected gluino lifetime . the central values of the standard model parameters given by pdg are used throughout this article @xcite , but there is sizeable sensitivity to the top yukawa coupling and @xmath48 .
[ fig : mhiggsuncertainty ] shows how the higgs mass varies with @xmath40 and @xmath44 for a benchmark value of @xmath41 and @xmath42 .
both the pdg @xcite central values and the value of @xmath44 from the recent global fit from @xcite are displayed . from any benchmark value of @xmath13 and @xmath28 there is at least an uncertainty in the higgs mass prediction of @xmath49 alternatively , this uncertainty in the higgs mass prediction can be stated as a variance on the squark mass @xmath50 a factor of 100 uncertainty in the gluino lifetime .
in addition to low scale uncertainties in extracting a prediction for the gluino lifetime , there are high scale uncertainties that substantially alter the predictions for the higgs mass @xcite .
an increased accuracy of both the top mass and @xmath44 will shrink the low scale uncertainties . with a measurement of the gluino lifetime
, it will be possible to determine if there are any additional contributions to the higgs quartic coupling beyond minimal split supersymmetry .
there is a wide range of gluino phenomenology that needs be explored .
the remaining portion of this section is divided into short - lived gluinos with nearly prompt decays ( subsec . [
sec : shortlived ] ) , massively displaced gluino decays ( subsec .
[ sec : moderatelived ] ) , and quasi - stable gluinos that leave the detector ( subsec .
[ sec : longlived ] ) . for the higgs mass range hinted at the lhc , the gluino lifetime predicted by split susy is greater than @xmath51 for all of its parameter space .
the resolution on impact parameters at the lhc is roughly @xmath52 and therefore the gluino lifetimes could be measured even at the low end of the gluino lifetime range .
impact parameters @xmath53 are reconstructed using normal tracking algorithms and will be tagged as heavy flavor candidates at both cms and atlas .
when the pair - produced gluinos decay , it gives events with multiple jets plus missing energy .
if searches do not require the displaced vertices to be compatible with @xmath54-hadron decays , then these events will fall into the heavy flavor plus missing energy searches .
if the heavy flavor analyses require compatibility with @xmath54-hadron decays , then these events will still be considered in unflavored jets plus missing energy searches because these searches do not exclude displaced vertices in their event selection criteria . in split
susy with unified boundary conditions for the gauginos , the low energy spectrum of gauginos can be significantly altered from the standard ratio of @xmath55 if the @xmath56-term is large , see sec .
[ sec : gauginomassunification ] for details .
for instance , it is possible to have a gluino that is lighter than the wino , though this is not a typical situation .
if that is the case , limits for normal msugra searches for jets and missing energy are not applicable . in order to remedy
that , simplified models have been proposed to capture such topologies in this region of phase space @xcite .
fortunately , both atlas and cms are performing searches in jets and missing energy for simplified models where the gluino directly decays to the bino with an arbitrary spectrum @xcite .
currently no results exist for simplified model searches where the gluino decays into a wino which subsequently decays into a bino . in @xcite ,
the 1-step cascade decay causes a loss of @xmath57 in gluino mass reach for @xmath58 of integrated luminosity . more generally , one of the primary reasons to assume gaugino mass unification is to avoid contributions to the gaugino masses from multiple susy breaking sources .
when that happens , it is likely that physical phases are induced for the gaugino masses .
normally , these phases give rise to unacceptably large cp violation , but with the squarks and sleptons decoupled to at least the @xmath15 scale , the cp problem is ameliorated .
thus in split susy there should be no prejudice for gaugino mass unification .
the incompatibility of displaced gluino decays with @xmath54-hadron decays is a striking handle that can be used to make jets and missing energy searches essentially background - free .
the easiest manner to separate these events from standard heavy flavor jets is to require a substantial invariant mass for the secondary vertex .
this will also remove any secondary interactions that give rise to displaced vertices since these are always low invariant mass vertices .
to date no lhc search has been performed in jets plus missing energy with high mass displaced vertices .
when gluino lifetimes become longer than @xmath59 , standard jet reconstruction starts failing .
there has been work over the past several years to study extremely displaced vertices .
these searches gain importance in light of the measurement of the higgs mass .
the range from @xmath60 to @xmath61 in @xmath62 corresponds to a factor of 5 in scalar masses , @xmath63 , or equivalently the @xmath64 higgs mass range at large @xmath13 .
recently , d0 looked for highly displaced jets in the range @xmath65 @xcite .
this analysis required an additional muon not necessarily linked to the primary vertex .
additional muons appear in the cascade decay of @xmath66 . with the assumptions that went into eq .
[ eq : gluinopairsintomuons ] , there is a @xmath67 cost to require a hard muon in these events .
atlas is currently studying jets with displaced vertices that are too distant from the interaction point to be reconstructed by standard tracking techniques .
such tracks are sufficiently displaced that they are either not reconstructed at all or are not associated with the calorimetric activity to which they belong .
this means that they can be interpreted as `` trackless jets . ''
the results from these searches are not currently available , but they are promising in being sensitive to decays of gluinos with highly - displaced vertices . in general , more effort is needed for this challenging range of gluino lifetimes to ensure that there is no loss in sensitivity to this promising region of the split susy parameter space .
if the gluino lifetime is greater than several meters , the primary way to discover split susy is through searches for charged massive particles , or
champs " @xcite .
the gluino does not carry any intrinsic electric charge , but instead can pick up a charge when it hadronizes and turns into an r - hadron .
r - hadrons can be classified into r - mesons and r - baryons .
gluinos are expected to fragment dominantly into r - mesons for similar reasons that more mesons than baryons are produced in jets .
the number of champs depends upon whether there is a stable isotriplet r - meson , @xmath68 .
if there is , then an @xmath69 fraction of produced gluinos will be champs that propagate from the interaction point , through the tracking and calorimetry , and finally leave through the muon chambers .
alternatively , if there is no stable isotriplet r - meson because there is an isosinglet r - meson , @xmath70 , with mass @xmath71 then the r - mesons will promptly decay to the isosinglet r - meson and the r - hadron will not appear as a champ in the tracking chamber .
there are still r - baryons that are in an isodoublet or higher representations of isospin , but these may make up only @xmath72 of r - hadrons , reducing the gluino mass reach by a factor of 2 .
there is an uncertainty that arises because a quasi - stable gluino that hadronizes is unlike any other hadronic state in nature
it is a static color octet source .
this means that there are no direct analogies to draw with low energy hadronic spectroscopy .
taking a non - relativistic , constituent quark potential model , _ e.g. _ such as de rujula - georgi - glashow ( dgg ) @xcite , may not translate into these new scenarios .
for instance , in dgg , the anomalously small mass for the pion arises from a large hyperfine splitting interaction . with a color octet source
, there is a new way to hadronize through gluons only .
the potential picture for constituent quarks is already strained and translating it to a constituent gluon may be misleading .
even if the lightest r - meson is an isosinglet , there is a matter induced process that converts r - mesons into r - baryons by picking up a baryon in the detector through the following reaction @xmath73 because the pion is anomalously light in qcd due to approximate chiral symmetry breaking , this reaction should be exothermic @xcite .
thus , as r - mesons propagate through the detector , they transmute into r - baryons .
r - baryons are charged and therefore can be champs .
the dominant region of the detector where this transmutation takes place is in the calorimeters .
thus the the way to see these converted r - mesons is as champs in the muon system , which is outside the calorimeter .
the most sensitive current search is by cms @xcite with an integrated luminosity of @xmath74 .
it places bounds on r - hadrons up to @xmath75 these limits eat up a small fraction of the parameter space for high susy breaking scale . by the end of the 7 tev run
, these bounds should advance to @xmath76 for charged mesons and @xmath77 for charged baryons .
dark matter plays a critical role in how observable split supersymmetry is . the dark matter candidate in split
susy is an electroweakino , and the correct yield for the relic abundance arises from the electroweak gauge and higgs interactions .
the phenomenology of electroweakino dark matter in split susy has been studied in detail and the relevant physics is reviewed below . in split
susy , the scalar superpartners receive masses at the susy breaking scale and can be integrated out .
the low energy parameters of the theory are the gaugino and higgsino masses and two irreducible cp phases in the higgs sector of the theory .
the non - gauge interactions of these particles are given by : @xmath78 the couplings @xmath79 and @xmath80 are given by susy boundary conditions at the scale where the scalars decouple and are parametrically @xmath81 and @xmath82 respectively .
there are two irreducible phases : @xmath83 and @xmath84 . at the supersymmetry breaking scale ,
the following supersymmetric relations are satisfied : @xmath85 however , these couplings run in a non - supersymmetric fashion from the supersymmetry breaking scale down to low energies . in split
susy , the thermal relic dark matter can be classified into the following categories : 1 .
nearly pure higgsino dark matter 2 .
nearly pure wino dark matter 3 .
well - tempered bino - higgsino dark matter 4 .
well - tempered bino - wino dark matter 5 .
bino dark matter annihilating through the higgs resonance .
the mapping of relative electroweakino masses to dark matter annihilation channels is shown in fig .
[ fig : splitdm ] @xcite .
the nearly pure higgsino and wino dark matter scenarios are unobservable at the lhc due to their heavy masses @xmath86 and @xmath87 @xcite .
the well - tempered scenario involves mixing of the sterile component ( bino ) with the active component ( wino / higgsino ) in order to get the proper relic abundance .
it approaches the pure higgsino or wino cases as the active component becomes sizable .
this means that a sizable fraction of the parameter space features heavy dark matter , which becomes a challenge for the lhc given that the gluino is typically a factor of several times heavier than the dark matter .
the dark matter in the higgs resonance region is approximately half the mass of the higgs , @xmath88 .
assuming that the bino mass is tied to the gluino mass , the gluino is not much heavier than @xmath89 .
much of this parameter space has already been tested at the lhc , and by the end of the 7 tev run the majority of this scenario can be ruled out .
the remainder of this article will explore the higgs resonance region and illustrate some mixing aspects of the phenomenology in the literature , but many of the observations will carry over to the well - tempered dark matter scenario as well , if they have a produceable gluino .
non - thermal dark matter in split susy is also an interesting possibility that will not be explored here ; however , @xmath15 squarks are motivated in these scenarios @xcite . in the higgs resonance region ,
the bino - like @xmath90 is a stable wimp dark matter candidate whose relic abundance is determined from near - resonant annihilation through the higgs .
the wimp miracle requires that the dark matter thermally averaged annihilation cross - section be approximately @xmath91 in order to yield the observed dark matter abundance .
the annihilation cross - section of @xmath90 can only be this small if its coupling to the higgs is suppressed .
the lsp interacts with the higgs through bino - higgsino mixing and suppressing this coupling requires heavy higgsinos , above @xmath92 . with at least an order of magnitude gap between the lsp and the higgsino scales , it is possible to integrate the higgsinos out in order to obtain the following effective lagrangian @xmath93 after electroweak symmetry breaking , the neutralino masses are @xmath94 with @xmath95 the @xmath96 term leads to a bino - wino mixing of order @xmath97 diagonalizing the above matrix gives the following spectrum @xmath98 the mass splitting between the charged and neutral winos is @xmath99 , @xmath100 additionally , loop - level electromagnetic induced splittings adds @xmath101 to the chargino mass @xcite
. electroweak symmetry breaking induces mixings between the electroweak gauginos and the higgsinos .
this results in effective couplings between a bino - like @xmath90 and the higgs given by @xmath102 with @xmath103 notice that the higgs yukawa coupling to @xmath104 is complex .
this phase is usually ignored because it gives rise to unacceptably large cp violation ; however , in this region of parameter space , the induced cp - odd observables are acceptable even with maximal cp violation and will be observable in the next generation of experiments .
the coupling @xmath105 is proportional to @xmath106 ; therefore , in order to keep the annihilation cross section fixed @xmath107 the relic abundance in the higgs resonance region depends dominantly on @xmath2-channel higgs exchange .
the thermally averaged cross section is sensitive to the dark matter s annihilation cross - section at small velocities when its interactions freeze - out .
majorana - scalar interactions are velocity suppressed while majorana - pseudoscalar interactions are not .
the resonant enhancement to the annihilation cross - section is larger in the presence of pseudoscalar interactions - a well known effect in the @xmath108 funnel region in the mssm . in split
supersymmetry , the higgs couplings to the neutralinos can have cp violating phases which give rise to majorana - pseudoscalar interactions . taking the cp violating phase into account , the annihilation cross - section of @xmath90
is given by : @xmath109 where @xmath2 is the center of mass energy of the annihilating lsps .
the dependence on the cp violating phase arises because @xmath110 is cp - odd . if @xmath111 , the annihilation at low energies is @xmath10-wave suppressed ; however , in the presence of cp violation there is an @xmath2-wave contribution .
this has an immediate impact on the phenomenology of the model . without cp violation , the higgs resonance region is fairly narrow and in order to achieve a large enough thermally - averaged annihilation cross section , the higgsinos can not be very decoupled : @xmath112 . with cp violation
, this higgs resonance region behaves like the @xmath108-funnel of the mssm , where the thermally - averaged annihilation cross section is large enough .
if the @xmath56-term scale is too low , the @xmath104 abundance is over - depleted , leaving room for other forms of dark matter @xcite .
this means that the phenomenologically acceptable region is much broader than what is generally considered . with maximal cp violation
, the @xmath56-term can be very large @xmath113 without over - predicting the relic abundance .
alternatively , @xmath22 can be small ( [ mutanbeta ] ) .
this is illustrated in fig .
[ fig : relic abundance ] , which shows the dependence of @xmath114 on the dark matter mass and cp - violation in the @xmath104-@xmath104-@xmath115 coupling , fixing the relic abundance to @xmath116 . on the dark matter mass and cp - violation in the @xmath104-@xmath104-@xmath115 coupling , fixing the relic abundance to @xmath116 .
the dependence is symmetric under @xmath117.,width=422 ] the phenomenology of the higgs resonance region is incomplete until the gluino and wino masses are specified .
a common relation for the gaugino masses which allows for an independent specification of the higgsino mass is gaugino mass unification at the gut scale .
the low energy masses are obtained by renormalization group evolution .
the free parameters of the theory at the gut scale are then the unified gaugino mass @xmath118 , the higgsino mass @xmath56 , and a single cp phase @xmath119 . the renormalization group flow of the gaugino masses from the gut scale to the supersymmetry breaking scale is determined by @xmath120 the flow below the supersymmetry breaking scale is governed by @xcite : @xmath121 at the supersymmetry breaking scale , the couplings @xmath122 , @xmath123 are given by their supersymmetric values and rg evolve down . fixing the low energy values of @xmath124 , @xmath56 and @xmath125 , the above equations can be numerically solved for @xmath126 . the wino and gluino masses
are then determined by rg evolution .
if cp violation is maximal , then the @xmath56-term can be quite large and more strongly affect the running of @xmath124 and @xmath127 .
this can significantly alter the gaugino mass unification prediction of @xmath128 regions of parameter space with @xmath129 are caused by rg - induced contributions to @xmath124 from the @xmath56-term at energies below the supersymmetry breaking scale .
similar phenomenology exists in string - inspired models , see e.g. @xcite .
the gluino mass varies between @xmath130 to several tev while the wino mass varies between the lep2 limit to around 500 gev .
the charginos are slightly lighter than the wino with @xmath131 .
in some regions of parameter space the gluino can become as light as @xmath90 , though this is not a generic prediction .
the limits on direct production of electroweakinos arise from lep2 .
there are two relevant channels that were searched for @xmath132 for chargino pair production , lep2 set a combined limit that excludes @xmath133 @xcite .
this is essentially model independent because of the relatively small mixing with @xmath134 and the absence of accessible neutrinos .
direct neutralino production is model dependent , particularly on the bino - wino mixing given in eq .
[ eq : bino - winomixing ] , @xmath135 .
the cross section for @xmath136 production is given by @xmath137 which is smaller than the delphi limit of @xmath138 @xcite .
therefore , the lep2 chargino bounds are the strongest throughout . the dominant electroweakino production channels at the lhc are @xmath139 and @xmath140 .
the portion of parameter space that leads to off - shell decays of the @xmath141 s and @xmath142 s @xmath143 is the most accessible at the lhc . in this case , the decay chains proceed as follows : @xmath144 the trilepton signature of the latter is the most visible once a @xmath142 dilepton veto is placed .
the dominant trilepton background comes from @xmath145 and @xmath146 . at low missing energy ,
@xmath147 is an important background .
the @xmath142 veto is necessary and sufficient to increase the signal to background ratio to @xmath148 for most of the available parameter space . in the higgs resonance region ,
wino states decaying through an off - shell @xmath142 are light enough to be discovered with @xmath149 from the 2012 7 tev lhc data .
the signal and backgrounds used in this study were calculated with ` madgraph 5 ` @xcite . showering and hadronization were done in ` pythia 6.4 ` @xcite , while ` pgs 4 ` was used for basic detector simulation @xcite .
nlo cross sections for the signal were calculated with ` prospino 2.1 ` @xcite .
one of the challenges in multilepton searches is the large contribution of fakes to the sm backgrounds .
the @xmath150 jets background fakes in this study are estimated based on the recent cms results on trilepton signatures @xcite .
the diboson backgrounds simulated for this study were checked to have reasonable agreement with cms s results .
the relative shape of the @xmath151 background with respect to the simulated diboson background is assumed to be the same as in the cms s analysis .
this background is effectively removed by applying a @xmath152 cut .
[ fig : electroweakinos ] shows the @xmath153 distribution for a benchmark signal point corresponding to 140 gev charginos and winos that decay via off - shell gauge bosons to a 70 gev bino , assuming 15/fb and @xmath154 7 tev .
three or more leptons are required in the selection .
any os - sf lepton pair is required to have an invariant mass outside of @xmath155 , in addition to requiring @xmath156 .
a @xmath157 gev cut gives @xmath148 making this signature potentially discoverable by the end of the 7 tev run .
distribution for a benchmark signal point ( blue ) consisting of 140 gev charginos and winos that decay via off - shell gauge bosons to the lightest neutralino .
the sm background contributions are shown in black , while the combined signal + background is displayed in red .
15 fb@xmath158 of integrated luminosity at 7 tev is assumed.,width=422 ] when the spectra are widely spaced and the @xmath142 goes on - shell , the trilepton searches become extremely challenging due to the sizable diboson background .
the primary handle to distinguish on - shell decays of electroweak bosons from the diboson background is the transverse mass of the unpaired lepton and the missing transverse energy .
once the @xmath142 goes on shell , the signal to background ratio decreases to @xmath159 therefore , electroweakino discovery through on - shell decays will only be possible with @xmath160 from the 14 tev lhc run @xcite .
[ fig : lhcreach ] summarizes the prospects for the higgs resonance region of split susy at the lhc .
( to the left of the red band ) give rise to discoverable trilepton signatures at the 7 tev lhc due to off - shell decays of the @xmath142 .
gluino lifetimes of a few centimeters may be constrained by standard jets plus met searches if the events pass quality cuts .
above many centimeters gluinos are constrained by champs searches at the lhc to be heavier than @xmath161.,width=422 ] direct dark matter detection experiments attempt to discover dark matter particles by measuring nuclear recoils caused by dark matter particles scattering off the detector s nuclear target .
prospects for direct detection in the context of split supersymmetry were studied in @xcite . in split
supersymmetry , the interaction between @xmath90 and the target nucleus is mediated through higgs boson exchange .
the wimp - nucleon scattering cross section is given by @xmath162 where @xmath163 is the nucleon mass and @xmath164 is the nuclear matrix element summed over all quarks @xcite . in the cp preserving case
@xmath105 is a factor of 2 to 10 larger than in the cp violating case .
therefore , the cp violating case has a direct detection rate a factor of 5 to 50 smaller than the cp preserving case .
xenon100 sets a cross - section limit of @xmath165 @xcite .
[ fig : directiondetection ] shows the wimp - nucleon scattering cross section when the dark matter abundance is fixed to @xmath166
. the upcoming xenon100 release should be sensitive to @xmath167 .
if @xmath168 is near resonance , then couplings to the higgs give sufficiently small cross sections to be at or below the atmospheric neutrino background , making direct detection discovery challenging even in the distant future .
-@xmath104-h coupling , fixing the dark matter relic abundance to @xmath169.,width=422 ] indirect signals of dark matter in split supersymmetry are discussed in @xcite .
the continuum spectrum of photons depends upon the level of cp violation . with sizeable cp violation ,
@xmath2-wave annihilation dominates and @xmath170 .
thermal averaging over a resonance provides a `` boost factor , '' @xmath171 , @xmath172)^{-1}}{\oo(10)}.\end{aligned}\ ] ] with @xmath173 one has @xmath174 @xcite , however the boost factor can be made larger by fine tuning @xmath104 to be nearly on resonance .
the current limits from fermi for @xmath175 in the @xmath176 annihilation channel are in the range @xmath177 depending on the assumptions of the dark matter distribution in the galaxy @xcite .
these limits can already rule out resonant dark matter annihilation . if @xmath178 , then the continuum annihilation cross section is smaller than the thermal annihilation cross section by a factor of @xmath179 hence without cp violation these models remain beyond the fermi limits on continuum gamma rays .
in addition to the continuum spectrum , @xmath90 can annihilate to @xmath180 or @xmath181z pairs leading to an excess flux in the @xmath181 ray power spectrum around the points @xmath182 and @xmath183 respectively .
the monochromatic photon annihilation cross section is approximately @xmath184 the current cross section limit on the most optimistic model of structure formation is @xmath185 @xcite . in split
susy , the dominant contribution to the electric dipole - moments of standard model fermions comes from 2-loop diagrams since the 1-loop contributions are suppressed by the susy breaking scale . for a standard model fermion @xmath186 with mass @xmath187 and charge @xmath188 , the split susy contribution to its edm was computed in @xcite : @xmath189 where @xmath190 with @xmath191 gev , @xmath192 tev and order one cp violating phases and mixing angles , the electron s electric dipole moment can be as large as @xmath193 , with the majority of the parameter space lying within the range @xmath194 .
this limit is close to the experimental upper bound of @xmath195 e - cm at the 90% confidence level @xcite .
if the evidence for a 142 gev to 147 gev standard model higgs boson holds , it gives an intriguing hint to the underlying principles of physics beyond the standard model . throughout the last decade
there has been no compelling experimental evidence for physics beyond the standard model , casting doubt over the whole principle of naturalness in the higgs sector .
if split supersymmetry is confirmed with sfermion masses at the pev scale , it will reveal an underlying simplicity to the structure of nature manifest by its minimal content .
pev supersymmetry solves the majority of the hierarchy problem , reducing the fine tuning from a factor of @xmath196 down to @xmath197 .
pev scale scalar masses also remove most of the constraints from squark and slepton flavor changing effects and cp violation . without the constraints from flavor , supersymmetry breaking can be mediated by planck - suppressed operators with @xmath198 .
disentangling the interplay between gravity mediation and anomaly mediation becomes the window into the origin of supersymmetry breaking . by studying the relative masses of the gauginos and the higgsinos ,
some of the effects can be disentangled . while it may not be possible to determine why the weak scale is nt completely natural
, it could be that approximate @xmath199-symmetries are more generic than appreciated in theories that give standard model - like field contents .
many of the simplest string constructions have approximate @xmath199-symmetries , e.g. @xcite .
alternatively , if the evidence for the higgs boson in this mass range disappears , the majority of the split susy parameter space is excluded at the same time because due to the exclusion of @xmath200 .
the remaining parameter space will be either at @xmath201 or for squark masses beneath @xmath15 where the flavor violation is no longer automatically safe .
jgw would like to thank andrew haas for helpful discussions about the lhc higgs results from the eps and for discussions about trackless jets and jet reconstruction with displaced vertices .
jgw would also like to thank surjeet rajendran for collaboration on this work in 2006 and 2007 , much of the inspiration came from the unpublished draft of work exploring the higgs resonance region of split supersymmetry . ei would like to thank ahmed ismail for valuable discussions .
dsma , ei and jgw are supported by the doe under contract de - ac03 - 76sf00515 .
jgw is partially supported by the doe s outstanding junior investigator award and the sloan fellowship .
ei is supported by a graduate lhcti fellowship .
cms collaboration , cms - pas - hig-11 - 004 ( 2011 ) .
atlas collaboration , `` search for the standard model higgs boson in the decay channel @xmath202 with @xmath203 of pp collisions at @xmath204 = 7 tev , '' in preparation ( 2011 ) .
see reference [ 14 ] of @xcite and also `` search for higgs to zz ( llll , llnunu , llqq ) '' by konstantinos nikolopoulos at europhysics conference on high - energy physics 2011 7/22/2011 .
j. r. espinosa and m. quiros , phys .
lett . * 81 * ( 1998 ) 516 , http://arxiv.org/pdf/hep-ph/9804235 [ arxiv : hep - ph/9804235 ] . n. polonsky and s. su , phys .
b * 508 * , 103 ( 2001 ) , http://arxiv.org/pdf/hep-ph/0010113 [ arxiv : hep - ph/0010113 ] .
j. a. casas , j. r. espinosa and i. hidalgo , jhep * 0401 * ( 2004 ) 008 , http://arxiv.org/pdf/hep-ph/0310137 [ arxiv : hep - ph/0310137 ] .
a. brignole , j. a. casas , j. r. espinosa and i. navarro , nucl .
b * 666 * ( 2003 ) 105 , http://arxiv.org/pdf/hep-ph/0301121[arxiv:hep-ph/0301121 ] .
r. harnik , g. d. kribs , d. t. larson and h. murayama , phys .
d * 70 * ( 2004 ) 015002 , http://arxiv.org/pdf/hep-ph/0311349 [ arxiv : hep - ph/0311349 ] .
s. chang , c. kilic and r. mahbubani , phys .
d * 71 * ( 2005 ) 015003 , http://arxiv.org/pdf/hep-ph/0405267 [ arxiv : hep - ph/0405267 ] .
p. batra , a. delgado , d. e. kaplan and t. m. p. tait , jhep * 0402 * , 043 ( 2004 ) , http://arxiv.org/pdf/hep-ph/0309149 [ arxiv : hep - ph/0309149 ] .
a. birkedal , z. chacko and y. nomura , phys .
d * 71 * ( 2005 ) 015006 , http://arxiv.org/pdf/hep-ph/0408329 [ arxiv : hep - ph/0408329 ] .
a. maloney , a. pierce and j. g. wacker , jhep * 0606 * ( 2006 ) 034 , http://arxiv.org/pdf/hep-ph/0409127 [ arxiv : hep - ph/0409127 ] .
k. s. babu , i. gogoladze and c. kolda , http://arxiv.org/pdf/hep-ph/0410085 [ arxiv : hep - ph/0410085 ] .
a. delgado and t. m. p. tait , jhep * 0507 * ( 2005 ) 023 , http://arxiv.org/pdf/hep-ph/0504224[arxiv:hep-ph/0504224 ] .
r. barbieri , l. j. hall , y. nomura and v. s. rychkov , phys .
d * 75 * , 035007 ( 2007 ) , http://arxiv.org/pdf/hep-ph/0607332 [ arxiv : hep - ph/0607332 ] .
a. c. kraan , eur .
j. * c37 * , 91 - 104 ( 2004 ) .
[ hep - ex/0404001 ] .
j. l. hewett , b. lillie , m. masip , t. g. rizzo , jhep * 0409 * , 070 ( 2004 ) .
[ hep - ph/0408248 ] .
k. cheung and w. y. keung , phys .
d * 71 * , 015015 ( 2005 ) [ arxiv : hep - ph/0408335 ] .
a. c. kraan , j. b. hansen , p. nevski , eur .
j. * c49 * , 623 - 640 ( 2007 ) .
[ hep - ex/0511014 ] .
m. fairbairn , a. c. kraan , d. a. milstead , t. sjostrand , p. z. skands , t. sloan , phys .
* 438 * , 1 - 63 ( 2007 ) .
[ hep - ph/0611040 ] .
a. arvanitaki , s. dimopoulos , a. pierce , s. rajendran , j. g. wacker , phys .
rev . * d76 * , 055007 ( 2007 ) . [ hep - ph/0506242 ] .
t. sjostrand , s. mrenna and p. z. skands , jhep * 0605 * , 026 ( 2006 ) [ arxiv : hep - ph/0603175 ] .
j. conway , _ pgs : pretty good simulator _ , http://www.physics.ucdavis.edu/ conway / research / software / pgs / pgs4-general.htm .
e. aprile _ et al .
_ [ xenon100 collaboration ] , [ arxiv:1104.2549 [ astro-ph.co ] ] .
l. anchordoqui , h. goldberg and c. nunez , phys .
d * 71 * , 065014 ( 2005 ) [ arxiv : hep - ph/0408284 ] .
a. arvanitaki , p.
w. graham , phys .
* d72 * , 055010 ( 2005 ) .
[ hep - ph/0411376 ] .
|
recent results from the lhc for the higgs boson with mass between @xmath0 points to pev - scale split supersymmetry .
this article explores the consequences of a higgs mass in this range and possible discovery modes for split susy .
moderate lifetime gluinos , with decay lengths in the to range , are its imminent smoking gun signature .
the @xmath1 lhc will be sensitive to the moderately lived gluinos and trilepton signatures from direct electroweakino production .
moreover , the dark matter abundance may be obtained from annihilation through an @xmath2-channel higgs resonance , with the lsp almost purely bino and mass @xmath3 .
the higgs resonance region of split susy has visible signatures in dark matter direct and indirect detection and electric dipole moment experiments .
if the anomalies go away , the majority of split susy parameter space will be excluded .
|
ever since being proposed by pauli , neutrinos have been a sort of fascinating puzzle for the physicists .
the recent observation of non zero leptonic mixing angle @xmath0 @xcite-@xcite has provided significant boost to the sharpening of implications of the neutrino oscillations and has added another dimension to neutrino physics by implying the possibility of cp violation in the leptonic sector , hence deepening the flavor puzzle further .
the non zero value of @xmath0 , on the one hand , restores the parallelism between the mixings of quarks and leptons , on the other hand , its unexpectedly ` large ' value signifies the differences between these , the leptonic mixing angles being large as compared to the quark counterparts . in the absence of any viable theory of flavor dynamics for explaining the fermion masses and mixings
, approaches followed on the theoretical front can broadly be catagorized into ` top - down ' and ` bottom - up ' . despite large number of attempts using the ` top - down ' perspective @xcite , we have yet to arrive at a viable approach which accounts for the vast amount of data related to flavor mixings .
therefore , in the present work , we follow the ` bottom - up ' approach consisting of finding the phenomenological fermion mass matrices which are in tune with the latest low energy data . in this context , texture
specific mass matrices have got a good deal of attention @xcite and play an important role in explaining the flavor mixing phenomena . in particular , texture
specific mass matrices provide valuable information to explain the quark and lepton masses and mixings , for details we refer the reader to @xcite . in the context of leptons
, it needs to be mentioned that with the texture approach , minimal textures for dirac neutrinos have recently @xcite been ruled out , however , a detailed and comprehensive analysis for the non - minimal textures in light of the recent measurement of @xmath0 is yet to be carried out .
the purpose of the present work is to carry out a systematic and comprehensive study of non - minimal lepton mass matrices using the texture zero approach . in the context of neutrinos ,
the issue of these being dirac like or majorana particles is still an open question for the physicists since the dirac neutrinos have not yet been ruled by experimental data . in the present work ,
considering the neutrinos to be dirac like particles , we have made an attempt to examine the compatibility of non - minimal lepton mass matrix textures with the recent neutrino mixing data for all the three neutrino mass hierarchies i.e. normal , inverted and degenerate scenario of neutrino masses .
it may be noted that while carrying out the analysis , we consider the charged lepton mass matrices and the effective neutrino mass matrices having parallel structures i.e. texture zeroes being located at identical positions , in consonance with some classes of family symmetries and grand unified theories @xcite wherein such parallel structures emerge naturally .
the detailed plan of the paper is as follows . in section ( 2 )
we discuss the general lepton mass matrices in the standard model ( sm ) .
inputs used in the analysis are given in section ( 3 ) .
results and discussion pertaining to normal hierarchy , inverted hierarchy and degenerate scenario of neutrino masses for texture 2 zero , texture 4 zero and texture 5 zero lepton mass matrices are presented in section ( 4 ) .
finally , section ( 5 ) summarizes our conclusions .
in the standard model ( sm ) @xcite-@xcite of particle physics , the lepton mass matrices are arbitrary @xmath1 complex matrices , thus containing a total of 36 real parameters . using the ` polar decomposition theorem ' ,
the lepton mass matrices in sm can be considered to be hermitian without any loss of generality bringing down the number of free parameters from 36 to 18 . to facilitate the formulation of phenomenological mass matrices , which perhaps are compatible with the gut scale mass matrices , it has been suggested @xcite that in order to avoid fine tuning amongst the elements of the mass matrices
, these should follow a ` natural hierarchy ' i.e. @xmath2 .
the number of free parameters in these mass matrices can further be reduced using the facility of weak basis ( wb ) transformations .
these wb approaches broadly lead to two possibilities for the texture zero lepton mass matrices . in the first possibility observed by branco _
@xcite , one ends up with a structure wherein one of the matrices is a texture two zero type and the other is a one zero type , i.e. ( m_l)_(1,1)=(m_l)_(1,3)=(m_l)_(3,1)=(m_d)_(1,1)=0 , + or ( m_l)_(1,1)=(m_d)_(1,1)=(m_d)_(1,3)=(m_d)_(3,1)=0 , [ branc]@xmath3 and @xmath4 correspond to the charged lepton and the dirac neutrino mass matrix respectively . in the second possibility , given by fritzsch and xing @xcite ,
one ends up with a texture two zero structure for the lepton mass matrices , wherein both the lepton mass matrices assume a texture one zero structure , viz .
, ( m_l)_(1,3)=(m_l)_(3,1)=(m_d)_(1,3)=(m_d)_(3,1)=0 .
[ branc ] although the two possibilities are equivalent , however for the present work we have followed the fritzsch - xing approach , this choice can be justified as the mass matrices obtained by fritzsch - xing approach not only exhibit parallel texture structures but can also be diagonalized exactly making the construction of the corresponding lepton mixing matrix easier .
therefore , the general lepton mass matrices within the framework of sm can be given as m_l= ( c_l & a _ l & 0 + a_l^ * & d_l & b_l + 0 & b_l^ * & e_l ) , m_d= ( c _ & a _ & 0 + a_^ * & d _ & b _ + 0 & b_^ * & e _ ) , [ t20]with @xmath5 . following the methodology connecting the lepton mass matrices to the mixing matrix detailed in @xcite , one can carry out diagonalization of a general mass matrix @xmath6 by expressing it as m_k = q_k m_k^r p_k , where @xmath7 , @xmath8 are diagonal phase matrices given as diag@xmath9 and diag @xmath10 respectively and @xmath11 is a real symmetric matrix .
@xmath11 can be diagonalized by an orthogonal transformation @xmath12 , e.g. , m_k^diag = o_k^t m_k^r o_k which can be rewritten as m_k^diag = o_k^t q_k^m_k p_k^o_k .
the elements of the general diagonalizing transformation can figure with different phase possibilities , however , these possibilities are related to each other through the phase matrices @xcite .
for the present work , we have chosen the possibility @xmath13 where @xmath14
@xmath15 @xmath16 @xmath17 @xmath18 @xmath19 @xmath20 @xmath21 o_k(33)= , [ diat20 ] @xmath22 , -@xmath23 , @xmath24 being the eigen values of @xmath6 . using the above expressions , one can easily obtain the elements of @xmath25 , the diagonalizing matrix for the charged lepton sector by replacing @xmath22 , -@xmath23 , @xmath24 with @xmath26 , -@xmath27 , @xmath28 , e.g. , the first element of @xmath25 can be given as o_l(11)=. in an analogous manner , diagonalizing matrix @xmath29 for dirac neutrinos for the normal mass hierarchy ( nh ) given by @xmath30 and for the corresponding degenerate scenario given by @xmath31 can be obtained from equation ( [ diat20 ] ) by replacing @xmath22 , -@xmath23 , @xmath24 with @xmath32 , -@xmath33 , @xmath34 .
for instance , first element of @xmath29 can be given as o_d(11)=. similarly , one can obtain the elements of diagonalizing transformation for the inverted hierarchy ( ih ) case defined as @xmath35 and the corresponding degenerate case given by @xmath36 by replacing @xmath22 , -@xmath23 , @xmath24 with @xmath37 , -@xmath38 , -@xmath34 in eqn .
( [ diat20 ] ) , e.g. , the first element of @xmath29 in this scenario can be given as o_d(11)=. the other elements of the diagonalizing transformations for the charged lepton as well as neutrinos can be obtained in a similar way .
the lepton mixing matrix , the ` pontecorvo - maki - nakagawa - sakata ( pmns ) ' matrix @xcite can be obtained from these orthogonal transformations using the relation u = o_l^q_l p_d o_d , where @xmath39 can be taken as diag(@xmath40 , 1 , @xmath41 ) .
the parameters @xmath42 and @xmath43 can be considered as free parameters and are related to the phases of mass matrices as @xmath44 , @xmath45 .
before getting into the details of the analysis , we would like to mention some of the essentials pertaining to various inputs . in the present analysis ,
we have made use of the results of a latest global three neutrino oscillation analysis @xcite , in table ( [ data ] ) we present the @xmath46 and @xmath47 ranges of the neutrino oscillation parameters .
c c c parameter & @xmath48 range & @xmath49 range + [ 0.5ex ] @xmath50 @xmath51 $ ] & ( 7.32 - 7.80 ) & ( 6.99 - 8.18 ) + @xmath52 @xmath53 $ ] & ( 2.33 - 2.49)(nh ) ; ( 2.31 - 2.49 ) ( ih ) & ( 2.19 - 2.62)(nh ) ; ( 2.17 - 2.61)(ih ) + @xmath54 @xmath55 $ ] & ( 2.16 - 2.66)(nh ) ; ( 2.19 - 2.67)(ih ) & ( 1.69 - 3.13)(nh ) ; ( 1.71 - 3.15 ) ( ih ) + @xmath56 @xmath57 $ ] & ( 2.91 - 3.25 ) & ( 2.59 - 3.59 ) + @xmath58 @xmath57 $ ] & ( 3.65 - 4.10)(nh);(3.70 - 4.31)(ih ) & ( 3.31 - 6.37)(nh);(3.35 - 6.63)(ih ) + while carrying out the analysis , the lightest neutrino mass , @xmath22 for the case of nh and @xmath24 for the case of ih , is considered as a free parameter .
for all the three possible mass hierarchies of neutrinos i.e. normal , inverted and degenerate scenario , the explored range of the lightest neutrino mass is taken to be @xmath59 , our conclusions remain unaffected even if the range is extended further . in the absence of any constraint on the phases , @xmath42 and @xmath43 have been given full variation from 0 to @xmath60 .
although @xmath61 and @xmath62 are free parameters , however , they have been constrained such that diagonalizing transformations @xmath25 and @xmath63 always remain real .
before discussing the results , we would like to briefly discuss the possible non - minimal textures which we plan to analyse and discuss . while considering such possibilities , we ignore wb related mass matrix structures , in particular those related through permutations , as in the case of parallel textures these are equivalent . in the case of texture two zero
matrices , therefore we have to consider one possibility only , which we discuss for the case of three hierarchies of neutrino masses . beyond texture two zero
, we have to make specific assumptions of taking particular elements of @xmath3 and @xmath4 being zero . in table ( [ 3t4 ] ) we have considered four classes with their permutations covering all possibilities .
texture five zero mass matrices can easily be derived from the corresponding texture four zero ones . to examine the compatibility of texture two zero lepton mass matrices given in equation ( [ t20 ] ) with the recent mixing data , we carry out a detailed analysis pertaining to all three possible neutrino mass hierarchies . to this end , in figure ( [ t2ih1 ] ) we present the plots showing the parameter space of two mixing angles , with the third one being constrained by its @xmath64 experimental bounds for inverted hierarchy .
the rectangular boxes in these figures show the @xmath49 ranges for the two mixing angles being considered in the figure .
as is evident from these figures , the parameter space for the mixing angles shows considerable overlap with the experimentally allowed @xmath49 region .
therefore , inverted hierarchy seems to be viable for texture two zero lepton mass matrices given in equation ( [ t20 ] ) .
cc after discussing the viability of inverted hierarchy for texture two zero mass matrices , we now examine the compatibility of these matrices for the normal hierarchy case . to this end , in figure ( [ t2nh1 ] ) we present the plots showing the parameter space allowed for two mixing angles when the third one is constrained by its @xmath64 experimental bound for normal neutrino mass hierarchy .
a general look at the figure ( [ t2nh1 ] ) reveals that the structure given in eqn.([t20 ] ) is compatible with the normal neutrino mass hierarchy .
cc further , in figures ( [ t2nh2 ] ) and ( [ t2nh3 ] ) we present the graphs showing variation of the mixing angles with the parameters @xmath65 and @xmath66 respectively for structure ( [ t20 ] ) pertaining to normal hierarchy of neutrino masses . while plotting these figures , all the three mixing angles have been constrained by their @xmath67 experimental bounds , while all the free parameters have been given full variation .
taking a careful look at these plots it is interesting to note that the leptonic mixing angles donot have much dependence on the parameters @xmath68 and @xmath69 .
further one can see that a fit for all the three mixing angles can be obtained for the values of parameters being @xmath70 and @xmath71 .
cc cc to further emphasise this conclusion , we present in figure ( [ t2ih2 ] ) the plots showing the dependence of leptonic mixing angle @xmath72 on @xmath65 and @xmath73 , constraining the other two angles by their @xmath49 experimentally allowed ranges for inverted hierarchy of neutrino masses .
the two parallel lines in these figures show the @xmath49 allowed range for the mixing angle @xmath72 .
figure ( [ t2ih2 ] ) clearly shows that in order to accomodate the latest @xmath49 ranges for the mixing parameters , one requires @xmath65 and @xmath73 @xmath74 1 .
therefore , one can conclude that very small values of the parameters @xmath68 and @xmath69 are required to fit the latest experimental data for the structure ( [ t20 ] ) corresponding to both normal as well as inverted neutrino mass orderings and the structure ( [ t20 ] ) thus essentially reduces to m_l= ( 0 & a _ l & 0 + a_l^ * & d_l & b_l + 0 & b_l^ * & e_l ) , m_d= ( 0 & a _ & 0 + a_^ * & d _ & b _ + 0 & b_^ * & e _ ) [ flt40]the structure ( [ flt40 ] ) is referred to as fritzsch - like texture four zero structure and is studied extensively in literature @xcite .
however , no such attempt has been made after the recent measurement of the mixing angle @xmath72 .
therefore , it becomes interesting to analyse this structure in detail for its compatibility with the latest lepton mixing data . as discussed in the previous section , starting with the most general lepton mass matrices having ` natural hierarchy ' @xcite among their elements , one essentially arrives at the fritzsch - like texture four zero structure as given in eqn.([flt40 ] ) . in this context , it is interesting to note that different four - zero texture parallel matrices , with zeros located in different positions , may have exactly the same physical content .
indeed , they can be related by a wb transformation , performed by a permutation matrix p , m_l^= p^t m_l p + m_^= p^t m _
p which automatically preserves the parallel structure , but changes the position of the zeros .
& class i & class ii & class iii & class iv + & & & & + a & @xmath75 & @xmath76 & @xmath77 & @xmath78 + & & & & + & & & & + b & @xmath79 & @xmath80 & @xmath81 & @xmath82 + & & & & + & & & & + c & @xmath83 & @xmath84 & @xmath85 & @xmath86 + & & & & + & & & & + d & @xmath87 & @xmath88 & @xmath89 & @xmath90 + & & & & + & & & & + e & @xmath91 & @xmath92 & @xmath93 & @xmath94 + & & & & + & & & & + f & @xmath95 & @xmath96 & @xmath97 & @xmath86 + the matrix p belongs to the group of six permutation matrices , which are isomorphic to @xmath98 .
the four - zero texture ansatz can then be classified into four classes , as indicated in table ( [ 3t4 ] ) .
it can easily be seen that the matrices given in eqn.([flt40 ] ) belong to class i of table ( [ 3t4 ] ) . in this context , it becomes interesting to confront all the classes of lepton mass matrices with the latest experimental data . from the table , it is clear that class iv is not viable as all the matrices in this class correspond to the scenario where one of the generations gets decoupled from the other two .
such classification of all possible parallel texture four zero structures was first presented in the analysis by branco _
@xcite wherein they carry out detailed study of lepton mass matrices corresponding to all the four classes pertaining to majorana neutrinos .
however , such analysis has not been carried out considering the neutrinos to be dirac like particles .
therefore , in this section , we carry out a detailed study of all classes of texture four zero lepton mass matrices assuming the neutrinos to be dirac type . to begin with , we carry out a detailed analysis for texture four zero mass matrices belonging to class i of table ( [ 3t4 ] ) , i.e. , m_i= ( 0 & a _ ie^i_i & 0 + a_ie^-i_i & d_i & b_ie^i_i + 0 & b_ie^-i_i & e_i ) , [ cl1 ] where @xmath99 corresponds to the charged lepton and dirac neutrino mass matrices respectively .
the diagonalizing transformations for this class can be obtained by substituting @xmath100 in eqn .
( [ diat20 ] ) . for the purpose of calculations ,
the elements @xmath101 , @xmath102 as well as the phases @xmath42 and @xmath43 have been considered as free parameters . following the methodology
as discussed in section ( [ metho ] ) , we attempt to carry out a detailed study pertaining to normal , inverted as well as degenerate neutrino mass orderings .
firstly , we examine the compatibility of mass matrices given in eqn .
( [ flt40 ] ) with the inverted hierarchy of neutrino masses .
for this purpose in figures ( [ t4ih1 ] ) and ( [ t4ih2 ] ) , we present the parameter space corresponding to any two mixing angles while the third one being constrained by its @xmath48 and @xmath67 range respectively . the blank rectangular regions in these figures represent the @xmath67 allowed ranges of the two mixing angles being considered .
as can be seen from all these plots , the parameter space of two angles does not show any overlap with the experimentally allowed region .
therefore , we find that the inverted hierarchy seems to be ruled out at both the @xmath46 as well as the @xmath103 level for class i ansatz of texture four zero dirac neutrino mass matrices . after ruling out the structure given in eqn.([cl1 ] ) for the inverted hierarchy , we now proceed to examine the compatibility of these matrices for the normal hierarchy case . to this end , in figure ( [ t4nh1 ] ) we present the plots showing the parameter space corresponding to any two mixing angles wherein the third one is constrained by its @xmath48 range .
interestingly , normal hierarchy seems to be viable in this case as can be seen from these plots , wherein the parameter space shows significant overlap with the experimentally allowed @xmath49 region shown by the rectangular boxes in each plot .
further , in figure ( [ t4nh2 ] ) we present the graphs showing the variation of the lightest neutrino mass with the mixing angles for normal hierarchy , keeping the other two mixing angles constrained by their @xmath67 bounds .
the parallel lines in each plot show the @xmath67 range of the mixing angle being considered . taking a careful look at these graphs
, one can find the range the lightest neutrino mass to be @xmath104 approximately .
next , in order to explore the possibilty of cp violation in the leptonic sector , in figure ( [ t4nh3 ] ) we study the dependence of jarlskog s rephasing invariant parameter @xmath105 on each of the three mixing angles , while keeping the other two mixing angles constrained by their @xmath67 ranges . from these graphs
, there appears to be significant possibility of cp violation in the leptonic sector and the range for the magnitude of @xmath105 seems to be @xmath106 approximately .
cc cc cc cc cc the degenerate scenario of the neutrino masses can be characterized by either @xmath107 or @xmath108 , corresponding to normal hierarchy and inverted hierarchy respectively .
since while carrying out the calculations pertaining to both the normal as well as inverted hierarchy cases , the range of lightest neutrino mass is taken to be @xmath109 ev , which includes the neutrino masses corresponding to the degenerate scenario ; therefore , by discussion similar to the one given for ruling out inverted hierarchy , class i ansatz seems to be ruled out for degenerate scenario of neutrino masses as well .
similarly , from figure ( [ t4nh2 ] ) the value of the lightest neutrino mass pertaining to the degenerate scenario , @xmath110 , seems to be outside the experimentally allowed region , thereby ruling out class i ansatz for both the cases of degenerate scenario of neutrino masses .
to analyse this class we follow the same procedure as for class i ansatz . the charged lepton and neutrino mass matrices which we choose to analyse for this class can be given as , m_i= ( d_i & a _ ie^i_i & 0 + a_ie^-i_i & 0 & b_ie^i_i + 0 & b_ie^-i_i & e_i ) , [ cl2 ] where @xmath99 corresponds to the charged lepton and dirac neutrino mass matrices respectively . as for class i , the elements @xmath101 and @xmath102 as well as the phases @xmath42 and @xmath43 are considered to be the free parameters .
firstly , we examine the viability of inverted hierarchy for the structure given in eqn.([cl2 ] ) . to this end in figures ( [ 4aih1 ] ) and ( [ 4aih2 ] ) , we present the plots showing the parameter space for two mixing angles wherein the third angle is constrained by its @xmath64 range and @xmath49 range respectively . the rectangular boxes in these plots show the @xmath67 ranges for the two mixing angles being considered .
it is interesting to note that the inverted hierarchy seems to be ruled at by the @xmath64 ranges for the present mixing data , while it seems to be compatible with the @xmath67 experimental bounds .
cc cc cc cc cc cc cc next , in order to examine the compatibility of structure given in eqn.([cl2 ] ) with the normal hierarchy , in figure ( [ 4anh1 ] ) we present the plots showing the parameter space for two mixing angles wherein the third angle is constrained by its @xmath64 experimental range . a general look at figure ( [ 4anh1 ] ) reveals that normal hierarchy is compatible with the structure ( [ cl2 ] ) at @xmath64 level .
thus , we find that both the normal as well as inverted neutrino mass hierarchies are compatible with the @xmath67 ranges for the lepton mixing data . as a next step , in figures ( [ 4anh2 ] ) and ( [ 4aih3 ] ) we attempt to study the implications of the @xmath67 ranges of the leptonic mixing angles on the lightest neutrino mass for this class for the normal and inverted neutrino mass hierarchy respectively . as can be seen from these figures , for both the neutrino mass hierarchies the @xmath67 ranges of the mixing angles provide only an upper bound on the lightest neutrino mass , viz .
@xmath111 and @xmath112 for the normal and inverted hierarchy respectively
. this bound seems to rule out both the degenerate neutrino mass ordering scenarios as the bound on the lightest neutrino mass does not include the value of the lightest neutrino mass pertaining to the degenerate scenario @xmath113 .
further , we study the possiblity of cp violation in the leptonic sector for this class by studying the variation of jarlskog s rephasing invariant @xmath114 with all the mixing angles for normal as well as inverted neutrino mass hierarchy in figures ( [ 4anh3 ] ) and ( [ 4aih4 ] ) respectively . while plotting these graphs the two mixing angles , other than the one being considered , are constrained by their @xmath67 ranges .
the parallel lines in these plots show the @xmath67 experimental ranges for the mixing angle being considered .
it is interesting to note that for the case of normal hierarchy one gets a broader range for @xmath114 as compared to the one for the case of inverted hierarchy . to be more explicit
, one finds @xmath115 and @xmath116 corresponding to normal and inverted hierarchy respectively . to study the lepton mass matrices for this class , we choose to analyse the following structure , m_i= ( 0 & a _
ie^i_i & b_ie^i_i + a_ie^-i_i & 0 & d_ie^i_i + b_ie^-i_i & d_ie^-i_i & e_i ) , [ cl3 ] where @xmath99 corresponds to the charged lepton and dirac neutrino mass matrices respectively
. in the case of factorizable phases , the lepton mass matrices belonging to this class can be analysed following a methodology similar to that for class i and class ii ansatz .
for the purpose of calculations , the ( 2,3 ) element in each sector , @xmath101 and @xmath102 , as well as the phases @xmath42 and @xmath43 are considered to be the free parameters .
it is interesting to note that in this class as well both the normal as well as inverted mass neutrino mass orderings seem to be viable as can be seen from figures ( [ 5anh1 ] ) and ( [ 5aih1 ] ) .
further , in figures ( [ 5anh2 ] ) and ( [ 5aih2 ] ) we attempt to study the dependence of the leptonic mixing angles on the lightest neutrino mass for this class for the normal and inverted neutrino mass hierarchy respectively .
as can be seen from these figures , the lightest neutrino mass seems to be unrestricted by the @xmath67 ranges of the mixing angles for both the the neutrino mass hierarchies .
since the value of the lightest neutrino mass pertaining to the degenerate scenario @xmath113 is included in the range allowed by the @xmath67 ranges of the mixing angles , therefore both the degenerate scenarios may be viable for this class of lepton mass matrices .
cc cc cc further , we study the possiblity of cp violation in the leptonic sector for this class by studying the variation of jarlskog s rephasing invariant @xmath114 with all the mixing angles for normal as well as inverted neutrino mass hierarchy in figures ( [ 5anh3 ] ) and ( [ 5aih3 ] ) respectively . while plotting these graphs the two mixing angles , other than the one being considered , are constrained by their @xmath67 ranges .
the parallel lines in these plots show the @xmath67 experimental ranges for the mixing angle being considered .
it is interesting to note that for the case of inverted hierarchy one gets a broader range for @xmath114 as compared to the one for the case of normal hierarchy , contrary to the observation in class ii ansatz . to be more explicit
, one finds @xmath115 and @xmath117 corresponding to normal and inverted hierarchy scenario respectively .
cc cc cc after studying all possible texture four zero lepton mass matrices , it becomes interesting to explore the parallel texture five zero structures for each class which can be derived by substituting either @xmath118 , @xmath119 or @xmath120 , @xmath121 in the corresponding texture four zero mass matrices . in this subsection
, we carry out a detailed study of all classes of texture five zero lepton mass matrices for dirac neutrinos pertaining to both the possibilities leading to texture five zero structures . the two possibilities for texture five zero lepton mass matrices for this class can be given as , m_l= ( 0 & a _ l & 0 + a_l^ * & 0 & b_l + 0 & b_l^ * & e_l ) , m_= ( 0 & a _ & 0 + a_^ * & d _ & b _ + 0 & b_^ * & e _ ) , [ cl1t51]or m_l= ( 0 & a _ l & 0 + a_l^ * & d_l & b_l + 0 & b_l^ * & e_l ) , m_= ( 0 & a _ & 0 + a_^ * & 0 & b _ + 0 & b_^ * & e _ ) . [ cl1t52 ]
we study both these possibilities in detail for all the neutrino mass orderings .
firstly , we examine the compatibility of matrices ( [ cl1t51 ] ) and ( [ cl1t52 ] ) with the inverted hierarchy of neutrino masses . for this purpose , in figures ( [ t5cl1ih1 ] ) and ( [ t5cl1ih2 ] ) , we present the plots showing the parameter space allowed by this ansatz for any two mixing angles wherein the third one is constrained by its @xmath67 experimental bound for inverted hierarchy of neutrino masses .
the rectangular regions in these plots represent the @xmath67 ranges for the two mixing angles being considered . a general look at these plots shows that inverted hierarchy is ruled out for both the texture five zero possibilities for this class .
cc cc cc cc cc cc after ruling out the structures ( [ cl1t51 ] ) and ( [ cl1t52 ] ) for the inverted hierarchy , we now proceed to examine the compatibility of these matrices for the normal hierarchy case . to this end , in figures ( [ t5cl1nh1 ] ) and ( [ t5cl1nh2 ] ) , we present the plots showing the parameter space corresponding to any two mixing angles wherein the third one is constrained by its @xmath67 range . interestingly , normal hierarchy seems to be ruled out for the case @xmath122 and @xmath123 , whereas for the case @xmath119 and @xmath118 it seems to be viable as can be seen from the figure ( [ t5cl1nh1 ] ) , wherein the parameter space shows significant overlap with the experimentally allowed @xmath67 region shown by the rectangular boxes in each plot . for the @xmath124 and @xmath125 case for texture five zero mass matrices , wherein normal hierarchy has been shown to be viable
, we proceed to study the dependence of the lightest neutrino mass and jarlskog s parameter on the the leptonic mixing angles . to this end , we present the plots showing variation of the lightest neutrino mass and jarlskog s parameter with the mixing angles in figures ( [ cl1t5nh2 ] ) and ( [ cl1t5nh3 ] ) respectively . while plotting these graphs , the other two mixing angles have been constrained by their @xmath67 ranges .
the parallel lines in these plots show the @xmath67 experimental ranges for the mixing angle being considered . having a careful look at these plots
, one can find the ranges for the lightest neutrino mass and the jarlskog s parameter for this case of texture five zero matrices , viz .
, @xmath126 , @xmath127 .
further , since inverted hierarchy is ruled out for both the cases for texture five zero and for the normal hierarchy the range of the lightest neutrino mass does not include that for the degenerate neutrino mass ordering , therefore both the cases for degenerate scenario seems to be ruled out for this class .
the two possibilities for texture five zero lepton mass matrices for this class can be given as , m_l= ( 0 & a _ l & 0 + a_l^ * & 0 & b_l + 0 & b_l^ * & e_l ) , m_= ( d _ & a _ & 0 + a_^ * & 0 & b _ + 0 & b_^ * & e _ ) , [ cl2t51]or m_l= ( d_l & a _ l & 0 + a_l^ * & 0 & b_l + 0 & b_l^ * & e_l ) , m_= ( 0 & a _ & 0 + a_^ * & 0 & b _ + 0 & b_^ * & e _ ) , [ cl2t52 ] we study both these possibilities in detail for all the neutrino mass orderings .
firstly , we examine the compatibility of matrices ( [ cl2t51 ] ) and ( [ cl2t52 ] ) with the inverted hierarchy of neutrino masses . for this purpose , in figures ( [ t5cl2ih1 ] ) and ( [ t5cl2ih2 ] ) , we present the plots showing the parameter space allowed by this ansatz for any two mixing angles wherein the third one is constrained by its @xmath67 experimental bound for inverted hierarchy of neutrino masses . the rectangular regions in these plots represent the @xmath67 ranges for the two mixing angles being considered .
interestingly , one finds that for the case @xmath124 and @xmath125 of texture five zero lepton mass matrices inverted hierarchy is ruled out , whereas for the case @xmath123 and @xmath122 of texture five zero lepton mass matrices inverted hierarchy scenario seems to be viable . for the @xmath123 and @xmath128 case of lepton mass matrices ,
wherein inverted hierarchy is shown to be viable , we proceed next to study the the dependence of the lightest neutrino mass and jarlskog s parameter on the the leptonic mixing angles . to this end , we present the plots showing variation of the lightest neutrino mass and jarlskog s parameter with the mixing angles in figures ( [ t5cl2ih3 ] ) and ( [ t5cl2ih4 ] ) respectively . while plotting these graphs , the other two mixing angles have been constrained by their @xmath67 ranges .
interestingly , for this case one finds a very narrow range for both the lightest neutrino mass as well the jarlskog s parameter as can be seen from figures ( [ t5cl2ih3 ] ) and ( [ t5cl2ih4 ] ) .
cc cc cc cc cc cc cc cc after studying both the cases for texture five zero mass matrices pertaining to class ii ansatz for inverted hierarchy , we now carry out a similar analysis pertaining to normal hierarchy . to this end , in figures ( [ t5cl2nh1 ] ) and ( [ t5cl2nh2 ] ) , we present the plots showing the parameter space corresponding to any two mixing angles wherein the third one is constrained by its @xmath67 range . interestingly , normal hierarchy seems to be viable for both the cases , @xmath122 and @xmath123 as well as @xmath119 and @xmath118 , of texture five zero lepton mass matrices as can be seen from these plots ( [ t5cl2nh1 ] ) and ( [ t5cl2nh2 ] ) , wherein the parameter space shows significant overlap with the experimentally allowed @xmath67 region shown by the rectangular boxes in each plot .
next , we study the dependence of the lightest neutrino mass and jarlskog s parameter on the the leptonic mixing angles for this case . to this end
, we present the plots showing variation of the lightest neutrino mass with the mixing angles in figures ( [ t5cl2nh3 ] ) and ( [ t5cl2nh4 ] ) respectively . while plotting these graphs , the other two mixing angles have been constrained by their @xmath67 ranges .
interestingly , for @xmath123 and @xmath129 , the lightest neutrino mass is largely unrestricted , whereas for the case @xmath118 and @xmath119 one obtains an upper bound @xmath130 for the lightest neutrino mass .
cc cc further , in figures ( [ t5cl2nh5 ] ) and ( [ t5cl2nh6 ] ) we present the graphs showing the variation of the jarlskog s parameter with each of the mixing angles , keeping the other two constrained by their @xmath67 ranges .
parallel lines in these graphs show the @xmath67 ranges for the mixing angles being considered .
interestingly , one finds that the range of j for @xmath131 and @xmath122 is quite narrow as compared to the @xmath132 and @xmath119 case of texture five zero lepton mass matrices for class ii ansatz . the two possibilities for texture five zero lepton mass matrices for this class can be given as , m_l= ( 0 & a _ le^i_l & b_l + a_le^-i_l & 0 & 0 + b_l & 0 & e_l ) , m_= ( 0 & a _ e^i _ & b
_ + a_e^-i _ & 0 & d_e^i _ + b _ & d_e^-i _ & e _ ) , [ cl3t51 ] or m_l= ( 0 & a _ le^i_l & b_l + a_le^-i_l & 0 & d_le^i_l + b_l & d_le^-i_l & e_l ) , m_= ( 0 & a _ e^i _ & b
_ + a_e^-i _ & 0 & 0 + b _ & 0 & e _ ) , [ cl3t52 ] we study both these possibilities in detail for all the neutrino mass orderings .
firstly , we examine the compatibility of matrices ( [ cl3t51 ] ) and ( [ cl3t52 ] ) with the inverted hierarchy of neutrino masses . for this purpose , in figures ( [ t5cl3ih1 ] ) and ( [ t5cl3ih2 ] ) , we present the plots showing the parameter space allowed by this ansatz for any two mixing angles wherein the third one is constrained by its @xmath67 experimental bound for inverted hierarchy of neutrino masses . the rectangular regions in these plots represent the @xmath67 ranges for the two mixing angles being considered .
interestingly , one finds that for the case @xmath124 and @xmath125 of texture five zero lepton mass matrices inverted hierarchy is ruled out , whereas for the case @xmath123 and @xmath122 of texture five zero lepton mass matrices inverted hierarchy scenario seems to be viable . for the @xmath123 and @xmath128 case of lepton mass matrices ,
wherein inverted hierarchy is shown to be viable , we proceed next to study the the dependence of the lightest neutrino mass and jarlskog s parameter on the the leptonic mixing angles . to this end , we present the plots showing variation of the lightest neutrino mass with the mixing angles in figures ( [ t5cl3ih3 ] ) . while plotting these graphs , the other two mixing angles have been constrained by their @xmath67 ranges .
interestingly , one finds that the lightest neutrino mass is unrestricted for this structure .
further , on carrying out the calculations for the jarlskog s parameter , one finds that the @xmath67 experimental bounds for the mixing angles allow an exteremely narrow range @xmath133 0.05 for jarlskog s parameter .
cc cc cc cc cc cc cc after studying both the cases for texture five zero mass matrices for inverted hierarchy pertaining to class ii ansatz , we now carry out a similar analysis pertaining to normal hierarchy . to this end , in figures ( [ t5cl3nh1 ] ) and ( [ t5cl3nh2 ] ) , we present the plots showing the parameter space corresponding to any two mixing angles wherein the third one is constrained by its @xmath67 range .
interestingly , normal hierarchy seems to be ruled out for the case @xmath123 and @xmath122 , whereas for the case @xmath118 and @xmath119 of texture five zero lepton mass matrices normal hierarchy seems to be viable .
next , we study the dependence of the lightest neutrino mass and jarlskog s parameter on the the leptonic mixing angles for the @xmath118 and @xmath119 case of texture five zero mass matrices corresponding to class iii . to this end
, we present the plots showing variation of the lightest neutrino mass and jarlskog s parameter with the mixing angles in figures ( [ t5cl3nh3 ] ) and ( [ t5cl3nh4 ] ) respectively . while plotting these graphs , the other two mixing angles have been constrained by their @xmath67 ranges .
interestingly , one finds that the lightest neutrino mass is largly unrestricted , whereas for jarlskog s parameter one approximately obtains a range , viz . , @xmath134 .
further , since the lightest neutrino mass is unrestricted , therefore the degenerate scenario pertaining to normal hierarchy can not be ruled out for this structure .
to summarize , for dirac neutrinos , we have carried out detailed calculations pertaining to non minimal textures characterized by texture two zero fritzsch - like structure as well as all possibilities for texture four zero and five zero lepton mass matrices . corresponding to these ,
we have considered all the three possibilities for neutrino masses i.e. normal , inverted as well as degenerate scenarios .
the compatibility of these texture specific mass matrices has been examined by plotting the parameter space corresponding to any two of the leptonic mixing angles .
further , for all the structures which seem to be compatible with the recent lepton mixing data , the implications of the mixing angles on the lightest neutrino mass as well as the jarlskog parameter have also been studied .
the analysis reveals that the fritzsch like texture two zero lepton mass matrices are compatible with the recent lepton mixing data pertaining to normal as well as inverted neutrino mass hierarchies .
interestingly , one finds that both the normal as well as inverted neutrino mass hierarchies are compatible with texture four zero mass matrices pertaining to class ii and iii contrary to the case for texture four zero mass matrices pertaining to class i wherein inverted hierarchy seems to be ruled out .
none of the two possibilities pertaining to degenerate neutrino mass scenario is compatible with texture four zero mass matrices in class i and ii , whereas degenerate scenario can not be ruled out for texture four zero mass matrices in class iii .
mass matrices in class iv are phenomenologically excluded . for texture five zero lepton mass matrices , we analyse both the cases , viz . @xmath135 as well as @xmath136 for all the three phenomenologically viable classes . for texture five
zero matrices pertaining to class i , inverted hierarchy is ruled out for both the cases , whereas normal hierarchy is viable for the @xmath137 case . for class ii ,
normal hierarchy is viable for both the cases while the inverted hierarchy is ruled out for the case @xmath137 . finally , for texture
five zero mass matrices pertaining to class iii we find that inverted hierarchy is viable for the case @xmath136 , while the normal hierarchy is compatible with the @xmath137 case
. 0.5 cm * acknowledgements * + s.s . would like to acknowledge ugc , govt . of india , for financial support .
g.a . would like to acknowledge dst , government of india ( grant no : sr / ftp / ps-017/2012 ) for financial support .
acknowledge the chairperson , department of physics , p.u .
, for providing facilities to work
. a. yu .
smirnov , hep - ph/0604213 ; c. d. froggatt and h. b. nielsen , nucl .
b * 147 * , 277 ( 1979 ) ; y. nir and n. seiberg , phys .
b * 309 * , 337 ( 1993 ) ; m. leurer , y. nir and n. seiberg , nucl .
b * 420 * , 468 ( 1994 ) ; l. e. ibanez and g. g. ross , phys .
b * 332 * , 100 ( 1994 ) ; g. altarelli and f. feruglio , hep - ph/1002.0211 ; a. j. buras , c. grojean , s. pokorski and r. ziegler , hep - ph/1105.3725 and references therein ; z. guo and b. ma , jhep * 0909 * , 091 ( 2009 ) and references therein .
h. fritzsch , z. z. xing , prog .
* 45 * 1 ( 2000 ) , and references therein ; z. z. xing , int . jour . of mod .
phys . a * 19 * 1 ( 2004 ) , and references therein , hep - ph/9912358 ; m. gupta , g. ahuja , int .
jour . of mod .
a * 26 * 2973 ( 2011 ) ; n. g. deshpande , m. gupta and p. b. pal , phys .
d * 45 * , 953 ( 1992 ) ; p. s. gill and m. gupta , pramana * 45
* , 333 ( 1995 ) ; _ ibid_. j. phys .
g * 23 * , 335 ( 1997 ) ; _ ibid .
d * 56 * , 3143 ( 1997 ) ; m. randhawa and m. gupta , phys .
d * 63 * , 097301 ( 2001 ) ; m. randhawa , g. ahuja and m. gupta , phys . lett .
b 643 , 175 ( 2006 ) ; g. ahuja , s. kumar , m. randhawa , m. gupta and s. dev , phys . rev .
d * 76 * , 013006 ( 2007 ) ; g. ahuja , m. gupta , m. randhawa and r. verma , phys .
d * 79 * , 093006 ( 2009 ) ; r. verma , g. ahuja and m. gupta , phys . lett .
b * 681 * , 330 ( 2009 ) .
stuart raby , phys.lett .
b * 561 * 119 ( 2003 ) ; k. s. babu , jogesh c. pati and parul rastogi phys
d * 71 * , 015005 ( 2005 ) ; k.s .
babu , jogesh c. pati and frank wilczek nucl .
b * 566 * , 33 ( 2000 ) ; s.m.barr phys .
rev.lett . * 64 * , 353 , 1990 ; k.s .
babu and s.m.barr , phys .
* 75 * , 11(1995 ) ; w. buchmller and d. wyler phys . lett .
b * 521 * 291 ( 2001 ) .
h. fritzsch , m. gell - mann and h. leutwyler , phys . lett .
* b 47 * , 365 ( 1973 ) .
glashow , nucl .
* 22 * , 597 ( 1961 ) ; s. weinberg , phys .
* 19 * , 1264 ( 1967 ) ; a. salam , in _ elementary particle theory _
n. svartholm ( almquist and wiksells , stockholm , 1969 ) . for excellent reviews on the standard model
see , j.f .
donoghue , e. golowich and b.r .
holstein , _ dynamics of the standard model _ , ( cambridge university press , 1992 ) .
g. ahuja , m. gupta , m. randhawa and rohit verma , phys .
d * 79 * , 093006 ( 2009 ) ; g. ahuja , s. kumar , m. randhawa , m. gupta and s. dev , phys .
d * 76 * , 013006 ( 2007 ) ; m. randhawa , g. ahuja and m. gupta , phys .
d * 65 * , 093016(2002 ) .
|
in the light of the recent measurement of the leptonic mixing angle @xmath0 , implications of the latest mixing data have been investigated for non - minimal textures of lepton mass matrices pertaining to dirac neutrinos .
all these texture specific lepton mass matrices have been examined for their compatibility with the latest data in the cases of normal hierarchy , inverted hierarchy and degenerate scenario of neutrino masses .
the implications of all the three lepton mixing angles have been investigated on the lightest neutrino mass as well as the jarlskog s cp violating parameter in the leptonic sector .
|
the heider balance @xcite is a final state of personal relations between members of a society , reached when these relations evolve according to some dynamical rules .
the relations are assumed to be symmetric , and they can be friendly or hostile . the underlying psycho - sociological mechanism of the rules is an attempt of the society members to remove a cognitive dissonance , which we feel when two of our friends hate each other or our friend likes our enemy . as a result of the process ,
the society is split into two groups , with friendly relations within the groups and hostile relations between the groups . as a special case ,
the size of one group is zero , i.e. all hostile relations are removed .
hb is the final state if each member interacts with each other ; in the frames of the graph theory , where the problem is formulated , the case is represented by a fully connected graph .
recently a continuous dynamics has been introduced to describe the time evolution of the relations @xcite . in this approach ,
the relations between nodes @xmath0 and @xmath1were represented by matrix elements @xmath2 , which were real numbers , friendly ( @xmath3 ) or hostile ( @xmath4 . as a consequence of the continuity , we observed a polarization of opinions : the absolute values of the matrix elements @xmath2 increase . here
we continue this discussion , but the condition of maximal connectivity is relaxed , as it could be unrealistic in large societies .
the purpose of first part of this work is to demonstrate , that even if hb is not present , the above mentioned polarization remains true . in section
ii we present new numerical results for a society of @xmath5 members , represented by barabsi - albert ( ba ) network @xcite .
although this size of considered social structure is rather small , it is sufficient to observe some characteristics which are different than those in the exponential networks . in second part ( section iii )
we compare the results of our equations of motion with some examples , established in the literature of the subject .
the section is closed by final conclusions .
the time evolution of @xmath2 is determined by the equation of motion @xcite @xmath6 where @xmath7 is a sociologically justified limitation on the absolute value of @xmath2 @xcite . here
initial values of @xmath2 are random numbers , uniformly distributed in the range @xmath9 . the equation is solved numerically with the runge - kutta iv method with variable length of timestep @xcite , simultaneously for all pairs @xmath10 of linked nodes .
the method of construction of ba networks was described in @xcite .
the connectivity parameter is selected to be @xmath11 , because in this case the probability @xmath12 of hb has a clear minimum for ba networks of @xmath5 nodes , and @xmath13 ( see fig .
this choice of @xmath14 is motivated by our aim to falsify the result on the polarization of opinions .
this polarization was demonstrated @xcite to be a consequence of hb ; therefore , the question here is if it appears also when hb is not present .
an example of time evolution of such a network is shown in fig .
our result is that the polarization is present in all investigated cases . as time increases ,
the distribution of @xmath2 gets wider and finally it reaches a stable shape , with two large peaks at @xmath15 and one smaller peak at the centre , where @xmath16 . in fig .
3 , we show a series of histograms of @xmath17 in subsequent times ( a - e ) .
particular networks differ quantitatively with respect to the heights of the peaks , but these differences are small .
we note here that when some links are absent , the definition of hb should be somewhat relaxed , because some other links , which do not enter to any triad @xmath18 , will not evolve at all .
therefore we should admit that some negative relations survive within a given group .
we classify a final state of the graph as hb if there are no chains of friendly relations between the subgroups . on the other hand
, more than two mutually hostile subgroups can appear .
these facts were recognized already in literature @xcite .
surprisingly enough , subgroups of @xmath19 nodes are never found in our ba networks . on the contrary , in the exponential networks groups of all sizes were detected . in figs .
4 and 5 we show diagrams for ba networks and exponential networks , respectively .
each point at these diagrams marks the value of @xmath2 and the size of the subgroup which contains nodes @xmath10 .
links between different subgroups are omitted .
we see that for ba networks ( fig.4 ) , the lowest value of @xmath20 is 97 .
the remaining three nodes are linked with all other nodes by hostile relations .
@xcite , an example of polarization of opinions on the lustration law in poland in 1999 was brought up .
the presented statistical data @xcite displayed two maxima at negative and positive opinions and a lower value at the centre of the plot . in our simulations performed for fully connected graphs @xcite , the obtained value for the center was zero .
however , it is clear that in any group larger than , say , 50 persons some interpersonal relations will be absent . taking this into account
, we can claim than the statistical data of @xcite should be compared to the results discussed here rather than to those for a fully connected graph .
here we reproduce a peak of the histogram at its centre , on the contrary to the results in @xcite .
this fact allows to speak on a qualitative accordance of the results of our calculations with the statistical data of @xcite .
next example is the set of data of the attendance of 18 southern women in local meetings in natchez , missouri , usa in 1935 @xcite .
these data were used to compare 21 methods of finding social groups .
the results were analysed with respect to their consensus , and ranked with consensus index from 0.543 ( much worse than all others ) to 0.968 . to apply our dynamics
, we use the correlation function @xmath21 as initial values of @xmath2 .
our method produced the division ( 1 - 9 ) against ( 10 - 18 ) , what gives the index value 0.968 .
as a by - product , the method can provide the time dynamics of the relations till hb and , once hb is reached , the leadership within the cliques @xcite .
we should add that actually , we have no data on the possible friendship or hostility between these women , then the interpretation of these results should be done with care .
last example is the set of data about a real conflict in the zachary karate club @xcite .
the input data are taken from @xcite .
all initial values of the matrix elements are reduced by a constant @xmath22 to evade the case of overwhelming friendship .
the obtained splitting of the group is exactly as observed by zachary : ( 1 - 8,11 - 14,17,18,20,22 ) against ( 9,10,15,16,19,21,23 - 34 ) .
these results were checked not to vary for @xmath22 between 1.0 and 3.0 .
the status of all group members can be obtained with the same method as in the previous example . to conclude , the essence of eq .
( 1 ) is the nonlinear coupling between links @xmath2 , which produces the positive feedback between the actual values of the relations and their time evolution .
we should add that the idea of such a feedback is not entirely new .
it is present , for example , in boltzmann - like nonlinear master equations applied to behavioral models @xcite . on the contrary
, it is absent in later works on formal theory of social influence @xcite . on the other hand ,
the theories of status @xcite are close to the method of transition matrix , known in non - equilibrium statistical mechanics @xcite . 99 f. heider , j. of psychology * 21 * ( 1946 ) 107 .
f. heider , _ the psychology of interpersonal relations _ , j.wiley and sons , new york 1958
. f. harary , r. z. norman and d. cartwright , _ structural models : an introduction to the theory of directed graphs _ , john wiley and sons , new york 1965 .
p. doreian and a. mrvar , social networks * 18 * ( 1996 ) 149 .
z. wang and w. thorngate , j. of artificial societies and social simulation , vol 6 , no 3 ( 2003 ) .
k. kuakowski , p. gawroski and p. gronek , int . j. mod
c ( 2005 ) , in print ( physics/0501073 ) see also ( physics/0501160 ) .
r. albert and a .-
barabsi , rev .
* 286 * ( 2002 ) 47 .
m. abramowitz and i. a. stegun ( eds . ) , _ handbook of mathematical functions _ ,
dover , new york , 1972 .
k. malarz and k. kuakowski , physica a * 345 * ( 2005 ) 326 ( see also cond - mat/0501545 ) .
report bs/152/99 of the public opinion research center , tab .
3 ( in polish ) . l. c. freeman , in r. breiger , k. carley and p. pattison ( eds . ) , _ dynamic social network modeling and analysis _ , washington , d.c.:the national academies press , 2003 .
bonacich and p. lloyd , social networks * 23 * ( 2001 ) 191 .
w. w. zachary , j. anthrop .
33 * ( 1977 ) 452 .
l. donetti and m. muoz , j. stat .
( 2004 ) 10012 .
m. girvan and m. e. j. newman , phys .
e * 69 * ( 2004 ) 026113 .
vlado.fmf.uni-lj.si/pub/networks/data/ucinet/ucidata.htm , dataset zachc d. helbing , physica a * 196 * ( 1993 ) 546 . n. e. friedkin and e. c. johnsen , social networks * 19 * ( 1997 ) 209 .
l. e. reichl , _ a modern course in statistical physics _ , j. wiley and sons , new york 1998 , p. 241
|
recently , a continuous dynamics was proposed to simulate dynamics of interpersonal relations in a society represented by a fully connected graph .
final state of such a society was found to be identical with the so - called heider balance ( hb ) , where the society is divided into two mutually hostile groups . in the continuous model , a polarization of opinions was found in hb .
here we demonstrate that the polarization occurs also in barabsi - albert networks , where the heider balance is not necessarily present . in the second part of this work
we demonstrate the results of our formalism , when applied to reference examples : the southern women and the zachary club . * heider balance in human networks * + p. gawroski and k. kuakowski + _ department of applied computer science , faculty of physics and applied computer science , agh university of science and technology + al .
mickiewicza 30 , pl-30059 krakw , poland _
_ pacs numbers : _ 87.23.ge _ keywords : _ numerical calculations ; sociophysics
|
coulomb disorder plays dominant role in determining thermodynamic and kinetic properties of doped semiconductors due to its long - range nature .
@xcite this physics becomes especially intriguing in the context of dirac materials , namely solid state systems that host quasiparticles with linear dispersion near the band - touching degeneracy points , such as that in @xmath0-wave superconductors , @xcite graphene , @xcite topological insulators @xcite and weyl semimetals .
@xcite while the former three classes of systems have been intensively studied over almost a decade by now , materials that can be identified as weyl semimetals have been discovered only fairly recently .
@xcite theoretical studies of the effect of disorder on a single weyl node have a long history , @xcite and this problem has been revisited in a number of recent works .
@xcite the prevailing point of view is that a weak disorder has negligible effect on the density of states , which vanishes quadraticaly with the energy counted from the nodal point .
this behavior persists up to a certain critical disorder strength beyond which the density of states acquires a finite value at zero energy .
obviously , such scenario should have serious implications for the transport properties of wsms that rely on their semimetallic nature ( a vanishing density of states at the nodal points ) .
@xcite at strong disorder singular transport features are eliminated .
it is thus important to have a controlled theory of disorder effects on a single nodal point .
conventional treatments based on the self - consistent born approximation , though can be qualitatively correct , are uncontrolled near dirac or weyl points .
@xcite magnetotransport properties of wsms are also expected to have a highly unusual character owing to the unconventional landau quantization for the three - dimensional dirac spectrum , and the multiband nature of weyl semimetals . in particular , a strong classical negative magnetoresistance was predicted for weyl systems with long intravalley relaxation times , @xcite as well as robust linear positive magnetoresistance in the quantum limit , when a single landau level is occupied .
@xcite both types of behavior were recently observed in experiments @xcite on a candidate dirac ( `` double - weyl '' ) material cd@xmath1as@xmath2 , yet at this point it can not be stated with confidence whether the experimental observations are in full correspondence with theoretical models . for various recent theoretical and experimental aspects of magnetotransport results in dirac semimetals see refs . .
in light of the recent experimental advances that triggered a flood of theoretical work , our essential motivation is to investigate the role of coulomb impurities on properties of wsms .
in contrast to the model of short - range disorder , which so far received most of theoretical attention , we obtain results that are controlled by a small parameter in the limit of _ strong _ disorder .
we also observe that due to the delicate interplay of landau quantization and screening effects , the low - field magnetoresistance may display nonmonotonic behavior .
the latter feature is highly sensitive to the strength of interaction and thus is not universal .
the quasiparticle dispersion in the vicinity of a single isotropic nodal point can be described by the effective hamiltonian ( @xmath3 throughout ) @xmath4 where @xmath5 is the fermi velocity , and @xmath6 and @xmath7 are vector of the pauli matrices , and momentum measured relative to the nodal point .
the total potential as seen by an electron at position @xmath8 due to randomly distributed charged impurities at positions @xmath9 is taken in the form of a yukawa - like potential @xmath10 here index @xmath11 labels impurities , @xmath12 is the dielectric constant of the host material , and @xmath13 is the inverse screening radius .
we assume that impurity positions are completely random , and calculate the correlator of the disorder potential , @xmath14 where @xmath15 is the concentration of impurities . as follows from eq .
( [ uu ] ) , the correlation length of the disorder potential is given by the screening radius , and the magnitude of its fluctuations is given by @xmath16 .
the gaussian approximation is sufficient to describe the disorder potential fluctuations when there are many impurities in the screening volume , @xmath17 .
the model of a disordered weyl semimetal described above will be used in the forthcoming calculations .
the validity of the gaussian approximation will be checked _ a posteriori _ , by considering the screening of coulomb impurities by electrons moving in a self - consistently screened disorder potential , see eq .
( [ nuandkappa ] ) , and the text below it .
a detailed account of self - consistent screening in the present problem is given in ref . .
in what follows we consider the density of states ( dos ) close to the nodal point , @xmath18 , where @xmath19 is the electron energy . dos at the nodal point crucially depends on the relation between the strength of disorder , and the energy scale associated with its finite correlation length , @xmath20 . for @xmath21 ,
the disorder is weak , and the born approximation can be used , with the expected result of zero dos at the nodal point . for strong disorder , @xmath22 ,
self - consistent born approximation predicts a finite dos at the nodal point,@xcite but the approximation itself is uncontrolled .
@xcite to obtain systematic results , controlled by a small parameter , for the dos and later for transport properties , we restructure the perturbation theory for the disorder potential similarly to the way it was done for heavily doped semiconductors .
@xcite a crucial observation is that for @xmath23 , the landscape of the disorder potential is smooth , in the sense that the typical length scale at which it changes , @xmath24 , is large compared to the typical electron wavelength @xmath25 .
( the latter estimate assumes that it is the _ total _ energy of the electron that is close to the nodal point , thus the kinetic energy is of the order of @xmath26 . )
this implies that the electron motion is semiclassical .
therefore , one may use the thomas - fermi approximation ( rather than the clean system ) as the `` reference point '' around which a perturbative expansion is developed . to illustrate the use of the thomas - fermi ( tf ) approximation in this case , we consider the single particle green s function for the hamiltonian ( [ h ] ) : @xmath27=\frac{1}{\varepsilon_\pm - v\bm{\sigma}\cdot\bm{p}-u(\rr)},\ ] ] where @xmath28 , and @xmath29 plays a role of the effective local chemical potential . in general ,
since the electron momentum is an operator , one can not simply average the green s function over the distribution of @xmath29 .
however , in the semiclassical limit we can treat @xmath29 as a number within tf - approximation and thus obtain @xmath30\frac{1}{\varepsilon_\pm - v{\bm \sigma}\cdot\pp - u},\ ] ] where the distribution function for the local values of the disorder potential , @xmath31 $ ] , is defined according to @xmath32=\langle\delta(u - u(\rr))\rangle.\ ] ] taking the imaginary part of the retarded component of @xmath33 gives the density of states , @xmath34\nu_0(\varepsilon - u),\ ] ] where @xmath35 is the dos of the clean system . in the case of a gaussian disorder potential defined in eq .
( [ uu ] ) , one obtains @xmath36=\frac{1}{\sqrt{2\pi}w}\exp(-u^2/2w^2).\ ] ] using @xmath37 , with @xmath38 being the degeneracy of the nodal point for a given weyl system , and performing an elementary integration , one finds @xmath39 that is , a nonzero value for @xmath40 . in the context of doped semiconductors , the semiclassical equation for the density of states , eq .
( [ eq : olddos ] ) , was suggested long ago , e.g. in refs . , and has been recently used to describe dos and nonlinear screening in dirac systems in ref .
we motivated equation ( [ g ] ) on physical grounds , but it can be rigorously shown that it is exact in the @xmath41 limit .
one can formally send @xmath42 while keeping @xmath26 constant by considering @xmath43 , @xmath44 limit with @xmath45 kept constant ( see below ) . in this limit
the disorder correlator is @xmath46 , and describes disorder with _
infinite _ correlation length .
the problem of finding the single - particle green s function for such a disorder potential ( the so - called `` keldysh model '' ) is exactly solvable , since the impurity lines in the standard technique do not transfer momentum , thus all diagrams to a given order have the same value , fig .
[ fig : diagrams ] .
the full resummation of the perturbation theory becomes a combinatorial problem , solving which @xcite requires determining the number of ways @xmath15 impurity lines can be attached to @xmath47 bare green s functions in the @xmath48 order of perturbation theory ( the second order diagrams are shown in fig .
[ fig : diagrams ] ) .
the results for the disorder - averaged single - particle green s functions and the dos are precisely the above eqs .
( [ g ] ) and ( [ eq : olddos ] ) . , all the diagrams have the same value of @xmath49 , where @xmath50 is the unperturbed green s function.,width=336 ] in what follows we consider the corrections to the result of eq .
( [ nu ] ) , following the treatment of ref . .
to build an improved perturbation theory for the potential defined by eq .
( [ uu ] ) , we formally rewrite the latter equation as @xmath51 if the second term in the right hand side of the above equation is neglected ( the corresponding condition to be formulated below ) , one recovers the case of an infinite - range disorder potential , which was discussed above . going beyond the keldysh model , when is it appropriate to treat the second term in the right hand side of eq .
( [ udecomposition ] ) as a small correction ?
it is clear that one should require that @xmath52 for the relevant values of @xmath53 . since the solution for the semiclassical problem
is represented by an ensemble average over samples having a random , but spatially uniform chemical potential of order @xmath26 , see eqs .
( [ g ] ) and ( [ p ] ) , the relevant value of @xmath53 is of the order of the wavelength corresponding to the energy @xmath26 , @xmath54 .
hence the corrections to the semiclassical approximations are expected to be small for @xmath55 , coinciding with the condition for the validity of the semiclassical approach formulated above .
our next immediate goal is to find the leading @xmath56 correction to eq .
( [ nu ] ) due to the short - range component of disorder potential .
as we are going to see , it is regular and free from divergencies near the dirac point .
the full disorder - averaged green s function can be written as @xcite @xmath57\frac{1}{\varepsilon_\pm - u - v{\bm \sigma}\cdot\pp-\sigma}.\ ] ] here , for every value of @xmath58 , the self - energy @xmath59 is determined from the usual diagram technique for disordered metals , in which the role of the bare green s function is played by eq .
( [ g - u ] ) with @xmath29 replaced with @xmath58 , and the impurity line ( in momentum space ) , @xmath60 , is determined _ only _ by the second term in the right hand side of eq .
( [ udecomposition ] ) : @xmath61 in eq .
( [ impurityline ] ) @xmath62 is the dirac delta - function .
its appearance signals that the long - range part of the impurity potential is subtracted from @xmath60 , as it has been already treated exactly by the averaging over the values of @xmath58
. we will see below that after such removal , a simple born approximation is valid for the short - range part of the disorder potential if @xmath63 . nonetheless , below we set up the self - consistent born approximation ( scba ) for the self - energy for the modified disorder correlator , eq .
( [ impurityline ] ) .
we will not go beyond the first born approximation in actual calculations , but having the full scba set up will allow us to contrast the present treatment with the conventional scba for long - range disorder . using the symmetry - suggested decomposition for the self - energy @xmath64 where @xmath65 is the unit vector in the direction of @xmath7 , and
introducing @xmath66 , for the retarded component we have @xmath67 where we have introduced @xmath68 to simplify equations , we first introduce new functions @xmath69 and @xmath70 .
second , we complete @xmath71-integration in eqs .
( [ scba-0 ] ) and ( [ scba-1 ] ) and observe that upon changing @xmath72 the corresponding integrands are even and odd functions respectively . we thus extend the @xmath73-integration to the whole axis such that @xmath74 is considered as an even function while @xmath75 is odd . as a result
we have to solve the following set of coupled integral equations @xmath76 where the kernels are given explicitly by @xmath77 .
\end{split}\ ] ] it is instructive to understand how the limit of very long correlation length , @xmath78 , is reached within various approximation schemes .
firstly , the results for the keldysh model , eqs .
( [ g ] ) and ( [ eq : olddos ] ) , are exact in this limit , as we have already discussed .
in particular , this means that the quantities @xmath79 and @xmath80 , that determine the self - energy at the scba level for the modified disorder correlator , must attain bare values @xmath81 , @xmath82 when eqs .
( [ eq : xy ] ) are solved . that this is indeed the case
can be seen noting that in the @xmath42 limit the kernels @xmath83 satisfy @xmath84 this implies that the second terms on the right hand sides of eqs .
( [ eq : xy ] ) are exactly cancelled by the third terms .
in fact , all self - energy corrections for the modified disorder correlator , not just the scba ones , vanish in the @xmath42 limit : this is the essence of the keldysh model , in which the full diagrammatic series is summed up _
exactly _ in the @xmath85 limit , to yield eq .
( [ g ] ) for the green s function .
@xcite we can contrast the keldysh model with the conventional scba , in which the full disorder correlator , eq .
( [ eq : fulldisorder ] ) , is used as the impurity line .
this approximation can easily be obtained from eqs .
( [ eq : xy ] ) by setting @xmath86 , and dropping the third terms on the right hand sides of the equations for @xmath79 and @xmath80 . to distinguish this case from the rest of the paper ,
we denote @xmath79 and @xmath80 for scba with full disorder correlator as @xmath87 and @xmath88 . to obtain simple analytic results ,
we limit ourselves to @xmath89 . in the @xmath85
limit the scba equations are @xmath90 the infinitesimal @xmath91 is kept as a reminder that @xmath92 .
@xmath93 and @xmath94 can be easily found noting that the former is purely imaginary , while the latter is real : @xmath95 the corresponding dos can be found as @xmath96 which upon using eqs .
( [ eq : solutionxy ] ) gives @xmath97 in this treatment inter - valley scattering is neglected , and the result eq .
( [ eq : nuborn ] ) should be understood as the dos per node .
we see that the scba with the full disorder correlator reproduces the exact answer for dos in @xmath42 limit .
this means that if one naively tries to correct scba result for dos by including the low - order interference corrections to the self - energy , @xcite the result will become worse , not better .
what the conventional scba does not capture is the form of the spectral weight that results from summation of the full diagrammatic series in the @xmath42 limit . as a function of momentum for a given energy @xmath12 , the spectral weight for the green s function of eq .
( [ g ] ) is a gaussian peak centered at @xmath98 , having a width of @xmath99 .
this exact result is clearly not reproduced by the conventional scba at low energies , eq .
( [ eq : solutionxy ] ) . for illustration purposes , we plot the total spectral weight ( for conduction and valence bands ) at zero energy energy for the retarded green s function within the keldysh model , eq .
[ g ] , and for the green s function in the conventional scba .
the latter is obtained using eqs .
( [ eq : solutionxy ] ) as @xmath100 .
limit : solid line - keldysh model , dashed line - scba with full disorder correlator.,width=288 ] in this section we calculate the leading order correction to the semiclassical dos , eq .
( [ nu ] ) . to this end
we use eqs .
( [ eq : xy ] ) to find the first perturbative correction to @xmath74 and @xmath75 , that is , restric ourselves to the first born approximation .
it is sufficient to find the leading @xmath101 ) term in the dos . to zero order in @xmath26
we take @xmath81 and @xmath82 and seek the first order correction by iterations : @xmath102 after performing the integrations and after some further algebra we find @xmath103,\\ & & \hskip-.45cm\delta y_p=\frac{w^2}{vp}\left[\frac{(vp)^2+(v\kappa)^2-iv\kappa\xi_+}{(\xi_++iv\kappa)^2-(vp)^2}-\frac{(vp)^2}{\xi^2_+-(vp)^2}\right]\nonumber\\ & & + \frac{iw^2v\kappa}{2(vp)^2}\ln\left(\frac{\xi_++iv\kappa+vp}{\xi_++iv\kappa - vp}\right).\end{aligned}\ ] ] these expressions appear very singular , but one must bear in mind that it is only the full disorder - averaged green s function that determines the dos , and all singularities of the present perturbation theory are removed by gaussian integration over @xmath58 in the expression for the retarded green s function , eq .
( [ gfull ] ) , expanded in the usual series in powers of @xmath59 .
knowledge of @xmath104 and @xmath105 allows one to find correction to the density of states .
we are only interested in the leading @xmath106 correction .
explicit calculation shows that the only term contributing to the dos correction is @xmath107 . following the standard path ( expanding the green s function to linear order in @xmath13 ,
taking its imaginary part , and performing the final momentum integrations ) , one finds @xmath108 finally , averaging this result with @xmath31 $ ] from eq .
( [ p ] ) we arrive at @xmath109 thus the first perturbative correction to the dos is regular and controlled by the smallness of @xmath110 , as expected from the preceding discussion .
we note that at no point did we have to introduce an artificial momentum cut - off , as is commonly done while treating the system in the self - consistent born approximation . taking into account the second born approximation , or the first interference correction to the self - energy all lead to dos - corrections that involve at least one additional power of @xmath111 compared to eq .
( [ nufinal ] ) , and can be ignored in the current treatment . in order to establish the limits of validity of our results we need to determine the typical value of @xmath26 .
this can be done with the help of self - consistent thomas - fermi approximation .
first recall that @xmath112 and @xmath113 , which follows from eq .
( [ uu ] ) . using now eq .
( [ nu ] ) for the density of states we find an equation that defines the typical magnitude of the disorder potential fluctuation as a function of density @xmath15 in the form @xmath114 where @xmath115 is the effective interaction parameter .
the most interesting case for us is the vicinity of the `` neutrality '' point where @xmath116 , so that @xmath117 which sets the typical scale of the density of states and inverse screening length , respectively , @xmath118 as discussed above , the gaussian model of disorder amounts to a condition @xmath119 .
perturbative treatment of the short - range disorder requires @xmath120 .
both of these conditions can be satisfied provided that @xmath121 . finally , we discuss validity of the semiclassical approximation that requires @xmath122 .
the latter can be estimated as follows , @xmath123 .
this semiclassical formalism can be readily extended to the case of finite magnetic field .
the generic analysis of the density of states is quite involved . here
we will be primarily interested in the regime of weak fields , when magnetic length @xmath124 is large compared to the correlation radius of the disorder potential @xmath125 . in this case
we can use the same approach we have used for the magnetic - field - free case . in the strong - field limit , @xmath126 , the treatment of ref . should be used .
( see also ref . for an alternative quasiclassical approach to transport in the strong - field limit . )
the spectrum of landau levels for the three - dimensional weyl equation has the form @xmath127 where we chose a particular chirality for the zeroth landau level without loss of generality .
we concentrate on the lowest energy limit where @xmath128 averaging this expression over @xmath58 with the help of eq .
( [ p ] ) , followed by the gaussian integration over momentum @xmath129 and the geometrical series summation over @xmath130 one finds @xmath131 expanding this result in the low field limit one recovers eq .
( [ nu ] ) to the lowest order with a _ positive _ correction term @xmath132 .
this finite field correction ultimately renders the change in the screening radius of the disorder potential , which in the case of coulomb impurities will have an important effect on magnetoconductivity .
in this section we briefly discuss what self - consistent theory of screening in wsms gives for the low - temperature transport , namely we concentrate on the regime of @xmath133 . in the semiclassical limit ,
provided that the mean free path @xmath134 for electron scattering is large , @xmath135 , where @xmath136 is the fermi momentum , the conductivity can be calculated from the drude formula with energy dependent scattering time @xmath137 , density of states @xmath138 , and cyclotron frequency @xmath139 , @xmath140 the ensemble - averaged transport scattering time for coulomb impurities within the boltzmann approximation can be calculated using the full disorder correlator , eq .
( [ eq : fulldisorder ] ) , since the @xmath141 part of @xmath60 from eq .
( [ impurityline ] ) can not lead to momentum transfer .
the transport scattering time is thus given by @xmath142,\label{tau}\end{aligned}\ ] ] where @xmath143 and @xmath144 is the fourier transform of eq .
( [ u ] ) taken at momentum @xmath145 , whereas the semiclassical expression for the cyclotron frequency for the linear spectrum is @xmath146 according to our earlier estimates of the screening radius one immediately concludes that @xmath147 , so it is sufficient to retain only the large logarithmic factor in eq .
( [ tau ] ) .
then , the zero field conductivity is estimated ( up to an overall numerical factor ) as follows @xmath148 where we took the typical value of potential @xmath149 and used eq .
( [ w ] ) .
one can check that derived expression indeed corresponds to the boltzmann limit .
we note in passing that a local conductivity tensor is appropriate for the situation considered here , since eqs .
( [ w ] ) , ( [ nuandkappa ] ) , and ( [ tau ] ) imply that @xmath150 ; in other words , the transport mean free path is much shorter than the screening length .
further , it is worth mentioning that the dependence of @xmath151 on the interaction parameter @xmath152 was a subject of controversy since different results exist in the literature .
the field dependence of the conductivity comes from two factors .
the first is obviously due to the cyclotron motion , which we estimate by expanding the denominator of eq .
( [ sigma ] ) @xmath153 ^ 2\simeq-\frac{1}{\alpha^7\ln^2(1/\alpha)}\frac{1}{l^4_bn^{4/3}}.\ ] ] curiously , there is another contribution , of the opposite sign , that stems from the already mentioned fact , that magnetic field modifies the screening radius .
since the product @xmath154 is inversely proportional to @xmath155 and @xmath156 then it follows from the low field asymptotic of eq .
( [ nu - b ] ) that for this mechanism @xmath157 for weak interactions , @xmath158 , eq .
( [ sigma - b - a ] ) always dominates by a parametrically large factor and leads to positive magnetoresistance at weak field that eventually crosses over to quantum regime at higher fields , @xmath159 .
@xcite however , for some material systems , where interaction is moderately strong , @xmath160 , one in principle could envision a scenario where the fate of the sign of magnetoconductivity will be determined by the overall numerical prefactors in eqs .
( [ sigma - b - a ] ) and ( [ sigma - b - b ] ) .
obviously the criterion @xmath161 limits the applicability of our estimates .
an elaborate theory of magnetoresistance in wsms was recently developed in ref . in various regimes determined by the relation between the relevant energy scales
@xmath162 , @xmath163 , and @xmath164 .
our estimates for @xmath165 at weak field are consistent with that of ref . with an addition of the contribution governed by eq .
( [ sigma - b - b ] ) , where in eqs .
( [ sigma - b-0])([sigma - b - b ] ) we also retained functional dependence on the interaction parameter @xmath152 .
to conclude , we have studied the density of states and transport properties of weyls semimetals with the long - range coulomb disorder .
provided that the semiclassical limit is satisfied , dos induced by the long - range disorder potential fluctuations can be found exactly .
in addition , we have developed systematic perturbation theory to account for corrections due to the short - range scattering . concerning the transport properties , for weak interactions and magnetic field the magnetoresistance
is expected to be positive for a single weyl node .
( as compared to a negative magnetoresitance for current flow along the magnetic field when the realistic multivalley situation is considered .
@xcite ) however , based on the presented analysis , one should not exclude a possibility of nonmonotonic magnetoresistance in systems with relatively strong interactions .
we would like to thank i. gornyi , j. klier , a. mirlin , and m. raikh for fruitful discussions .
we are grateful to m. dzero for reading and commenting on the paper .
this work was supported by nsf grant no .
dmr-1409089 ( d.a.p . ) , by nsf grant dmr-1401908 and in part by daad grant from german academic exchange services , and spp 1666 of the deutsche forschungsgemeinschaft ( a.l . ) , and doe basic energy sciences grant no
. de - fg02 - 06er46313 ( e.g.m . ) .
d.a.p and a.l .
acknowledge the hospitality of spin phenomena interdisciplinary center at johannes gutenberg - universitt mainz , where this work was finalized .
xu , n. alidoust , i. belopolski , c. zhang , g. bian , t .-
chang , h. zheng , v. strokov , d. s. sanchez , g. chang , z. yuan , d. mou , y. wu , l. huang , c .- c .
huang , b. wang , a. bansil , h .-
jeng , t. neupert , a. kaminski , h. lin , s. jia , m. zahid hasan , nat . phys .
* 11 * , 748 ( 2015 ) .
n. xu , h. m. weng , b. q. lv , c. matt , j. park , f. bisti , v. n. strocov , d. gawryluk , e. pomjakushina , k. conder , n. c. plumb , m. radovic , g. autes , o. v. yazyev , z. fang , x. dai , g. aeppli , t. qian , j. mesot , h. ding , m. shi , preprint arxiv:1507.03983 . c. shekhar , f. arnold , s .- c .
wu , y. sun , m. schmidt , n. kumar , a. g. grushin , j. h. bardarson , r. donizeth dos reis , m. naumann , m. baenitz , h. borrmann , m. nicklas , e. hassinger , c. felser , b. yan , preprint arxiv:1506.06577 .
|
we study the density of states and magnetotransport properties of disordered weyl semimetals , focusing on the case of a strong long - range disorder . to calculate the disorder - averaged density of states close to nodal points , we treat exactly the long - range random potential fluctuations produced by charged impurities , while the short - range component of disorder potential is included systematically and controllably with the help of a diagram technique .
we find that for energies close to the degeneracy point , long - range potential fluctuations lead to a finite density of states . in the context of transport
, we discuss that a self - consistent theory of screening in magnetic field may conceivably lead to non - monotonic low - field magnetoresistance .
|
during vertebrate development , segmentation of the continually elongating embryonic body axis occurs rhythmically and sequentially from head to tail in a process termed somitogenesis @xcite .
somites are regularly sized cell clusters that bud off periodically from the anterior end of the posterior - most unsegmented tissue , the pre - somitic mesoderm ( psm ) , with a species - specific frequency .
these transient , left - right symmetric structures are the embryonic precursors of adult bone and muscle segments , and defects in their formation lead to congenital birth defects @xcite . underlying the morphogenetic rhythm of somitogenesis , repeated waves of oscillating gene expression sweep through the cells of the psm from the posterior to the anterior @xcite , see fig .
[ fig : boundaries]a and supplementary movie 1 .
these genetic oscillations are thought to slow down and arrest at different phases of their cycles at an anteriorly positioned arrest front that moves in concert with embryonic elongation @xcite ( fig .
[ fig : boundaries]b ) , translating the temporal periodicity into a striped spatial pattern of gene expression .
representation of the pre - somitic mesoderm ( psm ) .
anterior is to the left and posterior to the right .
( a ) _ in situ _
hybridization @xcite showing the expression of _ deltac _ mrna in the zebrafish psm ( dorsal view ) .
( b ) schematic psm together with the already determined segments arrested and the most recently formed pair of somites .
the studied region lies between the arrest front and the posterior boundary .
( c ) schematic representation of the signal gradient spanning the psm ( broken line ) . the frequency profile @xmath0 related to this gradient is depicted as solid purple line , using eq .
( [ eq : freq_profile ] ) with the width @xmath1 given in table i. the length of the studied region is denoted by @xmath2 .
a linear array of @xmath3 coupled oscillators is indicated.,width=328 ] given the existence of genetic oscillators in the cells of the psm @xcite , several questions still remain unanswered : how does a collective segmentation period arise from the population of individual oscillators , and what is the relation between the overall pattern of gene expression observed in the psm and the oscillating expression at the cellular level . to investigate this
, we develop a theoretical description based on phase oscillators and four key ingredients motivated by the biology : ( i ) a frequency profile along the psm , slowing down the oscillations , ( ii ) coupling of oscillators , ( iii ) a time delay in the information transfer between neighboring oscillators , and ( iv ) the existence of a moving front that arrests the oscillations at the anterior end of the psm , while the posterior end moves due to embryonic outgrowth .
this delayed coupling theory provides an excellent fit to the existing biological data , allows perturbations to the system to be analyzed in terms of underlying processes , and predicts how intercellular communication affects the collective period of the segmentation clock .
below , we introduce the elements of the delayed coupling theory . to provide a simple picture of the segmentation process , cooke and zeeman proposed a clock and wavefront model more than thirty years ago @xcite .
however , to understand the role of collective processes in the emergence of dynamic gene expression patterns in the psm a more detailed analysis is needed , for which methods from other pattern forming systems can be borrowed @xcite . in particular , the periodic expression of genes in the oscillating psm cells can be described at tissue level using a set of phase oscillators , disregarding at this stage the underlying biochemical and genetic mechanisms that generate the oscillations and their pattern . in this phase description , each cell , or group of synchronous cells , is represented by an oscillator , and the state of each oscillator is characterized only by its phase in the cycle of periodic gene expression .
oscillators with the same phase represent cells with equivalent expression level of cyclic genes .
previous models described the psm as a continuous oscillatory medium with a phase defined at each point of the psm , see supplementary data of @xcite and @xcite . in this work
we show that a phase description is sufficient to compute the overall spatiotemporal patterns of gene expression and the collective period of the oscillations .
it has been suggested that the arrest of the oscillations and the observed oscillating gene expression patterns are shaped by a spatial dependence of the frequency of the individual oscillators @xcite .
a frequency profile could be controlled by the molecular gradients observed in the psm , see fig .
[ fig : boundaries]c , _ e.g. _ the gradients of the growth factor fgf @xcite or of wnt signaling @xcite .
several models have recently proposed regulatory mechanisms by which the genetic oscillations are affected by the gradients of signaling molecules along the psm @xcite . motivated by the changing width of the stripes of gene expression in the psm and the necessity that the oscillations slow down and finally stop at the arrest front , we include such a frequency profile in our theory . as with the assumption of cellular oscillators , our theory does not rely on the molecular origin of this frequency profile , and
hence its inclusion is purely phenomenological .
recent theoretical works seeking to describe spatiotemporal patterns in somitogenesis using phase oscillators have not included coupling between oscillators @xcite , although intercellular coupling has been considered in reduced models of regulatory circuits @xcite . here
, coupling means that oscillators can influence the phase of their neighbors .
coupling is essential to stabilize tissue - scale patterns against the unavoidable noise present in biological systems @xcite , and also to explain the re - synchronization of surgically inverted pieces of the psm @xcite .
thus , in this work we propose a description based on coupled phase oscillators .
coupling between cells via signaling macromolecules , _
e.g. _ through the notch pathway @xcite , involves synthesis and trafficking of such molecules within cells .
these dynamics imply the existence of time delays , which have been recently estimated to be in the range of tens of minutes in cell culture @xcite .
time delays in the coupling can have an impact on the self - organization of coupled oscillators @xcite , making their inclusion in our theory important . for simplicity we will use here a deterministic time delay
; a more realistic description would include a distribution of delays @xcite .
the embryo is a rapidly growing system , elongating about one somite length per oscillation cycle , which takes around 25 minutes in zebrafish , 90 minutes in chick and 120 minutes in mouse .
cells are continuously in transit from the tailbud through the psm , exiting it anteriorly as somites form .
additionally , cell proliferation plays a role during elongation , but since in the psm it has a stochastic character it can be considered as a potential noise source @xcite not otherwise significantly affecting the oscillatory dynamics @xcite , and we do not consider it further here . to correctly understand the formation of patterns of gene expression and how the frequency is regulated ,
it is necessary to consider the geometry and boundaries of the arena in which the process occurs . here
we neglect changes in the antero - posterior length of the psm or the rate of axial growth , which occur during development at time scales larger than the somitogenesis period @xcite . consequently
, both the arrest front and the posterior boundary move at the same velocity @xmath4 , see fig .
[ fig : boundaries]b . two different coordinate systems to describe genetic oscillations in the psm .
( a ) lab reference frame : oscillators labeled by index @xmath5 hold fixed positions @xmath6 , where @xmath7 is the distance between oscillators . the psm boundary moves posteriorly with velocity @xmath4 as the embryo extends .
( b ) psm reference frame : the psm boundaries do not move but oscillators move through the psm from posterior to anterior with velocity @xmath4 .
oscillators are constantly relabeled using symbols @xmath8 to denote discrete positions relative to the arrest front .
dashed lines indicate the state of the system at slightly earlier times.,width=328 ]
this section contains technical details of our theory .
readers who are more interested in the basic ideas and the biological justifications should note that a careful understanding of the equations is not a requisite to follow the arguments we introduce .
+ our model equations in the lab reference frame , fig .
[ fig : ref - frames]a , consist of a lattice of discrete phase oscillators .
this lattice comprises @xmath3 oscillators in the antero - posterior direction , labeled by an index @xmath5 .
each oscillator occupies a position @xmath6 along the psm axis in the lab reference frame , where @xmath7 is the characteristic distance between oscillators , _
i.e. _ the average cell diameter .
hence the total physical length of the considered system in the antero - posterior direction is @xmath9 .
as the embryo elongates with velocity @xmath4 , the arrest front is positioned at @xmath10 , where @xmath11 is time .
oscillators anterior to the arrest front , @xmath12 , are arrested .
new oscillators are added at the posterior boundary , situated at @xmath13 , as elongation proceeds .
the oscillators in the studied region , with indices @xmath14 are weakly coupled to their @xmath15 nearest neighbors , denoted by the index @xmath16 . in one dimension @xmath17 , while in a two - dimensional square lattice @xmath18 .
the phase dynamics of the coupled oscillators can be described by @xmath19+\zeta_i(t),\ ] ] where the dot denotes time derivative , @xmath20 is the phase of oscillator @xmath5 , @xmath21 its intrinsic frequency , @xmath22 the coupling strength , @xmath23 the time delay in the coupling and @xmath24 is a random variable with zero average representing different noise sources .
our objective in this work is to characterize the basic model in the synchronized state , and if not otherwise stated , we ignore the effects of noise . according to experimental evidence @xcite ,
coupling strength is weak compared to other timescales in the system .
this justifies the use of the sine function in the coupling , as it is the dominant term of any more general periodic coupling function @xcite . to specify the shape of the frequency profile we choose , for @xmath14 : @xmath25 where @xmath26 is a characteristic frequency scale of individual oscillators , and @xmath1 is a measure of the characteristic distance over which the frequency profile decreases from high to low values ( see fig . [
fig : boundaries]c ) .
below we will show that our choice of eq .
( [ eq : freq_profile ] ) is consistent with experimental observations , and we will determine @xmath1 from the width of the stripes of gene expression in the psm .
qualitatively similar choices for the frequency profile have been used before @xcite . for simplicity
we have chosen the frequency to be strictly zero at the arrest front .
note however that this is not biologically necessary : a very low value of the frequency at the arrest front means a very large period of oscillation .
if this period is much larger than any other time scale involved in the process , it determines in practice an arrested oscillation , which can specify the downstream fixed pattern that eventually sets the position of somite boundaries . furthermore , since the oscillations can be coupled to a bistable system arising from opposing signaling gradients in the psm , long period oscillations at the arrest front could be stopped by a bistable transition @xcite . for convenience
we introduce the parameter @xmath27 , that represents the intrinsic frequency of the oscillators at the posterior boundary of the system . based on in situ experiments that show a largely uniform spatial expression of cyclic genes in the tailbud at any given stage of the cycle ( fig .
[ fig : boundaries]a ) , we define a uniform phase ( fig . [ fig : boundaries]b ) and homogeneous frequency @xmath28 ( fig .
[ fig : boundaries]c ) in the tailbud , posterior to the notochord and the region we study here .
this homogeneity would be favored by the strong cell mixing @xcite , and high , potentially saturating uniform levels of the signaling molecules that establish the gradient anterior to this region @xcite .
furthermore , the shape of @xmath21 described by eq .
( [ eq : freq_profile ] ) resembles the posterior branch of fgf receptor saturation proposed in @xcite
. the coupling could also be position dependent . in particular
, since the oscillators anterior to the arrest front stop cycling , they can not influence the active oscillators in the interval @xmath29 .
we take this into account imposing @xmath30 for @xmath31 . for simplicity ,
in this work we consider the coupling strength @xmath32 and the time delay @xmath33 to be constant posterior to the arrest front . to gain insight into the role of delayed coupling in setting the period and the pattern of the genetic oscillations , as well as to illustrate the formation of realistic wave patterns within our theory , we performed computer simulations of eq .
( [ eq : discrete ] ) in a two dimensional geometry using different values of the time delay ( fig . [ fig:2d_simulations ] and supplementary movies 2 , 3 and 4 ) .
although the theory represents generic vertebrate segmentation , here we use parameters from the zebrafish embryo , table i. for the intrinsic frequency at the posterior we chose @xmath34
min@xmath35 , corresponding to an intrinsic period of @xmath36 min .
we show simulations with no time delay ( @xmath37 min ) , a short delay compared to the intrinsic period ( @xmath38 min ) , and a long delay close to the intrinsic period ( @xmath39 min ) .
the latter is consistent with the experimental observation of tens of minutes for intercellular communication times @xcite . numerical simulation of segmentation using eq .
( [ eq : discrete ] ) in a growing two - dimensional geometry .
color intensity indicates the value of @xmath40 of the phase @xmath41 : white is @xmath42 and dark ( red or blue ) is @xmath43 .
vertical line indicates the position of the arrest front , with the oscillating psm to its right ( blue ) and the arrested pattern to the left ( red ) .
intrinsic frequency is a decaying function of the distance to the black dot at the posterior boundary , causing the curvature of the stripes .
we have used parameters determined for zebrafish at 28@xmath44c , see table i , and @xmath45min@xmath35 , see main text for details .
we have chosen to display three illustrative values of the time delay ( a ) @xmath37 min , ( b ) @xmath46 min and ( c ) @xmath47 min .
delayed coupling affects the global frequency of oscillations according to eq .
( [ eq : omega - epsilon - tau ] ) .
the stripes of the oscillating psm pattern and the segment length of the frozen pattern change accordingly .
movies available as supplementary material.,width=328 ] fig .
[ fig:2d_simulations]a shows a snapshot of the simulation with no time delay .
not surprisingly , the collective period the time needed to form one new arrested segment , and also the time after which the oscillating pattern in the psm repeats itself is unchanged with respect to the intrinsic period at the posterior .
the arrested segments have a length of @xmath48 cell diameters .
the case of short delay with respect to the period , @xmath49 , qualitatively represents the situation in species with a relatively long segmentation period , such as mouse .
[ fig:2d_simulations]b shows that in this case , the effect of the delay in coupling is to slow down the collective period ( @xmath50 min ) with respect to the intrinsic period ( @xmath51 min ) .
further , the arrested segments are longer ( @xmath52 cell diameters ) than in the case without delay , and there is a smaller number of gene expression stripes in the psm , with increased size .
surprisingly , when the delay is made longer , the trends observed with the short delay are inverted . fig .
[ fig:2d_simulations]c shows that for the time delay of @xmath53 , the collective period ( @xmath54 min ) is shorter than the intrinsic period .
moreover , the arrested segments are also shorter ( @xmath55 cell diameters ) than in both previous cases , as are the stripes of oscillating gene expression in the psm .
this puzzling results show that delayed coupling introduces non - trivial effects to the system , large enough to be observable in real experiments . in order to understand these effects , in the following we perform an analysis of eq .
( [ eq : discrete ] ) , studying first the emergence of the collective period from the parameters of the theory and then turning our attention to the spatial pattern . for this purpose
, we will write eq .
( [ eq : discrete ] ) in a more convenient manner .
it is useful to consider the dynamics in the psm reference frame , fig . [
fig : ref - frames]b , where the oscillations can be characterized by a stationary phase profile and a collective frequency .
for simplicity from here on we use a one dimensional description of the system , with @xmath17 . in the lab reference frame ( fig .
[ fig : ref - frames]a ) the symbol @xmath5 represents a fixed oscillator . in the psm reference frame ( fig .
[ fig : ref - frames]b ) we introduce the symbol @xmath8 to label fixed discrete positions relative to the arrest front .
the label @xmath8 runs from @xmath56 at the arrest front to @xmath57 at the posterior boundary of the psm .
discrete position @xmath8 is occupied by different oscillators as the system evolves in time . for convenience
we have included in the description the last arrested oscillator , @xmath56 . in the psm reference frame , the frequency profile is stationary , @xmath58 .
re - expressing eq .
( [ eq : discrete ] ) in this psm reference frame , an extra term describes the drift of the phase due to the movement of the cells relative to the psm boundaries .
the resulting phase dynamics are given by : @xmath59+\\ \nonumber \frac{\varepsilon}{2a^2 } \!\ !
\sum_{k = j+p\pm 1 } \!\!\!\ !
\sin\left [ \varphi_{k } ( t-\tau)-\varphi_{j } ( t ) \right].\end{aligned}\ ] ] here @xmath60 is the phase at position @xmath8 relative to the arrest front and @xmath61 $ ] is the nearest integer to @xmath62 , representing the distance a cell moves during the time it takes for a signal from a neighbor to arrive .
note that now the coupling is non - local : due to the time delay and cell movement , the neighbors of an oscillator with position @xmath8 had positions @xmath63 and @xmath64 at the time the signal was sent .
global frequency @xmath65 of somitogenesis as a function of time delay and coupling strength .
( a ) dimensionless global frequency @xmath66 and time delay @xmath67 are displayed for constant coupling @xmath68 ( cell diameter)@xmath69min@xmath35 , and intrinsic frequency @xmath34 min@xmath35 .
analytical solutions of eq .
( [ eq : omega - epsilon - tau ] ) shown as blue lines : solid lines stable solutions of eq .
( [ eq : omega - epsilon - tau ] ) , dashed lines unstable solutions of eq .
( [ eq : omega - epsilon - tau ] ) .
blue dots correspond to numerical integration of the discrete model in two dimensions as given by eq .
( [ eq : discrete ] ) , for the two cases illustrated in fig .
[ fig:2d_simulations]c , d .
red lines : global frequency as a function of delay in the continuum limit , showing its ranges of validity .
( b ) global frequency @xmath65 as a function of time delay for different coupling strengths obtained from the solution of eq .
( [ eq : omega - epsilon - tau ] ) with @xmath70 ( cell diameter)@xmath69min@xmath35 ( green ) , @xmath71 ( cell diameter)@xmath69min@xmath35 ( blue ) and @xmath72 ( cell diameter)@xmath69min@xmath35 ( red ) .
solid lines are stable solutions , dashed lines are unstable solutions .
dotted line at @xmath73 corresponds to vanishing coupling , @xmath74 ( cell diameter)@xmath69min@xmath35.,width=328 ] the oscillating gene expression pattern in the psm repeats after a full period @xmath75 of oscillation @xcite , where @xmath65 is the collective frequency of the oscillation .
this leads to the steady state ansatz @xmath76 , where @xmath77 is the stationary phase profile describing the pattern in the psm . with this ansatz
we obtain from eq .
( [ eq : discrete - psm ] ) : @xmath78 + \frac{\varepsilon}{2a^2 } \!\
! \sum_{k = j+p\pm 1 } \!\!\!\!\ !
\sin\left [ \phi_{k}-\phi_{j } -\omega\tau \right].\ ] ] the collective frequency of oscillations @xmath65 is equivalent to the rate of somite formation .
note that the instantaneous frequencies @xmath79 of individual oscillators depend on position and are in general different from @xmath65 . to determine the collective frequency @xmath65 we need to specify the boundary conditions , namely the conditions that @xmath77 fulfills at the borders of the studied region ( @xmath56 and @xmath57 , fig .
[ fig : boundaries]b ) .
this boundary should not be confused with somite boundaries , which we do not discuss in this paper . at the arrest front ( @xmath56 ) ,
the fact that @xmath80 and @xmath81 implies with eq .
( [ eq : discrete - psm - spatial ] ) that @xmath82 .
thus , the anterior boundary condition determines the wavelength of the arrested pattern , which is the segment length @xmath83 : the segment length is the distance advanced by the arrest front during one oscillation period @xcite . at the posterior boundary of the psm
, we assume that new cells are added into the system with phase @xmath84 . to implement this
we impose in eq .
( [ eq : discrete - psm - spatial ] ) @xmath85 boundary conditions , @xmath86 for @xmath87 , accounting in this way for the effective non - locality of the coupling .
we base this choice on the experimental observation of cyclic gene mrna patterns , which maintain a smooth expression profile , and hence approximately homogeneous phase , across the interface between tailbud and posterior psm ( e.g. fig .
[ fig : boundaries]a ) . substituting the posterior boundary condition @xmath88 in eq .
( [ eq : discrete - psm - spatial ] ) we obtain a relation for the collective frequency of oscillations ( see also @xcite ) : @xmath89 the solutions to this equation are shown in figs .
[ fig : omega.tau]a and [ fig : omega.tau]b . results from numerical simulations of eq .
( [ eq : discrete ] ) in two spatial dimensions show that the collective frequency indeed fulfills eq .
( [ eq : omega - epsilon - tau ] ) , see blue dots in fig .
[ fig : omega.tau]a . for a given set of parameters @xmath28 , @xmath90 , and @xmath91 , eq .
( [ eq : omega - epsilon - tau ] ) allows for multiple solutions for the collective frequency @xmath65 .
independent measurement of coupling strength @xmath90 , collective frequency @xmath28 and collective frequency @xmath65 would allow the determination of possible values of the delay @xmath91 consistent with eq .
( [ eq : omega - epsilon - tau ] ) .
experimentally , this can be done studying situations where the intrinsic cellular oscillations are altered ( modified @xmath28 ) or where the coupling strength is altered ( modified @xmath90 ) and using the observed values of @xmath65 to fit @xmath91 . a linear stability analysis following @xcite reveals that when @xmath92 the solution to eq .
( [ eq : omega - epsilon - tau ] ) is stable , and unstable otherwise , see continuous and dashed lines in fig . [
fig : omega.tau ] .
consequently , multi - stability occurs for large values of @xmath90 and @xmath91 . as seen in fig .
[ fig : omega.tau]a , for the biologically plausible parameters that we use in the figure , there is a small gap of time delay values between the first and the second stable branches of the solution .
this happens around @xmath93 , which means that the delay is close to half the intrinsic period of the cellular oscillators in the posterior psm , @xmath94 .
for the parameters in fig .
[ fig : omega.tau]a , larger values of the delay always involve at least one stable solution .
note that values of the collective frequency equal to the intrinsic frequency , @xmath95 , are only possible for delays equal to integer and semi - integer multiples of the intrinsic period @xmath96 : these solutions are stable in the case of integer multiples ( @xmath97integer@xmath98 ) and unstable in the case of semi - integer multiples ( @xmath99integer@xmath100 )
. however , stable solutions are possible for these latter delays , albeit with the collective frequency @xmath65 different than the intrinsic frequency @xmath28 .
( [ eq : omega - epsilon - tau ] ) provides an explanation for the non - monotonic behavior of the collective period observed in fig . [ fig:2d_simulations ] .
moreover , the simulation results coincide quantitatively with the prediction of eq .
( [ eq : omega - epsilon - tau ] ) , as shown by the three dots in fig .
[ fig : omega.tau]a .
( [ eq : omega - epsilon - tau ] ) is biologically relevant : the collective frequency or period of somitogenesis emerges as a self - organized property , and depends not only on the intrinsic frequency of individual cells , but also on the coupling strength and the time delay ( fig .
[ fig : omega.tau ] ) .
note that @xmath65 does not depend on the specific shape of the frequency profile , and the period is set by the uniform phase cell population in the tail , which is the pacemaker of the whole oscillatory process .
effects of noise in the delayed coupling theory .
we include a white gaussian noise as discussed in the text .
( a ) delayed coupling is robust against the influence of noise .
parameters as in fig .
[ fig:2d_simulations]b .
( b ) impaired coupling results in segmentations defects .
after initial synchronization with resulting segments not shown , coupling is turned off ( @xmath101 ( cell diameter)@xmath69min@xmath35 ) .
the first segments have recognizable boundaries , but posterior segments are increasingly disrupted due to the effect of noise .
parameters as in ( a).,width=328 ] phase profile in the psm in the continuum limit .
( a ) phase profile as a function of relative position , given by eq .
( [ eq : solution.phase.nondimensional ] ) . left axis : phase relative to the arrest front .
right axis : corresponding number of gene expression stripes .
the green solid line corresponds to the set of parameters obtained from zebrafish data , see table i , using @xmath51 min and @xmath102 min for illustration .
orange dotted line corresponds to @xmath103 cell diameters .
( b , c ) waveform of the expression pattern represented as @xmath104 .
( d ) number of stripes in the psm as a function of @xmath1 from eq .
( [ eq : number - stripes ] ) , where @xmath1 is the parameter describing the decay length of the frequency profile .
black solid line corresponds to parameters in table i , with @xmath1 variable .
green square dot : @xmath1 obtained from zebrafish data , see table i. orange circular dot : mouse mode , see orange dotted curve in ( a , c ) . dotted and dashed curves correspond to higher and lower values of global frequency , which can potentially be affected by the intrinsic frequency , the coupling strength , or the time delay , see eq .
( [ eq : omega - epsilon - tau ] ) .
( e ) zebrafish mode , reproducing panel ( c ) of fig .
[ fig:2d_simulations ] for comparison with panel ( f ) .
( f ) mouse / chick mode : zebrafish parameters as in ( e ) , but with a sharper frequency profile , @xmath105 cell diameters . only one wave of expression appears in the psm , in contrast to the almost three waves in ( e ) .
movies available as supplementary material.,width=328 ] we have seen how the presence of time delay in the coupling can have an important effect in the spatiotemporal patterns of gene expression . a critical biological function of intercellular coupling is to keep neighboring cells oscillating in synchrony @xcite . to demonstrate that delays in the coupling allow this function , and to and to showcase the role of noise in our theory
, we simulate the phenotype of a class of mutant embryos in which coupling is strongly reduced @xcite . to do this we include an additive white gaussian noise in the simulations , with zero mean and correlations @xmath106 , and choose @xmath107 min@xmath35 for illustrative purposes . with coupling as in table
i , the process is not disrupted by noise , fig .
[ fig : noise.notch]a and supplementary movie 5 .
when coupling is disrupted , the simulation exhibits posterior segmentation defects , fig .
[ fig : noise.notch]b and supplementary movie 6 , resembling delta - notch mutant phenotypes in zebrafish @xcite . a more subtle feature of fig . [ fig : noise.notch]b is the change in segment length after coupling has been disengaged .
as soon as the coupling is removed , the effects of time delays are no longer present , and cells can oscillate at their intrinsic frequencies . because at the time the coupling is turned off there is an established pattern in the oscillating psm , it takes a few cycles for this information to be wiped out and to reach the new steady state value of the segment length .
although out of the scope of this work , our model provides a simple framework to study the effects of different kinds of noise on segmentation . this interesting possibility remains open for future work .
while the collective frequency @xmath65 describes the temporal regularity of somitogenesis , the spatial pattern of gene expression in the psm is characterized by the phase profile @xmath77 . to evaluate the phase profile
it is convenient to introduce a continuum limit where the spatial coordinate takes continuous values , denoted by @xmath108 , replacing the discrete index @xmath8 , see methods .
the stationary phase profile @xmath109 , see fig .
[ fig : phase.profile]a , can be compared to quantitative experimental measurements of the pattern , such as the width of the stripes of gene expression reported in @xcite .
we define the wavelength @xmath110 as the distance of two points in the psm with a phase difference of @xmath111 , see fig .
[ fig : lewis.fit]a .
the wavelength is large close to the tail and becomes smaller close to the arrest front where it matches the segment length .
using the continuum formalism we find an expression for the dependence of @xmath110 with the position @xmath108 of the stripe s center relative to the arrest front @xmath112.\ ] ] here @xmath113 and @xmath114 are dimensionless parameters relating intrinsic frequency , coupling , time delay , elongation speed and the frequency profile , as defined in the methods . in fig .
[ fig : lewis.fit]b we show the fit of eq .
( [ eq : wavelength2 ] ) to the wavelengths obtained from the raw data in @xcite : distances between consecutive points with equal level of _ her1 _
expression in zebrafish embryos around the ten somite stage and raised at 28@xmath44c .
the equation fits very well to the data , showing that our choice of eq .
( [ eq : freq_profile ] ) for the frequency profile is consistent with observations .
.parameters of the delayed coupling theory and their values in zebrafish embryo at 28@xmath44c near the ten somite stage .
the first five parameters have been determined before or come from our observations .
psm length @xmath2 and decay length of the frequency profile @xmath1 come from fits of our theory to experimental data in @xcite .
the parameters @xmath28 and @xmath91 could not be determined independently in this work , but they are related by eq .
( [ eq : omega - epsilon - tau ] ) .
note that parameters change with temperature and throughout development @xcite .
we choose as length unit one cell diameter ( cd ) , in terms of which the distance between neighbor oscillators is @xmath115cd . [
cols="^,<,<",options="header " , ] from the fit to data obtained from wildtype zebrafish shown in fig .
[ fig : lewis.fit]b we determine @xmath116 and @xmath117 .
we estimate the parameters @xmath2 and @xmath1 using the definitions of @xmath113 and @xmath114 and the measured values of @xmath118 and @xmath119 , see table i. time delay affects both the collective frequency and the wavelength of the gene expression patterns .
as we show in the methods section , delayed coupling introduces a renormalization of both frequency and coupling strength .
the effects of the time delay are thus included in the dimensionless renormalized parameters of eq .
( 6 ) , but the fit of spatial gene expression patterns does not allow the separation of the contribution of the time delay from that of the intrinsic frequency , and hence these two parameters remain undetermined from this fit .
the intrinsic frequency at the posterior @xmath28 , and the time delay @xmath91 , are related through eq .
( [ eq : omega - epsilon - tau ] ) .
thus experimental determination of one would suffice to calculate the other if the coupling strength and collective frequency are known . from our estimated parameters in table
i the value of the frequency @xmath28 can be up to 30% higher or lower than the collective frequency @xmath65 , see fig .
[ fig : omega.tau]a . for an intrinsic period @xmath120 around 28 min
, this implies that changing coupling strength or delay time could situate the collective period in a range between 21 and 40 min , in qualitative agreement with the magnitude of period change from simulations of the genetic regulatory network model in @xcite for two coupled cells @xcite .
note that this period change is only possible due to the presence of delays in the coupling , both in our theory ( see eq .
( [ eq : omega - epsilon - tau ] ) ) and in the model in @xcite .
this difference in period is large and should be accessible to experimental observation , allowing at the same time for numerical determination of the values of the time delay @xmath91 and the intrinsic frequency @xmath28 .
wavelength of the pattern as a function of the position @xmath108 in the psm , where @xmath2 is the length of the part of the psm considered .
( a ) schematic representation of the wavelength @xmath110 .
( b ) fit of eq .
( [ eq : wavelength2 ] ) to the experimental data obtained from wildtype zebrafish in @xcite .
best fit parameters are @xmath121 and @xmath117 .
( @xmath122 , @xmath113 and @xmath114 are dimensionless quantities defined in the methods.),width=328 ]
we have constructed a phenomenological theory describing the tissue - level dynamics of the vertebrate segmentation clock employing phase oscillators to represent cyclic gene expression in the cells of the psm . as key ingredients of the theory
, we considered ( i ) the existence of a frequency profile , ( ii ) coupling between oscillators , ( iii ) time delay in this coupling , and ( iv ) moving boundaries corresponding to embryonic elongation and the moving arrest front .
although these four elements have been considered before , here we combine them in a unified framework .
in this theory , tissue - level phenomena are generated by the interaction of cellular properties .
for example , the collective frequency of oscillation of the psm , related to the segmentation rate , depends on the intrinsic frequency at the posterior , the coupling strength and the time delay in the coupling , eq .
( [ eq : omega - epsilon - tau ] ) ; the spatial wavelength of gene expression stripes in addition depends on the shape of the frequency profile .
knowledge of the molecular underpinnings is not necessary for this mesoscopic description . by fitting the phase profiles obtained in our continuum limit to the existing data from a vertebrate embryo
, we obtained a description of the tissue- and cellular - level processes controlling period and pattern in the system that is both quantitative and predictive .
this framework can now be used to analyze experimental and evolutionary variants of embryonic segmentation or other permutations of growing , oscillating systems .
note that the basic relationship of a clock and wavefront type model for embryonic segmentation , as initially proposed by cooke and zeeman @xcite , is that the length of a segment is the product of the arrest wavefront velocity and the period of the clock . in our description
the population of oscillators create a pattern with a collective frequency , that together with the movement of the arrest front gives rise to a segment length consistent with the clock and wavefront picture .
we have compared our theory to zebrafish data , but it applies equally well to other vertebrate species , since it does not involve species - specific details .
the difference between what is termed a zebrafish mode of oscillation in somitogenesis and a mouse / chick mode , observed also in medaka @xcite , can be characterized as follows : in the zebrafish mode , several waves of gene expression sweep simultaneously through the psm , _ i.e. _ multiple stripes of expression are detected in _ in situ _ experiments ; in mouse / chick mode , only one wave is observed .
the zebrafish mode applies also to snakes , where up to nine waves of gene expression have been observed @xcite . within our theory these different modes are characterized by the phase difference between the arrest front and the posterior border : the number of stripes of gene expression in the psm is @xmath123 , see figs .
[ fig : phase.profile]a , b , c . from eq .
( [ eq : solution.phase.nondimensional ] ) in the methods we find : @xmath124 this expression can be written as a function of only two dimensionless parameters : the ratio @xmath125 between the system length and the decay length of the frequency profile and the ratio @xmath126 between the segment length and the system length .
the number of stripes is a decreasing function of both these ratios : smooth frequency profiles with long decay lengths , as well as small segment lengths , favor a large number of stripes of gene expression , as in the zebrafish mode , see fig .
[ fig : phase.profile]d .
the coupling strength and time delay do not appear in eq .
( [ eq : number - stripes ] ) because we have neglected for simplicity higher order terms in @xmath90 where they show up explicitly .
note however that the collective period @xmath75 in eq .
( [ eq : number - stripes ] ) does depend on both the coupling strength and the time delay through eq .
( [ eq : omega - epsilon - tau ] ) , see fig .
[ fig : phase.profile]d .
thus , in the same way that it may modify the collective period ( as discussed in section [ sec : param ] ) , the effect of delayed coupling can vary up to 30% the number of stripes of gene expression observed in the psm compared to a system without coupling , see fig .
[ fig:2d_simulations ] .
a similar formalism for calculating the number of stripes has recently been published as supplementary material in @xcite .
the underlying theory was previously proposed in @xcite , and is the same as our continuum theory , but without coupling , and hence without the effects caused by the delay in the coupling . however , in @xcite no explicit choice for the shape of the frequency profile is made , hence the resulting formula for the number of stripes is a function of an unknown integral , rather than a closed formula . our eq .
( [ eq : number - stripes ] ) allows for direct quantitative comparison with data . the choice of eq .
( [ eq : freq_profile ] ) for the frequency profile comes from phenomenological observations and it is not derived from the underlying molecular interactions of the signalling gradients with psm cells .
nevertheless , our function for the frequency profile is well supported by experimental data , see fig . [
fig : lewis.fit ] .
it is important to note that a switch between modes can be achieved while preserving the timing of somitogenesis by changing the shape of the frequency profile : in fig .
[ fig : phase.profile]f we show results of simulations of a _ mouse mode in zebrafish _ with all parameters given as in table i except for @xmath1 , which is 6 cell diameters instead of 36 cell diameters , see also supplementary movie 7 .
this implies that the number of stripes can change by changing the shape of the frequency profile while leaving the collective period and segment length unaffected .
previous hypotheses for the different modes include changes in period , loss of stripe specific cyclic gene enhancers , changes to the stability of cyclic mrna or different elongation velocities @xcite .
the delayed coupling theory indicates that changes to the frequency profile , potentially through changes to fgf or wnt signaling gradients in the psm , and different sizes of the psm must be considered as well .
this is consistent with recent experiments reported in @xcite , where extra stripes of gene expression appear in a mutant with an expanded psm .
current regulatory network models for the genetic oscillations in somitogenesis @xcite , undergo a hopf bifurcation a generic mechanism by which oscillations can appear in a dynamical system when varying some parameters of the models , as for instance the transcriptional delays @xcite . although it is also valid in more general settings , our eq .
( [ eq : discrete ] ) can be obtained as the phase equation associated to the normal form of a hopf bifurcation when variations in the amplitude of the oscillations can be neglected , and as such it can in principle be derived from any of the dynamical systems associated with these regulatory networks following standard procedures @xcite .
hence our formulation represents a simplification that captures general features and properties of more detailed models .
the mechanism arresting the oscillations at the arrest front is a different problem not addressed in our present work . while the above mentioned models undergo a hopf bifurcation when varying one of their parameters , something completely different ( another kind of bifurcation triggered by the variation of a different parameter of the models , for instance ) may be happening at the arrest front .
the possibility that the oscillations are coupled to a bistable switch related to the signaling gradients in the psm has been proposed @xcite . in this scenario
the arrest of the oscillators would not be a result of the intrinsic mechanism of the oscillations , but would result from an external signal
. only stable solutions of our theory can be biologically relevant .
in addition , we hypothesize that unique solutions are required to guarantee a robust behavior in the developing embryo . in the presence of multiple stable solutions for the collective frequency
, fluctuations could drive the system to switch between these different states , with dramatic consequences for healthy development . for this reason
, we conjecture that if several time delays are consistent with a fit to experimental data , those yielding a unique value of the collective frequency should be favored .
biochemical evidence indicates that coupling time delays should be relatively short compared to other signaling processes in the vertebrate segmentation clock @xcite , thus likely precluding the observation of multistability in such an embryonic system .
multistability has been observed in other systems where coupling delays can be large , with applications in biochemistry @xcite , chemistry @xcite , control theory @xcite or laser physics @xcite , for example .
key quantitative experiments in vertebrate segmentation include determination of segmentation rates @xcite , and the analysis of expression patterns from _ in situ _
experiments @xcite and fluorescent reporter genes @xcite .
our theoretical description allows for quantitative analysis of these experiments . the comparison to experimentally observed dynamic patterns of gene expression
permits the determination of the model parameters , which are provided for wildtype zebrafish in table i. future studies in mutant embryos or embryos treated with different inhibitors will reveal which parameters are affected .
the parameters in our model can be related to different cellular functions such as molecular synthesis and trafficking of signals in the cell ( coupling delay @xmath91 ) ; the strength of intercellular signaling ( coupling strength @xmath90 ) ; the speed of a cell autonomous oscillator ( intrinsic frequency @xmath28 ) ; changes in the signaling gradients responsible for the frequency profile ( decay length @xmath1 ) ; and changes in the position of the arrest front ( reflected by the system length @xmath2 ) .
thus , analysis of experimental results using our theory can provide a deeper understanding of how molecular changes lead to new phenotypes from the altered collective dynamics of tissues .
our framework can be extended to other developmental processes that combine growth with a molecular clock .
these are for instance fore - limb autopod outgrowth and patterning @xcite , or segmentation in short germ band insects , spiders , centipedes , and other invertebrates that might form segments by a mechanism similar to the one we described @xcite .
the delayed coupling theory describes spatiotemporal patterns of gene expression during morphogenesis in agreement with experimental observations .
most importantly , our phenomenological theory provides a unified quantitative framework relating the segmentation period and cyclic patterns of gene expression to underlying properties , such as the characteristics of intercellular communication , cell autonomous oscillations , the spatial profile of the slowing of the oscillators in the psm , the rate of axial growth and the size of the psm .
our results indicate that the specific spatial pattern of cyclic gene expression in the psm does not affect the overall timing of somitogenesis , but intercellular communication should be considered as a fundamental mechanism in setting the collective frequency of the segmentation clock .
starting from eq . ( [ eq : discrete - psm ] ) a continuum limit describing the evolution of the phase can systematically be derived for any value of the time delay .
this continuum limit is valid when the typical length scale of the modulations of the pattern is much larger than the distance between oscillators , @xmath7 .
the limit is obtained by letting the distance @xmath7 tend to zero , while the total number of oscillators @xmath3 tends to infinite , in such a way that the length of the psm , @xmath127 , remains finite and constant . in the continum limit ,
we require a finite coupling strength @xmath128 to exist , which implies that @xmath90 scales as @xmath129 .
the description based on discrete oscillators with phase @xmath130 at a distance @xmath131 from the arrest front ( where @xmath8 is a discrete label ) is substituted by a description defined in a continuous field spanning from @xmath132 to @xmath133 , where @xmath108 is a real positive value giving the distance to the arrest front of a point of the field with phase @xmath134 .
the resulting continuum equation reads : @xmath135 where @xmath4 is the velocity of the arrest front , @xmath136 denotes spatial derivatives ( @xmath137 in one dimension , @xmath138 in two dimensions , and son on ) , @xmath139 is a position dependent effective frequency given by @xmath140 and @xmath141 is the effective coupling strength .
the effect of the time delay appears through @xmath91 and @xmath142 $ ] , the nearest integer to @xmath143 . in analogy with @xmath144 in the discrete case ,
the intrinsic frequency is defined as @xmath145 .
note that for simplicity we have assumed that the intrinsic coupling @xmath146 is constant throughout the psm ( as we did with @xmath90 ) ; it is straightforward to include a positional dependence by substituting @xmath146 by @xmath147 in all the previous expressions .
we can simplify eq .
( [ eq : continuum ] ) using the steady state ansatz @xmath148 as we did in the discrete case : @xmath149 the boundary conditions for eq .
( [ eq : continuum.ansatz ] ) are @xmath150 and @xmath151 . as in the discrete case , we assume that the phase is defined and uniform in the tailbud , @xmath152 .
this implies that at @xmath133 all the derivatives in eq .
( [ eq : continuum.ansatz ] ) vanish and @xmath153 .
in fact the right hand side of eq .
( [ eq : frequency.cont.approx.1 ] ) coincides with the expression for @xmath65 obtained from solving eq .
( [ eq : omega - epsilon - tau ] ) after linearization around values of the delay @xmath154 . in fig
[ fig : omega.tau]a we show in red the dependence of @xmath153 with @xmath91 given by eq .
( [ eq : frequency.cont.approx.1 ] ) for several values of @xmath155 ; note that almost the whole range of stable solutions of eq .
( [ eq : omega - epsilon - tau ] ) ( solid blue ) can be well approximated by the continuum limit ( red ) .
( [ eq : continuum ] ) with @xmath139 given by eq .
( [ eq : frequency.cont.approx.1 ] ) is valid when @xmath156 for an integer @xmath155 .
a different equation can be obtained in the cases where @xmath157 : it corresponds to the continuum approximation of the unstable solutions of eq .
( [ eq : omega - epsilon - tau ] ) shown by the broken blue lines in fig [ fig : omega.tau]a .
( [ eq : continuum.ansatz ] ) can be solved and the corresponding phase profile reads : @xmath158 + \eta^2 e^{-\mu \xi / \eta } - e^{-\mu \xi } \big\}.\end{aligned}\ ] ] where we have defined the dimensionless coordinate @xmath159 and parameters @xmath160 , @xmath161 , and @xmath162 ; @xmath163 has been set to @xmath164 to fix an arbitrary constant .
[ fig : phase.profile]a shows the shape of this phase profile .
the wavelength of the patterns of gene expression can be measured as a function of the relative position @xmath165 . in @xcite
this is done experimentally , using a definition of the wavelength @xmath110 that in our notation can be expressed as the condition @xmath166 . in the limit of small coupling strength @xmath167 , we obtain a simple relation between the local wavelength of the pattern @xmath110 and the position @xmath165 along the psm given by eq .
( [ eq : wavelength2 ] ) of the main text .
we thank ingmar riedel - kruse , lola bajard , ewa paluch and the oates and jlicher groups for enlightening discussions , and the mpi - cbg fish facility for providing healthy fish .
we also thank julian lewis for sending us the raw experimental data from @xcite and for insightful comments on an earlier version of the manuscript .
lgm acknowledges support from conicet .
this work was supported by the max planck society .
|
rhythmic and sequential subdivision of the elongating vertebrate embryonic body axis into morphological somites is controlled by an oscillating multicellular genetic network termed the segmentation clock .
this clock operates in the presomitic mesoderm ( psm ) , generating dynamic stripe patterns of oscillatory gene - expression across the field of psm cells . how these spatial patterns , the clock s collective period , and the underlying cellular - level interactions are related is not understood .
a theory encompassing temporal and spatial domains of local and collective aspects of the system is essential to tackle these questions .
our delayed coupling theory achieves this by representing the psm as an array of phase oscillators , combining four key elements : a frequency profile of oscillators slowing across the psm ; coupling between neighboring oscillators ; delay in coupling ; and a moving boundary describing embryonic axis elongation .
this theory predicts that the segmentation clock s collective period depends on delayed coupling .
we derive an expression for pattern wavelength across the psm and show how this can be used to fit dynamic wildtype gene - expression patterns , revealing the quantitative values of parameters controlling spatial and temporal organization of the oscillators in the system .
our theory can be used to analyze experimental perturbations , thereby identifying roles of genes involved in segmentation . the hfsp journal , in press .
|
a realistic model of the galactic magnetic field ( gmf ) is needed for various applications , such as direct dark matter searches , studies of cosmic rays , and others .
in particular , the gmf plays a crucial role in the propagation of ultra - high energy cosmic rays ( uhecrs)those with energies in excess of @xmath1 ev . their deflections ( assuming these are charged particles ) in the gmf are large enough to prevent the identification of some , perhaps most sources directly from the uhecr data .
however , in the case of protons these deflections are sufficiently small to be corrected for , provided the gmf is known with enough accuracy .
thus , the possibility of astronomy with charged particles at ultra - high energies may depend crucially on our knowledge of the gmf . the magnetic field ( mf ) of our galaxy
is thought to contain both regular and turbulent components ( see , e.g. , @xcite ) . while the regular component is subdominant in strength by a factor of a few , it is expected to give a dominant contribution in integral quantities such as total deflections of ultra - high energy particles . unlike the turbulent component
, the regular component adds up coherently and comes to dominate @xcite when integrated over many coherence lengths . estimating the strength and global structure of the regular mf of our galaxy is the main purpose of this paper .
our primary interest is on mf outside the thin galactic disk , since that is the field that determines the deflections for most arrival directions of uhecrs .
the regular component of the gmf has been studied previously e.g. @xcite , and in particular in the context of uhecr propagation @xcite .
the advance of this paper is that it is based on improved data on rotation measures ( rms ) and the ionized gas distribution .
we also develop improved numerical model commensurate with this improved data .
observations of other spiral galaxies reveal mfs of a few @xmath2 g strength , coherent over distances of several kiloparsecs ( see , e.g. , @xcite ) .
observations from the face - on and edge - on perspectives indicate that these fields consist of at least two components : the disk one which is concentrated towards the galactic plane and has largely a spiral pattern , and the halo one that resides at some distance from the disk plane and has radial size comparable with the size of the galaxy . as the milky way seems to be a typical member of the entire class of spiral galaxies , mfs of similar strength and configuration
should be expected in our galaxy .
there are several methods to estimate the mf in the galaxy , such as zeeman splitting measurements @xcite , infrared polarization studies @xcite , synchrotron polarization and intensity surveys @xcite , starlight polarization studies @xcite and , finally , observation of the faraday rms of different galactic ( pulsars ) and extragalactic radio sources ( e.g.r.s . )
except for the zeeman measurements all these methods are somewhat indirect as they involve additional assumptions such as density of warm ionized medium or concentrations of cosmic ray electrons . in this paper
we use recent data on the faraday rms .
when propagating through a magnetized plasma , the polarization plane of a linearly polarized electromagnetic wave of wavelength @xmath3 rotates by the angle @xmath4 proportional to the square of the wavelength , @xmath5 multi - frequency observations thus allow one to measure the value of rm .
the rm can be expressed in terms of the properties of the intergalactic / interstellar medium ( igm / ism ) and permeating mfs : @xmath6 where @xmath7 is the density of free electrons measured in @xmath8 , @xmath9 is the component of the mf parallel to the line of sight measured in @xmath2 g ( positive when directed towards the observer ) , and @xmath10 is the distance from the observer to the source in pc . combined with data on the galactic free electron distribution , equations ( [ rm1 ] ) and ( [ rm2 ] ) can be used to estimate the gmf from the faraday rms of a large number of sources .
two types of sources are normally used to measure galactic faraday rotation : galactic pulsars and extragalactic radio sources .
pulsars have an advantage that their _ intrinsic _ rm values are negligibly small .
however , the use of pulsars requires the knowledge of their distances ( usually inferred from their dispersion measures dm ) , which introduces additional assumptions about the ism density distribution . in addition , pulsars are concentrated around the galactic plane , and their total number ( @xmath11 ) is currently much smaller than the number of extragalactic radio sources with measured rms . for these reasons in this paper ,
we concentrate on the rms of the extragalactic radio sources .
our general strategy is the following .
we consider several types of analytical gmf models , each containing a finite number of parameters . for each model
we calculate the expected rms , and fit them to the observations by the binned @xmath12 method .
the quality of the fit is then used to determine the best values of the gmf parameters and their allowed ranges .
the paper is organized as follows .
section [ sec : data ] describes the data .
sections [ sec : models ] and [ sec : method ] describe the implemented models and our method .
the results are presented in section [ sec : results ] .
section [ sec : conclusions ] contains the results and conclusions .
in the present work we use two sets of the rm data . the first one is the compilation of rms recently obtained by @xcite who reanalyzed nrao vla sky survey ( nvss ) data .
nvss is the largest by number survey of polarized radio sources at declinations @xmath13 @xcite .
it was made in two nearby bands , 1364.9 and 1435.1 mhz ; each having a width of 42 mhz .
simultaneous observations in these two different bands give estimates of the rms of the sources .
the total number of the observed sources was 37,543 with the mean error , estimated by the authors , of @xmath14 .
the sky map of the nvss set is shown in figure [ figure_nvss ] , which displays the location and rm sign of each source . the data have a large blind spot corresponding to declinations below @xmath15 .
( adopted from @xcite).,width=80 ] ( adopted from @xcite).,width=80 ] an advantage of the nvss rm catalog is the large number of sources .
however , observations being made in two close bands lead to relatively large individual rm errors .
also because in general rm is not always constant in @xmath16 , it can have modest variations over a wider range of @xmath3 .
thus an nvss rm , measured in just one small wavelength range at 21 cm ( @xmath17 20.9 - 22.0 cm ) , can differ from an rm measured over a wider wavelength range which tends to average out fluctuations in rm(@xmath16 ) .
such systematic differences will add to the mean error of @xmath18 quoted above by up to @xmath19 rad m@xmath20 , and will contribute to the widths of the rm comparison plots with the results from the @xcite compilation ( knm11 ) shown in figure [ figure_nvss - vs - kr ] below . on the other hand , because @xmath21 is small , ambiguities of @xmath22 in the nvss rm s are unlikely to occur , especially at @xmath23 .
it can be easily seen that the nvss data set should be treated with care when considering the sources with large observed rms , which mainly reside near the galactic plane and in the region of inner galaxy . in order to avoid this problem
, we use the nvss data only at galactic latitudes @xmath24 .
the second all - sky data set we used is the compilation by @xcite ( 2257 sources ) .
it consists of @xmath25 revised and statistically more accurate extragalactic radio source rms of p.
p. kronberg et al .
( in preparation ) .
these are derived from polarization measurements made over a large `` baseline''of @xmath16 .
the remaining rm s are taken from the canadian galactic plane survey @xcite , the southern galactic plane survey @xcite , and two smaller published lists from @xcite ( 108 rms ) and @xcite ( 68 rms ) .
the extragalactic radio sources in the knm11 set were carefully observed at multiple frequencies and give rms largely free from ambiguities .
errors of the rm measurements in the knm11 set are typically of order @xmath26 .
we use this set for control .
the sky map of the knm11 rms is shown in figure [ figure_kronberg ] , upper panel .
the lower panel shows the same data smoothed over the circle of @xmath27 centered at the location of each source as described by @xcite .
one may observe a generally good agreement in the rm sky distributions of the nvss and knm11 .
this is particularly visible when comparing figure [ figure_nvss ] with the lower panel of figure [ figure_kronberg ] .
the rm sky structure is rather peculiar .
the regions of positive and negative rms are roughly symmetric with respect to the galactic plane in the outer galaxy @xmath28 and anti - symmetric in the inner galaxy . note , however , that the boundary between positive and negative rms in the inner galaxy is displaced away from the galactic plane ( toward negative @xmath29 for @xmath30),that is , rms do not change sign across the galactic plane itself . this fact was observed , e.g. , in @xcite . this overall structure of rms suggests , qualitatively , a galactic field structure that consists of two components , with the disk field being symmetric in @xmath29 with respect to the galactic plane and the halo field anti - symmetric .
although the second data set is significantly smaller , it has much smaller individual rm errors and can be used in two ways .
first , it allows for an independent evaluation of the quality of the nvss data . in order to assess the accuracy of the nvss rms we identified , by positional cross - correlation , the sources that are common to both catalogs .
we did it separately for the two regions @xmath24 and @xmath31 .
we found 1338 pairs and 306 pairs of sources , respectively , separated by less than @xmath32 .
for these sources , we built distributions of differences between rms given in the nvss catalog and in the set by knm11 .
the difference distributions are shown in figure [ figure_nvss - vs - kr ] .
, shows a gaussian fit for the range ( 200 , 200 ) rad m@xmath20 .
the lower curve , for @xmath33 , shows the corresponding result for the full range ( 1000 , 1000 ) rad m@xmath20 .
, title="fig:",width=302 ] , shows a gaussian fit for the range ( 200 , 200 ) rad m@xmath20 .
the lower curve , for @xmath33 , shows the corresponding result for the full range ( 1000 , 1000 ) rad m@xmath20 . , title="fig:",width=302 ] in both regions , the distributions of @xmath34 have peaks centered at zero and well fitted by gaussians with dispersions @xmath35 and @xmath36 , respectively . in the region
@xmath24 the tails contain @xmath37% of points and are confined within @xmath38 .
the effective error introduced by these tails is of the order of the gaussian part .
thus , in this region the nvss data can be used for our purposes . on the contrary , in the region
@xmath31 the tails ( containing similar fraction of points ) are spread over a very wide range @xmath39 .
such tails would introduce errors in the low-@xmath29 bins which are too large to constrain the fit .
for this reason we exclude the bins with @xmath31 from the nvss fit , as will be described in section [ sec : method ] .
the knm11 rm set provides statistically more accurate rms in the galactic plane , including areas where the nvss data do not give sky coverage .
this allows us to check whether the models which fit the nvss data also fit the rm data in the galactic plane .
we adopt a general model of the gmf consisting of two different components : a disk field and a halo field @xcite .
each of these two components is parameterized independently . according to the qualitative features of the rm distribution discussed above , for the most part of this work we take the disk field to be symmetric with respect to the galactic plane and the halo field anti - symmetric
( other possibilities will be discussed in section [ sec : results ] ) .
the general layout of the model is shown in figure [ sketch_model ] .
.,width=302 ] [ sketch_model ] + + we consider several disk gmf models .
the first one is the widely used logarithmic spiral model , e.g. , @xcite .
there are two versions of this model depending on whether the direction of the field in two different arms is the same ( axisymmetric , or ass model ) or opposite ( bisymmetric , or bss model ) .
they are defined as follows : @xmath40 @xmath41 or @xmath42 for ass and bss models , respectively . here
@xmath43 @xmath44 where @xmath45 is the pitch angle and @xmath46 is the distance to the first field reversal .
negative @xmath46 means that the nearest reversal occurs in the direction to the galactic center , positive corresponds to the opposite direction .
@xmath47 is the distance to the galactic center ( we adopt @xmath48 kpc in this paper ) .
the amplitude of the gmf is a function of the radial coordinate @xmath49 , @xmath50 where @xmath51 is the radius of the central region where the disk field is assumed to have constant magnitude . in the course of the simulations
, the parameters were varied within the following ranges : l|l|l parameter & min & max + @xmath51 & @xmath52 kpc & @xmath53 kpc + @xmath54 & @xmath55 kpc & @xmath56 kpc + @xmath46 & @xmath57 kpc & @xmath58 kpc + @xmath45 & @xmath59 & @xmath60 + the second model of the disk field was the axisymmetric field with reversals in concentric rings .
it was taken from @xcite : @xmath61 @xmath62 @xmath63 where @xmath64 is radial scale of the mf .
we studied the following parameter space : l|l|l parameter & min & max + @xmath51 & @xmath65 kpc & @xmath66 kpc + @xmath64 & @xmath67 kpc & @xmath68 kpc + @xmath54 & @xmath55 kpc & @xmath56 kpc + we also varied slightly the radial positions of the inner mf reversals ( the boundaries of regions of different sign in equation ( [ eq : d2 ] ) ) in the ranges @xmath69 kpc , @xmath70 kpc and @xmath71 kpc .
the magnitude of the disk mf at the vicinity of the solar system @xmath72 was taken to be @xmath73 g .
note that , in the total gmf , the disk and halo components will be combined with arbitrary coefficients ( to be determined by fitting ) , so the choice @xmath74 g at this point is merely a matter of definition .
the basic halo model was taken from @xcite : @xmath75 @xmath76^{-1}\frac{r}{r_0^h } \exp{\left ( 1- { r\over r_0^h } \right ) } , \label{halo1}\ ] ] where the direction of the field is reversed in the southern hemisphere @xcite .
we repeat here the following definitions from @xcite : @xmath77 is the halo mf strength , @xmath78 is its radial scale , parameter @xmath79 defines the vertical position of the halo , @xmath80 is its vertical scale ; this scale could differ in directions to and away from the galactic plane , @xmath81 and @xmath82 respectively .
we will refer to these parameters as @xmath83 and @xmath84 , correspondingly . here and below the index `` h '' denotes parameters belonging to the halo .
the original values of the parameters in the model of @xcite were : @xmath85 g , @xmath86 kpc , @xmath87 and @xmath88 for @xmath81 , otherwise @xmath89 ( @xmath90 and @xmath91 ) .
we also tested halo models with slightly different radial profiles : @xmath75 @xmath92^{-1 } \exp{\left(-\left(\frac{r -r_0^h}{r_0^h}\right)^2\right ) } , \label{halo2}\ ] ] and @xmath75 @xmath93^{-1 } \exp{\left(-\left|\frac{r -r_0^h}{r_t}\right|\right)}. \label{halo3}\ ] ] we investigated the following parameter space : l|l|l parameter & min & max + @xmath94 & @xmath95 kpc & @xmath68 kpc + @xmath96 & @xmath97 kpc & @xmath98 kpc + @xmath99 & @xmath100 kpc & @xmath95 kpc + @xmath101 & @xmath102 kpc & @xmath103 kpc + @xmath104 & @xmath105 kpc & @xmath55 kpc + given that the halo field has not been firmly established as entirely azimuthal , we further studied the basic halo model , varying the pitch angle within the range @xmath106 .
calculation of the rms , equation ( [ rm2 ] ) , requires a knowledge of the galactic electron density @xmath7 . in this paper
we adopted the ne2001 model developed by @xcite .
however , there is mounting evidence that the original vertical scale of this model is insufficient and pulsar observations are better described by the electron density with larger vertical scale @xcite .
we tried both the original ( @xmath107 , @xmath108 at @xmath109 ) and modified versions ( @xmath110 , @xmath111 at @xmath109 ) of ne2001 .
the preliminary simulations with both versions of ne2001 were made with a coarse grid of gmf parameters .
the original model performed systematically worse than the modified one , so we proceeded to the full simulation with the modified version of ne2001 .
the results presented below were obtained with this modified version .
apart from the test of two vertical scales as described in section [ sec : results ] , we did not vary any other parameters of ne2001 during the fits .
rms of individual sources are affected by uncertainties due to the intrinsic source contributions to the rm and by small - scale fluctuations of the gmf .
the effect of the coherent component of the gmf can be revealed by averaging over a large number of nearby sources .
the angular bin size should be small enough so that variations of the coherent component of gmf within the bin can be neglected , and it should include a sufficient number of sources to suppress source - to - source fluctuations . applying these requirements , we chose the following binning : the celestial sphere is divided into 18 bands of 10@xmath112 width , from 90@xmath112 to + 90@xmath112 in galactic latitude @xmath29 .
each band is further divided into bins so that individual bins would have roughly equal area , about 100 deg@xmath113 , as shown in figure [ figure_bin ] . for details ) .
, width=302 ] so , there are 36 bins in the 9th band ( @xmath114 ) , but only four in the first one ( @xmath115 ) .
the total number of bins is equal to 422 . after averaging ,
all the nvss data are reduced to 422 values of average rm in each bin .
no outliers were removed at this stage .
figures [ figure_bands_north ] , and [ figure_bands_south ] shows the averaged rms in the eight bands with @xmath116 and @xmath117 , respectively .
the coherent structure is clearly visible . .
the error bars show the errors used in the fit .
, width=283 ] .
the error bars show the errors used in the fit.,width=283 ] the gmf models were treated in a similar way . for each model we simulated 20,000 values of rms at randomly chosen locations and then applied the same binning and averaging procedure to these sets .
the resulting averaged rms were compared to observations .
a somewhat different procedure was used for the set of sources from knm11 . to check the disk field
, we used the sources residing within @xmath118 from the galactic plane .
as their number is more limited ( about 240 ) and they are located irregularly within their respective bins ( 10 deg@xmath113 in this case ) , we simulated model rm values in their exact locations to avoid an additional error that would be produced by different distributions of simulated and real sources within the bin .
important for this analysis is the weighting of different bins or , equivalently , the bin - by - bin estimation of errors .
there are several sources of scatter of individual rms within the bin : internal rm of the source , an anomalous extragalactic rm component , random component of the galactic field , variation of the regular gmf over the bin , fluctuations of the galactic electron density @xmath7 , and the measurement error .
all these factors contribute to deviations of the observed average value of rm from that predicted by the gmf model , and determine the error that has to be assigned to the bin . if all the above factors were random , it would be possible to estimate the error by statistical means .
note , however , that some of these factors ( e.g. , random component of gmf with coherence length comparable to the bin size , fluctuations of @xmath7 ) may act coherently over the bin .
these factors are much more difficult to treat rigorously .
we therefore take a phenomenological approach and assume that these factors produce an error in the bin that is some fraction of the root mean square ( rms ) of the rms in that bin . in appendix [ excludedbins ]
we develop a toy model where such proportionality takes place .
taking account of this complication , we have estimated the errors of rms in the individual bins as follows . by considering the sources in the directions of north and south galactic poles where the effect of the regular mf is expected to be suppressed ,
the dispersion in the observed rms of quasars is estimated as @xmath119 rad@xmath120m@xmath20 ( see appendix [ excludedbins ] ) .
if the rms @xmath121 of rms in the @xmath122th bin did not exceed this value , we set the error of average rms to @xmath123 , where @xmath124 is the number of sources in the bin .
the reason is that in this case there is no indication of coherence effects , so one may assume that the error is due to stochastic factors such as the internal scatter , measurement error , and short - scale random mfs . in the opposite case
we assume that the variance in the observed rms is ( partly ) due to the presence of some unknown features that could act coherently over the bin , but can not be accurately modeled .
we assigned the error of @xmath125 to these bins , where the factor @xmath126 is taken from the toy model of appendix [ excludedbins ] .
we have checked that the best - fit parameters do not change significantly when this factor is varied in the range @xmath127 .
this uncertainty in the absolute normalization of errors we can not reduce further .
after a bin - by - bin examination , we have increased the errors in bins where there are clearly visible anomalies in the observed data ( see the list of these bins below in the appendix [ excludedbins ] ) .
these problems are : ( 1 ) existence of small cluster of sources with rms sharply different from the surrounding values ( 2 ) a strongly irregular distribution of the observed sources across the bin ( due to , e.g. , deep surveys within small regions ) that could bias the average compared to the case of a random distribution of the simulated sources , and ( 3 ) the presence of sources with absolute values of rms larger than 300 rad m@xmath20 which should be treated with suspicion due to the @xmath128 ambiguity in the definition of polarization angle .
the total number of bins with errors increased by hand is about 10% .
the algorithm is described in appendix [ excludedbins ] .
some bins were completely removed from the analysis .
first , these are the bins in polar regions with @xmath129 ( 50 bins ) and @xmath130 ( 77 bins ) . in these regions
the model values of rm are small because the projection of the mf onto the line of sight is suppressed ( recall that in our model both disk and halo fields are parallel to the galactic plane regardless of the parameters ) . on the other hand ,
there is strong evidence that some sort of screen is present in the direction of high latitudes @xcite which may provide a dominant contribution in these regions .
the averaged rms in polar regions , together with the assigned errors , are shown in figures [ northern_cap ] , and [ southern_cap ] where some bias toward positive values of rms can indeed be seen .
second , all bins with @xmath131 were excluded because we could not trust the accuracy of the nvss values in this region as explained in section [ sec : data ] . finally , we excluded all bins where the number of observed sources was smaller than 30 ( this primarily concerns bins overlapping with the blind spot ) .
figure [ figure_bin ] shows the excluded bins in gray .
a similar prescription was adopted for the analysis of the 1@xmath112 strip along the galactic plane using the rm set by knm11 , with the following changes .
as the area in this case is 10 times smaller ( @xmath132 deg@xmath113 instead of @xmath133 deg@xmath113 ) we set our lower limit to three sources in the bin . also , we increased the assigned error value to @xmath134 instead of @xmath125 , thus attempting to reproduce the enhanced uncertainty in the disk due to the presence of a large number of local structures and very long propagation path length inside the disk ( of order of tens of kpc ) . in the 1@xmath112 strip ,
after visual examination , we removed 20 sources out of total number of 259 .
rm values of the removed sources clearly demonstrate large differences with rms of neighboring sources .
the best values of the gmf parameters were identified by the maximum likelihood method .
this amounts to minimizing the value of the @xmath12 defined as follows , @xmath135 where @xmath136 is the total number of bins that we have included in the fit , @xmath137 is the difference between the observed and simulated average rm values for the @xmath138th bin and @xmath139 is the adopted error of the observed average .
since the rms ( equation ( [ rm2 ] ) ) are linear in the mf , the total rm of a given source is a simple sum of the disk and halo contributions . making use of this property
, we separately simulated more than 30,000 halo rm maps and 10,000 disk rm maps , corresponding to different parameters varied within the ranges described in section [ sec : models ] . at this stage the strengths of the disk and halo components were held fixed and set to values given in section [ sec : models ] .
then each disk configuration was combined with each halo configuration with two arbitrary coefficients @xmath140 and @xmath141 . for a global gmf model thus obtained , the value of @xmath12 in equation ( [ chi2 ] ) is a quadratic polynomial in @xmath140 and @xmath141 , whose minimum with respect to @xmath140 and @xmath141 we found analytically .
the resulting values of @xmath140 and @xmath141 determined the strength of the disk and halo fields .
the parameters of the gmf field were varied , within the ranges given in section [ sec : models ] , by two different methods : by uniformly covering a given parameter range with a regular grid ( grid - type simulation ) and by randomly picking parameter values from a uniform distribution over the desired range ( monte carlo type simulations ) .
the disk field was simulated using the grid method , whereas for the halo we used both methods in roughly equal proportion .
because of the uncertainty in the absolute normalization of @xmath12 resulting from uncertainties in the error estimation ( see section [ sec : errors ] ) , we did not combine the northern and southern galactic hemispheres in a single fit .
the two hemispheres do not come on equal footing because of the blind spot in the southern sky . instead
, our general fitting strategy was as follows .
first we performed the fit using the nvss data for the northern hemisphere only , thus obtaining the preliminary set of models .
next , we fitted the nvss data in the southern hemisphere .
we chose only the models that performed well in both tests and had the same value of the disk field .
note that this allowed for different halo fields in the northern and southern hemispheres . in these tests
we imposed the additional constraint that @xmath142 be greater than 1.8 @xmath2 g in both hemispheres , to conform with the results of other observations .
models were selected as acceptable if they satisfied the following requirement , @xmath143 where @xmath144 ( @xmath145 ) is a minimal value of @xmath12 ( @xmath146 ) .
these are the models having parameters within @xmath147 from the best - fit ones .
note that the criterion ( [ eq : chi2 ] ) is insensitive to the absolute normalization of the errors .
we choose such a criterion to reduce the effect of the ambiguities in the error estimation discussed in section [ sec : errors ] .
independently , we fitted the sources from the compilation by knm11 in the narrow strip of the galactic plane @xmath148 .
this imposed additional constraints on the structure of the field in the thin disk to which our fit of nvss data is insensitive . requiring that fits to both the nvss and knm11 data
be good , we performed the final selection of the models .
overall , we required that a successful model passes three separate fits : to the nvss data in the northern and southern hemispheres , and to the knm11 data in the disk .
note that this is , in general , more restrictive than performing a single combined fit .
figure [ fig : bin_nvss ] shows a comparison between one of the best fit models and the binned nvss data .
positive average rms are shown by red circles , and negative by blue squares .
the intensity of the color indicates the absolute value of rm .
it is seen that the general structure of the field is reproduced quite well .
the results of the fits are as follows : _ north_. the absolute minimum value of @xmath12 was equal to 232 for 104 degrees of freedom ( dof ) , @xmath149 .
a bin - by - bin comparison between the data and the best - fit model is shown in figure [ north_fit ] .
parameter ranges that were obtained from the fits are presented in table [ param_ranges ] .
the fit was most sensitive to the pitch angle @xmath45 .
the vertical scale of the disk mf @xmath54 and the distance to the nearest reversal @xmath46 affect the fit less .
there was no significant influence of the @xmath51 parameter , so we excluded it from further studies fixing its value at @xmath150kpc .
the same concerns the parameters @xmath101 and @xmath104 which were fixed at @xmath151 kpc and @xmath152 kpc .
we could not make a successful fit using the disk component only .
the addition of the halo component decreases @xmath144 by a factor of 1.5 . due to a weaker effect of the halo mf ,
its parameters are estimated with larger uncertainties than the disk ones .
in addition , there is a degeneracy between the height of the halo @xmath99 , the amplitude of the halo field and , to some extent , the radial size @xmath94 .
even unrealistically large values of the halo field of order tens of @xmath2 g could give an acceptable fit if the halo is placed at a large distance from the galactic plane where the concentration of the electrons greatly decreases ( cf .
figure [ degeneracy ] ) .
there is also a degeneracy between the vertical scales of the gmf model ( both halo and disk ) and the vertical scale of the electron density ( recall that the latter was held fixed during the fits ) .
the strength of the disk field corresponding to the allowed set of models resides in the @xmath153 g range , largely concentrating around @xmath154 g .
the amplitude of the allowed halo field is in the range @xmath155 g .
note that the halo field strength is constrained by the synchrotron radio - emission from the population of galactic relativistic electrons @xcite .
these constraints favor halo fields with the magnitude in the range @xmath156 g .
ring models of the disk field performed rather poorly in the northern hemisphere no matter what was the halo field , giving the best value of @xmath157 for 101 dof , which corresponds to @xmath158 as compared to @xmath159 for the best - fit model with the spiral disk .
thus , these models are disfavored by the nvss data , confirming the earlier finding of simard - normandin & kronberg(1980 ) .
finally , our attempts to improve the quality of the fit by introducing a non - zero halo pitch angle were unsuccessful : there was no significant change of @xmath12 when non - zero pitch angle of the halo field was varied within the range @xmath160 $ ] .
we also could not improve the fit by choosing the alternative model of the halo described by equations ( [ halo2 ] ) and ( [ halo3 ] ) . _ south_. the data from the southern hemisphere are more sparse because of the blind spot , and show more scatter , so the fits are generally worse .
the best - fit ass - type model gives @xmath161 for @xmath162 dof , which corresponds to @xmath163 .
the results for the best - fit bss model are @xmath164 for @xmath162 dof ( @xmath165 ) .
the bin - by - bin comparison between the data and the best - fit model in the southern hemisphere is shown in figure [ south_fit ] . _
disk_. finally , we fitted the rms from knm11 at very low latitudes @xmath166 .
the best - fit ass - type model gives @xmath167 for @xmath168 dof ( @xmath169 ) , while the best - fit bss model gives @xmath170 for @xmath168 dof ( @xmath171 ) .
the comparison between these data and the best - fit model is shown in figure [ plane_fit ] . .
areas not included into the fit are shaded with light - blue.,width=302 ] the results are fully compatible with the other fits .
they also agree with the results of @xcite .
unfortunately , there is not enough information about the region around the direction to the galactic center where the discrepancy between predictions of different models is the largest .
this reduces the discriminative power of the fit .
nevertheless , the fit was very sensitive to the field strength in the disk , so its results impose most stringent constraints on this parameter .
also , it was shown by @xcite that the knm11 data favor the pitch angle near the disk of @xmath172 .
taking this constraint into account allows us to exclude positive pitch angles , which is not possible from the fits to the nvss data alone . in both north and south
fits of the nvss data , the reduced @xmath12 values significantly exceed unity ( especially in the southern hemisphere ) .
this could result either from our underestimation of errors , or from features in the data that can not be accounted for by our models .
visual examination of the fits ( figures [ north_fit ] and [ south_fit ] ) shows that there exists a systematic discrepancy between the data and the best - fit models in the region @xmath173 which corresponds to directions away from the galactic center .
the discrepancy is present in both northern and southern hemispheres .
in fact , the observed variation of rm in the outer galaxy is too strong to be accounted for in _ any _ model that has a thick disk , or halo with a nearly tangential field in the outer galaxy ( i.e. , small pitch angle ) if the field magnitude decays away from the galactic center .
so , either these models are completely wrong in this region , or the large rms are caused by some nearby anomaly in the mf or the electron density , which is sufficiently close to subtend a large angle of about @xmath174 .
the latter option is supported by the fact that the variation of rm which is responsible for this feature does not visibly decrease away from the galactic plane , as it would if it were due to an additional galactic arm or another global ( and more distant ) feature .
interestingly , the anomaly in the nearby region ( the so - called region a ) was pointed out previously @xcite . for an additional check of this assumption we made a separate fit of rms of the sources located in the direction of the inner galaxy ,
i.e. , at @xmath175 . excluding this apparent anomaly should improve the fit and show the extent to which the best - fit parameters were affected by its presence .
the results of this additional test are as follows .
_ north - inner_. the best fit gives @xmath176 for 46 dof , considerably better than for the full sky .
the allowed values of parameters obtained from the restricted fit largely coincide with those obtained from the full fit .
this indicates that the full fit is not dominated by the anomalous region . _ south - inner_. @xmath177 for 22 dof .
the quality of the fit has considerably increased as compared to the fit in the full range of longitudes , but is still worse than in the northern hemisphere .
all our attempts to fit the outer galaxy separately ( @xmath178 ) , even those including an extra disk component , were unsuccessful in the sense that we could not get @xmath146 smaller than five .
this confirms our conclusion that it is not possible to fit the anomalous region with the gmf models of the type considered here .
the results of individual fits are presented in table [ param_ranges ] in the form of acceptable ranges of model parameters .
it should be stressed that there are dependencies and degeneracies among them ; table [ param_ranges ] only shows the maximum variation of each parameter within the whole acceptable set of models . in other words ,
not every combination of the acceptable parameters gives an acceptable model .
parameters of the models which fit the data best and are in agreement with all fits are summarized in tables [ disktable ] , and [ halotable ] .
ccccc ass & @xmath179\cup [ 3,7]$ ] & @xmath180 $ ] & @xmath181 $ ] & @xmath182 $ ] + & & & & + bss & @xmath183\cup[5,8]$ ] & @xmath184$]&@xmath185 $ ] & @xmath186 $ ] + cccc & & north & + vs. ass & [ 1.2 , 2.4 ] & [ 6.0 , 15 ] & [ 4.0 , 12 ] + vs. bss & [ 1.0 , 2.0 ] & [ 8.0 , 15 ] & [ 3.0 , 12 ] + & & south & + vs. ass & [ 1.0 , 2.5]&[3.5 , 15 ] & [ 1.5 , 6.0 ] + vs. bss & [ 1.0 , 2.2]&[4.0 , 15]&[2.5 , 5.0 ] + finally , let us comment briefly on the other tests that we have made .
first , as has been already mentioned in section [ sec : electron - density ] , we have run preliminary simulations with the original , lower vertical scale of electron density . for the realistic strength of the disk field @xmath187 and the original electron density , we found only @xmath12 values larger than @xmath188 as compared to @xmath189 for the best - fit case .
if we vary the magnitude of the disk field , the value of @xmath12 becomes close to the best - fit value for field strengths @xmath190 , which is clearly unacceptable . in conclusion ,
our simulations disfavor the original vertical scale of the electron density .
second , we have tested different combinations of symmetries ( with respect to the galactic plane ) of the halo and disk fields in order to check whether the nvss data alone favor the symmetric disk and antisymmetric halo as we have chosen in our model . with our fitting procedure which treats the northern and southern hemispheres independently ,
no additional simulations are required for this test .
for instance , for an anti - symmetric disk field we simply have to require that the coefficients with which the disk contribution enters the northern and southern hemispheres are opposite in sign .
we could not find a satisfactory fit neither with ass nor with bss anti - symmetric disk field , obtaining the values of @xmath12 more than four times larger than the best - fit value . in the case of the symmetric disk and symmetric halo
, the best fit gives @xmath12 2.5 times larger than the best - fit model .
thus , the combination of the symmetric ( with respect to @xmath29 ) disk and antisymmetric ( with respect to @xmath29 ) halo is favored by the nvss data .
the latest faraday rotation data show an unmistakable coherent component of the rms over the sky .
the rms are correlated in sign and value over the largest angular scales . moreover ,
the distribution of rms is correlated with the galactic plane and the direction to the galactic center .
this strongly indicates the galactic origin of this coherence and provides firm evidence for the existence of a regular gmf .
apart from the galactic plane region @xmath191 , the global pattern of the rms has different symmetry with respect to the galactic plane in the inner and outer galaxy : symmetric outside and anti - symmetric inside ( see figure [ fig : bin_nvss ] ) .
this change of the symmetry can not be achieved with a single ( symmetric or antisymmetric ) component of the gmf .
at least two components disk and halo are required ( figure [ sketch_model ] ) .
since the study of the mf in the disk shows that it does not change sign across the galactic plane itself , the combination of the symmetric disk field and the anti - symmetric halo field is favored .
we have demonstrated that such a combination , with properly chosen amplitudes of components , reproduces the observed pattern of rms very well .
several further conclusions can be made . *
the cross check between the nvss and the smaller knm11 set shows that the quality of the nvss is sufficient for our purposes all over the sky except , in the @xmath192 band around the galactic plane .
* there is a feature in the observed rms near the direction to the galactic anti - center that can not be accounted for by models of the type considered in this paper ( nearly tangential field decaying away from the galactic center ) .
this feature may be due to some unspecified local structure . *
a ring model of the gmf is disfavored as compared to the model with logarithmic spiral arms . in the former case
the best value of @xmath146 is a factor of @xmath193 larger than in the latter .
* our fits favor the increased vertical scale of the electron density distribution as proposed by @xcite .
we were unable to find any fit with the original vertical scale of ne2001 model and strong disk field .
this is in agreement with the results of @xcite .
* our fits favor somewhat smaller values of the pitch angle @xmath45 than have been discussed in the literature previously .
these smaller values agree with the recent results of @xcite where the reference value of @xmath194 was obtained by a totally different method .
our fitting procedure could not discriminate between the ass and bss models because a major contribution to the extragalactic source rms ( especially at @xmath195 ) comes from the near vicinity of the solar system ; in this region both models behave quite similarly .
a more detailed analysis of the galactic plane region may be more sensitive to the differences between the ass and bss models .
two `` benchmark '' models ( one with the ass and one with the bss disk field ) are presented in table [ tab : benchmark ] .
these models fit the rm data well and have values of parameters favored by other studies ( not too small pitch angle @xmath45 , not too strong halo field and the magnitude in the vicinity of the earth close to @xmath73 g ) .
they can be used , e.g. , for propagation of cosmic rays through the galaxy . at ultra - high energies ,
an accurate three - dimensional description of the galactic magnetic field may be crucial for identification of extragalactic cosmic ray sources .
lcc disk : & & + @xmath45 & @xmath196 & @xmath197 + @xmath54 & @xmath198 kpc & @xmath198 kpc + @xmath46 & @xmath199 kpc & @xmath199 kpc + @xmath72 & @xmath200 g & @xmath200 g + @xmath51 & @xmath201 kpc & @xmath201 kpc + + halo ( north ) : & & + @xmath99 & @xmath202 kpc & @xmath202 kpc + @xmath94 & @xmath67 kpc & @xmath67 kpc + @xmath203 & @xmath204 g & @xmath204 g + @xmath101 & @xmath205 kpc & @xmath205 kpc + @xmath104 & @xmath103 kpc & @xmath103 kpc + + halo ( south ) : & & + @xmath99 & @xmath202 kpc & @xmath202 kpc + @xmath94 & @xmath67 kpc & @xmath67 kpc + @xmath203 & @xmath206 g & @xmath204 g + @xmath101 & @xmath205 kpc & @xmath205 kpc + @xmath104 & @xmath103 kpc & @xmath103 kpc + in this paper we used only the faraday rms of extragalactic radio sources .
more precise understanding of the gmf can be achieved by taking into account of other data , also sensitive to the mf , such as rms of pulsars , measurements of the galactic synchrotron radiation , etc .
this would permit better constraints on the unknown quantities such as the three - dimensional electron density profiles and other properties of the ism .
it will also help to resolve some degeneracies inherent in our models .
the authors are grateful to d.d .
sokoloff and jinlin han for fruitful discussions .
also , we thank an anonymous referee for careful reading of the manuscript and valuable remarks . this work is supported by iisn project no .
4.4509.10 and the natural sciences and engineering research council of canada ( nserc ) , the australian research council , the u.s .
department of energy , and the use of nasa s astrophysics data system .
in our fitting procedure , each bin included in the analysis was characterized by an average rm and its error @xmath207 .
here we describe the algorithm by which these errors were assigned on a bin - by - bin basis .
first , we estimated the intrinsic dispersion of rms of extragalactic sources . to this end
we investigated the regions on the sky where the influence of the gmf is expected to be suppressed .
since rms are proportional to the mf component that is parallel to the line of sight , these are the regions around the galactic poles . considering the sources with @xmath208 we found that the dispersion of rms of individual sources in these regions was about @xmath209 . comparing to figure [ figure_nvss - vs - kr ]
we see that the dominant contribution to this number comes from the measurement errors . in the error assignment algorithm
, we adopted the value of @xmath210 as an indicator of whether a given bin is perturbed by effects other than the random scatter of sources . if a bin showed the dispersion of rms @xmath121 smaller than @xmath210 we considered it unperturbed and assigned its error on a statistical basis .
since the average of a sample of @xmath211 elements varies around the true average with the width @xmath212 , where @xmath213 is the variance of the distribution , we assigned to such bins an error of @xmath214 , @xmath124 being the number of sources in the bin . in practice
, this situation occurs only for high - latitude bins . in the opposite case , i.e. , when the dispersion @xmath121 was larger than @xmath210 , we considered the bin as perturbed by coherence effects .
these may be due to , e.g. , random mf with the coherence length comparable to the bin size , fluctuations of electron density , strong variation of the regular component across the bin , etc . in this case
the statistical approach is not correct , and a precise determination of the error is difficult .
we have assumed that the error in such bins is some fraction of the bin variance @xmath121 , which we have adopted to be @xmath126 , so that the error in perturbed bins was taken @xmath215 .
we can not rigorously derive this coefficient ( although it can be motivated by a toy model presented below ) .
since this uncertainty propagates directly into the value of @xmath12 , it results in some uncertainty of the absolute normalization of the latter . note , however , that the ratios of @xmath12 are less affected by this uncertainty .
the following toy model shows how the proportionality between the error and the dispersion in the bin may come about .
let the random part of the mf consist of cells of a typical size @xmath216 pc , such that their contributions to rm have zero mean and the dispersion @xmath217 .
if the cells were statistically independent , individual sources would have an rm dispersion given by @xmath218 , where @xmath136 is the total number of cells along the line of sight ( we have neglected the internal scatter ) .
however , for sources in one bin a number @xmath219 of nearby cells contributes _
coherently_. if @xmath220 this will not change the dispersion @xmath221 over the bin .
the error , on the contrary , will be dominated by the coherent contribution , @xmath222 .
thus , the error and the dispersion are proportional to one another with the coefficient @xmath223 .
the latter coefficient is purely geometrical . taking @xmath224 , where @xmath225 is the bin size , and @xmath226 , where @xmath227 is the length of the trajectory inside gmf , we find @xmath228 .
interestingly , similar conclusions ( proportionality and a similar coefficient ) are obtained by considering a perturbation caused by a close group of a few sources having similar rms substantially different from the rest of the sources in the bin . as a last step ,
all bins were subject to visual scrutiny , with the result that some of the bins were completely removed from the analysis .
these are the bins showing obvious pathologies like small number of sources or non - uniform distribution across the bin ( e.g. , some bins at the boundary of the nvss blind spot ) , compact aggregates of sources with rms drastically different from the surrounding value , etc .
finally , some bins with such pathologies were assigned large errors ( 23 times larger than prescribed by the above algorithm ) by hand .
this was done in cases where the corresponding bin might serve as a good discriminator between models , even with a large error .
a constant error of @xmath229 was used in this case .
this value was chosen in such a way that these bins do not contribute significantly into the value of @xmath12 , at least near the best - fit point .
we have checked that the results of the fits do not depend on the precise choice of these errors .
the number of such bins is about 10% ; they are easily identifiable in figures [ figure_bands_north ] and [ figure_bands_south ] .
here is the complete list of excluded bins and bins with manually assigned errors .
_ northern hemisphere .
_ an error of @xmath229 was assigned to the bins in the region @xmath230 ( local structures described in @xcite ) ; to the bins in the area @xmath231 ( a prominent structure with large positive rms , possibly related with the gum nebula ) ; to the bin located at @xmath232 ( strong non - uniformity : only 12 sources are located at @xmath233 , while 38 at @xmath234 ) .
excluded from the analysis were the bins in the area @xmath235 ( unresolved structure in rm and a large positive spot at @xmath236 ) ; bins in the area @xmath237 ( large negative spot centered at @xmath238 ; bin affected by gum nebula at ( @xmath239 ) ; the bin centered at @xmath240 ( strongly non - uniform distribution ) .
_ southern hemisphere .
_ an error of 40 rad m@xmath20 was assigned to the bin @xmath241 ( local structure resulting in positive excess ) ; to the bin @xmath242 ( positive spot ) ; to the bin centered at ( 180@xmath112 , 15@xmath112 ) ( negative spot near the anti - center direction ) ; and to the bin @xmath243 ( large positive spot ) .
cccccccccc + & & & & & north & & & & + + a & 240 & 2.30 & [ -10 , -3 ] & [ 0.5 , 1.3 ] & [ -1.1 , -0.3 ] & [ 0.9 , 2.0 ] & [ 1.0 , 1.4 ] & [ 4 , 15 ] & [ 0.4 , 0.7 ] + b & 232&2.23 & [ -6 , -10 ] & [ 0.5 , 1.3 ] & [ -1.1 , -0.5 ] & [ 0.9 , 1.5 ] & [ 1.0 , 1.4 ] & [ 6 , 15 ] & [ 0.3 , 1.0 ] + + + & & & & & south & & & & + + a&243 & 3.74 & [ -10 , 10 ] & [ 0.5 , 1.5 ] & [ -1.1 , 1.1 ] & [ 0.9 , 2.0 ] & [ 1.0 , 3.5 ] & [ 3.5 , 15 ] & [ 0.2 , 2.0 ] + b & 235&3.62 & [ -10 , 10 ] & [ 0.5 , 1.1 ] & [ -1.0 , 1.0 ] & [ 0.2 , 2.0 ] & [ 1.0 , 3.5 ] & [ 3.5 , 15 ] & [ 0.2 , 2.0 ] + + + & & & & & plane & & & & + + a&30 & 1.03 & [ -7 , 10 ] & [ 0.5 , 1.4 ] & [ -1.4 , 1.4 ] & [ 0.9 , 1.1 ] & & & + b & 37&1.27 & [ -8 , -3]@xmath244[5 , 8 ] & [ 0.5 , 1.5 ] & [ -0.8 , 1.2 ] & [ 1.0 , 1.2 ] & & & + + + & & & & & north - inner & & & & + + a&43 & 0.93 & [ -5 , 6 ] & [ 0.5 , 1.2 ] & [ -1.0 , 0.9 ] & [ 0.9 , 3.0 ] & [ 1.2 , 2.4 ] & [ 6.0 , 8.0 ] & [ 0.4 , 1.2 ] + b & 45&0.98&[-7 , 9 ] & [ 0.5 , 1.4 ] & [ -0.6 , 1.1 ] & [ 0.9 , 1.1 ] & [ 1.0 , 2.0 ] & [ 8.0 , 15 ] & [ 0.41.2 ] + + + & & & & & south - inner & & & & + + a&40 & 1.81 & [ -7 , 11 ] & [ 0.5 , 1.3 ] & [ -0.2 , -0.9 ] & [ 0.9 , 3.5 ] & [ 1.0 , 2.5 ] & [ 8.5 , 15 ] & [ 0.1 , 0.6 ] + b & 35&1.59&[-5 , 9 ] & [ 0.5 , 1.2 ] & [ -0.3 , -0.8 ] & [ 0.9 , 1.2]&[1.2 , 2.2 ] & [ 3.5 , 5.5 ] & [ 0.2 , 0.5 ] + vs. its strength @xmath245 . each dot represents an acceptable model . the degeneracy could be easily seen : a very strong halo field gives an acceptable fit provided the vertical scale @xmath246 is high .
[ degeneracy ] , width=302 ]
|
we made use of the two latest sets of rotational measures ( rms ) of extra - galactic radio sources , namely the nrao vla sky survey rotation measures catalog , and a compilation by kronberg and newton - mcgee , to infer the global structure of the galactic magnetic field ( gmf ) .
we have checked that these two data sets are mutually consistent .
given the existence of clear patterns in all - sky rm distribution we considered gmf models consisting of two components : disk ( spiral or ring ) and halo .
the parameters of these components were determined by fitting different model field geometries to the observed rms .
we found that the model consisting of a symmetric ( with respect to the galactic plane ) spiral disk and anti - symmetric halo fits the data best and reproduces the observed distribution of rms over the sky very well .
we confirm that ring disk models are disfavored .
our results favor small pitch angles around @xmath0 and an increased vertical scale of electron distribution , in agreement with some recent studies . based on our fits
, we select two benchmark models suitable for studies of cosmic ray propagation , including the ultra - high energies .
|
in his pioneering work , alan turing @xcite showed that stable spatial patterns can develop in reaction - diffusion systems which include chemical species ( morphogens ) with different diffusion constants . considering a system of two chemical species with concentrations @xmath0 and @xmath1 in one - dimensional interval
@xmath2 $ ] , the underlying deterministic model of turing patterns can be written as a system of two reaction - diffusion partial differential equations ( pdes ) @xmath3 where @xmath4 and @xmath5 are diffusion constants of morphogens @xmath6 and @xmath7 , respectively , and @xmath8 and @xmath9 describe chemical reactions . then the standard analysis proceeds as follows @xcite : a homogeneous steady state @xmath10 , @xmath11 is found by solving @xmath12 and @xmath13 .
it is shown that the homogenous steady state is stable when @xmath14 , and conditions on @xmath15 , @xmath16 , @xmath4 and @xmath5 are obtained which guarantee that the homogeneous steady state will become unstable for @xmath17 . then turing patterns are observed at the steady state .
the above argument was extensively analysed in the mathematical biology literature and conditions for turing patterns have been determined @xcite .
experimental studies with chemical systems ( chlorite - iodide - malonic acid reaction ) demonstrated turing type patterns @xcite .
there has also been experimental evidence that a simple turing patterning mechanism can appear in developmental biology , for example , in the regulation of hair follicle patterning in developing murine skin @xcite .
one of the criticism of turing patters is their lack of robustness @xcite .
the pde system ( [ upde])([vpde ] ) can have several stable non - homogeneous solutions which the system can achieve with relatively small perturbations to the initial condition . considering pdes in a suitably growing domain , one can obtain an additional constraint on the system which restricts the set of accessible patterns , increasing the robustness of pattern generation with respect to the initial conditions @xcite .
however , to assess the sensitivity of patterns with respect to fluctuations , stochastic models have to be considered @xcite .
one of the most common approaches to stochastic reaction - diffusion modelling is formulated in the compartment - based ( lattice - based ) framework @xcite . in the one - dimensional setting , the compartment - based analogue of the pde model ( [ upde])([vpde ] )
can be formulated as follows : the computational domain @xmath18 $ ] is divided into @xmath19 compartments of length @xmath20 .
we denote the number of molecules of chemical species @xmath21 ( resp .
@xmath22 ) in the @xmath23-th compartment @xmath24 by @xmath25 ( resp .
@xmath26 ) , @xmath27 .
then the diffusion of @xmath21 and @xmath22 is described by the following chains of chemical reactions " @xcite : @xmath28 @xmath29 where @xmath30 reactions are localized to each compartment .
for example , considering the commonly studied schnakenberg reaction system @xcite , chemical reactions in the @xmath23-th compartment are described by @xcite : @xmath31 the above formulation ( [ diffgillu ] ) , ( [ diffgillv ] ) and ( [ schnak ] ) describes the stochastic reaction - diffusion model as a system of ( @xmath32 ) chemical reactions : we have ( @xmath33 ) diffusive jumps of @xmath21 molecules to the left ( resp .
right ) , ( @xmath33 ) diffusive jumps of @xmath22 molecules to the left ( resp . right ) , and @xmath34 reactions ( [ schnak ] ) .
this system can be simulated using the gillespie algorithm @xcite , or its equivalent formulations @xcite . in figure
[ figure1 ] , we present an illustrative simulation of the reaction - diffusion system ( [ diffgillu ] ) , ( [ diffgillv ] ) and ( [ schnak ] ) .
we clearly see that turing patterns can be observed for the chosen set of dimensionless parameters : 4.5 cm -5 mm , @xmath35 and @xmath36 .
( a ) numbers of molecules of chemical species @xmath21 in each compartment at time @xmath37 ; ( b ) the same plot for chemical species @xmath22 .
the initial condition was the homogeneous steady state @xmath38 and @xmath39 for the parameters given in the text .
the values of @xmath40 and @xmath41 are denoted by dashed lines . adapted from @xcite with permission._,title="fig:",height=170 ] 4.5 cm -5 mm , @xmath35 and @xmath36 .
( a ) numbers of molecules of chemical species @xmath21 in each compartment at time @xmath37 ; ( b ) the same plot for chemical species @xmath22 .
the initial condition was the homogeneous steady state @xmath38 and @xmath39 for the parameters given in the text .
the values of @xmath40 and @xmath41 are denoted by dashed lines . adapted from @xcite with permission._,title="fig:",height=170 ]
@xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46 and @xmath47 .
compartment values above ( resp .
below ) the homogeneous steady state values @xmath38 and @xmath39 are coloured black ( resp .
light gray ) to visualize stochastic turing patterns .
let us note that the rate constants @xmath48 and @xmath49 are production rates per unit of area .
the stochastic model uses the production rates per one compartment which are given as @xmath50 and @xmath51 , respectively .
more details of this stochastic simulation are given in section [ sec2 ] where we introduce the corresponding propensity functions ( [ firstfouralpha])([lastfouralpha ] ) .
the compartment - based approach has been used for both theoretical analysis and computational modelling @xcite .
the regions where stochastic turing patterns can be expected were calculated using the linear noise analysis @xcite .
these studies were also generalized to growing domains @xcite , to stochastic reaction - diffusion models with delays @xcite , to non - local trimolecular reactions @xcite and to stochastic turing patterns on a network @xcite .
compartment - based software packages were developed @xcite and applied to modelling biological systems @xcite .
computational approaches were also generalized to non - regular compartments ( lattices ) and complex geometries @xcite .
stochastic simulations of turing patterns @xcite and excitable media @xcite were also presented in the literature . however , these theoretical and computational studies use the same discretization for each chemical species . in this paper , we will demonstrate that , in the case of turing patterns , this simplifying assumption can undesirably bias the obtained theoretical and computational results .
one of the assumption of the compartment - based modelling is that compartments are small enough so that they can be assumed well - mixed . in particular , the relative size of diffusion and reaction constants determine the appropriate size of the compartment @xcite .
it can be shown that there exists a limitation on the compartment size from below whenever the reaction - diffusion system includes a bimolecular reaction @xcite .
there are also bounds on the compartment size from above @xcite , again the diffusion constant plays an important role in these estimates . in the case of turing patterns , we have chemical species with different diffusion constants . for example , in the illustrative simulation in figure [ figure1 ] , we have @xmath52 , i.e. the diffusion constant of @xmath22 is 100-times larger than the diffusion constant of @xmath21 .
however , we used the same discretization for both @xmath21 and @xmath22 which is schematically denoted in figure [ figure2](a ) . 1.4 cm -5 mm and @xmath22 where @xmath53 defined by @xmath54 is equal to 5._,title="fig:",height=52 ] 1.4 cm -5 mm and @xmath22 where @xmath53 defined by @xmath54 is equal to 5._,title="fig:",height=52 ] if we take into account that @xmath22 diffuses much faster , then one could also consider the discretization in figure [ figure2](b ) where one compartment in the @xmath22 variable corresponds to several compartments in the @xmath21 variable . in this paper
, we will study differences between discretizations in figure [ figure2](a ) and figure [ figure2](b ) .
we will show that these discretizations lead to different parameter regimes for stochastic turing patterns .
the paper is organized as follows . in section [ sec2 ]
we introduce and analyse a simple test problem which will be used to illustrate our results
. it will be based on the above model ( [ diffgillu ] ) , ( [ diffgillv ] ) and ( [ schnak ] ) . in section [ sec3 ]
we analyse both types of discretizations , considering a simple two - compartment discretization in @xmath21 .
illustrative numerical results are presented in section [ sec4 ] .
we conclude this paper with the discussion of our results in section [ sec5 ] .
we will consider a simple one - dimensional schnakenberg model ( [ schnak ] ) where the reaction rate constants are given by @xcite @xmath55 and @xmath56 is a scale factor .
we used @xmath57 in the illustrative simulation in figure [ figure1 ] . when there is no diffusion involved , the dynamics of this system can be represented as the system of reaction rate ordinary differential equations ( odes ) @xmath58 which has a unique stable steady state at @xmath59 and @xmath60 .
when we consider diffusion , the reaction - diffusion pdes ( [ upde])([vpde ] ) are given by @xmath61 we are implicitly assuming homogeneous neumann boundary conditions ( zero - flux ) in the whole paper , but both the pde model ( [ updeschnak])([vpdeschnak ] ) and its stochastic counterparts could also be generalized to different types of boundary conditions @xcite . using standard analysis of turing instabilities @xcite , one can show that the turing patterns are obtained for @xmath62 for the parameter values ( [ omegascaling ] ) .
this condition is independent of @xmath63 the illustrative simulation in figure [ figure1 ] was computed for @xmath52 , i.e. the condition for ( deterministic , mean - field ) turing patterns was satisfied .
when we are concerned with the stochastic effects , the reaction - diffusion system can be simulated by the gillespie stochastic simulation algorithm with the one - dimensional computational domain @xmath18 $ ] discretized .
considering uniform discretization in figure [ figure2](a ) , the stochastic model is given as a set of chemical reactions " ( [ diffgillu ] ) , ( [ diffgillv ] ) and ( [ schnak ] ) . denoting the compartment length by @xmath64 , we have the following propensity functions in the @xmath23-th compartment @xcite : @xmath65 @xmath66 where @xmath67 and @xmath68 are given by ( [ dudvformula ] ) .
the first four propensities ( [ firstfouralpha ] ) are for the four chemical reactions in .
the propensities ( [ lastfouralpha ] ) are for the diffusive jumps ( left and right ) for @xmath21 ( indices 5 and 6 ) and @xmath22 ( indices 7 and 8) which correspond to ( [ diffgillu ] ) and ( [ diffgillv ] ) , respectively . in the illustrative simulation in figure [ figure1 ] , we divided interval @xmath69 $ ] into @xmath70 compartments , i.e. @xmath71 .
in particular , the production rate of @xmath21 molecules in one compartment was equal to @xmath72 .
the homogeneous steady state in compartments corresponded to values @xmath73 and @xmath74 the compartmentalization in figure [ figure2](b ) generalizes ( [ diffgillu ] ) and ( [ diffgillv ] ) to the case where different discretizations are used for @xmath21 and @xmath22 .
we will denote by @xmath75 ( resp .
@xmath76 ) the number of compartments in the @xmath21 ( resp .
@xmath22 ) variable .
we define the compartment lengths by @xmath77 where @xmath53 is the ratio of compartment sizes in the @xmath22 and @xmath21 variable . in what follows
, we will consider that @xmath53 is an integer .
for example , a schematic diagram in figure [ figure2](b ) used @xmath78 .
then the diffusion model is formulated as follows @xmath79 @xmath80 where @xmath81 in the standard comparment - based model ( [ diffgillu ] ) and ( [ diffgillv ] ) , we have @xmath82 .
one option to choose @xmath53 in the generalized model ( [ diffgillum ] ) and ( [ diffgillvm ] ) is to ensure that @xmath83 which implies @xmath84 then the jump rates @xmath67 and @xmath68 from the corresponding compartments are equal for molecules of @xmath21 and @xmath22 .
however , we will not restrict to the case ( [ gammaformula ] ) and consider general choices of @xmath53 in this paper .
the generalization of the first three propensities in ( [ firstfouralpha ] ) is straightforward .
propensities @xmath85 and @xmath86 in ( [ firstfouralpha ] ) correspond to chemical species @xmath21 and we have the following propensities in the @xmath23-th compartment , @xmath87 : @xmath88 and @xmath89 .
the propensity @xmath90 in ( [ firstfouralpha ] ) is considered in the @xmath91-th compartment corresponding to the @xmath22 species , i.e. in the compartment @xmath92 .
it is given as @xmath93 .
to generalize @xmath94 , we have to consider the occurrences of the trimolecular reaction @xmath95 in every small compartment in discretization of the @xmath21 variable . in the @xmath23-th compartment ,
the propensity function @xmath94 is : @xmath96 where @xmath97 corresponds to the @xmath91-th compartment in the @xmath22 variable to which the @xmath23-th compartment belongs , i.e. @xmath98 the main idea of the compartment - based model is that the molecules of @xmath22 are considered to be well - mixed in the compartments of the size @xmath99 .
thus the propensity function ( [ generalizationofalpha4 ] ) correctly generalizes the propensity of trimolecular reaction @xmath94 in the smaller compartment of length @xmath100 . in figure
[ figure3 ] , we present an illustrative simulation of the generalized compartment - based model ( [ diffgillum])([generalizationofalpha4 ] ) .
we use the same parameters as in figure [ figure1 ] to enable direct comparisons , i.e. @xmath48 , @xmath101 , @xmath49 , @xmath102 are given by ( [ omegascaling ] ) where the scale factor @xmath57 .
we use ( [ gammaformula ] ) to select the value of @xmath53 . since @xmath46 and @xmath47 , the formula ( [ gammaformula ] ) implies @xmath103 .
we use the same number of compartments for @xmath21 variable as in figure [ figure1 ] : @xmath104 . using @xmath103
, we obtain that @xmath22 is discretized into @xmath105 compartments . in figure
[ figure3 ] , we see that the turing pattern can still be clearly observed . as in figure
[ figure1 ] , 4.5 cm -5 mm @xmath106 .
( a ) numbers of molecules of chemical species @xmath21 in each compartment at time @xmath37 ; ( b ) the same plot for chemical species @xmath22 .
the initial condition was the homogeneous steady state @xmath38 and @xmath107 for the parameters given in the text .
the values of @xmath40 and @xmath41 are denoted by dashed lines._,title="fig:",height=170 ] 4.5 cm -5 mm @xmath106 .
( a ) numbers of molecules of chemical species @xmath21 in each compartment at time @xmath37 ; ( b ) the same plot for chemical species @xmath22 .
the initial condition was the homogeneous steady state @xmath38 and @xmath107 for the parameters given in the text .
the values of @xmath40 and @xmath41 are denoted by dashed lines._,title="fig:",height=170 ] compartment values above ( resp .
below ) the homogeneous steady state values @xmath38 and @xmath108 are coloured black ( resp .
light gray ) to visualize stochastic turing patterns .
since the compartments in @xmath22 variable are 10-times larger in figure [ figure3](b ) then in figure [ figure1](b ) , it is not suprising that the numbers of molecules of @xmath22 ( per compartment ) increased by the factor of 10 .
however , we can also notice that the numbers of molecules of @xmath21 per compartment quantitatively differ in figure [ figure1](a ) and figure [ figure3](a ) ( black peaks are twice taller ) .
an open question is to quantify these differences . in this paper , we will study even more fundamental issue : we will see that we can find parameter regimes where the generalized compartment - based model exhibits turing patterns , while the original discretization does not .
the generalized compartment - based model ( [ diffgillum ] ) and ( [ diffgillvm ] ) can be used to construct computational approaches to speed - up simulations of the standard compartment - based model , because it does not simulate all diffusion events for chemical species with large diffusion constants @xcite .
for example , the illustrative simulation in figure [ figure3 ] simulates ten times less compartments for @xmath22 and is less computationaly intensive than the original simulation in figure [ figure1 ] .
however , in this work , we are interested in a different question than discussing different numerical errors with different discretization strategies
. we will investigate the turing pattern formation under different discretizations .
we will argue that the classical compartment - based approach is not the best starting point to analyse noise in systems which have chemical species with different diffusion constants .
this conclusion can be already demonstrated if we consider a simple two - compartment model as we will see in the next section .
we will consider that the domain @xmath18 $ ] is divided into two compartments in the @xmath21 variable , i.e. @xmath110 then we have two possible options for the discretization of the quickly diffusing chemical species @xmath22 : -5 mm \(i ) @xmath111 which corresponds to the classical compartment - based model where @xmath112 ; -6 mm \(ii ) @xmath113 which corresponds to the generalized compartment - based model where @xmath114 .
we will start with the latter case which includes three variables @xmath115 , @xmath116 and @xmath117 and is easier to analyse . in section [ seccla2 ]
we compare our results with the classical compartment - based approach .
we consider the case where the whole interval @xmath118 $ ] is divided into two compartments for @xmath21 and one compartment for @xmath22 .
the discretization is illustrated in figure [ fig4](a ) .
-5 mm and @xmath114 : the interval is divided into two compartments for @xmath21 and remains as one compartment for @xmath22 . _
( b ) _ classical compartment - based model : the interval is divided into two compartments for both @xmath21 and @xmath22._,title="fig:",height=75 ] -5 mm and @xmath114 : the interval is divided into two compartments for @xmath21 and remains as one compartment for @xmath22 . _ ( b ) _ classical compartment - based model : the interval is divided into two compartments for both @xmath21 and @xmath22._,title="fig:",height=75 ] we will denote by @xmath119 , @xmath120 and @xmath121 the average numbers of molecules of @xmath115 , @xmath116 and @xmath117 as predicted by the corresponding mean - field model . they satisfy the following system of three odes @xcite @xmath122 we will study the stability of its steady states . in order to find the steady state , we let the left hand side terms be zero . the corresponding algebraic equations can be written in the following form : @xmath123 where we used @xmath124 and @xmath125 adding all three equations we have @xmath126 where we used the parameter choice ( [ omegascaling ] ) .
let @xmath127 and @xmath128 . solving ( [ equil3 ] ) for @xmath121 ,
we obtain @xmath129 substituting back to , we have @xmath130 using the parameter choice ( [ omegascaling ] ) , we can simplify it to @xmath131 = 0.\ ] ] the system will have a non - homogeneous solution @xmath132 if and only if the equation has a non - zero solution , and that requires @xmath133 . using ( [ dudvformulam ] ) and @xmath134 , we obtain @xmath135 if this condition is satisfied than the system has two non - nonhomogeneous steady - state solutions @xmath136 where @xmath137 in figure [ figure5 ] , we illustrate this result .
we use @xmath138 , @xmath139 and @xmath140 .
then @xmath141 and the steady state values of @xmath119 ( resp .
@xmath120 are ) : @xmath142 in figure [ figure5](a ) , we present the time evolution of @xmath115 computed by the gillespie algorithm .
we initialize the system at the steady state @xmath143=[634,366,350]$ ] .
we clearly see that the system is capable of switching between this state and the second non - homogeneous state . in figure [ figure5](b ) , we visualize the corresponding time - dependent pattern . as in figures [ figure1 ] and [ figure3 ] , we plot the values which are larger than the homogeneous steady state @xmath144 in black .
light gray colour denotes the values which are lower than @xmath144 .
we plot both @xmath115 and @xmath116 values in figure [ figure5](b ) to visualize the resulting pattern
. 4.5 cm -5 mm computed for the generalized compartment - based model with @xmath109 and @xmath114 .
the homogeneous steady state @xmath144 is plotted using the dashed line .
( b ) the time - dependent pattern given by the values of @xmath115 and @xmath116 computed for the same realization of the gillespie algorithm as in the panel ( a)._,title="fig:",height=170 ] 4.5 cm -5 mm computed for the generalized compartment - based model with @xmath109 and @xmath114 .
the homogeneous steady state @xmath144 is plotted using the dashed line .
( b ) the time - dependent pattern given by the values of @xmath115 and @xmath116 computed for the same realization of the gillespie algorithm as in the panel ( a)._,title="fig:",height=170 ] next we consider the case where the whole interval @xmath118 $ ] is divided into two compartments for both @xmath21 and @xmath22 .
the discretization is illustrated in figure [ fig4](b ) . denoting @xmath119 , @xmath120 , @xmath121 and
@xmath145 the average numbers of molecules obtained by the corresponding mean - field model , they satisfy the following system of four odes @xcite @xmath146 again letting the left hand side terms be zero and using @xmath147 , we obtain the following system of algebraic equations @xmath148 adding all equations together , we have @xmath149 adding ( [ algebra23 ] ) and ( [ algebra24 ] ) , we also have @xmath150 adding ( [ algebra21 ] ) and ( [ algebra23 ] ) , we obtain @xmath151 using ( [ sum1 m ] ) , we have @xmath127 and @xmath128 for a suitable @xmath152 . thus ( [ pom1 ] ) can be rewritten as @xmath153 where we denoted @xmath154 . substituting into and denoting @xmath155 , we have @xmath156 similarly from we have @xmath157 substituting both and to , we obtain @xmath158 which can be simplified to the equation @xmath159 we are looking for the non - homogeneous solution where @xmath160 . denoting @xmath161
, we have a quadratic equation @xmath162 we will look for conditions such that the equation has a solution @xmath163 ( since @xmath164 ) .
let @xmath165 then we have @xmath166 .
one can verify that if @xmath167 , it is impossible for the equation @xmath168 to have a solution between 0 and 1 . on the other hand ,
if @xmath169 , we will definitely have a solution between 0 and 1 .
thus we have a necessary and sufficient condition @xmath170 which corresponds to the condition for @xmath67 and @xmath68 : @xmath171 we note that @xmath172 and @xmath173 , where @xmath174 .
thus the necessary and sufficient condition for patterns becomes @xmath175 if @xmath176 , then the condition ( [ condition2_1 ] ) becomes the condition which was derived for the case of the generalized compartment - based model .
the condition is a necessary condition for but not sufficient .
we illustrate it in figure [ fig6 ] for @xmath138 .
-@xmath5 plane .
the shaded area is the region where the standard compartment - based model does not yield patterns and the generalized compartment - based model has patterns .
the ( blue ) circle is the parameter regime used in figure [ fig7](a ) and the ( red ) square is the parameter regime used in figure [ fig7](b)._,height=226 ] the condition corresponds to all parameter values to the left of the dashed line in figure [ fig6 ] .
the condition corresponds to the values of @xmath4 and @xmath5 which are above the ( blue ) solid line .
the shaded area are parameter values for which the generalized compartment - based model yields non - homogeneous patterns and the standard compartment - based model does not .
next , we will use the same value of @xmath4 as in figure [ figure5 ] , namely @xmath139 .
we choose two values of @xmath5 which are denoted as the ( blue ) circle and ( red ) square in figure [ fig6 ] .
we use the gillespie algorithm to simulate the standard compartment - based model for @xmath177 .
the results are shown in figure [ fig7 ] .
the top panels show the time evolution of @xmath115 and @xmath116 .
4.5 cm -5 mm ( blue line ) and @xmath116 ( red line ) for @xmath177 is shown in top panels for ( a ) @xmath139 , @xmath178 and ( b ) @xmath139 , @xmath179 .
the corresponding time - dependent pattern is shown in bottom panels._,title="fig:",height=170 ] 4.5 cm -5 mm ( blue line ) and @xmath116 ( red line ) for @xmath177 is shown in top panels for ( a ) @xmath139 , @xmath178 and ( b ) @xmath139 , @xmath179 .
the corresponding time - dependent pattern is shown in bottom panels._,title="fig:",height=170 ] ( blue line ) and @xmath116 ( red line ) for @xmath177 is shown in top panels for ( a ) @xmath139 , @xmath178 and ( b ) @xmath139 , @xmath179 .
the corresponding time - dependent pattern is shown in bottom panels._,title="fig:",height=170 ] ( blue line ) and @xmath116 ( red line ) for @xmath177 is shown in top panels for ( a ) @xmath139 , @xmath178 and ( b ) @xmath139 , @xmath179 .
the corresponding time - dependent pattern is shown in bottom panels._,title="fig:",height=170 ] we clearly see the switching between two patterns for @xmath179 , but there is no bistability for @xmath180 the resulting patterns are visualized in the bottom panels .
as in figures [ figure1 ] , [ figure3 ] and [ figure5 ] , we plot the values which are larger than the homogeneous steady state @xmath144 in black .
light gray colour denotes the values which are lower than @xmath144 .
let us note that we are comparing the generalized compartment - based model with @xmath109 and @xmath114 with the classical compartment - based model .
in particular , the generalized compartment - based model uses @xmath181 if we substitute @xmath113 in formula ( [ gammaformula ] ) , we obtain @xmath182 .
in particular , the parameter values @xmath139 and @xmath180 are compatible with the choice ( [ gammaformula ] ) .
however , the standard comparment - based model does not exhibit patterns for this parameter choice as we observed in figure [ fig7](a ) .
* let @xmath183 .
then the inequality ( [ condition2_1 ] ) becomes @xmath184 which is possible for some values of @xmath185 if and only if @xmath186 thus patterns are possible for some values of @xmath185 provided that @xmath187 this condition is also the condition for the turing patterns to show for the original system of mean - field partial differential equations ( [ updeschnak])([vpdeschnak ] ) .
the condition for the generalized compartment - based model is only a necessary condition for the condition for the classical case as we showed in figure [ fig6 ] .
the bistability condition difference suggests that , if we use different discretizations for @xmath21 and @xmath22 , the stability of the homogeneous system may change . in this section
, we compare the generalized and classical compartment - based models for @xmath189 in figure [ fig8 ] , we use @xmath190 and @xmath191 . in this case
the condition for ( deterministic ) turing patterns is not satisfied .
the classical compartment - based model also does not show turing patterns as it is demonstrated in figure [ fig8](a ) ( with @xmath192 compartments ) and figure [ fig8](b ) ( with @xmath193 compartments ) . in both cases , no spatial turing pattern
is observed except noise from stochastic effect .
however , if the generalized compartment - based model is used , then the turing pattern may appear . in figure [ fig8](c ) , a result for the generalized compartment - based model with @xmath194 and @xmath195 is presented .
there is a clear turing pattern . in figure [ fig8](c ) , we have @xmath196 .
we also tested cases when @xmath113 and @xmath197 and obtained turing patterns .
the case @xmath197 is plotted in figure [ fig8](d ) . 4.5 cm -5 mm at time @xmath198 for @xmath191 , @xmath199 and @xmath190 with _ ( a ) _
@xmath192 ; _ ( b ) _ @xmath193 ; _ ( c ) _ @xmath194 and @xmath195 ; _ ( d ) _ @xmath200 and @xmath195 .
there is no turing pattern in the top panels ( classical compartment - based model ) .
turing patterns appear in the bottom panels ( generalized compartment - based model ) .
_ , title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath191 , @xmath199 and @xmath190 with _
@xmath192 ; _ ( b ) _ @xmath193 ; _ ( c ) _ @xmath194 and @xmath195 ; _ ( d ) _ @xmath200 and @xmath195 .
there is no turing pattern in the top panels ( classical compartment - based model ) .
turing patterns appear in the bottom panels ( generalized compartment - based model ) .
_ , title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath191 , @xmath199 and @xmath190 with _ ( a ) _
@xmath192 ; _ ( b ) _ @xmath193 ; _ ( c ) _ @xmath194 and @xmath195 ; _ ( d ) _ @xmath200 and @xmath195 .
there is no turing pattern in the top panels ( classical compartment - based model ) .
turing patterns appear in the bottom panels ( generalized compartment - based model ) .
_ , title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath191 , @xmath199 and @xmath190 with _
@xmath192 ; _ ( b ) _ @xmath193 ; _ ( c ) _ @xmath194 and @xmath195 ; _ ( d ) _ @xmath200 and @xmath195 .
there is no turing pattern in the top panels ( classical compartment - based model ) .
turing patterns appear in the bottom panels ( generalized compartment - based model ) .
_ , title="fig:",height=170 ] in figure [ fig9 ] , we demonstrate that both discretizations strategies clearly show turing patterns when we increase the ratio of diffusion constants to @xmath201 . in this case , the condition for ( deterministic ) turing patterns is satisfied . finally , we present results for @xmath202 in figure [ fig10 ] . in the deterministic pde system , when @xmath202 , turing pattern should still appear .
but in the classical compartment - based model , it is hard to claim that there is a visible turing pattern ( see figures [ fig10](a ) and [ fig10](c ) ) . considering the generalized compartment - based model , turing patterns
can be clearly observed ( see figures [ fig10](b ) and [ fig10](d ) ) .
4.5 cm -5 mm at time @xmath198 for @xmath203 .
both discretization strategies clearly show turing patterns .
we use @xmath199 , @xmath190 with _ ( a ) _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195._,title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath203 . both discretization strategies clearly show turing patterns .
we use @xmath199 , @xmath190 with _ ( a ) _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195._,title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath205 .
the generalized compartment - based model clearly shows turing patterns , while it is difficult to see whether turing patterns appear in the classical compartment - based model .
we use @xmath199 , @xmath190 with _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195 ; _ ( c ) _ @xmath192 ; _ ( d ) _ @xmath194 and @xmath195._,title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath205 .
the generalized compartment - based model clearly shows turing patterns , while it is difficult to see whether turing patterns appear in the classical compartment - based model .
we use @xmath199 , @xmath190 with _ ( a ) _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195 ; _ ( c ) _ @xmath192 ; _ ( d ) _ @xmath194 and @xmath195._,title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath205 .
the generalized compartment - based model clearly shows turing patterns , while it is difficult to see whether turing patterns appear in the classical compartment - based model .
we use @xmath199 , @xmath190 with _ ( a ) _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195 ; _ ( c ) _ @xmath192 ; _ ( d ) _ @xmath194 and @xmath195._,title="fig:",height=170 ] 4.5 cm -5 mm at time @xmath198 for @xmath205 .
the generalized compartment - based model clearly shows turing patterns , while it is difficult to see whether turing patterns appear in the classical compartment - based model .
we use @xmath199 , @xmath190 with _ ( a ) _
@xmath204 ; _ ( b ) _ @xmath200 and @xmath195 ; _ ( c ) _ @xmath192 ; _ ( d ) _ @xmath194 and @xmath195._,title="fig:",height=170 ]
we showed that two choices of compartments illustrated in figure [ figure2 ] can give different parameter regions for stochastic turing patterns .
an obvious question is which one is correct .
one possibility to address this question is to consider a more detailed molecular - based approach which would be written in the form of brownian dynamics @xcite .
we are currently working on such a simulation and we will report our findings in a future publication . although our results might look like a warning against the use of compartment - based methods for patterns based on the turing mechanism , there are very good reasons to use the compartment - based model in other situations @xcite .
compartment - based models are often less computationally intensive than detailed brownian dynamics simulations @xcite . they can be used for developing efficient multiscale methods where parts of the domain are simulated using the detailed brownian dynamics while the rest of the domain is simulated using compartments @xcite .
they can also be used to bridge brownian dynamics simulations with macroscopic pdes @xcite , because direct multiscale methods for coupling brownian dynamics with pdes are challenging to implement @xcite .
we showed in figure [ fig9 ] that the resulting patterns are comparable when the ratio of diffusion constants is sufficiently large . in this case , the generalized compartment - based model could also be used to construct computational approaches to speed - up simulations of the standard compartment - based model , because it does not simulate all diffusion events for chemical species with large diffusion constants @xcite .
the research leading to these results has received funding from the european research council under the _
european community s _ seventh framework programme _
( fp7/2007 - 2013 ) _
/ erc _ grant agreement _ no .
this publication was based on work supported in part by award no kuk - c1 - 013 - 04 , made by king abdullah university of science and technology ( kaust ) .
radek erban would also like to thank the royal society for a university research fellowship ; brasenose college , university of oxford , for a nicholas kurti junior fellowship ; and the leverhulme trust for a philip leverhulme prize .
yang cao s work was supported by the national science foundation under awards dms-1225160 and ccf-0953590 , and the national institutes of health under award gm078989 .
f. li and y. cao .
multiscale discretization for reaction diffusion systems .
proceedings of the 2012 international conference on bioinformatics and computational biology ( editors : hamid r. arabnia , quoc - nam tran associate editors : andy marsh , ashu m. g. solo ) , las vegas , nevada , usa , 2012 , 305 - 311 .
|
turing patterns can be observed in reaction - diffusion systems where chemical species have different diffusion constants . in recent years , several studies investigated the effects of noise on turing patterns and showed that the parameter regimes , for which stochastic turing patterns are observed , can be larger than the parameter regimes predicted by deterministic models , which are written in terms of partial differential equations for species concentrations . a common stochastic reaction - diffusion approach is written in terms of compartment - based ( lattice - based ) models , where the domain of interest is divided into artificial compartments and the number of molecules in each compartment is simulated . in this paper ,
the dependence of stochastic turing patterns on the compartment size is investigated .
it has previously been shown ( for relatively simpler systems ) that a modeller should not choose compartment sizes which are too small or too large , and that the optimal compartment size depends on the diffusion constant .
taking these results into account , we propose and study a compartment - based model of turing patterns where each chemical species is described using a different set of compartments .
it is shown that the parameter regions where spatial patterns form are different from the regions obtained by classical deterministic pde - based models , but they are also different from the results obtained for the stochastic reaction - diffusion models which use a single set of compartments for all chemical species . in particular , it is argued that some previously reported results on the effect of noise on turing patterns in biological systems need to be reinterpreted .
|
network structures have appeared for a long time , and along with them are those who behave abnormally within the system .
we refer to these people or their illegal activities as anomalies . with respect to financial transactional networks
, anomalies can include those who execute fraudulent transactions . in these networks ,
a common goal is to detect those anomalies to prevent future illegal actions .
bitcoin is a special type of transaction system ; more information about it can be found in [ 1 ] .
we seek to detect anomalies or suspicious activities in this anonymous network , where nodes ( i.e. users , transactions ) are unlabeled and there is no confirmation as to whether or not a given node is actually conducting illicit activities . in this project ,
we focus particularly on the problem of detecting anomalies in the bitcoin transaction network , which is related to the study of fraud detection in all types of financial transaction systems .
since this problem can be generalized to those in other network settings , which may or may not involve financial transactions , we are examining the more general problem of anomaly detection in networks . in this paper , we use three unsupervised learning methods including @xmath0-means clustering , mahalanobis distance based method , and unsupervised support vector machines ( svm ) on two graphs generated by the bitcoin transaction network : one graph has users as nodes , and the other has transactions as nodes . the rest of the paper is organized as follows .
section @xmath1 describes everything related to methods , including data collection and parsing , feature extraction , and mathematical descriptions of the machine learning techniques .
section @xmath2 discusses evaluation methods for our proposed algorithms .
section @xmath3 presents the results we obtained by running these techniques on the network - type data set we generated .
section @xmath4 evaluates our methods and results .
section @xmath5 describes future work .
section @xmath6 concludes our study .
in this section , we describe data collection and parsing . then , we describe feature extraction , and finally we provide mathematical explanations for the unsupervised learning techniques we use .
we use the bitcoin transaction data set obtained from the university of illinois urbana- champaign .
all bitcoin transactions are documented in a public ledger and are in the currency unit called the bitcoin ( btc ) .
the data set contains all bitcoin transactions beginning from the network s creation until april 7th , 2013 . for each transaction
, there can be multiple sender and receiver addresses .
furthermore , multiple addresses can belong to a single user .
finally , users are also anonymous in that there are no names or personal information associated with a given user .
the data set is quite large : there are @xmath7 users with @xmath8 transactions .
we parse the data in two ways .
the first way , which we will call the _ user graph _ , is quite intuitive : users ( where each user owns a list of addresses ) are nodes and transactions between users are edges . the second way , which we will call the _ transaction graph _
, models transactions as nodes and bitcoin flow between transactions as edges . in our analysis , we will use both graph types to investigate the bitcoin network .
the user graph will help us detect suspicious users , while the transaction graph will help us detect suspicious transactions .
using these two graph representations , we can not only find out both abnormal users and abnormal activities , but also check if our methods are consistent in the sense that suspicious transactions should belong to suspicious users . in order to use @xmath0-means as a baseline ,
calculate the mahalanobis distance , use an unsupervised svm , for each node in the graph we extract a set of features . in each of the graph representations of the data ( mentioned above ) , we extract the following @xmath9 features .
keep in mind that even with the same names , the features mean different things for the two representations . * in - degree , out - degree , unique in - degree , unique out - degree , clustering coefficient * average in - transaction , average out - transaction
* average time interval between in - transactions , average time interval between out - transactions , * balance , creation date , active duration the purpose of the @xmath0-means clustering method is to partition @xmath10 points ( i.e. @xmath10 nodes in the graph ) into @xmath0 groups of similar characteristics . technically speaking , @xmath0-means clustering itself is not a method for anomaly detection ; however , it can be useful .
because we expect outliers to stay far away from the centroids found by @xmath0-means , @xmath0-means can be used to assess our true methods . for this method to work , we first represent each node as a multi - dimensional vector in the euclidean space ; each dimension of a node is a feature that we choose from the list described in part @xmath11 . for runtime purposes ,
we select only a subset of features . for the user graph
, we use six features : in - degree , out - degree , mean incoming transaction value , mean outgoing transaction value , mean time interval , and clustering coefficient . for transaction graph
, we use three features : in - degree , out - degree , total amount of each transaction .
this method produces a set of @xmath10 points @xmath12 in which @xmath13 ( where @xmath14 or @xmath2 depending on graphs ) for each @xmath15 .
we seek to partition these @xmath10 points into @xmath0 clusters @xmath16 to solve @xmath17 where @xmath18 is the mean of the points in @xmath19 for each @xmath20 .
we use the @xmath0-means clustering algorithm as a heuristic method to solve this problem . the algorithm in details
can be found in @xmath21 $ ] .
our final note in this part is that we use the normalized log of feature values to account for different value scales of different features .
this method is based on the multivariate gaussian distribution assumption .
specifically , we assume the training set @xmath12 where @xmath13 ( again , @xmath14 for user graph and @xmath22 for transaction graph ) drawn from multivariate normal distribution @xmath23 the normal parameter fitting method ( mle ) implies that @xmath24 and @xmath25 can be estimated by @xmath26 then , we will flag @xmath27 as an anomaly if @xmath28 for some chosen threshold @xmath29 .
note that our data are unlabelled , so we will train and detect on the same data set . the usual svm method does not work here because our data are unlabeled , so we will use a modified version of it .
we also take advantage of the kernel trick so that we can use an infinite dimensional feature space .
we also assume the training set @xmath12 where @xmath13 for all @xmath30 .
we start with the primal optimization problem : @xmath31 a new test example @xmath27 will be labeled @xmath32 .
the point @xmath27 with label @xmath33 is flagged as an anomaly .
note that the data are not linearly separable , so we need to use a soft - margin svm
. however , we will not use the constant @xmath34 as in the usual soft - margin svm due to its difficulty in interpretation .
instead , we follow @xmath35 $ ] to use @xmath36-svm method as follows . @xmath37 @xmath38 here , @xmath39 $ ] can be interpreted as an upper bound for the fraction of outliers ( i.e. anomalies ) . in the results section
, we will determine which @xmath36 is optimal .
following @xmath35 $ ] again , we can write the corresponding dual problem using the kernel trick : @xmath40 @xmath41 here , we choose @xmath42 where the hyper - parameter @xmath43 will be fine - tuned later . to solve for this optimization problem efficiently , we can use smo method ( @xmath44 $ ] ) . finally for @xmath45
, we can recover @xmath46 : @xmath47 then a new point @xmath27 is flagged anomaly if @xmath48 .
again , since our data are all unlabelled , we will will use the same data set for training and detecting .
with unlabeled data , evaluating our methods is a difficult challenge . due to the nature of the network - type data
set we have , we propose three evaluation methods . * using @xmath0-means as a baseline
, we can calculate the relative distances between the detected outliers and the centroids .
if these values are small , then we conclude that our methods are not good enough .
we call this `` visualization evaluation . ''
* since we represent our data in two ways with nodes and edges somehow exchanging , we can test for our methods consistency by checking if detected suspicious users own detected suspicious transactions .
we call this `` dual evaluation . ''
+ specifically , with the user graph we can get the top @xmath49 user outliers and with the transaction graph we can get the top @xmath50 transaction outliers . in this paper , we choose @xmath51 .
we then determine @xmath52 - the set of transactions corresponding to the top @xmath49 node outliers and @xmath53 - the set of users corresponding to the top @xmath50 transaction outliers defined above .
we define + @xmath54 + and @xmath55 + finally , we define the dual evaluation metric @xmath56 by @xmath57 note that @xmath58 $ ] , and the bigger it is the more accurate our method is . *
finally , there are roughly @xmath59 revealed thieves in the bitcoin network
. we can check if they belong to our detected suspicious user and transaction sets . for this reason
, these users and their illegal transactions will be included in the test sets of our methods .
the @xmath36-svm method takes a long time to run , and since we are not able to use techniques like gpu - parallelized computation at the moment , we will limit our data set to @xmath60 data points for all methods . using the @xmath0-means clustering metric in @xmath61
$ ] we find that setting @xmath62 minimizes cross - cluster entropy for the user graph and @xmath63 for the transaction graph .
( figure @xmath64 . ) for the sake of the dual evaluation method , we choose @xmath62 for both graph types .
now , we run the mahalanobis distance based method for two types of graphs .
( figures @xmath65 . )
the detected anomalies seem to appear at the border of the plot , which indicates that abnormal activities are usually extreme .
we now run the unsupervised ( one - class ) @xmath36-svm method for two graph types .
we start with identifying the optimal hyper - parameter @xmath36 . based on the evaluation section
, we will choose the @xmath36 which gives the largest values of @xmath66 and @xmath67 .
it turns out that the optimal @xmath36 is around @xmath68 .
( figure @xmath3 . ) and @xmath67 over @xmath36 in @xmath36-svm method . ]
now with @xmath69 , the @xmath36-svm method gives us the detected anomalies for both types of graph representations in figures @xmath4 and @xmath5 .
the suspicious users and transactions suggested by the unsupervised svm method appear to be quite similar to those suggested by the mahalanobis distance based method .
they both usually appear on the outer borders of the graphs .
* using @xmath0-means clustering method , we get @xmath0 clusters with corresponding @xmath0 centroids . for each graph type ( user and transaction ) and for each method ( mahalanobis and unsupervised svm )
, we calculate the average of the ratios of detected anomaly distances to corresponding centroids over max distances from those centroids to their assigned points for the top @xmath70 outliers .
+ for the mahalanobis method , we get @xmath71 for the user graph and @xmath72 for the transaction graph . for the unsupervised svm method
, we get @xmath73 for the user graph and @xmath74 for the transaction graph .
these values are as large as we expect them to be since by observations in the result section , detected anomalies appear to be extreme points . * for the mahalanobis distance based method , we calculate @xmath66 and @xmath67 to be @xmath75 and @xmath76 respectively , which gives @xmath77 .
though this value is quite small , it comes out not as a surprise given the simplicity of our method and our small set of data .
+ for the unsupervised svm method , we get @xmath78 and @xmath79 which gives @xmath80 .
this value is much higher than the one produced by the mahalanobis distance based method , even given our small set of data .
* for the mahalanobis distance based method , we detect one known theft that occurred in june 2011 .
the anomalous transaction obtained a total of over 4000 btc from 620 various addresses and funneled them to a single address . + for the unsupervised svm based method
, we detect one known loss that occurred in october @xmath81 .
the anomalous transaction was one of 23 transactions that caused a user to lose over @xmath82 btc due to corruption in a hashing function .
we propose to parallelize computation in order to enable faster outlier detection using a gpu .
this involves significant work , such as introducing thread - safety in the learning methods , but will allow us to analyze the full dataset of @xmath83 million transactions .
in this paper , we have investigate the bitcoin network .
we first represent the data with two focuses : users and transactions .
we then use three main social network techniques to detect anomalies , which are potential anomalous users and transactions
. while the agreement metrics are not high , we are able to detect two known cases of theft and one known case of loss , out of the 30 known cases we have .
[ 1 ] https://bitcoin.org/en/ + [ 2 ] lloyd , stuart p. ( 1982 ) , _ least squares quantization in pcm _ , ieee transactions on information theory 28 ( 2 ) : 129 - 137 . + [ 3 ] b. schlkopf , j. platt , j. shawe - taylor , a. j. smola , and r. c. williamson._estimating the support of a high - dimensional distribution . _ neural computation , 13 , 2001 , 1443 - 1471 . + [ 4 ] j. platt .
_ sequential minimal optimization : a fast algorithm for training support vector machines .
_ technical report msr - tr-98 - 14 , microsoft research , 1998a .
+ [ 5 ] xiong , h. et al . _
k - means clustering versus validation measures : a data distribution perspective .
_ kdd06 , august 20 - 23 , 2006 .
|
the problem of anomaly detection has been studied for a long time . in short , anomalies are abnormal or unlikely things . in financial networks ,
thieves and illegal activities are often anomalous in nature .
members of a network want to detect anomalies as soon as possible to prevent them from harming the network s community and integrity .
many machine learning techniques have been proposed to deal with this problem ; some results appear to be quite promising but there is no obvious superior method . in this paper , we consider anomaly detection particular to the bitcoin transaction network .
our goal is to detect which users and transactions are the most suspicious ; in this case , anomalous behavior is a proxy for suspicious behavior . to this end , we use three unsupervised learning methods including @xmath0-means clustering , mahalanobis distance , and unsupervised support vector machine ( svm ) on two graphs generated by the bitcoin transaction network : one graph has users as nodes , and the other has transactions as nodes .
|
cosmology , especially its observational sector , is currently a thriving field of physics . on
the theoretical side opinions have converged to what is nowadays dubbed _ cosmological concordance model _ ( ccm ) .
but , despite of all the successes of this model in describing different cosmological observations , we should not fool ourselves to believe that the grand picture of cosmology stands on a firm basis .
the reason for this is simple : interpretation of the data within the concordance model leads inevitably to the introduction of the concepts of _ dark matter _ and _ dark energy_. we surely could live with such concepts by stating that they depend on some peculiar details which yet have to be added to the description of our universe . unfortunately , dark matter and energy make up the complete energy budget within our simple picture and therefore can not be treated as some minor details which remains to be worked out .
this is clearly an embarrassing situation which needs to be addressed by cosmologists . in the following
we will have a glance at the theoretical landscape of cosmology and pay special attention to the non - riemannian approach .
non - riemannian extensions of our current gravity theory , i.e. general relativity ( gr ) , represent a well motivated framework and have been discussed extensively in the literature . in this short review
we only address questions which are related to the possible cosmological significance of such an approach .
readers who want to learn more about the fundamentals of non - riemannian gravity theories , the gauge theoretical approach to gravity , and metric - affine gravity ( mag ) should consult the excellent reviews @xcite .
with the right amount of crudeness one could summarize the reasons to consider a drastic step , like the change of the gravity theory which underlies cosmology , as follows : * large amounts of dark matter / energy necessary to fit current observations within the ccm . * no direct observation of a dark matter particle in the laboratory . * no theoretical explanation for the smallness of the dark energy component when compared to quantum field theory . *
no reason to believe that gr is valid in the early universe , i.e. at high energies . * no test of newtonian / general relativistic gravity on cosmological scales . in the following
we have a glance at some of the proposed remedies for this situation .
although we know for sure that gr has to be modified in order to make it compatible with quantum theory @xcite , we do not have any final form of this new gravitational theory .
additionally , we do not know how possible low - energy modifications , and thus modifications that may play an important role for the aforementioned observational problems in cosmology , caused by such a new theory will look like . in table
[ table_models ] we provided a very rough overview over current theoretical approaches to extend / replace our current gravity theory and thereby also our cosmological model .
the separation into different model classes is sometimes not unique .
for example , one could also count the non - symmetric gravity models as non - riemannian models and all of the models listed in table [ table_models ] could in principle also have a non - trivial topology .
.some examples for different classes of models recently used to explain observations of cosmological significance . [ cols="<,<",options="header " , ] [ table_models ] one of the general frameworks for non - riemannian gravity theories is metric - affine gravity ( mag ) as reviewed by hehl et al . in @xcite . in the following we focus on the new geometrical notions in metric - affine gravity and try to explain their possible impact on cosmology . the reasons to pick mag as starting point are varied : ( i ) among the different alternative gravity theories mag represents a very well motivated and natural generalization , c.f .
the introduction of @xcite for a list of arguments , and ( ii ) there exists a general lagrangian formulation of mag according to which many other non - riemannian theories may be systematically classified .
( iii ) there exist several exact ( also non - cosmological ) solutions for mag which rank it among the best studied alternative gravity theories during the last years .
( iv ) the idea to couple intrinsic features of matter to new geometrical quantities can be viewed as the natural prolongation of the line of reasoning which led to the formulation of the so far most successful gravity theory , namely gr . in metric - affine gravity
the spacetime continuum which contains matter carries both stresses @xmath0 ( or momentum currents ) and hyperstresses @xmath1 ( or hypermomentum currents ) .
the geometry of spacetime is described by means of a metric @xmath2 and an independent affine connection @xmath3 .
the metric is still symmetric , i.e. @xmath4 , but the connection is no - longer given by the metric compatible connection @xmath5 . ]
known from gr and may be asymmetric @xmath6 . if we define cartan s torsion tensor @xmath7}$ ] and the nonmetricity tensor @xmath8 then , cf .
@xcite , the affine connection might be split up as follows @xmath9 furthermore , one assumes that the momentum current @xmath10 of matter essentially couples to the metric @xmath2 whereas the hypermomentum current @xmath1 couples to the affine connection @xmath11 of the spacetime . from the last assumption and the splitting of the connection as given in eq .
( [ gencon ] ) it becomes clear that mag incorporates several other alternative gravitational theories , as well as gr itself .
for example it is well known from einstein - cartan ( ec ) theory that the torsion of spacetime couples to the intrinsic spin of particles . in figure
[ fig_spacetimes ] we sketched how different spacetimes , and thereby different alternative theories which make use of these richer spacetime concepts , maybe classified with respect to the torsion and nonmetricity .
the gravitational field lagrangian of mag is expected to be of the form @xmath12 and a matter lagrangian minimally coupled to the new geometrical fields @xmath13 .
the gravitational field equations are given by the variational derivatives with respect to the metric @xmath14 and the connection @xmath15 .
we only mention here that a very general suggestion for the dynamics of this theory , which makes use of a slightly different but equivalent notation than the one used here , has been made in @xcite .
in @xcite we provided a brief chronological guide to the literature on non - riemannian cosmological models .
therein the developments in cosmology were traced back to the early seventies and were given in table form .
most of the early non - riemannian cosmological models were based on einstein - cartan theory .
investigations mainly revolved around the construction of exact solutions and the question of whether or not an initial singularity can be avoided in such models . in the 1980s
more general types of lagrangians were considered .
the inclusion of quadratic terms in the lagrangian , leading to dynamical degrees of freedom , was mainly motivated by the framework of poincar gauge theory ( pgt ) and led to new classes of exact solutions .
the story continued with the advent of the inflationary model , which led to investigations which tried to mimic or justify this new idea within different non - riemannian scenarios .
till the end of the 1990s most of the works in non - riemannian cosmology ( nrc ) were focused on the description of the early stages of the universe .
this bias can mainly be ascribed to the estimates for the new spin - spin contact interaction encountered in einstein - cartan theory .
this interaction shows up at extremely high energy densities . ] and might therefore play only a crucial role in the early universe .
this focus has changed during recent years , mostly due to the persisting need for the large amount of dark matter and more recently also dark energy .
the requirement of dark energy at late stages of the cosmic evolution might be taken as an indicator for the presence of new physics possibly due to some non - riemannian relics in cosmology .
since the field equations of the flrw model are extremely simple and the main evidence ( see @xcite for the latest snia samples ) for the new dark energy component comes from cosmological tests which are related to the expansion history of the universe , it is natural to study the impact of changes in this history and their possible origin .
there have been several suggestions for modifications during the last years coming from different directions , cf .
table [ table_models ] .
non - riemannian models , especially mag , provide a very good starting point for the study of changes of the cosmological field equations which are justified by a grander theory , in the case of mag by the general lagrangian provided in @xcite .
let us now come to the rhs , i.e. the matter side , of the field equations .
a very promising approach to construct cosmological models in a non - riemannian setup is related to the availability of more sophisticated fluid models . in the case of metric - affine gravity obukhov & tresguerres
@xcite devised a fluid model termed the _
hyperfluid_. this kind of fluid can be used as a natural source for the hypermomentum current which appears on the rhs of the field equations of mag .
the new degrees of freedom in such a fluid model can be coupled to the new geometrical properties , i.e. torsion and nonmetricity . in the hyperfluid picture , which can viewed as a generalization of early spin - fluid models @xcite ,
the motion of the fluid is described by usual four - velocity and a triad attached to each fluid element , which can undergo arbitrary deformations during the motion of the fluid .
this is analogous to the description of continua with microstructure @xcite in the theory of elasticity . in figure [ fig_dislocbub ]
we sketched two examples of such media , namely one with dislocations and another one with finely dispersed voids .
the hyperfluid and special cases of it were used in several cosmological models in the past , a systematic treatment for a fairly general lagrangian of mag will be published in @xcite .
as we sketched in figure [ fig_modelbuild ] , the list of different cosmological tests is quite long , and still growing .
usually one starts with a theoretical model from the upper portion of the figure and then compares it to some of the observations in the lower portion of the figure in order to falsify it . in view of the sometimes very different theoretical approaches this can become quite cumbersome , i.e. one has to spend a lot of time to work out the single tests in a scenario which significantly deviates from the cosmological concordance model .
therefore one of the most pressing tasks in cosmology is the development of a post - newtonian framework which allows us to compare different theoretical approaches in a systematic and somewhat standardized way , and hopefully allows for the a fast backreaction of the cosmological tests on the theoretical model . in figure [ fig_modelbuild ]
we denoted such a framework , which has yet to be developed , by the connecting middle part .
up to now there seems to be no real competitor model in the non - riemannian context which can replace the current cosmological concordance model and at the same time explain the effects caused by dark matter / energy in a purely gravitational way .
most of the models proposed so far are either not worked out to a sufficient level of detail , fail one or more of the cosmological tests , or are not distinguishable from the ccm with the currently available data .
but , and this can not be stressed enough , we are clearly only beginning to explore the different possibilities of non - riemannian models .
this is especially true for the cosmological sector of metric - affine gravity which currently only covers a very small region of the theoretically permissible parameter space .
the author wants to thank f.w .
hehl for constant advice and support .
the financial support by m. pohl and slac is gratefully acknowledged .
additionally , the author wants to thank s. lebohec for the hospitality during his stay at university of utah . used acronyms
: bbn = big bang nucleosynthesis , ccm = cosmological concordance model , cmb = cosmic microwave background , ec = einstein - cartan , friib = fanaroff - riley radio galaxy of type iib , gr = general relativity , lss = large scale structure , mag = metric - affine gravity , ngt = non - symmetric gravity theory , nrc = non - riemannian cosmology , sg = string inspired gravity , snia = supernova of type ia
|
in this work we provide the motivation for considering non - riemannian models in cosmology .
non - riemannian extensions of general relativity theory have been studied for a long time . in such theories
the spacetime continuum is no longer described by the metric alone but endowed with additional geometric quantities .
these new quantities can be coupled to the intrinsic properties of matter in a very natural way and therefore provide a richer gravitational theory , which might be necessary in view of the recent cosmological evidence for dark matter and dark energy . in this work we mainly focus on the concepts in metric - affine gravity and point out their possible significance in the process of cosmological model building .
|
suppose the president of a bank , alice , wants to give access to a vault to two vice presidents , bob and charlie , who are not entirely trusted . instead of giving the combination to any one individual , it may be desirable to distribute information in such a way that no vice president alone has any knowledge of the combination , but both of them can jointly determine the combination .
classical cryptography provides an answer which is known as secret sharing@xcite .
alice creates two coded messages and one of them is sent to bob and the other to charlie .
each of the encrypted message contains no information about her original message , but together they contain the complete message .
however , either a fourth party or the dishonest member of the bob - charlie pair gains access to both of alice s transmissions can learn the contents of her message in this classical procedure .
quantum secret sharing protocols have been proposed to accomplish this work securely@xcite where multi - photon entanglement is employed .
recently , many kinds quantum secret sharing with entanglement have been proposed@xcite . as the quantum key distribution and classical sharing protocol can be used straightforward to accomplish secret sharing safely ,
the aim of the quantum secret sharing protocols is to allow one to determine whether an eavesdropper has been active during the secret sharing and reduce the resources necessary to implement such multi - party secret sharing tasks@xcite .
but the efficiency of those quantum secret sharing protocols using entanglement can only approach 50% in principle , the lack of the multi - parties entanglement also holds their experimental implementation .
the efficiency of preparing even tripartite or four - partite entangled states is very low@xcite .
there is another thing that should be noted in this quantum secret sharing scheme with ghz states .
it uses the correlation of the measurement value of the three qubits of the ghz state@xcite : @xmath0where @xmath1 means the measurements value @xmath2 of qubit @xmath3 along basis @xmath4 with @xmath5 and @xmath6 so only the cases that even of the three qubits are measured along the @xmath7 direction are kept to share secret in this scheme and its theory efficiency is only @xmath8 in the security analysis of the case that bob eavesdrops both qubits @xmath9 and @xmath10 , the authors assume that he measures them jointly in the @xmath11or @xmath12basis .
but we find if bob measures qubits @xmath9 and @xmath10 separately ( measures qubits @xmath13 and @xmath10 in product states ) , he can cheat and get alice s secret without being detected .
for example , suppose bob measures the qubits @xmath9 and @xmath10 in the product basis @xmath14 , he can thus infer alice s value if she measures in the basis @xmath15 .
then bob can send a qubit in arbitrary state to charlie . in the procedure of announcing basis , bob tries to announce his direction after charlie and cheats his announcement to let alice keep the choose that she has measured his particle in the basis @xmath15 .
different from the eavesdropping analysis in the paper@xcite , this eavesdropping will not introduce a higher than usual failure rate .
for example , if charlie announces that he has chosen his measurement in the basis @xmath16 , bob can announce that his measurement is also in the basis @xmath17 .
then if alice also makes her measurement in the basis @xmath15 , she will believe that there is correlation between those three man s measurement value and keep this measurement value as secret .
but bob has successfully eavesdrop this secret value .
although bob has changed the state of qubit sending to charlie , he can cheat again in the check procedure if charlie announces his check bit before bob and then escapes detection . for example , when he had measured the qubits @xmath9 and @xmath10 in the product basis such as @xmath14 , bob can infer that if alice measures her particle in the basis @xmath15 will gets the value @xmath18 . when charlie that his measurement is @xmath19 , bob can announces that his measurement is also @xmath19 . then no error will occur in this check procedure .
thus bob successfully eavesdrops the secret without being detected by alice and charlie .
it means that if one party ( say bob ) can reveal his directions and values after the other party ( say charlie ) , he is able to discover alice s secret bit value , without any assistance from charlie , and escape being detected .
the feature of this eavesdropping is that bob directly measures qubits @xmath9 and @xmath10 in product basis to learn alice s value .
he cheats in the basis announcing procedure and check procedure again .
bob need nt know alice measurement direction , but he can let alice only keep the measurements he desired by announcing his direction according to charlie s measurement direction . then this eavesdropping is more powerful than those having been analyzed in the paper@xcite . in order to detect this kind of eavesdropping , alice should randomly require one party to reveal his directions before the other one , sometimes bob before charlie , sometimes charlie before bob .
and in the check procedure , she also randomly requires one party to reveal his value before the other .
this random can ensure the security of this quantum secret sharing protocol based on ghz state correlation .
it has also been pointed out in the paper@xcite , alice and bob can prevent this eavesdropping by releasing the outcomes for the check bits before the announcement of their measurement directions .
we can see that the paper@xcite introduce the epr state correlations to split the secret .
the efficiency of those quantum secret sharing protocols with entanglement can only reach 50% in principle . in this paper ,
we propose the idea to directly encode the qubits of quantum key distributions classically and present a quantum secret sharing scheme employing product states to achieve the aim mentioned above .
its security is based on the quantum no - cloning theory just as the bb84 quantum key distribution . comparing with the efficiency 50% limiting for the existing quantum secret sharing protocols with quantum entanglement
, the present scheme can be 100% efficient in principle .
it has been pointed out in the paper@xcite that the quantum key distribution can accomplish the task of secret sharing , the resources necessary can be much smaller with our present protocol although it is very simple and alike the bb84 quantum key distribution scheme .
our emphasis is that the classically encoding the qubits of quantum key distributions simply may be more economical than the idea of the paper@xcite to use epr states correlations to split secret .
this may suggests that the introduce of entanglement into the quantum secret sharing may be unworthiness .
now we present a quantum secret sharing protocol where no entanglement is employed .
its essence is to encode the sources of two bb84 key distribution schemes .
the particular process of this quantum secret sharing is as follows : \1 .
alice creates two random @xmath20-bit string @xmath21 and @xmath22 . for each bit of @xmath21 and @xmath22
, she creates a two - qubit product state @xmath23 in the basis @xmath24 ( if the corresponding bit of @xmath21 is 0 ) or the basis @xmath25 ( if the bit of @xmath21 is 1 ) where the xor result of the two qubits @xmath9 and @xmath10 equals to the bit value of the string @xmath22 .
the two - qubit vectors of the basis @xmath24 and @xmath26 are @xmath27 and @xmath28 respectively where @xmath29 and @xmath30 .
table 1 summarizes the coding two - qubit states in the corresponding basis .
for example , if the bits of @xmath21 and @xmath22 are 1 and 0 respectively , then the two - qubit state alice prepared can be @xmath31 or @xmath32 each with 50% probability .
but she knows exactly which pair he prepares .
alice sends the resulting two strings @xmath33 and @xmath34 to bob and charlie separately .
it should be emphasized that every bit of the two strings @xmath33 and @xmath34 are corresponding and their xor result equals to the corresponding bit of the string @xmath22 .
when both bob and charlie have announced the receiving of their strings @xmath35 and @xmath34 , alice announces the string @xmath21 .
\4 . bob and charlie
then measure each qubit of their string in the basis @xmath36 or @xmath26 according to the corresponding bit value of string @xmath21 .
\5 . in the check procedure , bob and charlie
are required to announce the values of their check bits .
if alice finds too few of these values agree , they abort this round of operation and restart from the first step .
the rest unchecked bits of strings @xmath22 , @xmath33 and @xmath34 can be used as raw keys for secret sharing .
alice s secret encrypted by her keys can only be decrypted by bob and charlie when they cooperate with each other .
obviously , the qubits @xmath9 and @xmath10 need not be sent out simultaneously in this quantum secret sharing scheme .
alice can send them separately to bob and charlie .
they can hold their strings until alice wants them to extract out her information and announces her the string @xmath21 .
it is virtually a combination of two modified bb84 quantum key distribution protocols , alice to bob and alice to charlie , where the states alice sent out have been classically encoded so that her keys are shared in the strings @xmath33 and @xmath34 .
either of them can learn nothing about alice s keys without the cooperation of the other one .
furthermore , this secret sharing protocol is almost 100% efficient as all the keys can be used in the ideal case of no eavesdropping , while the quantum secret sharing protocols with entanglement states@xcite can be at most 50% efficient in principle . however , in this procedure of making measurements after the announcement of bases , quantum memory is required to store the qubits which has been shown available in the present experimental technique@xcite .
adjusting the measurements order in the similar way the bb84 quantum key distribution protocol can also double its efficiency of to nearly 100%@xcite .
when no quantum memory is employed , bob and charlie measure their qubits before alice s announcement of basis in our protocol , the efficiency of the present protocol falls to 50% .
as the present scheme is more efficient than employing two bb 84 procedures , one between alice and bob and the other between alice and charlie , which has been briefly discussed in the references@xcite , it allows the possible eavesdropper , who obtains both of the particles that alice sent , to gain more information than the two - bb-84 scheme , when the qubits @xmath9 and @xmath10 are sent simultaneously .
suppose an eavesdropper , eve , obtains both of the particles that alice sent , and measures them in the bell basis .
then when the two qubits @xmath9 and @xmath10 are measured in the state @xmath37 , eve can infers that the secret is @xmath38 . and
if they in the state @xmath39 , he knows that the secret is @xmath40 .
but we can prevent this kind of information leakage causing by the jointed measurements , either by sending the two qubit un - synchronously or by encoding the two qubit as table 2 . then in the same security degree ,
the present protocol is double efficient as the two - bb-84 scheme .
in this quantum secret sharing scheme , only product states are employed and one can not get any information about the value of one qubit from measurement on the other qubit .
in fact it is exactly two independent bb84 quantum key distribution protocols , which has been proved security generally@xcite .
the only difference is that the qubits alice sent out have been classical encoded .
obviously , any eavesdropping on individual quantum channel for example alice to charlie will get nothing useful and will unavoidably introduce errors .
now suppose an adversary ( who could also be either bob or charlie ) eavesdrops the qubits @xmath9 and @xmath10 jointly .
his object is to extract out alice s information without any assistance from charlie in a way that can not be detected .
but what ever he does , if bob eavesdrops qubit @xmath10 , alice will find some error in the check procedure although bob could cheat again in this check procedure .
then the security for the present quantum secret sharing is guaranteed .
the generalization of this quantum secret sharing scheme to multi - parties is straightforward .. for each bit of @xmath21 and @xmath22 , alice can creates an @xmath41-qubit product state @xmath42 in the basis @xmath36 ( if the corresponding bit of @xmath21 is 0 ) or the basis @xmath26 ( if the bit of @xmath21 is 1 ) where the xor result of the @xmath41 qubits @xmath43 equals to the bit value of the string @xmath22 .
the two - qubit vectors of the basis @xmath24 and @xmath26 are @xmath44 and @xmath45 . obviously , if there are odd numbers @xmath46 in @xmath47 ( @xmath48 in @xmath49 ) , the bit value of @xmath22 is 1 .
the case of even numbers @xmath46 stands for 0 .
the following steps are the same as the above two - party sharing case . as no entanglement , especially inaccessible multi - parties entangled states , are required , it may be more experimentally realizable than the original one@xcite .
finally , let us discuss the resources necessary to implement this quantum secret sharing protocol .
in the most obvious way of sharing secret , alice first must establish mutual keys among different pairs of parties , and then use quantum cryptographic protocols to send each of the bit strings which result from the classical procedure . while in our secret sharing protocol of direct encoding the qubits of the two bb84 quantum key distribution , once the key has been established , alice needs to send only one string of classical bits to either bob or charlie .
then in order to share a classical bit(cbit ) information between two parties , bob and charlie , quantum cryptography as bb84 protocol ideally needs 2 qubits , 2 cbit and no ebit on average .
the quantum secret sharing with epr states needs 4 qubits , 1cbit and 2ebits . while our protocol with product states needs only 2 qubits , 1cbit and no ebit . in general , the more parts into which the secret is split
, the greater the difference between the number of classical bits which be sent in those two ways .
we see the direct encoding to the source qubits is able to act as a substitute for transmitted random bits .
furthermore , the efficiency of the existing entangled - state scheme can only reach 50% in principle while the present secret sharing protocol can be 100% efficient .
in this paper , we analyze the main quantum secret sharing protocol based on the ghz - state and propose a quantum secret sharing scheme where only product states are employed . as entanglement , especially the inaccessible multi - party entangled state , is not necessary in the present quantum secret sharing protocol , it may be more applicable when the number of the parties of secret sharing is large .
its theoretic efficiency is also doubled to approach 100% .
although those quantum secret sharing with entanglement can be a demonstrator of concepts of various entanglement applications , the inherent efficiency limiting of these protocols and the difference for the preparation of multi - party entanglement may suggest that the employing of the quantum entanglement in quantum secret sharing be unworthiness in some cases .
after all the practicality is an important pursuit of the quantum information theory .
this work was funded by national fundamental research program(2001cb309300 ) , national natural science foundation of china , the innovation funds from chinese academy of sciences , and also by the outstanding ph .
d thesis award and the cas s talented scientist award entitled to luming duan .
|
after analysing the main quantum secret sharing protocol based on the entanglement states , we propose an idea to directly encode the qubit of quantum key distributions , and then present a quantum secret sharing scheme where only product states are employed . as entanglement , especially the inaccessable multi - entangled state ,
is not necessary in the present quantum secret sharing protocol , it may be more applicable when the number of the parties of secret sharing is large .
its theoretic efficiency is also doubled to approach 100% .
pacs number(s ) : 03.67.hk , 89.70.+c
|
rs ophiuchi ( rs oph ) is a well - observed recurrent nova ( recurrence period @xmath9 20 years ) and is one of the ten confirmed recurrent novae that belong to our galaxy ( kato @xmath10 hachisu 2012 , schaefer 2010 ) .
the rs oph system is composed of a massive ( @xmath9 1.35 m@xmath8 , kato , hachisu @xmath10 luna 2008 ) white dwarf ( wd ) primary accompanied by a red giant secondary of estimated spectral class around m2 iii ( worters et al .
2007 and references therein ) .
brandi et al .
( 2009 ) estimated the orbital period to be 453.6 days , the red giant mass , m@xmath11 = 0.68 - 0.80 m@xmath8 and the orbital inclination , i = 49@xmath12 - 52@xmath12 for the system .
the outburst takes place due to a thermonuclear runaway ( tnr ) on the wd surface that accretes matter from the secondary red giant companion .
the outburst causes ejection of mass @xmath9 10@xmath7 - 10@xmath13 m@xmath8 at a high speed of @xmath9 4000 km s@xmath5 ( e.g. buil 2006 ) .
previous studies of outbursts indicate that the wd mass of rs oph is possibly increasing due to the accumulation of a percentage of the accreted matter on its surface .
consequently , the mass of the wd in rs oph may gradually reach the chandrasekhar limit and explode as a type ia supernova - this has made rs oph an object of great significance to the astrophysicists .
however , there have been considerable debates about this hypothesis ( starrfield et al .
2004 , wood - vasey @xmath10 sokoloski 2006 ) .
+ rs oph was detected in outburst previously in 1898 , 1933 , 1958 , 1967 , 1985 ; its latest outburst was discovered on 2006 february 12.83 ut ( hirosawa 2006 ) .
the reason for the much shorter recurrence period in the rs oph system , in comparison to classical novae ( cne ) , is due to combined effect of the high wd mass and a high accretion rate ( starrfield et al .
1985 , yaron et al . 2005 ) .
the two recent outbursts of rs oph in 1985 and 2006 have been observed intensively over a wide range of wavelengths , from x - rays to the radio regions ( bode 1987 ; evans , bode , obrien @xmath10 darnley 2008 ) .
detailed studies display very similar characteristics of the outbursts . in the early phase of the outburst , the spectra show broad , low - ionization emission features of h
, he , n , o and fe ; the nova enters quickly ( about a month after outburst ) to the nebular phase with the emergence of strong coronal ( e. g. , [ fe xiv ] 0.5303 @xmath14 m , [ ar x ] 0.5535 @xmath14 m , [ fe x ] 0.6374 @xmath14 m , [ si vi ] 1.9641 @xmath14 m , [ al ix ] 2.0444 @xmath14 m , [ mn xiv ] 2.0894 @xmath14 m ) and nebular lines ( e.g. , [ o iii ] ( 0.4363 , 0.4959 and 0.5007 @xmath14 m ) and [ n ii ] 0.5755 @xmath14 m ) ( iijima 2009 , banerjee et al . 2009 ) .
the nova light curves also behave similarly ; they decline fast with @xmath15 @xmath9 6 and @xmath16 @xmath9 17 days ( rosino 1987 , munari et al .
the key result of the 1985 and 2006 observations was the detection of a shock that is generated while the ejecta interacts with the surrounding wind of the red giant secondary ( bode @xmath10 kahn 1985 , das et al .
2006 and references therein ) and a non - spherical bi - polar shape of the nova ejecta ( e.g. , taylor et al .
1989 , chesneau et al .
2007 , bode et al . 2007 ) .
further investigations also helped to determine a few important parameters viz .
determinations of the distance , d = 1.6 @xmath6 0.3 kpc ( hjellming et al .
1986 ) , the interstellar hydrogen column density , n @xmath9 2.4 @xmath2 10@xmath17 @xmath18 ( hjellming et al .
1986 ) , and an interstellar reddening of e(b - v ) = 0.73 ( snijders 1987 ) .
+ however , despite plenty of observations of rs oph , the abundance analysis of the nova ejecta has not been done adequately .
a few values have been calculated , for example , from optical studies of the 1985 outburst , anupama @xmath10 prabhu ( 1985 ) derived a helium abundance of @xmath19 = 0.16 ; evans et al .
( 2007 ) estimated the o / ne ratio ( by number ) to be @xmath20 0.6 from ir studies of the 2006 outburst .
a complete knowledge of elemental abundances in the ejecta is of crucial importance for several reasons , for example , to understand the tnr process that leads to the nova explosion , the composition of material of the wd * * , * * as there is a possibility of mixing of wd material with the ejecta , the contribution of novae to the chemical evolution of galaxy etc . in this paper , we report the results of an elemental abundance analysis of the ejecta of rs oph by modeling its available optical and near - infrared ( nir ) spectra observed during the 2006 eruption .
we have used the photoionization code cloudy ( version 13.02 ; ferland et al . , 2013 ) to generate spectra , by varying the parameter values .
model generated spectra are then compared with the observed emission line spectra , the best fit model is chosen by calculating the corresponding @xmath21 values .
the procedure of modeling is described in section 3 ; results obtained from the analysis is described in section 4 .
we use the cloudy photoionization code , c13.02 ( ferland et al . , 2013 ) for the abundance analysis in rs oph .
the benefit of using photoioinization models is that in addition to elemental abundances , they also provide estimate of several other parameters , e.g. , density , source luminosity , source temperature etc .
previously , this method was used to determine the elemental analysis and physical characteristics of a few novae by modeling the observed spectra , for example , lmc 1991 ( schwarz 2001 ) , qu vul ( schwarz et al .
2002 ) , v1974 cyg ( vanlandingham et al .
2005 ) , v838 her @xmath10 v4160 sgr ( schwarz et al .
2007a ) , v1186 sco ( schwraz et al .
2007b ) , v1065 cen
( helton et al . 2010 ) .
the photoionization code cloudy uses a set of parameters that specify the initial physical conditions of the source and the ejected shell .
the source is described by the spectral energy distribution of the continuum source , its temperature and luminosity .
the physical condition of the shell is described by the density , inner and outer radii , geometry , covering factor ( fraction of 4@xmath22 sr enclosed by the model shell ) , filling factor ( ratio of the contribution of the dense shell to the diffuse shell ) and elemental abundances ( relative to solar ) .
the density of the shell is set by a hydrogen density parameter and the elemental abundances , relative to hydrogen , are set by the abundance parameters .
the hydrogen density , n(r ) , and filling factor , f(r ) , may vary with the radius as given by the following relations , @xmath23 where , r@xmath24 is the inner radius , @xmath25 and @xmath26 are exponents of power laws .
we choose @xmath25 = -3 , the filling factor = 0.1 and the filling factor power - law exponent , ( @xmath26 ) = 0 , which are the typical values used in similar kind of studies ( e.g. , schwarz 2002 , vanlandingham et al .
2005 , helton et al .
+ cloudy solves the equations of thermal and statistical equilibrium using the above mentioned set of input parameters to generate output spectra from the non - lte ejecta illuminated by the central source .
its calculations incorporate effects of important ionization processes , e.g. , photo , auger , collisional @xmath10 charge transfer and recombination process viz .
radiative , dielectronic , three - body recombination , and charge transfer .
we assume the continuum shape to be a blackbody of a high temperature @xmath27 @xmath28 10@xmath3 k , as done in the previous investigations , to ensure that it supplies the correct amount of photons for photoionization .
the output predicts the flux of emission lines , which is compared to the measured line fluxes in the observed spectra .
+ [ cols="<,^,^,^,^,^,^,^ " , ] relative to h @xmath26 in optical , pa @xmath26 in @xmath29 band , br 12 in @xmath30 band and br @xmath31 in @xmath32 band .
the results of our analysis are presented in table 2 @xmath10 3 .
table 2 shows the relative fluxes of the best - fit model predicted and observed lines along with corresponding @xmath0 values . here ,
we have considered only the lines which are present both in the model**-**generated and observed spectra .
we assume a distance of 1.6 kpc for rs oph to match the predicted luminosities with the reddening - corrected observed flux .
the measured line fluxes were dereddened using e(b - v ) = 0.73 ( snijders 1987 ) and compared to the output of each cloudy model to calculate @xmath21 for the fit .
the line intensities are determined from direct integration of the line ; for features with multiple components , the profiles were decomposed with multiple gaussians using iraf tasks . to minimize the errors associated with flux calibration between different epochs and wavelength regions , we calculate the modeled and observed flux ratios relative to prominent hydrogen lines , which are not blended with any other features , within a given wavelength region , e. g. , relative to h@xmath26 in the optical region , relative to paschen @xmath26 in the @xmath29-band , relative to brackett 12 in the @xmath30-band and relative to brackett @xmath31 in the @xmath32-band . + l c c c c c c c parameters & d31 & d49 & predicted abundances + + @xmath27 ( @xmath210@xmath3 k ) & 5.8 & 5.5 & ... + source luminosity ( @xmath210@xmath4 erg s@xmath5 ) & 6.3 & 8.0 & ... + clump hydrogen density ( @xmath210@xmath33 @xmath34 & 10.0 & 6.3 & ... + diffuse hydrogen density ( @xmath210@xmath33 @xmath34 & 1.6 & 1.0 & ...
+ @xmath25 & -3 & -3 & ... + inner radius ( @xmath210@xmath35 cm ) & 2.1 & 2.8 & ... + outer radius ( @xmath210@xmath35 cm ) & 4.8 & 6.8 & ... + clump to diffuse covering factor & 90/10 & 85/15 & ... + filling factor & 0.1 & 0.1 & ...
+ @xmath26 & 0.0 & 0.0 & ...
+ he / he@xmath36 & 1.8 ( 11 ) & 1.9 ( 16 ) & 1.8 @xmath6 0.1 + n / n@xmath36 & ... & 12.0 ( 2 ) & 12.0 @xmath6 1.0 + o / o@xmath36 & 1.0 ( 1 ) & 1.0 ( 1 ) & 1.0 @xmath6 0.4 + ne / ne@xmath36 & ... & 1.5 ( 1 ) & 1.5 @xmath6 0.1 + si / si@xmath36 & 0.3 ( 1 ) & 0.5 ( 1 ) & 0.4 @xmath6 0.1 + fe / fe@xmath36 & 3.0 ( 12 ) & 3.5 ( 11 ) & 3.2 @xmath6 0.2 + ar / ar@xmath36 & 4.9 ( 1 ) & 5.2 ( 2 ) & 5.1 @xmath6 0.1 + al / al@xmath36 & 0.9 ( 1 ) & 1.1 ( 1 ) & 1.0 @xmath6 0.1 + ejected mass ( @xmath210@xmath7 m@xmath36 ) & 3.4 & 4.9 & ... + number of observed lines ( n ) & 42 & 51 & ... + number of free parameters ( n@xmath37 ) & 11 & 13 & ... + degrees of freedom ( @xmath38 ) & 31 & 38 & ... + total @xmath0 & 38.0 & 69.1 & ... + @xmath39 & 1.2 & 1.8 & ... + + radial dependence of the density @xmath40 radial dependence of filling factor @xmath41 abundances are given in logarithmic scale , relative to hydrogen .
all other elements which are not listed in the table were set to their solar values .
the number in the parentheses represents number of lines used in determining each abundance .
the best fit modeled spectra ( dotted line ) together with the observed optical and nir spectra ( solid lines ) on d31 and d49 , are shown in figure 1 .
we have marked the strong spectral lines on the figure , a detailed observed line list is presented in banerjee et al .
( 2009 ) and ( iijima 2009 ) .
the spectra on d31 are dominated by prominent features of low ionization lines of h@xmath42 ( 0.3970 @xmath14 m ) , h@xmath43 ( 0.4103 @xmath14 m ) , h@xmath31 ( 0.4341 @xmath14 m ) , he ii ( 0.4666 @xmath14 m ) , h @xmath26 ( 0.4863 @xmath14 m ) in the optical region ; he i ( 1.0830 @xmath14 m ) , pa @xmath31 ( 1.0938 @xmath14 m ) , o i ( 1.1287 @xmath14 m ) , pa @xmath26 ( 1.2818 @xmath14 m ) , brackett series lines , fe ii ( 1.6872 @xmath14 m ) , he i ( 1.7002 @xmath14 m ) in the @xmath30-band ; and he i ( 2.0581 @xmath14 m ) , br @xmath31 ( 2.1655 @xmath14 m ) in the @xmath32-band .
there are also higher ionization lines , viz .
[ fe vii ] ( 0.3760 @xmath14 m ) , a blended feature of [ o iii ] , he i @xmath10 fe ii ( 0.5017 @xmath14 m ) , * a * blended feature of [ fe ii ] @xmath10 [ fe vii ] ( 0.5158 @xmath14 m ) , [ ar x ] ( 0.5535 @xmath14 m ) , [ fe x ] ( 0.6374 @xmath14 m ) in the optical ; and [ si vi ] ( 1.9641 @xmath14 m ) , [ al ix ] ( 2.0444 @xmath14 m ) , [ mn xiv ] ( 2.0894 @xmath14 m ) in the nir region . as mentioned in the earlier section , a single shell of ejecta could not generate all of these lines . for example , if we consider a clumpy shell of high density , e.g. 10@xmath44 @xmath45 only , the modeled spectra fit the majority of the lines in the observed spectra , but systematically under represents the higher ionization lines , for example ,
[ al ix ] ( 2.0444 @xmath14 m ) , fe ii ( 1.7406@xmath14 m ) , [ fe x ] ( 0.6374 @xmath14 m ) and [ fe vii ] ( 0.3760 @xmath14 m ) .
this one component shell also increases the strength of the k - band he i line enormously . on the other hand
, only a diffuse shell of lower density can produce the higher - ionization lines successfully but can not produce the he i and fe ii lines sufficiently to match the observed lines .
hence , for a better fitting of the lines , it was required to consider a two - component model consisting of a dense shell ( hydrogen densities , by number : 10 x 10@xmath33 @xmath45 and 6.3 x 10@xmath33 @xmath45 on d31 and d49 , respectively ) and a diffuse shell ( 1.6 x 10@xmath33 @xmath45 and @xmath10 1.0 x 10@xmath33 @xmath45 on d31 and d49 , respectively ) .
the ratio of the clump to diffuse components are 90:10 and 85:15 , respectively , on d31 and d49 .
this indicates that the ejecta volume was dominated more by the dense gas .
the best - fit models use a blackbody temperature ( t@xmath1 ) of 5.8 @xmath2 10@xmath3 k and 5.5 @xmath2 10@xmath3 k and a luminosity of 6.3 @xmath2 10@xmath46 ergs s@xmath5 and 8.0 @xmath2 10@xmath46 ergs s@xmath5 , respectively , on d31 and d49 .
this is in agreement with the wd temperature of around 8 @xmath2 10@xmath3 k derived from x - ray studies by nelson et al .
the low @xmath39 values of 1.2 and 1.8 for the two dates , respectively , indicate that the fits are satisfactory .
the best - fit model parameters for each epoch are described in table 3 .
+ despite the low values of @xmath47 , the model still has some problems .
an inspection of table 2 shows that a few lines make the highest contributions to the total @xmath0 , for example , he ii ( 0.4666 @xmath14 m ) , the blended feature of [ o iii ] , he i @xmath10 fe ii ( 0.5017 @xmath14 m ) on d31 and h @xmath42 @xmath10 [ n iii ] ( 0.3970 @xmath14 m ) , h @xmath43 , [ fe x ] ( 0.6374 @xmath14 m ) , he i ( 0.7065 , 1.0830 and 2.0581 @xmath14 m ) , pa @xmath31 , br 16 and 17 on d49 . the possible reason for these misfits is that a two - component model is not sufficient to adequately reproduce the complex density structure of the nova ejecta ; consequently , the exact conditions inside the ejecta in which these lines are formed , have not been not reproduced correctly .
in addition , cloudy was unable to adequately reproduce a few of the observed lines , e.g. , si ii ( 0.5041 , 0.5056 @xmath14 m ) , fe ii ( 0.5235 , 0.5276 , 0.5317 @xmath14 m ) , [ fe xiv ] ( 0.5303 @xmath14 m ) , n iii ( 0.4641 @xmath14 m ) , raman emission band ( 0.6830 @xmath14 m ) , he i ( 1.1969 , 1.2527 @xmath14 m ) , n i ( 1.2074 , 1.2096 , 1.2470 @xmath14 m ) , o i ( 1.1287 , 1.3164 @xmath14 m ) , [ mn xiv ] ( 2.0894 @xmath14 m ) and he i ( 2.1120 @xmath14 m ) .
so , we have excluded these lines from our analysis .
the problem of reproducing the nir o i emission lines ( 1.1287 @xmath14 m and 1.3164 @xmath14 m ) and he i ( 2.0581 @xmath14 m ) has also been reported by helton et al .
they attributed these difficulties to related excitation mechanisms which are not properly included in the present cloudy code .
+ overall abundance values for rs oph , calculated from the mean of these modeled results , are presented in table 3 .
abundances are given as logarithm of the numbers relative to hydrogen and relative to solar .
the derived values show that the helium , nitrogen , neon , iron and argon abundances are all enhanced relative to solar , while the oxygen and aluminium abundances are solar with respect to hydrogen , and silicon abundances are subsolar with respect to hydrogen .
we note that there is no prominent line of oxygen in the observed spectra .
our determination of the oxygen abundance is based on the feature at 0.5017 @xmath14 m that is blended with he i and fe ii .
therefore the calculated abundance value of oxygen may incorporate a significant error .
similarly , the calculation of abundances of silicon , argon and aluminium is based on one or two observed lines only .
modeling of multi - wavelength spectra , observed over a longer time scale may improve the level of accuracy . using the parameter values determined from the present analysis
, we can estimate the hydrogen ejected mass predicted by the two component models using the following relation ( e.g. , schwarz 2001 , 2002 ) , @xmath48 the resulting mass is then multiplied by corresponding covering factors to obtain mass of the dense and diffuse shells ; the final mass is calculated by adding the masses of these two shells . using this method , we find an ejected mass of 3.4 @xmath2 10@xmath7 m@xmath8 and 4.9 @xmath2 10@xmath7 m@xmath8 for d31 and d49 , respectively .
similar values of 1 - 3 @xmath2 10@xmath7 m@xmath8 and 3 @xmath2 10@xmath7 m@xmath8 were estimated from other studies ( e.g. , das et al .
2006 ; kato , hachisu @xmath10 luna 2008 ) of the 2006 outburst of rs oph . from the light curve analysis ,
hachisu @xmath10 kato ( 2001 ) estimated an ejecta mass of @xmath9 2 @xmath2 10@xmath7 m@xmath8 for a wd of mass of 1.35 m@xmath8 .
therefore the present result also favors the possibility of a high mass of @xmath9 1.35 m@xmath8 in the rs oph system .
we have used the photoionization code cloudy to model the observed optical and nir emission - line spectra of rs oph observed on two epochs , 31 and 49 days after the outburst , during 2006 .
we generated a set of spectra by varying several parameters and assuming a spherical geometry of the ejecta that consists of two different shells of different densities .
the model**-**generated spectra were then compared with observed spectra , the best fit parameters were chosen using the @xmath0 technique .
the best - fit model parameters are in agreement with a hot wd with temperature of @xmath9 5.5 - 5.8 @xmath2 10@xmath3 k and luminosity of 6 - 8 @xmath2 10@xmath4 ergs s@xmath5 .
the abundance analysis shows that the ejecta are significantly enhanced , relative to solar , in helium , nitrogen , neon , iron and argon as well as silicon was found to be subsolar .
we estimate an ejected mass in the range of 3.4 - 4.9 @xmath2 10@xmath7 m@xmath8 which fits a high mass @xmath9 1.35 @xmath49 of the associated wd in the rs oph system . in continuation of the present work ,
a complete and detailed analysis of abundances and related parameters , covering a larger wavelength region and evolution period , will be the subject of a future work .
the research work at s n bose national centre for basic sciences is funded by the department of science and technology , government of india .
kato m. , hachisu i. & luna g. j. m. , 2008 , in rs ophiuchi ( 2006 ) and the recurrent nova phenomenon , astronomical society of the pacific conference series , eds .
evans a. , bode m. f. , obrien t. j. @xmath10 darnley m. j. , vol .
401 , 308 skopal a. , pribulla t. , buil c. et al . , 2008 , in rs
ophiuchi ( 2006 ) and the recurrent nova phenomenon , astronomical society of the pacific conference series , eds .
evans a. , bode m. f. , obrien t. j. @xmath10 darnley m. j. , vol .
|
we present an analysis of elemental abundances of ejecta of the recurrent nova rs oph using published optical and near - infrared spectra during the 2006 outburst .
we use the cloudy photoionization code to generate synthetic spectra by varying several parameters , the model generated spectra are then matched with the observed emission line spectra obtained at two epochs .
we obtain the best fit model parameters through the @xmath0 minimization technique .
our model results fit well with observed optical and near - infrared spectra .
the best - fit model parameters are compatible with a hot white dwarf source with t@xmath1 of 5.5 - 5.8 @xmath2 10@xmath3 k and roughly constant a luminosity of 6 - 8 @xmath2 10@xmath4 ergs s@xmath5 . from the analysis we find the following abundances ( by number ) of elements with respect to solar : he / h = 1.8 @xmath6 0.1 , n / h = 12.0 @xmath6 1.0 , o / h = 1.0 @xmath6 0.4 , ne / h = 1.5 @xmath6 0.1 , si / h = 0.4 @xmath6 0.1 , fe / h = 3.2 @xmath6 0.2 , ar / h = 5.1 @xmath6 0.1 , and al / h = 1.0 @xmath6 0.1 , all other elements were set at the solar abundance .
this shows the ejecta are significantly enhanced , relative to solar , in helium , nitrogen , neon , iron and argon . using the obtained parameter values , we estimate an ejected mass in the range of 3.4 - 4.9 @xmath2 10@xmath7 m@xmath8 which is consistent with observational results .
novae ; abundances .
|
with the rapid progress in the preparation and measurement techniques for isolated cold gases @xcite investigating the fundamental questions of if and how a closed quantum many - body system prepared in a nonequilibrium initial state approaches a stationary one is within experimental reach . studying the physics of the steady state itself
is of particular interest if it is nonthermal,@xcite that is expectation values of observables differ from those computed using a canonical statistical operator with the temperature fixed by the excess energy .
one - dimensional ( 1d ) interacting fermi systems are promising candidates for realizing such unusual nonequilibrium states as in many of those the dynamics is restricted not only by energy conservation but by additional conservation laws.@xcite an often studied protocol , which we also employ , is an abrupt quench of the amplitude @xmath0 of the two - particle interaction : the system is prepared in the canonical ensemble of an initial hamiltonian @xmath1 , while the time evolution is performed with @xmath2 , @xmath3 . after taking the thermodynamic limit local observables might become stationary at long times @xmath4.@xcite in equilibrium the exactly solvable tomonaga - luttinger ( tl ) model @xcite is the infrared fixed point model under a renormalization group ( rg ) flow of a large class of interacting 1d models in their ungapped metallic phase.@xcite the low - energy physics of models out of this luttinger liquid ( ll ) universality class @xcite is given by that of the tl model . for spinless models
the fixed point is characterized by the velocity @xmath5 of the elementary excitations and the ll parameter @xmath6 entering exponents of algebraically decaying correlation functions .
both depend on the parameters of a given model , in particular on @xmath0 .
one of the hallmarks of lls is their sensitivity towards inhomogeneities . for repulsive interactions with @xmath7
the ground state density response function of a ll diverges as @xcite @xmath8 , with the fermi momentum @xmath9 , indicating that even a single weak impurity acts as a strong perturbation .
the homogeneous perfect chain fixed point is unstable.@xcite the system flows towards an open chain one with strong consequences for the equilibrium low - energy properties ; e.g. the linear conductance vanishes as @xmath10 for temperature @xmath11 . we provide evidence that the steady state of a microscopic 1d lattice model after an interaction quench is characterized by the same power laws as found in the steady state of the tl model after a similar quench @xcite with the @xmath6 taken for the considered model parameters .
as the rg arguments leading to this type of ll universality in equilibrium @xcite can not directly be transferred to nonequilibrium @xcite this finding is far from obvious .
it complements earlier indications of ll universality in the time evolution towards the steady state.@xcite we compute the time evolution of the density as a function of the distance from an open boundary as well as that of the conductance across a single impurity as a function of @xmath12 and take @xmath4 .
the dynamics of the tl model is solved exactly using standard bosonization.@xcite to study the time evolution of the lattice model we use an approximate functional rg @xcite based approach which so far was only applied to open quantum systems.@xcite for small two - particle interactions this technique allows controlled access to time scales large enough such that the physics is dominated by the steady state .
it complements calculations using the density - matrix renormalization group ( dmrg ) which provide ` exact ' results at small @xmath13 but abruptly become unreliable beyond a characteristic time scale;@xcite the latter might be smaller than the one on which the steady state is reached.@xcite we show that the fixed point structure of a single impurity in a nonequilibrium steady - state ll is similar to the one in equilibrium.@xcite the rest of the paper is structured as follows . in the following section
we discuss the exact solution of the quench dynamics of the tl model via bosonization . in sect .
[ sect : mm ] we introduce the lattice model considered and show how its relaxation dynamics can be treated approximately within the functional rg .
we then compare the prediction of the tl model for observables and correlation functions of the steady state with results obtained for the lattice model . finally , in sect .
[ sect : oq ] we briefly discuss the relation of our results to those obtained by field - theoretical methods and hint towards open questions .
the starting point of our investigation of ll universality is the exact computation of the desired observables and correlation functions within the ( spinless ) tl model . starting out from the 1d electron gas the tl model
is obtained by linearizing the single - particle dispersion and keeping only the marginal two - particle scattering terms .
in contrast to earlier studies on interaction quenches in the tl model @xcite we consider the one with open boundaries at @xmath14 and @xmath15.@xcite this allows us to distinguish between bulk and boundary ll exponents.@xcite while in equilibrium the former are quadratic in the two - particle interaction the latter are linear .
the model is given by @xmath16 , \label{htl}\end{aligned}\ ] ] with the fermi velocity @xmath17 , @xmath18 , the two - particle potential @xmath19 , and bosonic operators @xmath20 associated with the density of the fermions . employing a bogoliubov transformation @xmath21 can be written as a diagonal quadratic form in eigenmodes with ( bosonic ) ladder operators @xmath22 and energy @xmath23 . to keep the formulas transparent we here take the noninteracting canonical statistical operator @xmath24/\operatorname{tr}\ { \exp[-\beta h_{\rm tl}(u=0)]\}$ ] as the initial state with @xmath25 ; in appendix [ ap : bos ] we describe the changes when starting in the canonical state with @xmath26 . using the bogoliubov transformation and standard bosonization of the field operator @xcite it is straight forward to derive closed analytical expressions for
the density @xmath27 , with @xmath28 being the distance from the boundary , the density of states ( dos ) @xmath29 , and the density response @xmath30 ( see appendix [ ap : bos ] ) .
the latter two functions can be used to compute the conductance in the limits of small and large impurities ( see below ) .
the expectation values are obtained by taking @xmath31 $ ] , where @xmath32 stands either for an observable or the operator product defining a correlation function .
after performing the thermodynamic limit the steady state values follow by taking @xmath4 ; all the above quantities converge and their steady - state limits are indicated by dropping the index @xmath13 .
we verified that the same @xmath4 expectation values can be computed using the statistical operator of a generalized gibbs ensemble ( gge).@xcite at @xmath33 , with the initial state given by the noninteracting ground state , the steady state reached after the quench is ` critical ' , that is characterized by power - law scaling.@xcite as a consequence different observables show characteristic power - law behavior with exponents which in general are different to the exponents found in equilibrium.@xcite they can all be expressed in terms of the models ll parameter @xmath34^{-1/2}$ ] after the quench . the access density @xmath35 , where @xmath36 denotes the filling , for large distances from the boundary @xmath28 falls off as @xmath37 where superscripts @xmath38 and @xmath39 refer to the ( @xmath33 ) equilibrium or the steady state reached after the interaction quench , respectively .
both cases show damped friedel oscillations with frequency @xmath40 .
one finds that as the ( repulsive ) interaction strength is increased and thus @xmath6 becomes smaller ( starting at @xmath41 for the noninteracting case ) , the access density in the presence of a boundary in both cases falls off slower than for vanishing two - particle interaction , but with different exponents .
the difference between ground- and steady - state exponents in the tl model was emphasized before considering other observables and correlation functions.@xcite .equilibrium and steady - state scaling exponents .
[ exponents ] [ cols="^,^,^,^",options="header " , ] additionally , we consider the frequency resolved local dos @xmath42 at small @xmath43 and close to the boundary . in the ground state
it is suppressed as @xmath44 which changes to @xmath45 in the nonequilibrium steady state .
in contrast to the equilibrium case in the steady state reached after the interaction quench the dos takes a finite value @xmath46 at @xmath47.@xcite this incomplete suppression is reminiscent of the equilibrium dos at finite temperatures for which the zero frequency value scales as @xmath48.@xcite the exponents of the power law behavior with which the @xmath47 spectral weights are reached differ between the equilibrium and steady - state situation . both increase with increasing interaction .
finally , we compute the zero frequency bulk charge susceptibility @xmath49 for wave vectors close to the backscattering condition @xmath50 .
the divergence ( @xmath7 for repulsive interactions ) changes from the ground - state result @xmath51 to @xmath52 in the steady state .
the comparison of the equilibrium and steady state exponents is summarized in table [ exponents ] . in the next section
we directly compare the decay of the densities friedel oscillations off the boundary in the steady state of a microscopic lattice model with the tl model prediction .
the scaling behavior of the bulk static density response @xmath53 allows us to make predictions for how the steady state reacts to a single impurity . for repulsive interactions
@xmath54 and @xmath55 diverges for @xmath56 . as in equilibrium
even a weak single impurity strongly disturbs the homogenous system . when applying an infinitesimal bias voltage @xmath57 across the impurity the steady - state linear conductance @xmath58 ( with the stationary current @xmath59 ) is expected to scale as @xmath60 , with the constant homogenous chain conductance @xmath61 .
the power law holds as long as the right hand side stays small , that is for not too small @xmath12 . using the language of equilibrium
rg this indicates that the perfect chain fixed point is unstable .
in contrast , the steady - state analog of the open chain one is stable as follows from the scaling of the local dos .
fermis golden rule - like arguments lead to a tunneling conductance across a weak link connecting two semi - infinite chains which scales as @xmath62 in a temperature regime which at the lower end is cut off by the finite dos at @xmath47 [ see eq . ] .
these arguments do not rule out intermediate impurity fixed points . provided the concept of ll universality holds for the steady state we expect to find those weak and strong impurity scaling laws of @xmath63 for a lattice model with the @xmath6 of the model considered .
we consider the lattice model of spinless fermions with nearest - neighbor hopping @xmath64 as well as interaction @xmath0 and open boundaries terminating the @xmath65-site chain given by @xmath66 \label{hlatticemodel } \end{aligned}\ ] ] in standard second quantized notation . in equilibrium
the model is ( a ) bethe ansatz solvable and ( b ) shows universal ll physics with @xmath6 and @xmath5 exactly known.@xcite it is commonly believed that because of ( a ) the steady state after an interaction quench is described by a gge but the corresponding statistical operator was so far neither constructed nor was a proof of its existence given .
our analysis does not rely on any such assumption .
when discussing the impurity physics @xmath67 is supplemented by a hopping impurity @xmath68 of strength @xmath69 $ ] located in the middle of the chain . to compute the time evolution of the density @xmath70 as well as the conductance @xmath71 we use an approximate functional rg
@xcite based approach . here
, we employ the lowest order truncation scheme in the two - particle interaction . to this order
the self - energy acquires a rg flow , which is crucial to capture the impurity physics,@xcite while the two - particle vertex remains the bare one .
renormalization of the latter is a higher order effect .
the same truncation level was earlier shown to capture the equilibrium ll properties of inhomogeneous lattice models including the characteristic power - law scaling , with exponents agreeing to the exact ones to leading order in @xmath0.@xcite .
motivated by the functional rg s success in describing the equilibrium properties of inhomogeneous lattice models , we extend it to tackle the quench dynamics in such systems . as the functional rg can directly be applied to the microscopic model , i.e. without the need of mapping it to an effective low - energy field - theory , the information about the high energy modes is not lost and one can hope to find reliable results also for the relaxation at short times ( being influenced by the high energy characteristics of the underlying lattice model ) as well as the crossover behaviour .
we study the relaxation dynamics and the steady state of a closed many - body system described by a lattice model of spinless fermions with nearest neighbor hopping and interaction .
compared to the functional rg approach to time evolution for open quantum systems introduced in ref . , some minor amendments need to be made .
those are outlined next .
we can treat hamiltonians of the form @xmath72 written in standard second quantization . here
@xmath73 is the antisymmetrized two - particle interaction . for our lattice model @xmath74 and @xmath75 the indices @xmath76 stand for the quantum numbers , e.g. the @xmath65 wannier states in our lattice model .
we assume an initial density matrix @xmath77}\ ] ] which allows for the application of wick s theorem .
for our calculations we always choose the noninteracting canonical statistical operator @xmath78 $ ] .
we introduce a cutoff in the noninteracting keldysh @xcite green functions @xmath79 ( as motivated in ref . ) by considering @xmath80^\dagger , \\
\label{eq : gk } g^{{{\ensuremath{\text{k}}}},\lambda}(t , t ' ) & = -i g^{{{\ensuremath{\text{ret}}}},\lambda}(t,0 ) ( 1 - 2 \bar n ) g^{{{\ensuremath{\text{adv}}}},\lambda}(0,t'),\end{aligned}\ ] ] with @xmath81 being the @xmath82 matrix with entries @xmath83 and @xmath84 $ ] .
the self - energy is obtained by solving a set of coupled differential flow equations @xmath85 with the initial conditions at @xmath86 @xmath87 the right hand sides of the flow equations contain @xmath88 with the full cutoff dependent keldysh component of the green function @xmath89 the star differential operator @xmath90 acts only on the free green function @xmath91 , not on @xmath92 , in the dyson series expansion @xmath93 used to calculate @xmath94 .
an approximation to the self - energy of the cutoff - free problem is obtained at @xmath95 . how to efficiently evaluate eq .
is summarized in appendix [ ap : num_im ] .
an approximation for the occupancy of site @xmath96 can directly be obtained from the keldysh green function at the end of the rg flow as @xmath97 to calculate the current flowing from the left to the right half of the lattice in our microscopic model of interacting spinless fermions , we need to determine @xmath98 where @xmath99 is defined as @xmath100 , with @xmath101 being the occupancy operator of site @xmath96 and @xmath102 denotes the expectation value with respect to @xmath103.@xcite for simplicity we assume an even number of lattice sites @xmath65 .
furthermore , we use @xmath104(t)\right\rangle_{\rho^0}=(h - j)g^<_{n/2 n/2 + 1}(t , t)+{\rm c.c}.\label{eq : it}\ ] ] with @xmath105 being the equal - time lesser green function of the interacting system.@xcite the functional rg method used here provides an approximation for this given by @xmath106.\ ] ] therefore , plugging this lesser green function into eq . allows to compute an approximation to the current and from this the conductance by numerical differentiation . at half filling @xmath107 after a quench in the interaction amplitude from @xmath108 to @xmath109 for @xmath110 , @xmath33 , and different sites @xmath96 .
left inset : friedel oscillations induced by the boundary and the propagation of a main signal from the boundary to @xmath111 for two values of @xmath13 .
right inset : velocity of the main signal for different @xmath0 ( symbols ) .
the exact bethe ansatz @xmath5 ( line ) is in excellent agreement with our data . ] in fig .
[ fig1 ] we show the access density @xmath112 , with the filling @xmath36 , for fixed @xmath96 starting out of the noninteracting impurity free ground state ( @xmath33 ) .
we can reach times of the order of a few @xmath113 which has to be contrasted to the dmrg approach which becomes unreliable for times of the order of @xmath114.@xcite as shown in the main plot and the left inset a signal originating from the left boundary travels through the system .
a similar one is generated at the right one . for a spatial region
in which the left signal passed through and the right one did not enter yet the density becomes stationary .
the physics for @xmath115 and @xmath96 up to @xmath116 is thus dominated by the steady state and does barely suffer from finite size effects .
the two signals propagate with the ll velocity @xmath5,@xcite with our method providing an excellent approximation to the exact @xmath5 ( right inset of fig .
[ fig1 ] ) . at half
filling @xmath107 after a quench in the interaction amplitude from @xmath108 to @xmath117 for @xmath118 , @xmath33 , and different sites @xmath96 .
functional rg and the numerically exact dmrg are in excellent agreement for times reachable by the dmrg calculation . ] in fig .
[ fig1_dmrg ] we compare the functional rg data to numerically exact dmrg results for the time evolution after a quench in the interaction amplitude from @xmath108 to @xmath117 .
this value of @xmath0 is of the same order as taken for all further analysis . on the scale of the plot the two datasets are indistinguishable for all sites @xmath96 and all times reachable with dmrg.@xcite within our dmrg approach to prepare the noninteracting ground state an iterative single site algorithm in matrix product state formulation was employed .
the resulting wavefunction was than subjected to a real time evolution using a fourth order suzuki - trotter decomposition ( @xmath119 ) ensuring that the discarded weight stays below a certain @xmath81 ( different @xmath81 ranging from @xmath120 to @xmath121 were tested to yield coinciding results ) .
the technical details are described in length in ref . .
the excellent agreement of the results obtained by both methods does not only strengthen our confidence in the functional rg approach for the following steady state analysis , but also shows that indeed within functional rg we correctly incorporate also the high energy physics of the underlying lattice model , which is crucial for short times .
the numerically exact solution of the time evolution within dmrg can however not be pushed to times large enough to allow for the sensitive analysis of power - law scaling conducted in the following with our functional rg approach . for @xmath107 , @xmath110 , and @xmath33 determined by taking the log - derivative .
the predictions from the tl model with the exact lattice model @xmath6 ( dashed lines ) are consistent with our results .
the inset shows @xmath122 at @xmath123 and @xmath124 ( symbols ) .
the line is the tl model prediction . ] in analogy to the ground state density @xcite the stationary one @xmath125 shows friedel oscillations with frequency @xmath126 ( see the @xmath127 region of the lower part of the left inset of fig .
[ fig1 ] ) .
we next analyze their decay .
figure [ fig2 ] shows the log - derivative of @xmath128 , that is an effective exponent .
the dashed lines is the prediction from the tl model of table [ exponents ] with the exact lattice model @xmath6 .
our data are consistent with a power - law decay and the tl model exponent .
this finding is our first indication of ll universality of the steady state .
the differences between the exact exponent and our result is of order @xmath129 . on the right hand side of our rg flow equations
we do not fully capture terms @xmath130 and thus control exponents only to order @xmath131 .
the discussed behavior is not restricted to the case of half filling .
the inset of fig .
[ fig2 ] shows @xmath122 for @xmath123 on a log - log scale and the corresponding tl model prediction as the envelope . to compute the steady - state linear conductance of the lattice model we take the canonical density matrix @xmath132 ( with @xmath133 ) corresponding to @xmath134 as the initial state .
the time evolution is performed with @xmath135 supplemented by onsite energies @xmath136 ( @xmath137 ) for all @xmath138 ( @xmath139 ) .
the current @xmath140 across the impurity bond is computed .
following the same reasoning as for the density @xmath141 becomes stationary for @xmath13 of the order of @xmath113 .
we take @xmath57 to be the smallest energy scale of the problem ( typically @xmath142 ) to ensure that we are in the linear regime @xmath143 . for @xmath144
we find @xmath145,@xcite which is a band effect ( see the inset of fig . [ fig4 ] ) .
the universal scaling of the conductance discussed in the last section can only be expected for @xmath146 . , filling @xmath107 , and @xmath110 ( symbols ) .
the prediction from the tl model with the exact lattice model @xmath6 is shown as the solid lines . to emphasize the sizable differences in the exponents we have added a power law with the @xmath147 exponent as the dashed line to the @xmath109 data . ]
we first analyze the case of weak impurities . to eliminate the constant @xmath61 we take the derivative of @xmath71 with respect to @xmath12 . based on our above considerations we expect to find @xmath148
; see fig .
[ fig3 ] . over roughly one order of magnitude
the functional rg data follow the tl model prediction with the exact @xmath6 of our lattice model .
the deviations for @xmath149 indicate the crossover to the nonuniversal @xmath145 regime .
the ones for @xmath150 have two reasons .
as discussed above the scaling only holds as long as @xmath12 does not become too small .
furthermore , the energy level spacing @xmath151 ( @xmath152 for the parameters of the plot ) is an energy scale of the problem which cuts off any universal behavior.@xcite this is an artefact of our treatment of finite systems . for small @xmath153 and @xmath154 $ ] , @xmath155 .
our analysis thus requires very accurate data . to minimize the error due to small residual oscillations of @xmath140 present even for @xmath13 of the order of @xmath113 we averaged the data at large @xmath13 over a small time interval . of the lattice model for @xmath109 , @xmath107 , and @xmath110 .
different symbols stand for different @xmath153 increasing from left to right .
no fixed point in between the perfect and the steady - state analog of the open chain ones exist .
the crosses were computed for @xmath156 , @xmath157 giving the same @xmath6 .
the scaling function thus depends on @xmath0 and @xmath36 only via @xmath6.@xcite the prediction of the tl model for @xmath158 and @xmath159 with the exact lattice model @xmath6 is shown as dashed lines .
inset : the effective exponent of @xmath63 for a strong impurity @xmath160 determined by a log - derivative .
dashed lines show the prediction of the tl model with the exact lattice model @xmath6 . ] in the inset of fig .
[ fig4 ] we present our results for @xmath63 across a strong impurity . even without any @xmath13 averaging
our data are accurate enough such that the log - derivative , that is the effective exponent , gives a smooth curve .
the data clearly show the crossover from the nonuniversal @xmath161 behavior at large @xmath12 to the tl model prediction @xmath162 at low ones . for @xmath11
the scaling is cut off by both the finite size scale @xmath163 as well as the finite dos at @xmath47 [ see eq .
( [ eq : bla ] ) ] . in the limits of strong and weak impurities our results for the linear conductance of the lattice model
thus agree to the tl model prediction providing the second indication of ll universality of the steady state .
we finally show that in the steady state of the lattice model no fixed point in between the perfect and the steady - state analog of the open chain ones exist . to this end
we compute @xmath63 for a variety of @xmath153 at fixed @xmath0 and @xmath36 . using a one - parameter scaling ansatz of the form @xmath164,@xcite with @xmath165 and the nonuniversal scale @xmath166 ,
all data can be collapsed on a single curve continuously connecting the weak ( @xmath167 ) and strong ( @xmath168 ) impurity fixed points ; see the main plot of fig .
we provided evidence that the steady state of an interacting 1d lattice model after a quantum quench shows ll universality .
our analysis relies on the functional rg approach in its lowest - order truncation,@xcite which is sufficient to obtain ll power laws with @xmath0-dependent exponents .
an obvious first question arising is if higher order terms in @xmath131 might change this picture .
in fact , a series of rg studies of the field - theoretical tl model complemented by ` perturbations ' indicates , that power - law scaling is destroyed on long times by certain such terms.@xcite in nonequilibrium it is not established if and how the field theory studied in those papers is related to microscopic lattice models considered by us.@xcite currently , the results of refs . and our findings should thus be viewed as complementary and not contradicting .
we emphasize that the notion of ll universality involves lattice models and not only field theories.@xcite the numerically ` exact ' results of ref . for the time - evolution towards the steady state are consistent with our findings .
for the @xmath169 considered by us the corrections of order @xmath129 are small .
even if they would destroy the ll scaling on very large time scales , we expect that remnants of the predicted ll steady state can be found up to this scale .
the second apparent open question is if and how the picture changes if a lattice model is considered which is not bethe ansatz solvable ( ` nonintegrable ' ) .
we here merely note that for the time dependence indications of universal ll power laws were found even for such models.@xcite
we thank s. andergassen , e. dalla torre , c. karrasch , a. mitra , k. schnhammer , d. schuricht , and g. uhrig for discussions as well as s. blgel and the jlich supercomputing centre for access to the judge gpu cluster .
this work was supported by the dfg via for 723 .
to compute observables and correlation functions in the steady state of the tl model with open boundaries after an interaction quench we use ` open boundary bosonization ' for the hamiltonian and the field operator.@xcite we are interested in the scaling behavior with all energy scales send to zero and all length scales send to infinity .
thus subtleties resulting out of the momentum dependence of the two - particle potential @xmath19 become irrelevant @xcite and the ultraviolet regularization can be implemented at will . to illustrate the procedure we consider the density @xmath27 .
we first study the quench from @xmath170 to @xmath171 . for simplicity
we focus on temperature @xmath33 .
the density is given by the green function @xmath172 with the noninteracting ground state @xmath173 which corresponds to the vacuum with respect to the @xmath174 s [ see eq . ] .
the fields @xmath175 contain the open boundary conditions . using auxiliary fields
@xmath176 which are identical to the ones obtained for periodic boundary conditions @xcite and are e.g. given in eqs .
( 18)-(20 ) of ref .
, the green function reads @xmath177 .\end{aligned}\ ] ] those expectation values can be computed following the usual steps @xcite which involve the multiple use of the bogoliubov transformation @xmath178 and the baker - campbell - hausdorff relation .
the coefficients @xmath179 and @xmath180 depend on the two - particle potential @xmath181 and are e.g. given in eq .
( 9 ) of ref . .
the first two terms of the green function provide the homogenous background density while the latter two oscillate in space with frequency @xmath126they contain the friedel oscillations induced by the boundaries .
after taking the thermodynamic limit and the limit @xmath4 we find for the steady - state access density @xmath182 [ c(0 ) + s(0 ) ] } \ , \cos{(2 k_{\rm f } x ) } .
\label{friedel}\end{aligned}\ ] ] at zero momentum the coefficients of the bogoliubov transformation can be expressed in terms of the models ll parameter @xmath6 given in the main text as @xmath183 using those relations the exponent of eq .
( [ friedel ] ) can be written as @xmath184 ( see table i of the main text ) .
the other scaling exponents of the last column of table i of the main text can be obtained in a similar fashion . in any ( equilibrium ) ll the ll parameter is given by @xmath185 where @xmath186 is a dimensionless measure for the interaction strength ; e.g. @xmath187 for the tl model and @xmath188/[2 \pi j \sin(k_{\rm f})]$ ] for our lattice model at filling @xmath189 . using this expansion
it is evident that all scaling exponents discussed by us ( see table i of the main text ) have a leading order contribution in @xmath186 .
this is crucial as within our approximate treatment of the lattice model we control exponents only to leading order.@xcite we next briefly discuss the case when starting in the ground state of @xmath21 with @xmath190 and performing the time evolution with a different interaction @xmath191 ; all quantities depending on the interaction strength acquire indices i or f. we have to consider two bogoliubov transformations [ corresponding to the transformations from zero interaction to @xmath192 and two sets of eigenmode ladder operators . the initial state is the vacuum with respect to one of those while the hamiltonian with which the time evolution is performed is a diagonal quadratic form in the other one . repeatedly applying the bogoliubov transformations and the baker - campbell - hausdorff relation gives for the steady - state access density @xmath193 with the scaling exponent @xmath194 . \label{gammaexp}\end{aligned}\ ] ] for @xmath195 , that is if we start in the noninteracting ground state , it becomes equal to @xmath196 ( see table i of the main text ) .
using the expansion eq .
( [ expansion ] ) it is easy to see that @xmath197 and @xmath196 agree to leading order in the two - particle interaction , that is @xmath198 only contributes to order @xmath199 and @xmath200 or higher . to leading order in the interaction strength
the scaling exponent of the density is thus exclusively given by @xmath201 .
the same holds for the other exponents considered by us ( see table i of the main text ) . in our computations for the lattice model
we control exponents only to leading order which explains why in the main text we exclusively consider quenches out of the noninteracting ground state .
we can calculate @xmath202 very efficiently by using an iterative procedure .
first , we discretize time in steps such that during one small step @xmath203 the time dependent self - energy can be set constant . for our results shown in the main text we made sure that @xmath203 is always chosen small enough such that further reducing it does not lead to any changes visible on the scale of the respective plots .
we use @xmath204^\dagger,\end{aligned}\ ] ] to write @xmath205 as a product of green functions@xcite @xmath206 \delta t } e^{-i\lambda(t - t')},\end{aligned}\ ] ] where @xmath207 is the self - energy time averaged over the interval @xmath208 .
the interacting keldysh green function @xmath209 can then be found iteratively employing @xmath210 in every time step two matrix exponentials of @xmath211 matrices have to be performed and multiplied with the keldysh component of the green function of the previous one .
this renders the problem a natural candidate for graphics processing unit ( gpu ) supported algorithms .
we use such to compute the results shown in the main text .
the number @xmath212 of time steps needed to obtain sufficient accuracy ( and resolution ) as well as to reach times which are large enough such that the physics is dominated by the steady state enters the number of equations to be solve . for the hamiltonian considered in the main text
we solve sets of @xmath213 coupled differential equations . due to the nearest neighbor structure of the interaction @xmath214 components of the self energy flow for each of the @xmath212 time steps .
typical numbers considered are @xmath215 lattice sites and @xmath216 time steps .
|
we study the dynamics resulting out of an abrupt change of the two - particle interaction in two models of closed one - dimensional fermi systems : ( a ) the field theoretical tomonaga - luttinger model and ( b ) a microscopic lattice model . using a nonperturbative approach which is controlled for small two - particle interactions we are able to reach large times allowing us to access the properties of the steady state of the lattice model . comparing those to the exact solution of the full dynamics in the tomonaga - luttinger model we provide evidence for universal luttinger liquid behavior .
|
electrical detection of ferromagnetic resonance ( fmr ) in ferromagnets ( fm ) is a powerful new experimental tool which has transformed the research on spin and magnetization dynamics .
@xcite over the past few years , this technique has generated a great deal of interest in the communities of magnetism , spintronics , and microwave technologies .
it has been broadly applied for studying diverse material structures , ranging from ferromagnetic thin films such as py ( permalloy , ni@xmath0fe@xmath1),@xcite , cro@xmath2,@xcite fe@xmath3o@xmath4,@xcite single crystal fe,@xcite gamnas,@xcite and la@xmath5sr@xmath6mno@xmath3,@xcite bilayer devices such as py / pt,@xcite py / au,@xcite py / gaas,@xcite and y@xmath3fe@xmath7o@xmath8/pt,@xcite to a variety of magnetic tunneling junctions ( mtj ) based on magnetic multilayers.@xcite from a technical standpoint , its high sensitivity has made it possible to quantitatively determine spin boundary conditions@xcite and to directly measure non - linear magnetization damping@xcite , the quasiparticle mass for the domain wall@xcite , the phase diagram of the the spin - transfer driven dynamics@xcite and various kinds of parametric spin wave excitation @xcite . its capability to probe the interplay of spins , charges , and photons has been utilized for studying spin rectification@xcite , spin pumping@xcite , spin torque@xcite , and spin hall effects@xcite , which have led to the proposing and realization of novel dynamic spintronic devices such as the spin battery,@xcite spin diode,@xcite spin dynamo,@xcite and spin demodulator @xcite . very recently , its ability to detect coherent processes@xcite has enabled electrical probing of the spin - resonance phase and the relative phase of electromagnetic waves@xcite , which pave new ways for microwave sensing@xcite , non - destructive imaging,@xcite and dielectric spectroscopy@xcite .
such a coherent capability is especially exciting as it resembles the latest achievement in semiconductor spintronics , where a new platform for coherent optical control of spin / charge currents has been developed by using nonresonant quantum interferences .
@xcite from the physical standpoint , many different effects may generate a time - independent dc voltage in magnetic materials via the fmr .
reported mechanisms involve spin rectification@xcite , spin pumping@xcite , spin torque@xcite , spin diode@xcite , spin hall@xcite and inverse spin hall effects@xcite .
two major issues stand out here : ( 1 ) a unified picture clarifying the relations and distinctions between such diverse mechanisms has not been established , which leads to increasing controversy and confusion in interpreting and understanding experimental results .
a stunning example of this issue is found in the very recent studies of the spin hall effect via electrically detected fmr , where two similar experiments performed on similar devices were interpreted completely differently.@xcite ( 2 ) when more than one mechanism simultaneously plays a role in the fmr generated dc voltage , proper interpretation requires a quantitative analysis of the fmr line shape . in our opinion
, this has remained the `` achilles heel '' in recent studies of spin pumping and the spin hall effect which utilize electrically detected fmr .
the purpose of this article is to address these two critical issues with a brief review of the key physics of this subject , followed by systematically measured experimental data with detailed theoretical analysis .
this paper is split into three main sections .
first we provide a brief review of different mechanisms which may generate the photovoltage via the fmr .
then we use the dynamic susceptibility obtained from a solution of the landau - lifshitz - gilbert equation to derive analytical formulae for analyzing the line shape and the symmetry properties of the photovoltage generated through spin rectification .
finally we present experimental results measured from different samples , at different frequencies , and in different experimental configurations , showing that the fmr line shape is determined by the relative phase of microwaves which is sample and frequency dependent .
[ t ] under microwave excitation at angular frequency @xmath9 , the rf electric ( @xmath10 ) and magnetic ( @xmath11 ) fields inside a ferromagnetic material can be described as @xmath12 and @xmath13 , respectively .
note that in general , due to the inevitable losses of microwaves propagating inside the ferromagnetic material , there is a phase difference @xmath14 between the dynamic @xmath10 and @xmath11 fields .
such a relative phase is determined by the frequency - dependent wave impedance of the materials@xcite . as shown in fig .
[ fig1 ] , the rf @xmath10 field drives a rf current @xmath15 , while the rf @xmath11 field exerts a field torque on the magnetization and drives it to precess around its equilibrium direction [ fig .
[ fig1](e ) ] .
such a magnetization precession is described by the non - equilibrium magnetization @xmath16 . here
@xmath17 and @xmath18 are the high - frequency conductivity and polder tensor , respectively .
note that due to the resonance nature of the precession , @xmath19 lags @xmath11 by a spin resonance phase @xmath20 . however , despite the phase of @xmath14 and @xmath20 , the dynamic @xmath21 and @xmath19 keep the coherence of their respective driving fields , so that the product of any combination of their components may generate a time independent signal proportional to @xmath22 , where @xmath23 denotes the time average .
the amplitude of such a signal depends on the phase difference of @xmath21 and @xmath19 , which can be easily understood from the trigonometric relation : @xmath24/2 $ ] .
this is the spin rectification@xcite as we highlight in table i. for transport measurements on magnetic structures under microwave irradiation , various magnetoresistance effects such as anisotropic magnetoresistance ( amr ) , giant magnetoresistance ( gmr ) and tunneling magnetoresistance ( tmr ) make corrections to ohm s law via their corresponding magnetoresistance terms@xcite .
such non - linear terms typically lead to the product of @xmath21 and @xmath19 .
spin rectifications induced by such magnetoresistance effects are listed in table i by the terms labeled @xmath25 .
the general feature of @xmath25 is that its amplitude depends on both the relative phase @xmath14 and the spin resonance phase @xmath20 , which leads to a characteristic phase signature of the fmr line shape@xcite .
similar to the effect of the rf @xmath11 field torque , a spin torque induced by a spin polarized current may also drive magnetization precession .
for example , in a bilayer [ fig . [ fig1](b ) ] made of a ferromagnetic layer and a nonmagnetic layer with spin - orbit coupling@xcite , in addition to the rf current @xmath21 flowing in the ferromagnetic layer , the rf @xmath10 field also induces a rf charge current flowing in the nonmagnetic layer . via the spin hall effect in such a nonmagnetic layer with spin - orbit coupling
, the rf charge current can be converted into a spin current @xmath26 , which may flow into the ferromagnetic layer and then drive the magnetization precession via the spin torque .
such a spin torque induced non - equilibrium magnetization can be described by @xmath27 , where the spin - torque susceptibility tensor @xmath28 introduces a spin resonance phase @xmath29 that is different from @xmath20 in @xmath18 . following a similar consideration for the magnetoresistance
induced spin rectification , a photovoltage depending on the spin hall effect may be generated in the ferromagnetic layer .
this is the physical origin of the spin hall induced spin rectification effect,@xcite which is listed in table i by the term labeled @xmath30 . in mtj
[ fig1](c ) ] , the spin polarized current @xmath26 can be directly generated in the ferromagnetic layer where the magnetization is pinned along a different direction than that of the free layer .
it tunnels into the free layer and drives the magnetization precession via the spin torque [ fig . [ fig1](f ) ] .
the induced spin rectification signal has been measured in spin diodes@xcite , which is listed in table i by the term labeled @xmath31 . over the past few years
, systematic studies on spin rectifications induced by the field ( @xmath25 ) and spin torque ( @xmath30 , @xmath31 ) have been performed , respectively , at the university of manitoba@xcite and cornell university@xcite .
it has been found that due to the coherent nature of spin rectification , @xmath25 , @xmath30 and @xmath31 all depend on the phase difference between @xmath21 and @xmath19 .
however , only the field torque spin rectification ( @xmath25 ) can be controlled by the relative phase @xmath14 of the microwaves.@xcite in addition to such coherent spin rectification effects , it is known that at the interface between a ferromagnetic and a nonmagnetic layer , microwave excitation may generate a spin polarized current flowing across the interface via the spin pumping effect@xcite .
this effect has been observed in a few striking experiments by measuring either transmission electron spin resonance@xcite or enhanced magnetization damping@xcite .
it involves fmr , exchange coupling and non - equilibrium spin diffusion . in our opinion
the physical picture of spin pumping was best explained in the classical paper of silsbee @xmath32 [ ref .
] , which highlighted the key mechanism of dynamic exchange coupling between the precessing magnetization and the spin polarized current .
such a dynamic coupling significantly `` amplifies '' the effect of the rf @xmath11 field in generating non - equilibrium spins .
it was later proposed that the spin current generated via spin pumping may also induce a photovoltage , either across the interface in a spin battery@xcite , or within the nonmagnetic layer via the inverse spin hall effect @xcite .
recent experiments performed on magnetic bilayers@xcite have found that spin - pumping induced dc voltage ( the term @xmath33 in table i ) should be about two orders of magnitude smaller than spin hall induced spin rectification ( the term labeled @xmath30 ) . in contrast to phase sensitive coherent spin rectification effects , the proposed spin - pumping photovoltage is based on incoherent spin diffusion and fmr absorption .
hence , the anticipated fmr line shape is symmetric and phase - independent .
[ h ] [ table1 ] ccccccc ac driving & @xmath34 & @xmath35 & @xmath36 & @xmath37 & & + effect & ohm s law & spin hall & field torque & spin torque & spin rectification & spin pumping + + dc voltage & & & & & @xmath38 & @xmath39 + + + thin film & @xmath40 & & @xmath41 & & @xmath42 & + + + bilayer & @xmath40 & @xmath43 & @xmath41 + & @xmath44 & @xmath45 & @xmath46 + + + mtj & @xmath47 & & @xmath41 + & @xmath44 & @xmath48 & + + + + + from the above discussion , it is clear that the line shape analysis plays the essential role in distinguishing the microwave photovoltage generated by different mechanisms .
this issue has been partially addressed by a number of theoretical@xcite and experimental works@xcite studying nanostructured mtjs where the photovoltage is dominated by the spin torque induced spin rectification .
enlightened by these works and also based on our own previous studies @xcite , we discuss in the following the critical issue of fmr line shape analysis in microstructured devices , where the field and spin torque induced spin rectification may have comparable strength .
our theoretical consideration and experimental data demonstrate the pivotal role of the relative phase @xmath14 , which was often under - estimated in previous studies . via systematic studies with different device structures , measurement configurations and frequency ranges ,
we find that @xmath14 has to be calibrated at different microwave frequencies for each device independently . hence
, our results are in strong contradiction with the recent experiment performed on microstructured magnetic bilayers for quantifying the spin hall angles , where @xmath14 was declared to be zero for all devices at different microwave frequencies@xcite .
from table i , the role of the phase in the fmr line shape symmetry can be understood by considering the spin rectified voltage @xmath49 . for spin rectification induced by the field torque , depending on the experimental configuration ,
at least one matrix component @xmath50 of the polder tensor @xmath51 will drive the fmr ; whether an on or off - diagonal component is responsible for the magnetization precession depends on the measurement configuration . since @xmath52 , @xmath53 . therefore after time averaging a time independent dc voltage is found @xmath54 $ ] .
it is well known that for diagonal matrix elements , @xmath55 has a dispersive line shape while @xmath56 has a symmetric line shape .
however since the on and off - diagonal susceptibilities differ by a phase of @xmath57 , if the fmr is driven by an off - diagonal susceptibility , the roles are reversed and @xmath55 has a symmetric line shape while @xmath56 has a dispersive line shape .
based on the simple argument leading to the above @xmath58 expression , one can see that the line shape symmetry has a characteristic dependence on the relative phase @xmath14 between electric and magnetic fields .
thus when measuring fmr based on the field torque induced spin rectification effect , it is important to consider the relative phase , whereas for a spin pumping measurement which measures @xmath59 , or for a spin torque induced spin rectification which involves @xmath60 , the relative phase does not influence the experiment . in the next two sections , a detailed analysis is given by solving the landau - lifshitz - gilbert equation , which leads to analytical formulae describing the symmetric and dispersive line shapes for different measurement configurations .
the landau - lifshitz - gilbert equation provides a phenomenological description of ferromagnetic dynamics based on a torque provided by the internal magnetic field @xmath61 which acts on the magnetization * m * , causing it to precess@xcite @xmath62 here @xmath63 is the effective electron gyromagnetic ratio and @xmath64 is the gilbert damping parameter which can be used to determine the fmr line width @xmath65 , according to @xmath66 .
for the case of microwave induced ferromagnetic resonance eq .
( [ llg ] ) can be solved by splitting the internal field into dc and rf components and taking the applied dc field * h * , along the @xmath67-axis .
we can relate the internal field @xmath68 , to the applied field through the demagnetization factors @xmath69 , @xmath70 , @xmath71 , where @xmath72 is the relative phase shift between the electric and magnetic fields in the @xmath73 direction and @xmath74 is the dc magnetization also along the @xmath67-axis . with the magnetization separated into dc and rf contributions @xmath75 , the solution of eq .
( [ llg ] ) yields the dynamic susceptibility tensor @xmath51 which relates the magnetization * m * to the externally applied rf field * h * @xmath76 where @xmath77 $ ] is the spin resonance phase@xcite which describes the phase shift between the response and the driving force in terms of the line width @xmath65 and the resonance field @xmath78 which are constant for a fixed frequency .
@xmath20 will change from 180@xmath79 ( driving force out of phase ) to 0@xmath79 ( driving force in phase ) around the resonance position , in a range on the order of @xmath65 , passing through 90@xmath79 at resonance .
this represents the universal feature of a resonance ; the phase of the dynamic response always lags behind the driving force.@xcite to emphasize the resonant feature of the susceptibility tensor elements we define the symmetric lorentz line shape @xmath80 , and the dispersive line shape @xmath81 as @xmath82 clearly the spin resonance phase can also be written in terms of @xmath80 and @xmath81 as @xmath77 = \arctan(l / d)$ ] so that @xmath83 and @xmath84
. therefore @xmath80 and @xmath81 carry the resonant information of the susceptibility tensor .
using @xmath80 and @xmath81 allows the elements of @xmath51 to be written as @xmath85 .
@xmath86 and @xmath87 are real amplitudes which are related to the sample properties @xmath88 since these amplitudes are real all components of @xmath51 include both a dispersive and a lorentz line shape determined solely from the @xmath89 term . however , in a transmission experiment performed using a resonance cavity @xmath90 is measured .
this product removes the phase dependence carried by @xmath80 and @xmath81 and leaves only the lorentz line shape . for the same reason , the microwave photovoltage induced by spin pumping ( the @xmath33 term in table i )
has a symmetric line shape .
the susceptibility for the two cases of in - plane and perpendicularly applied dc magnetic fields can easily be found from eq .
( [ as ] ) by using the appropriate demagnetization factors .
when the lateral dimensions are much larger than the thickness , n@xmath6 = n@xmath91 = 0 and n@xmath92 = 1 for an in - plane field and n@xmath6 = n@xmath92 = 0 and n@xmath91 = 1 for a field applied at a small angle from the perpendicular . in this paper , we focus on the in - plane case .
the line shape analysis for the perpendicular case can be found in ref . .
in both cases the form of the susceptibility , @xmath93 , describes the ferromagnetic resonance line shape where each element of @xmath51 is the sum of an antisymmetric and symmetric lorentz line shape . as we describe in the next section , via the @xmath25 term of the spin rectification effect , the symmetry properties of the dynamic susceptibility influence the symmetry of the electrically detected fmr which can be controlled by tuning the relative electromagnetic phase @xmath14 . the field - torque spin rectification effect results in the production of a dc voltage from the non - linear coupling of rf electric and magnetic fields .
for example , it may follow from the generalized ohm s
law@xcite @xmath94 where @xmath17 is the conductivity , @xmath95 is the resistivity change due to amr and @xmath96 is the extraordinary hall coefficient .
[ t ] as shown in fig .
[ fig2 ] , we use two coordinate systems to describe a long narrow strip under the rotating in - plane magnetic field @xmath97 .
the sample coordinate system @xmath98 is fixed with the sample length along the @xmath99 direction and the sample width in the @xmath100 direction .
the measurement coordinate system @xmath101 rotates with the @xmath97 direction which is along the @xmath102 axis .
we define @xmath103 as the angle between the direction of the strip and the in - plane applied static magnetic field ( _ i.e. _ , between the @xmath99 and @xmath67 directions ) . in both coordinate systems , the @xmath104 axis is along the normal of the sample plane . in the case of a sample length much larger than the width , the rf current , @xmath105 flows along the strip direction @xmath99 . in this geometry
the field due to the hall effect will only be in the transverse direction and will not generate a voltage along the strip . taking the time average of the electric field integrated along the @xmath99 direction ,
the photovoltage is found as@xcite @xmath106 where @xmath107 is the resistance change due to the amr effect and the @xmath108 term is a result of the amr effect which couples * j * and * m*. the susceptibility tensor given by eqs .
( [ chifinal ] ) and ( [ as ] ) can be used to write @xmath109 in terms of the rf @xmath11 field .
since @xmath74 and @xmath97 are both along the @xmath67-axis , only the components of * h * perpendicular to * z * will contribute to * m*. however , since the rf current flows in the @xmath99 direction , to calculate the rectified voltage , @xmath109 must be transformed into the @xmath110 coordinate system by using the rotation @xmath111 , which introduces an additional @xmath103 dependence into the photovoltage .
we find that the photovoltage can be written in terms of the symmetric and antisymmetric lorentz line shapes , @xmath80 and @xmath81 , as @xmath112 where @xmath113 , \\
\nonumber a_d & = \sin(2\theta_h)[a_{xx}h_{x^\prime}\cos(\theta_h)\cos(\phi_{x^\prime } ) \\ & \qquad -a_{xy}h_{y}\sin(\phi_{y})- a_{xx}h_{z^\prime}\sin(\theta_h)\cos(\phi_{z^\prime})],\end{aligned}\ ] ] and @xmath114 and @xmath115 are the relative phases between electric and magnetic fields in the @xmath116 and @xmath117 directions , respectively .
the amplitudes of the lorentz and dispersive line shape contributions show a complex dependence on the relative phases for the @xmath118 and @xmath99 directions and in general both line shapes will be present .
however , depending on the experimental conditions , this dependence may be simplified .
for instance when @xmath119 is the dominate driving field as shown in fig .
[ fig2](a ) , we may take @xmath120 = @xmath121 and @xmath122 , which results in @xmath123.\end{aligned}\ ] ] from eq .
( [ vx ] ) we see that the photovoltage line shape changes from purely symmetric to purely antisymmetric in @xmath124 intervals of @xmath14 , being purely antisymmetric when @xmath125 and purely symmetric when @xmath126 . as shown in fig .
[ fig2](b ) and ( c ) , the photovoltage in eq .
( [ vx ] ) also shows symmetries depending on the static field direction @xmath103 . since *
h * @xmath127 -*h * corresponds to @xmath128 , @xmath129 .
furthermore at @xmath130 the voltage will be zero .
similarly when @xmath120 dominates as shown in fig .
[ fig2](c ) , we take @xmath119 = @xmath131 and @xmath132 which results in a voltage @xmath133.\end{aligned}\ ] ] the symmetry properties are now such that the line shape is purely symmetric when @xmath125 and purely antisymmetric when @xmath134 . also the photovoltage determined by eq .
( [ vy ] ) is now symmetric with respect to @xmath135 under @xmath136 so that @xmath137 as shown in fig .
[ fig2](e ) .
therefore , experimentally the different symmetry of the fmr at @xmath135 and @xmath138 can be used as an indication of which component of the * h * field is dominant . both eq .
( [ vx ] ) and eq .
( [ vy ] ) demonstrate that a change in the relative electromagnetic phase is expected to result in a change in the line shape of the electrically detected fmr .
it is worth noting that when the relative phase @xmath139 , the line shape is purely antisymmetric for fmr driven by @xmath119 and purely symmetric for fmr driven by @xmath140 as illustrated in fig . [ fig2](b ) and [ fig2](e ) , respectively . in the general case
when @xmath109 is driven by multiple @xmath11 components , eq .
( [ fullv ] ) must be used in combination with angular ( @xmath103 ) dependent measurements in order to distinguish different contributions .
it is clear therefore that for field torque induced spin rectification , the relative phase @xmath14 between the microwave electric and magnetic fields plays the pivotal role in the fmr line shape .
note that @xmath14 is a material and frequency dependent property which is related to the losses in the system.@xcite when a plane electromagnetic wave propagates through free space the electric and magnetic fields are in phase and orthogonal to each other.@xcite however when the same electromagnetic wave travels through a dispersive medium where the wave vector is complex , the imaginary contribution can create a phase shift between electric and magnetic fields .
the most well known example is that of a plane electromagnetic wave moving in a conductor @xcite where faraday s law gives a simple relation between electric and magnetic fields , @xmath141 .
therefore the complex part of the wave vector * k * will induce a phase shift between electric and magnetic fields .
although the field will exponentially decay inside a conductor , it will still penetrate a distance on the order of the skin depth , and in a perfect conductor the conductivity , which produces an imaginary dielectric constant , will result in a phase shift of @xmath142 between the electric and magnetic fields.@xcite in a complex system such as an experimental set up involving waveguides , coaxial cables , bonding wires and a sample holder , which are required for electrical fmr detection , the relative phase can not be simply calculated .
nevertheless losses in the system which can be characterized in a variety of ways , such as through the wave impedance,@xcite will lead to a phase shift between electric and magnetic fields which will influence the fmr line shape .
although the physics of @xmath14 is in principle contained in maxwell s equations , due to the lack of technical tools for simultaneously and coherently probing both @xmath10 and @xmath11 fields , the effect of the relative phase had often been ignored until the recent development of spintronic michelson interferometry@xcite . in the following we provide systematically measured data showing the influence of the relative phase @xmath14 on the line shape of fmr which is driven by different @xmath11 field components .
in order to use the @xmath120 field to drive fmr a first generation spin dynamo was used where a cu / cr coplanar waveguide ( cpw ) was fabricated beside a py microstrip with dimension 300 @xmath144 m @xmath145 20 @xmath144 m @xmath145 50 nm on a sio@xmath2/si substrate as shown in fig . [ fig3](a ) .
a microwave current is directly injected into the cpw and flows in the @xmath99 direction inducing a current in the py strip also along the @xmath99-axis . in this geometry the dominant rf @xmath11 field in the py will be the oersted field in the @xmath146 direction produced according to ampre s law .
this field will induce fmr precession with the same cone angle independent of the static * h * orientation .
[ t ] the amr resistance depends on the orientation of the magnetization relative to the current and follows the relation @xmath147 , where @xmath148 ( not shown ) is the angle between the magnetization and the current direction . for py the amr effect , which is responsible for the spin rectification ,
is observed to produce a resistance change of @xmath149 % .
when * h * is applied along the @xmath100-axis , _
i.e. _ , the in - plane hard axis , the magnetization * m * tends to align toward the static field * h * and the angle @xmath148 is determined by @xmath150 for @xmath151 , where @xmath152 is the in - plane shape anisotropy field .
the measured data ( symbols ) shown in fig .
[ fig3](c ) is fit ( solid curve ) according to @xmath153 with @xmath154 , @xmath155 , @xmath156 mt , and @xmath157=0.004 .
[ fig3](d ) shows that the line shape at @xmath158 and @xmath159 = 5 ghz is almost purely dispersive , indicating that at this frequency @xmath160 according to eq .
( [ vy ] ) . the @xmath103 dependence of @xmath78 is shown in fig .
[ fig3](e ) and can be well fit by the function @xmath161}$ ] by taking the shape anisotropy field @xmath162 along the @xmath100-axis into account.@xcite as expected the amplitude of these oscillations is @xmath163 = 4.0 mt .
the frequency dependence of @xmath78 at @xmath164 is shown in fig .
[ fig3](f ) and is fit using @xmath165 with @xmath166ghz / t and @xmath167=1.0 t. [ t ] by systematically measuring the line shape as a function of the microwave frequency , we observe the interesting results of fig .
the fmr line shape is observed to change from almost purely dispersive at @xmath159 = 5 ghz to almost purely symmetric at @xmath159 = 5.56 ghz . as discussed before ,
the line shape may be affected by the * h * orientation , @xmath168 , different * h * vector components will affect the line shape differently .
hence , if changing the microwave frequency changes the dominant driving field , the line shape may change . to rule out such a possibility
an angular dependent experiment was performed to measure the line shape at different @xmath103 for each frequency @xmath9 .
the results are plotted on the right panel of fig .
[ fig4 ] which shows the sinusoidal curves for the lorentz , @xmath169 , and dispersive , @xmath170 , amplitudes ( dashed and solid curves respectively ) as a function of the static field angle @xmath103 . both the lorentz and dispersive amplitudes
are found to follow a @xmath108 dependence on the field angle in agreement with eq .
( [ vy ] ) indicating that the magnetization precession is indeed dominantly driven by the @xmath120 field .
therefore the line shape change indicates that the relative phase @xmath14 is frequency dependent . as shown in fig .
[ fig5](a ) , at @xmath159 = 5 ghz the amplitude of @xmath170 is approximately one order of magnitude larger than @xmath169 , while at @xmath159 = 5.56 ghz @xmath170 is one order of magnitude less than @xmath169 .
such a large change in @xmath171 shows that in a microwave frequency range as narrow as 0.6 ghz , the relative phase @xmath14 can change by 90@xmath79 .
[ fig5](b ) shows @xmath14 determined by using eq .
( [ vy ] ) , which smoothly changes with microwave frequency except for a feature near 5.18 ghz , which is possibly caused by a resonant waveguide mode at this frequency .
[ t ] such a large change of @xmath14 within a very narrow range of microwave frequency indicates the complexity of wave physics .
note that microwaves at @xmath172 5 ghz have wavelengths on the order of a few centimeters which are much larger than the sub - millimeter sample dimensions .
consequently the microwave propagation depends strongly on the boundary conditions of maxwell s equations which physically include the bonding wire , chip carrier , as well as the sample holder .
this is similar to the microwave propagation in a waveguide where the field distribution @xmath168 the waveguide modes , are known to depend strongly on boundary conditions and frequency.@xcite despite the complex wave properties , the key message of our results is clear and consistent with the consideration of the physics of the relative phase : it shows that in order to properly analyze the fmr line shape , @xmath14 has to be determined for each frequency independently .
[ t ] in order to drive the fmr using the rf field in the @xmath100 direction , @xmath119 , a second generation spin dynamo was fabricated with the py strip underneath the cpw as shown in fig .
[ fig6 ] . in this case
the 300 @xmath144 m @xmath145 7 @xmath144 m @xmath145 100 nm py strip is underneath the cu / cr coplanar waveguide which is fabricated on a sio@xmath2/si substrate . again
a microwave current is directly injected into the cpw and induces a current in the @xmath99 direction in the py strip .
the dominant rf field in the py is still the oersted field , but due to the new geometry it is in the @xmath100 direction . due to the smaller width and larger thickness , the demagnetization factor ,
@xmath174 is twice that in the first generation sample .
this corresponds to @xmath163 = 8.0 mt as indicated by the broader amr curve in fig .
[ fig6](c ) .
this value is further confirmed by the @xmath78 vs @xmath103 plot shown in fig . [ fig6](e ) .
[ fig6](f ) shows the frequency dependence of @xmath78 for fmr ( circles ) and for the first perpendicular standing spin wave resonance ( swr ) ( triangles ) measured at @xmath164 .
the frequency dependence of @xmath78 follows @xmath175 where @xmath176 is the exchange field . in fig .
[ fig6](f ) the standing swr is fit using @xmath177ghz / t , @xmath178 mt and @xmath179 t. similar to the results presented in the previous section , the line shape of fmr measured on the second generation sample is also found to be frequency dependent ( not shown ) .
hence , @xmath14 is found to be non - zero in the general case .
for example , at @xmath9/2@xmath180 = 8 ghz , the line shape is found to be nearly symmetric , as shown in fig . [ fig6](d ) for the fmr measured at @xmath181 , which indicates @xmath14 is close to @xmath182 at this frequency .
note that our result is in direct contrast with the recent study of ref .
and , where experiments were measured in the same configuration and where it was suggested that @xmath14 = 0@xmath79 for all samples at all frequencies .
[ t ] while the line shape and hence the relative phase is found to be frequency dependent , @xmath14 is expected to be independent of the static field direction @xmath103 .
this is confirmed in fig .
[ fig7](a ) which shows the line shape measured at several values of @xmath103 in 10@xmath79 increments .
the data can be fit well using eq .
( [ vx ] ) with a constant @xmath183 for all @xmath103 .
it confirms that the fmr is driven by a single * h * component , in this case the @xmath119 field , and that @xmath14 does not depend on @xmath103 . in fig .
[ fig7](b ) the @xmath103 dependence of @xmath169 and @xmath170 ( solid / circles and dashed / squares respectively ) is shown .
the circles and squares are experimental data while the solid and dashed lines are fitting results using a @xmath184 function according to eq .
( [ vx ] ) .
it provides further proof that the @xmath119 field is responsible for driving the fmr in this sample .
while the results from both the 1st and 2nd generation spin dynamos show consistently that @xmath14 is sample and frequency dependent , the 2nd generation spin dynamos exhibit special features in comparison with the 1st generation spin dynamos : the reduced separation between the py strip and cpw enhances the @xmath119 field so that the line width @xmath65 is enhanced by non - linear magnetization damping@xcite , which depends on the cone angle @xmath185 of the precession via the relation @xmath186 . as shown in fig .
[ fig7](c ) , @xmath65 is found to oscillate between 4.0 and 9.0 mt as @xmath103 changes . at @xmath187 , @xmath188 and
the cone angle is at its largest ( about 4@xmath79 ) .
as @xmath103 increases from @xmath189 and moves toward 90@xmath79 , @xmath185 decreases so that the non - linear damping contribution to @xmath65 decreases . using the cone angle calculated from fig .
[ fig7](c ) , we plot in fig .
[ fig7](d ) @xmath190 as a function of the cone angle .
it shows that @xmath191 has a quadratic dependence on the precession cone angle , which is in agreement with our previous study in the perpendicular @xmath97-field configuration@xcite .
we note that for cone angles above only a few degrees , the non - linear damping already dominates the contribution to @xmath191 .
again , this is in direct contrast with the result of mosendz _
et al._,@xcite , where @xmath65 was found to be constant by varying @xmath103 , indicating no influence from non - linear damping , but the cone angle @xmath185 was estimated to be as high as 15@xmath79 based on the line shape analysis assuming relative phase @xmath14 = 0 .
next we consider the most general case which is described by eq .
( [ fullv ] ) where all components of * h * may contribute to the fmr line shape .
the sample used here is a single py strip where a waveguide with a horn antennae provided both the electric and magnetic driving fields .
the sample chip is mounted near the centre , at the end of a rectangular waveguide and the py strip is directed along the short axis of the waveguide .
[ t ] in a waveguide , the electromagnetic fields are well known and in general three components , @xmath192 and @xmath193 exist.@xcite figure [ fig8](a ) shows both the fmr and perpendicular standing swr at @xmath103 = 45@xmath79 .
indeed both the amplitude and the line shape are different for the two fmr peaks located at @xmath135 and @xmath138 , which indicates the existence of multiple * h * field components and eq .
( [ fullv ] ) and eq .
( [ ld ] ) are needed to separate them .
this separation is done using the lorentz and dispersive amplitudes determined from a fit to the fmr which are plotted as a function of @xmath103 in fig . [
fig8](b ) and ( c ) for @xmath194 and 11.2 ghz , respectively .
a fit using eq .
( [ ld ] ) allows a separation of the contributions from each of the @xmath192 and @xmath193 fields based on the their different contributions to the @xmath103 dependence of the line shape .
[ h ] .angular separation of * h * field components for 12 and 11.2 ghz . [ cols="^,^,^",options="header " , ] the results of the fit have been tabulated in table [ table2 ] where @xmath195ghz / t , @xmath196 = 0.97 t and @xmath197 = 152 mt were used .
the amplitudes of the different * h * field components have been normalized with respect to the @xmath119 component . at both 11.2 and 12 ghz
the @xmath119 field is much larger than @xmath120 or @xmath193 , which is expected based on the wave propagation in a horn antennae . in changing from 11.2 to 12 ghz the relative phase for each component
is seen to change .
therefore even in the case of a complex line shape produced by multiple * h * field components , by separating the individual contributions of the rf magnetic field via angular dependence measurements , the relative phase @xmath14 of each field component is found to be frequency dependent .
[ h ! ] in addition to the frequency and sample dependencies , the relative phase @xmath14 may also depend on the lead configuration and wiring conditions of a particular device , as we have mentioned in section a. here , we address such additional influences by using the first generation spin dynamos@xcite shown in the inset of fig .
[ fig9](a ) .
two spin dynamos with the same lateral dimensions but different py thickness @xmath198 are studied .
each spin dynamo involves two identical py strips denoted by s1 and s2 , one in each center of the g - s strips of the cpw , which are placed symmetrically with respect to the s strip .
the current and rf @xmath11 field are induced in the py via a microwave current directly injected into the cpw .
similar to the sample discussed in section a , @xmath140 is the dominant field which drives the fmr . as shown in fig .
[ fig9](a ) , fmr measured at @xmath159 = 5 ghz on the sample s1 with @xmath198 = 100 nm shows a nearly symmetric lorentz line shape and a field symmetry of @xmath137 . from the fmr line shape
fitting , @xmath14 = -11@xmath79 is found .
interestingly , as shown in fig .
[ fig9](b ) , the fmr of the sample s2 of the same spin dynamo measured under the same experimental conditions shows a different line shape , from which a different @xmath14 = 22@xmath79 is found .
we can further compare @xmath14 measured on the other spin dynamo with a different py thickness of @xmath198 = 60 nm , also at @xmath159 = 5 ghz . here for s1 , @xmath14 = -29@xmath79 while for s2 , @xmath14 = 27@xmath79 .
again , the relative phase is found different for s1 and s2 .
these results demonstrate that due to additional influences such as a different lead configuration and wiring conditions , even for samples with the same lateral dimensions , @xmath14 in each device is not necessarily the same .
it demonstrates clearly that the relative phase @xmath14 can not be simply determined by analyzing the fmr line shape measured on a reference device . the experimental data presented above show that regardless of the fmr driving field configuration , the relative phase between the rf electric and magnetic field is sample and frequency dependent and non - zero .
this non - zero phase results in both symmetric and antisymmetric lorentz line shapes in the fmr detected via field - torque induced spin rectification .
the @xmath14 dependence of the line shape symmetry changes based on which component of the rf @xmath11 field is responsible for driving the fmr precession .
for instance a purely antisymmetric line shape could correspond to @xmath199 if the fmr is driven by @xmath119 , or to @xmath200 if the fmr is driven by @xmath120 , therefore the line shape itself can not be used to determine @xmath14 directly . to separate the * h * field components an angular ( @xmath103 )
dependence measurement is necessary , which allows both * h * as well as the phase to be determined .
using such a measurement @xmath14 has been observed to change from 0@xmath79 to 90@xmath79 in a narrow frequency range ( 0.6 ghz ) resulting in a change from an antisymmetric to symmetric line shape demonstrating the large effect the relative phase has on the fmr line shape .
furthermore @xmath14 is not identical even in samples with the same geometrical size .
therefore in our opinion @xmath14 can not be simply determined from a reference sample but should be calibrated for each sample , at each frequency and for each measurement cycle .
spin rectifications caused by the coupling between current and magnetization in a ferromagnetic microstrip provide a powerful tool for the study of spin dynamics . in order to distinguish different mechanisms which enable the electrical detection of fmr via microwave photovoltages , it is essential to properly analyze the fmr line shape . for spin rectification caused by a microwave field torque , due to the coherent nature of this coupling ,
the resulting dc voltage depends strongly on the relative phase between the rf electric and magnetic fields used to drive the current and magnetization , respectively .
therefore not only does electrical fmr detection provide a route to study the relative phase , but it also necessitates calibrating the relative phase prior to performing electrically detected fmr experiments . based on a systematic study of the electrically detected fmr , the line shape is observed to depend strongly on the microwave frequency , driving field configuration , sample structure and even wiring conditions .
it is in general a combination of lorentz and dispersive contributions .
these effects have been quantitatively explained by accounting for the relative phase shift between electric and magnetic fields .
analytical formula have been established to analyze the fmr line shape .
our results imply that for electrically detected fmr which involves both spin hall and spin rectification effects , the pivotal relative phase must be calibrated independently in order to properly analyze the fmr line shape and quantify the spin hall angle .
this can not be done by using a reference sample but could be achieved through such techniques as spintronic michelson interferometry@xcite .
we would like to thank b. w. southern , a. hoffmann , and s. d. bader for discussions .
this work has been funded by nserc , cfi , cmc and urgp grants ( c .- m .
zxc was supported by the national natural science foundation of china grant no .
|
this work reviews and examines two particular issues related with the new technique of electrical detection of ferromagnetic resonance ( fmr ) .
this powerful technique has been broadly applied for studying magnetization and spin dynamics over the past few years .
the first issue is the relation and distinction between different mechanisms that give rise to a photovoltage via fmr in composite magnetic structures , and the second is the proper analysis of the fmr line shape , which remains the `` achilles heel '' in interpreting experimental results , especially for either studying the spin pumping effect or quantifying the spin hall angles via the electrically detected fmr .
|
suppose @xmath14 is an elliptic surface defined over a number field @xmath1 , so @xmath2 is a smooth projective curve and all but finitely many fibers @xmath15 , @xmath16 , are smooth elliptic curves .
we let @xmath17 denote the nron - tate canonical height of @xmath18 viewed as an elliptic curve over the function field @xmath19 ; we let @xmath20 denote the canonical height on the fibers for ( all but finitely many ) @xmath21 .
suppose that @xmath22 is a section defined over @xmath1 for which @xmath23 , so , in particular , the points @xmath8 on the fiber are not torsion in @xmath15 for all @xmath24 .
tate showed that the function @xmath25 is a weil height on @xmath6 , up to a bounded error @xcite .
more precisely , there exists a divisor @xmath26 of degree equal to @xmath27 so that @xmath28 where @xmath29 is a weil height on @xmath6 associated to @xmath30 . in a series of three articles @xcite , silverman refined statement ( [ elliptic variation ] ) by analyzing the nron decomposition of the canonical height on the fibers @xmath31 where @xmath32 denotes the set of places of the number field @xmath1 , and @xmath33 are the integers appearing in the product formula @xmath34 for all @xmath35 . in this article , we explain how silverman s conclusions about the local functions @xmath36 are precisely the input needed to show that @xmath37 is a good " height function on the base curve @xmath2 , from the point of view of equidistribution . combining his work with methods from complex dynamics , as in @xcite , and the inequalities of zhang on successive minima @xcite , we prove : [ good height ] let @xmath1 be a number field and @xmath38 for a smooth projective curve @xmath2 defined over @xmath1 .
fix any elliptic surface @xmath14 defined over @xmath1 and point @xmath39 satisfying @xmath40 .
then @xmath41 for @xmath24 with smooth fibers , is the restriction of a height function on @xmath6 induced from an adelically metrized ample line bundle @xmath42 , with continuous metrics of non - negative curvature , satisfying @xmath43 theorem [ good height ] implies that our height function on @xmath2 satisfies the hypotheses of the equidistribution theorems of thuillier and yuan for points of small height on curves @xcite , and we deduce the following : [ equidistribution on b ] let @xmath1 be a number field and @xmath38 for a smooth projective curve @xmath2 defined over @xmath1 . fix any elliptic surface @xmath14 defined over @xmath1 and point @xmath39 satisfying @xmath40 .
there is a collection of probability measures @xmath44 on the berkovich analytifications @xmath45 such that for any infinite , non - repeating sequence of @xmath46 such that @xmath47 as @xmath48 , the discrete measures @xmath49 converge weakly on @xmath45 to the measure @xmath50 at each place @xmath51 of @xmath1 .
[ measures ] the measures @xmath50 of corollary [ equidistribution on b ] are not difficult to describe , at least at the archimedean places . at each archimedean place @xmath51
, there is a canonical positive @xmath52-current @xmath53 on the surface @xmath54 ( with continuous potentials away from the singular fibers ) which restricts to the haar measure on each smooth fiber @xmath55 .
the measure @xmath50 on @xmath9 is just the pull - back of this current by the section @xmath12 .
moreover , at every place , the measure @xmath50 is the laplacian of the local height function @xmath56 , away from its singularities .
we give more details about ( and a dynamical perpective on ) the construction of the current @xmath53 in section [ dynamics ] . as a consequence of theorem [ good height ] , and combined with the work of masser and zannier @xcite , we obtain the so - called bogomolov extension of their theorems
. suppose @xmath57 and @xmath58 are elliptic surfaces over a curve @xmath2 , defined over @xmath59 , and assume that at least one is non - isotrivial .
we consider sections @xmath12 of the fiber product @xmath60 defined over @xmath59 .
we say that a section @xmath61 is _ special _ if either * @xmath62 is torsion on @xmath63 for some @xmath64 , or * @xmath65 and @xmath66 are isogenous and there exists a pair of nonzero integers @xmath67 so that @xmath68 on a common cover @xmath18 of @xmath63 .
if a family of abelian surfaces @xmath13 is isogenous to a fiber product ( after performing a base change @xmath69 if needed ) , we say that a section of @xmath70 is special if it is special on the fiber product .
it is well known , assuming both @xmath65 and @xmath66 are non - isotrivial , that a special section will always pass through infinitely many torsion points in the fibers @xmath71 .
that is , there are infinitely many @xmath72 for which @xmath73 if one of the curves , say @xmath66 , is isotrivial then a special section will pass through infinitely many torsion points unless the component @xmath74 is torsion and @xmath75 is constant and non - torsion on @xmath66 .
is isotrivial if there exist a finite @xmath69 , an elliptic curve @xmath76 , and a birational @xmath77 .
now let @xmath12 be a section of @xmath78 and denote by @xmath79 its extension to @xmath80 .
then we say that @xmath12 is constant if @xmath81 is constant on @xmath80 .
] for proofs of these statements , see ( * ? ? ?
* chapter 3 ) or , for dynamical proofs , see @xcite .
the converse statement is also true , but it is much more difficult : masser and zannier proved that if @xmath82 for infinitely many @xmath83 , then the section @xmath12 must be special @xcite .
we extend these results of masser - zannier from points of height 0 to points of small height : [ zhang language ] let @xmath2 be a quasiprojective smooth algebraic curve defined over @xmath59 .
suppose @xmath84 is a non - isotrivial family of abelian varieties of dimension 2 defined over @xmath59 with a non - simple generic fiber .
let @xmath42 be a line bundle on @xmath70 which restricts to an ample and symmetric line bundle on each fiber @xmath85 , and let @xmath86 be the induced nron - tate canonical height on @xmath85 , for each @xmath83 . for each non - special section @xmath87 defined over @xmath59 , there is a constant @xmath88 so that @xmath89 is finite . to put theorem [ zhang language ] in context ,
recall that ullmo and zhang proved the bogomolov conjecture @xcite , extending the manin - mumford conjecture ( a theorem of raynaud @xcite ) to points of small height on abelian varieties defined over @xmath59 .
that is , for any abelian variety @xmath90 and choice of ample , symmetric line bundle @xmath42 on @xmath70 , and for any subvariety @xmath91 which is not itself an abelian subvariety or torsion translate of one , there is a positive constant @xmath92 so that @xmath93 is not zariski dense in @xmath94 .
a key ingredient in their proofs is the equidistribution theorem of szpiro , ullmo , and zhang @xcite . in his 1998 icm lecture
notes @xcite , zhang presented a conjecture about geometrically simple families of abelian varieties , which stated , in its most basic form : [ zhang ] [ zhang conj ] let @xmath2 be a quasiprojective smooth algebraic curve defined over @xmath59 .
suppose @xmath84 is a non - isotrivial family of abelian varieties with fiber dimension @xmath95 , defined over @xmath59 with a simple generic fiber .
let @xmath42 be a line bundle on @xmath70 which restricts to an ample and symmetric line bundle on each fiber @xmath85 , and let @xmath86 be the induced nron - tate canonical height on @xmath85 , for each @xmath83 .
for each non - torsion section @xmath87 defined over @xmath59 , there is a constant @xmath88 so that @xmath89 is finite . when the dimension of the fibers @xmath85 is equal to 2 , the finiteness of @xmath96 for sections as in conjecture [ zhang conj ] was established recently by masser and zannier in @xcite .
it is well known that the conclusion of conjecture [ zhang conj ] can fail to hold if @xmath70 is not simple and certainly fails if it is a family of elliptic curves , as mentioned above .
however , the results of masser and zannier in their earlier work @xcite suggested a formulation of zhang s conjecture for the non - simple case in relative dimension 2 ; this is what we proved in our theorem [ zhang language ] .
theorem [ good height ] , corollary [ equidistribution on b ] , and theorem [ zhang language ] were obtained in the special case of the legendre family @xmath97 over @xmath98 and the abelian variety @xmath99 , for sections @xmath12 with @xmath100-coordinates in @xmath101 in @xcite , using methods from complex dynamical systems , without appealing to silverman and tate s results on the height function . moreover , restricting further to sections @xmath12 with constant @xmath100-coordinate ( in @xmath102 ) , theorem [ zhang language ] was obtained without relying on the theorems of masser and zannier and gave an alternate proof of their result .
this includes the special case treated by masser and zannier in their article @xcite . for sections with constant @xmath100-coordinate , the hypothesis on @xmath12 ( that @xmath40 ) is equivalent to asking that @xmath103 ( * ? ? ?
* proposition 1.4 ) .
the result of theorem [ zhang language ] extends to non - isotrivial families of abelian varieties @xmath13 in higher dimensions that split as a product @xmath104 of @xmath105 elliptic curves .
indeed , if @xmath106 is a section , and if there exists an infinite sequence @xmath107 so that @xmath108 , then in fact @xmath109 for all @xmath64 , and we can reduce to theorem [ zhang language ] by considering pairs @xmath110 , @xmath111 . we conclude that every pair @xmath112 is special as a section of @xmath110 . * comments and acknowledgements .
* this project was motivated , in part , by experiments to visualize silverman s results on the variation of canonical height @xcite in terms of the measures @xmath50 at archimedean places , and to examine their dependence on @xmath12 .
in particular , the measure detects the failure of the local height function @xmath56 to be harmonic ; compare the comments on non - analyticity preceding theorem i.0.3 of @xcite .
the images appearing in section [ experiment ] were first presented at the conference in honor of silverman s birthday , august 2015 .
we thank charles favre , dragos ghioca , robert rumely , joseph silverman , and amaury thuillier for helpful suggestions and inspiring conversations .
our research was supported by the national science foundation and the simons foundation .
let @xmath113 be a product formula field of characteristic 0 , so there exists a family @xmath114 of non - trivial absolute values on @xmath113 and a collection of positive integers @xmath33 for @xmath115 so that @xmath116 for all @xmath117 .
let @xmath118 be an elliptic curve with origin @xmath119 , expressed in weierstrass form as @xmath120 with discriminant @xmath121 .
denote by @xmath122 the nron - tate canonical height function ; it can be defined by @xmath123p))}{n^2}\ ] ] where @xmath124 is the naive weil height on @xmath125 and @xmath126 is the degree 2 projection to the @xmath100-coordinate . for each @xmath127
, we let @xmath128 denote a minimal , algebraically closed field containing @xmath113 which is complete with respect to @xmath129 . for each @xmath51
, we fix an embedding of @xmath130 into @xmath128 .
the canonical height has a decomposition into local heights , as @xmath131 for @xmath132 , with the local heights @xmath133 characterized by three properties ( * ? ? ?
* chapter 6 , theorem 1.1 , page 455 ) : 1 .
@xmath133 is continuous on @xmath134 and bounded on the complement of any @xmath51-adic neighborhood of @xmath119 ; 2 .
the limit of @xmath135 exists as @xmath136 in @xmath137 ; and 3 . for all @xmath138 with @xmath139 p \not= o$ ] , @xmath140p ) = 4 \hat{\lambda}(p ) - \log|2y+a_1x+a_3|_v + \frac{1}{4 } \log|\delta|_v.\ ] ]
now let @xmath1 be a number field and @xmath14 an elliptic surface defined over a number field @xmath1 with zero section @xmath141 .
let @xmath142 be a non - zero section defined over @xmath1 , and assume that @xmath143 when viewing @xmath12 as a point on the elliptic curve @xmath18 defined over @xmath144 . for each @xmath16
such that the fiber @xmath15 is non - singular , we have point @xmath145 .
we will investigate the function @xmath25 which is well defined at all but finitely many @xmath146 .
furthermore , via the embedding of @xmath147 into @xmath128 for each place @xmath148 , we may view @xmath78 as defined over @xmath128 and consider the nron local heights @xmath36 on the non - singular fibers @xmath15 as functions of @xmath149 .
let @xmath150 be the @xmath151-divisor on @xmath2 defined by @xmath152 here , @xmath153 is the local canonical height of the point @xmath12 on the elliptic curve @xmath18 over @xmath144 at the place @xmath154 , for each @xmath155 .
the degree of @xmath150 is equal to @xmath27 .
it follows from the definitions that @xmath156 is a subset of the finite set @xmath157 by enlarging @xmath1 , we may assume that the support of @xmath150 is contained in @xmath158 . that @xmath150 is a @xmath151-divisor is standard , following from the fact that the numbers @xmath153 can be viewed as arithmetic intersection numbers on a nron local model .
see ( * ? ? ?
* chapter iii , theorem 9.3 ) for a proof that @xmath159 ; see ( * ? ? ?
* section 6 and p. 203 ) and ( * ? ? ?
* chapter 11 theorem 5.1 ) for proofs that each local function @xmath133 also takes values in @xmath151 ; see ( * ? ? ?
* theorem 1.1 ) for a dynamical proof .
let @xmath160 be an analytic weil height on @xmath6 as defined in ( * ? ? ?
* example 1(a ) ) .
that is , we let @xmath161 be the genus of @xmath2 , and for each point @xmath162 , we choose an element @xmath163 of @xmath164 which has a pole of order @xmath165 at @xmath166 and no other poles . for each non - archimedean place @xmath51 of @xmath1 ,
set @xmath167 for all @xmath168 . for archimedean places
@xmath51 , the local height is defined by @xmath169 we set @xmath170 for all @xmath7 . for fixed choices of @xmath163 , we will call the associated height @xmath160 our reference height " for the divisor @xmath150 .
silverman proved : [ silverman triviality ] ( * ? ? ?
* theorem [email protected] ) for any choice of reference height @xmath160 , there is a finite set @xmath171 of places so that @xmath172 for all @xmath173 and all @xmath174 .
fix a point @xmath175 and a uniformizer @xmath176 for @xmath177 , and consider the function @xmath178 which is not _ a priori _ defined at @xmath177 .
theorem [ silverman triviality ] implies that @xmath179 for all but finitely many places @xmath51 in a @xmath51-adic neighborhood of each @xmath177 .
silverman also proved the following : [ silverman continuity ] ( * ? ? ?
* theorem [email protected] ) fix @xmath180 and a uniformizer @xmath181 at @xmath177 . for all @xmath182 ,
there exists a neighborhood @xmath183 containing @xmath177 so that the function @xmath184 of extends to a continuous function on @xmath185 .
recall that the nron - tate height @xmath17 and its local counterparts @xmath133 can be defined dynamically .
letting @xmath18 be an elliptic curve defined over a number field @xmath1 , the multiplication - by-2 endomorphism @xmath186 on @xmath18 descends to a rational function of degree 4 on @xmath125 , via the standard quotient identifying a point @xmath12 with its additive inverse : @xmath187_\pi \ar[r]^\phi & e\ar[d]^\pi \\
{ \mathbb{p}}^1 \ar[r]^{f_\phi } & { \mathbb{p}}^1 } \ ] ] an elementary , but key , observation is that a point is torsion on @xmath18 if and only if its quotient in @xmath125 is preperiodic for @xmath188 .
the height @xmath17 on @xmath189 satisfies @xmath190 where @xmath124 is the standard logarithmic weil height on @xmath191 .
now let @xmath78 be an elliptic surface defined over a number field @xmath1 , and let @xmath192 be a section , also defined over @xmath1 . in this section , we use this perspective to give a proof of subharmonicity of the local height functions @xmath193 and the extensions @xmath194 of .
we will present this fact as an immediate consequence of now - standard complex - dynamical convergence arguments , at least when the fiber @xmath15 is smooth and the local height @xmath56 is finite .
near singular fibers , we utilize the maximum principle and standard results on removable singularities for subharmonic functions .
the same reasoning applies in both archimedean and non - archimedean settings . in [
subsection : measures ] we provide the background to justify the explicit description of the limiting distribution @xmath50 at the archimedean places @xmath51 of @xmath1 , as mentioned in remark [ measures ] . as in [ preliminaries
] , we let @xmath18 be an elliptic curve in weierstrass form , defined over a product - formula field @xmath113 of characteristic 0 . we define a rational function @xmath195 on @xmath125 by the formula @xmath196p)\ ] ] for all @xmath197 . here
@xmath198 is the @xmath199coordinate for a point @xmath200 ; this function @xmath100 plays the role of @xmath201 in . in coordinates , we have @xmath202 and @xmath203 for @xmath204 . by a _ lift _ of @xmath205 , we mean any homogeneous polynomial map f on @xmath206 , defined over @xmath207 , so that @xmath208 , where @xmath209 is the tautological projection .
a _ lift _ of a point @xmath210 is a choice of @xmath211 so that @xmath212 .
the _ standard lift _ of @xmath205 will be the map @xmath213 defined by @xmath214 for each @xmath215 , the @xmath51-adic _ escape rate _ is defined by @xmath216 where @xmath217 any other lift of @xmath205 is of the form @xmath218 for some @xmath219 ; observe that @xmath220 note that @xmath221 for any choice of lift @xmath222 .
furthermore , @xmath223 is continuous on @xmath224 , as proved in the archimedean case by @xcite . for non - archimedean absolute values
@xmath51 , @xmath223 extends continuously to the product of berkovich affine lines @xmath225 ( * ? ? ?
* chapter 10 ) .
[ lambda and g ] for the standard lift @xmath222 of @xmath205 , and for each place @xmath51 of @xmath207 , the local canonical height function satisfies @xmath226 where @xmath227 .
the proof is immediate from the properties of @xmath223 by checking the three characterizing conditions for @xmath133 .
now let @xmath1 be a number field and @xmath14 an elliptic surface defined over a number field @xmath1 with zero section @xmath141 .
let @xmath144 ; viewing @xmath18 as an elliptic curve defined over @xmath228 , we also fix a point @xmath39 .
recall the function @xmath229 defined in .
[ subharmonic ] for every @xmath180 and uniformizer @xmath181 in @xmath228 at @xmath177 , the function @xmath230 extends to a continuous and subharmonic function on a neighborhood of @xmath177 in the berkovich analytification @xmath231 .
the continuity was already established in theorem [ silverman continuity ] , though it was not explicitly stated for the berkovich space .
the argument below takes care of that .
we begin with a lemma . [ harmonic ] fix @xmath232 and @xmath180 .
let @xmath233 be a uniformizer at @xmath177 .
for each place @xmath51 of @xmath1 , the function @xmath234 is harmonic in a neighborhood of @xmath177 in the berkovich analytification @xmath231 .
this is silverman s ( * ? ? ?
* lemma [email protected](c ) ) plus a removable singularities lemma for harmonic functions .
see also ( * ? ? ?
* proposition 7.19 ) for the extension of a harmonic function to a disk in the berkovich space @xmath231 .
fix @xmath39 .
let @xmath222 and @xmath94 be lifts of @xmath205 and @xmath198 to @xmath235 , respectively . iterating @xmath222 , we set @xmath236 and observe that @xmath237 from the definition of the escape rate .
we let @xmath238 and @xmath239 denote the specializations of @xmath222 and @xmath94 at a point @xmath7 ; they are well defined for all but finitely many @xmath24 .
observe that if @xmath222 is the standard lift for @xmath18 then so is @xmath238 for all @xmath24 .
[ escape rate subharmonic ] fix @xmath39 , @xmath180 , and @xmath148 . for any choice of lifts @xmath222 of @xmath205 and @xmath94 of @xmath198
, the function @xmath240 extends to a continuous and subharmonic function in a neighborhood of @xmath177 in @xmath231 .
first observe that the conclusion does not depend on the choices of @xmath222 and @xmath94 .
indeed , @xmath241 for any @xmath242 .
so by lemma [ harmonic ] the function @xmath243 is continuous and subharmonic for one choice if and only if it is continuous and subharmonic for all choices .
let @xmath222 be the standard lift of @xmath205 .
suppose that @xmath244 .
since @xmath245 , we compute that @xmath246 now suppose that @xmath247 .
fix @xmath248 and local uniformizer @xmath181 at @xmath177 . choose a lift @xmath222 of @xmath205 so that the coefficients of @xmath222 have no poles at @xmath177 , with @xmath249 .
choose lift @xmath94 of @xmath198 so that @xmath239 is well defined for all @xmath24 near @xmath177 and @xmath250 .
as above , we write @xmath251 and put @xmath252 so that @xmath253 for all @xmath254 and @xmath255 .
set @xmath256 for each place @xmath51 , we set @xmath257 by construction , the limit of @xmath258 ( for @xmath24 near @xmath177 with @xmath259 ) is exactly the function @xmath260 for these choices .
in fact , for @xmath24 in a small neighborhood of @xmath177 , but with @xmath261 , the function @xmath262 on @xmath125 is a well - defined rational function of degree 4 ; so the specialization of the homogeneous polynomial map @xmath238 satisfies @xmath263 . furthermore , as the coefficients of @xmath238 depend analytically on @xmath24 , the functions @xmath258 converge locally uniformly to the function @xmath260 away from @xmath264 .
this can be seen with a standard telescoping sum argument , used often in complex dynamics , as in ( * ? ? ?
* proposition 1.2 ) .
in particular , @xmath260 is continuous on a punctured neighborhood of @xmath177 . at the archimedean places @xmath51 , and for each @xmath254 ,
the function @xmath258 is clearly continuous and subharmonic in a neighborhood of @xmath177 . at non - archimedean places
@xmath51 , this definition extends to a berkovich disk around @xmath177 , setting @xmath265_t , \log[b_n(t)/t^{a_n}]_t\}\ ] ] where @xmath266_t$ ] is the seminorm on @xmath267 $ ] associated to the point @xmath24 .
each of these functions @xmath258 is continuous and subharmonic for @xmath24 in a berkovich disk around @xmath177 .
( compare @xcite example 8.7 , proposition 8.26(d ) , and equation ( 10.9 ) . )
[ upper bound ] for all @xmath51 , and by shrinking the radius @xmath268 if necessary , the functions @xmath258 are uniformly bounded from above on the ( berkovich ) disk @xmath269 . as observed above , the functions @xmath258
converge locally uniformly away from @xmath264 to the continuous function @xmath270 .
choose a small radius @xmath268 , and let @xmath271 which is finite by the convergence . because the functions are subharmonic
, the maximum principle implies that @xmath272 throughout the disk of radius @xmath268 , for all @xmath254 .
for the non - archimedean places , there is also a maximum principle on the berkovich disk , where the role of the circle of radius @xmath268 is played by the type ii point associated to the disk of radius @xmath268 ( see @xcite proposition 8.14 ) .
we can now complete the proof of proposition [ escape rate subharmonic ] . as each @xmath258 is subharmonic , and the functions are uniformly bounded from above on the disk by lemma [ upper bound ] , we know that the ( upper - semicontinuous regularlization of the ) limsup of these functions is subharmonic .
see @xcite proposition 8.26(f ) for a proof in the non - archimedean case .
subharmonicity now follows from proposition [ lambda and g ] , lemma [ harmonic ] , and proposition [ escape rate subharmonic ] .
the continuity at each archimedean place is the content of theorem [ silverman continuity ] .
the continuity at each non - archimedean place is a combination of the continuity on the punctured berkovich disk ( as in the proof of proposition [ escape rate subharmonic ] ) and the continuity on type i ( classical ) points given in theorem [ silverman continuity ] .
here we provide more details about the description of the measures appearing in the statement of corollary [ equidistribution on b ] , as discussed in remark [ measures ] .
fix an archimedean place @xmath51 and any point @xmath180 . choosing a uniformizer @xmath181 at @xmath177 ,
recall the definition of @xmath194 from .
we define @xmath273 on a neighborhood of @xmath177 in @xmath231 ; note that this is indepedent of the choice of @xmath181 .
note that @xmath50 can be expressed as @xmath274 for @xmath24 outside of the finitely many points in the support of the divisor @xmath150 or where the fiber @xmath15 is singular .
note , further , that @xmath50 assigns no mass to any individual point @xmath177 , because the potentials are bounded by theorem [ subharmonic ] .
the details on the metric and the equidistribution theorem in section [ sec : metric ] will show that these are exactly the measures that arise as the distribution of the points of small height in corollary [ equidistribution on b ] .
it is well known that the local height function on a smooth elliptic curve is a potential for the haar measure .
that is , for fixed @xmath24 we have @xmath275 where @xmath276 is the normalized haar measure on @xmath15 and @xmath277 is a delta - mass supported at the origin of @xmath15 ; see , e.g. , ( * ? ? ?
* theorem ii.5.1 ) .
we present an alternative proof of this fact related to dynamics as part of proposition [ measure description ] , as a consequence of proposition [ lambda and g ] .
[ measure description ] let @xmath14 be an elliptic surface and @xmath3 a section , both defined over a number field @xmath1 .
let @xmath278 be the union of the finitely many singular fibers in @xmath18 .
for each archimedean place @xmath51 of @xmath1 , there is a positive , closed ( 1,1 ) current @xmath53 on @xmath279 with locally continuous potentials so that @xmath280 is the haar measure on each smooth fiber , and @xmath281 is equal to the measure @xmath50 . as @xmath53 has continuous potentials , the restriction @xmath280 and the pullback @xmath281
are well defined .
that is , we have @xmath282 where @xmath181 is a locally defined potential of @xmath53 , and @xmath283 locally on @xmath2 .
the measure @xmath50 has no atoms , so it is determined by @xmath53 along the image of @xmath12 in @xmath279 .
let us fix any small neighborhood @xmath185 in the base curve @xmath9 so that all fibers @xmath15 are smooth for @xmath284 .
let @xmath262 be the map on @xmath125 defined in
[ ss : dynamics ] ; by shrinking @xmath185 if necessary , we can find lifts @xmath238 of @xmath262 that are holomorphic in @xmath285 . from @xcite ( or the proof of ( * ? ? ?
* proposition 1.2 ) ) , we know that the escape rate @xmath286 is continuous and plurisubharmonic as a function of @xmath287
. therefore @xmath288 projects to a closed and positive ( 1,1)-current @xmath289 on the complex surface @xmath290 , with locally continuous potentials .
this current @xmath289 has the property that , restricted to each fiber @xmath125 , its total mass is 1 ; and the measure on the fiber is the measure of maximal entropy for the rational map @xmath262 @xcite .
the restriction @xmath291 of the elliptic surface @xmath18 over @xmath185 maps with degree 2 to the complex surface @xmath292 by the projection @xmath201 of .
the current @xmath289 can be pulled back to @xmath18 as @xmath293 where @xmath161 is a locally - defined continuous potential for @xmath289 . covering the base of @xmath279 by sets of the form @xmath185 , the local definitions glue to form the closed , positive @xmath52-current @xmath53 on @xmath279 . if @xmath3 is a section defined over the number field @xmath1 , then @xmath281 has potential given locally by @xmath294 for any lift @xmath239 of @xmath295 .
proposition [ lambda and g ] yields that @xmath281 must coincide with @xmath50 . finally , to conclude that @xmath280 is equal to the normalized haar measure @xmath276
, we may use the well - known dynamical fact that for each fixed @xmath24 in the base , the measure @xmath276 projects by @xmath201 to @xmath125 to the unique measure of maximal entropy for the map @xmath262 ; see , e.g. , @xcite .
in this section we give the proofs of theorem [ good height ] and corollary [ equidistribution on b ] .
we first outline the proofs .
let @xmath14 be an elliptic surface defined over a number field @xmath1 with zero section @xmath141 , and let @xmath142 be a section also defined over @xmath1 so that @xmath40 .
recall from [ set up ] that we introduced a @xmath151-divisor @xmath296 on @xmath2 . by enlarging @xmath1
, we may assume that @xmath156 lies in @xmath158 .
we will define an adelic metric on the ample line bundle @xmath42 associated to the divisor @xmath150 , inducing a height function @xmath297 such that @xmath298 and @xmath299 applying silverman s results on the variation of canonical height , theorems [ silverman triviality ] and [ silverman continuity ] , we will deduce that the metric is continuous and adelic . from theorem [ subharmonic ] , we will conclude that the metric is also semi - positive in the sense of zhang @xcite .
we will use zhang s inequalities @xcite to deduce that @xmath300 consequently , we will be able to apply the equidistribution results of chambert - loir , thuillier , and yuan @xcite to complete our proofs .
let @xmath301 be such that @xmath302 is an integral divisor .
let @xmath303 be the associated line bundle on @xmath2 .
note that @xmath304 so @xmath305 is ample ; by replacing @xmath306 with a multiple , we may assume that @xmath303 is very ample .
fix a place @xmath51 of @xmath1 .
let @xmath185 be an open subset of @xmath231 .
each section @xmath307 is identified with a meromorphic function @xmath205 on @xmath185 satisfying @xmath308 we set @xmath309 taking the locally - defined uniformizer @xmath310 at @xmath24 in the definition of @xmath311 from .
[ metric properties ] the metric @xmath312 on @xmath303 is continuous , semipositive , and adelic .
the continuity and semipositivity follows from theorem [ subharmonic ] .
( in @xcite , semipositivity of a continuously metrized line bundle on a curve is defined terms of subharmonicity of potentials for the curvature form at each archimedean place , and as a uniform limit of smooth semipositive " metrics at each non - archimedean place . in @xcite , it is established that subharmonicity of potentials is a sufficient notion at all places , and he proves in ( * ? ? ?
* theorem 4.3.3 ) that this notion of semipositivity coincides with that of zhang @xcite .
see also ( * ? ? ?
* lemma 3.11 , theorem 3.12 ) where this same argument is applied in a dynamical context . )
the adelic condition follows from theorem [ silverman triviality ] .
a height function on @xmath6 is defined by setting @xmath313 where @xmath186 is any global section of @xmath303 which is nonvanishing along the galois orbit of @xmath24 , and @xmath314 is the metric of [ metric definition ] .
recall that @xmath315 ; we may assume that our sections @xmath186 are defined over @xmath1 , and the product formula guarantees our height is independent of the choice of @xmath186 .
our next goal is to prove the following two important facts about this height function @xmath316 .
[ heightvsheight ] the height function @xmath316 satisfies @xmath317 for all @xmath21 such that the fiber @xmath15 is smooth .
[ height is nonnegative ] the height function @xmath316 satisfies @xmath318 for all @xmath7 . first fix @xmath319 with smooth fiber @xmath15 .
choose a section @xmath186 defined over @xmath1 that does not vanish along the galois orbit of @xmath24 , and let @xmath205 be the associated meromorphic function on @xmath2 .
then @xmath205 takes finite and nonzero values along the galois orbit of @xmath24 .
we have , @xmath320 where the second equality follows from the product formula . for @xmath321 such that @xmath322 is smooth , it is necessarily the case that @xmath323 , and therefore @xmath324 . to compute @xmath325 ,
observe that @xmath326 so its galois orbit is trivial ; fixing a uniformizer @xmath327 at @xmath177 , we have @xmath328 where @xmath329 is the function of associated to the uniformizer @xmath181 . we can compute @xmath325 using the dynamical interpretation of the local heights , described in section [ ss : dynamics ] . fix a weierstrass equation for @xmath18 in a neighborhood of @xmath177 and write @xmath330 .
the assumption that @xmath331 is equivalent to @xmath332 .
after possibly shrinking @xmath185 , write @xmath333 for the chosen uniformizer @xmath181 and a function @xmath334 that does not vanish in @xmath185 .
we choose a lift @xmath94 of @xmath335 on @xmath185 defined as @xmath336 , where @xmath337 . notice that @xmath338 and @xmath339 are regular at @xmath177 .
let @xmath222 be the standard lift in these coordinates , defined in ; it satisfies @xmath340 , and we have @xmath341 . since @xmath342 , proposition [ lambda and g ] implies that @xmath343 for all @xmath285 .
therefore , @xmath344 the product formula now yields that @xmath345 , as claimed . to prove proposition [ height
is nonnegative ] , we first reduce to the case that the elliptic surface @xmath78 has semi - stable reduction ; that is all of its fibers are either smooth or have multiplicative reduction .
the next lemma describes how the height associated with the divisor @xmath150 behaves under base extensions of the elliptic surface @xmath78 .
it is adapted from ( * ? ? ?
* reduction lemma [email protected] ) .
we include it here for completeness .
[ reductionlemma ] let @xmath346 be a finite map of smooth projective curves , let @xmath347 be a minimal model for @xmath348 , and let @xmath349 be the extension of the section @xmath12 . for each @xmath180 and @xmath350
, there is a neighborhood @xmath185 of @xmath351 in @xmath352 and a regular non - vanishing function @xmath205 on @xmath185 such that @xmath353 on @xmath354 .
in particular , @xmath355 let @xmath181 be a uniformizer at @xmath177 , @xmath356 a uniformizer at @xmath351 and @xmath357 .
since local heights are invariant under base extension we have @xmath358 notice that for all @xmath359 in a punctured neighborhood of @xmath360 the fibers @xmath361 are smooth . hence the map @xmath362 gives an isomorphism between the fibers @xmath363 . under this isomorphism @xmath364
is mapped to @xmath365 .
invoking now the uniqueness of the nron local heights , we have @xmath366 combining and we get that for @xmath359 in a punctured neighborhood of @xmath351 , @xmath367 the definition of @xmath254 yields that the function @xmath368 is regular and non - vanishing at @xmath351 . the first part of the lemma follows
. finally , theorem [ silverman continuity ] allows us to conclude that @xmath369 at the point @xmath360 , as claimed .
the following lemma will allow us to prove proposition [ height is nonnegative ] in the case that a fiber has multiplicative reduction .
the proof is lengthy , but it is merely a collection of computations using the explicit formulas for the local height functions , as in ( * ? ? ?
* theorem [email protected] , [email protected] ) .
[ bad fibers ] let @xmath78 be an elliptic surface and let @xmath142 be a non - zero section defined over @xmath1 . then there exists a finite extension @xmath370 of the number field @xmath1 so that , for each @xmath175 such that @xmath322 has multiplicative reduction , there exists an @xmath371 so that @xmath372 at all places @xmath51 of @xmath370
. we let @xmath373 be a minimal weierstrass equation for @xmath18 over an affine subset @xmath374 defined over @xmath1 with @xmath375 .
here @xmath376 are regular functions at @xmath177 . using this weierstrass equation we write @xmath377 where @xmath378 .
since @xmath78 has multiplicative reduction over @xmath379 , we have @xmath380 let @xmath51 be a place of @xmath1 ( archimedean or non - archimedean ) .
we denote by @xmath381 the @xmath382invariant of @xmath383 , given by @xmath384 notice that equation yields that @xmath381 has a pole at @xmath177 .
hence , we can find a @xmath385adic open neighborhood @xmath185 of @xmath177 and an analytic map @xmath386 such that the following holds : if @xmath387 is the modular @xmath382invariant ( * ? ? ?
* chapter slowromancap5@ ) , then we have @xmath388 the function @xmath389 is given as @xmath390.\end{aligned}\ ] ] in the following , we choose a uniformizer @xmath391 at @xmath177 , and we identify @xmath392 with its expression @xmath393 $ ] and write @xmath394 equation yields that @xmath395 and @xmath396 $ ] . following the proof of ( * ? ? ?
* section 6 ) and after possibly shrinking @xmath185 we have isomorphisms @xmath397 for @xmath398 . under these isomorphisms , we have @xmath399 here @xmath400 are the modular invariants , given by their usual @xmath401series @xmath402 and @xmath403 is the weierstrass @xmath404function given by @xmath405 in view of ( * ? ? ?
* lemma [email protected] ) , after possibly replacing @xmath12 by @xmath406 , we may assume that @xmath407 is an analytic map satisfying @xmath408 in the following we identify @xmath409 with its series in @xmath410 $ ] and write @xmath411 where @xmath412 and @xmath413 $ ] .
we claim that @xmath414 $ ] . to see this ,
notice that from ( * ? ? ?
* slowromancap3@ ) we have that for @xmath284 @xmath415 where @xmath416 in the equation above @xmath121 denotes the modular discriminant given by @xmath417 since the functions @xmath418 and @xmath121 are defined over @xmath1 , we have that @xmath419 is also defined over @xmath1 . since @xmath420
$ ] and @xmath421 $ ] we get that @xmath422 $ ] .
therefore , there are non - zero constants @xmath423 , non - negative integers @xmath424 and functions @xmath425 $ ] such that for all @xmath284 @xmath426 next , we aim to express @xmath427 ( as in the statement of the lemma ) in terms of @xmath428 . using the isomorphisms in [ uni - iso ] , the uniqueness of the local canonical heights and the explicit formulas for the local canonical heights ( * ? ? ?
* theorem [email protected] , [email protected] ) , we get @xmath429 where @xmath430 is the second bernoulli polynomial . since @xmath431 and using ,
we get @xmath432 in what follows , for @xmath433 $ ] we write @xmath434 in view of ( * ? ? ?
* lemma [email protected] ) , we have @xmath435 using equations and , equation yields @xmath436 finally , notice that ( * ? ? ?
* theorem [email protected] ) implies @xmath437 therefore @xmath438 where @xmath439 belongs to a finite extension of @xmath1 , denoted by @xmath370 . by (
* lemma [email protected] ) there is a finite map of smooth projective curves @xmath440 such that if @xmath347 is a minimal model for @xmath441 , then @xmath442 has semi - stable reduction over the singular fibers of @xmath78 .
moreover , we may choose @xmath80 so that everything is defined over @xmath1 .
thus , by lemma [ reductionlemma ] and using the product formula , we may assume that the singular fibers of our elliptic surface @xmath14 have multiplicative reduction . for all @xmath7
for which @xmath15 is smooth , we know from proposition [ heightvsheight ] that @xmath443 .
the canonical height is always non - negative , so we may conclude that @xmath444 for all such @xmath24 .
assume now that @xmath180 has a fiber with multiplicative reduction .
enlarging the number field @xmath1 if necessary we may assume that @xmath379 and that its corresponding @xmath427 defined in the statement of lemma [ bad fibers ] is in @xmath445 .
then , on using the product formula , lemma [ bad fibers ] implies that @xmath446 .
this completes the proof .
let @xmath447 be the line bundle on @xmath2 induced from the divisor @xmath150 . from theorem [ metric properties
] , we know that its @xmath306-th tensor power can be equipped with a continuous , adelic , semipositive metric , so that the corresponding height function is ( a multiple of ) the canonical height @xmath448 on the smooth fibers . thus , by pulling back the metric to @xmath447 , we obtain a continuous , semipositive , adelic metric on @xmath447 inducing the desired height function .
it remains to show that this height @xmath316 satisfies @xmath449 .
this is a consequence of propositions [ heightvsheight ] and [ height is nonnegative ] and zhang s inequalities on successive minima @xcite .
recall that , since @xmath40 , we know that there are infinitely many @xmath21 for which @xmath450 ( for a complex - dynamical proof , see ( * ? ? ? * proposition 1.5 , proposition 2.3 ) . ) therefore , from proposition [ heightvsheight ] , we may deduce that the essential minimum of @xmath316 on @xmath2 is equal to 0 .
on the other hand , from proposition [ height is nonnegative ] , we know that @xmath451 for all @xmath7 .
therefore , from ( * ? ? ?
* theorem 1.10 ) , we may conclude that @xmath449 . when combined with the equidistribution theorems of yuan and thullier @xcite
, we immediately obtain the corollary from theorem [ good height ] .
the measures @xmath50 are the curvature distributions associated to the metrics @xmath314 at each place @xmath51 . from the definition of the metric in [ metric definition ]
, we see that they are given locally by @xmath452 in a @xmath51-adic neighborhood of any point @xmath175 , and for any choice of uniformizer @xmath181 at @xmath177 .
we first show that , to prove the theorem , it suffices to prove the result for sections of the fiber product @xmath453 of elliptic surfaces @xmath78 and @xmath454 over the same base , and to assume that the line bundle @xmath42 is generated by the divisor @xmath455 .
let @xmath2 be a quasiprojective smooth algebraic curve defined over @xmath59 .
suppose @xmath84 is a non - isotrivial and non - simple family of abelian varieties of dimension 2 defined over @xmath59 . from the poincar reducibility theorem @xcite , we know that @xmath70 is isogenous to a fibered product of elliptic curves . that is , there is a branched cover @xmath69 and a pair of elliptic surfaces @xmath456 and @xmath347 that give rise to an isogeny @xmath457 over @xmath80 .
now let @xmath42 be a line bundle on @xmath70 which restricts to an ample and symmetric line bundle on each fiber @xmath85 for @xmath458 .
then the line bundle @xmath42 pulls back to a line bundle @xmath459 on @xmath460 , and it again restricts to an ample and symmetric line bundle on each fiber over @xmath461 .
now suppose that we have a section @xmath87 .
the section @xmath12 pulls back to a section @xmath462 , and this in turn pulls back to a ( possibly multi - valued ) section of @xmath460 . if multi - valued , we can perform a base change again , passing to a branched cover @xmath463 , so that the induced section @xmath464 is well defined . by definition , the assumption that @xmath12 is non - special on @xmath70 means that it is non - special as a section of @xmath465 . finally , we observe that the conclusion of theorem [ zhang language ] does not depend on the choice of line bundle .
( we thank joe silverman for his help with this argument . ) recall that , on any abelian variety defined over @xmath59 , the notion of a small sequence " of points is independent of the choice of ample and symmetric line bundle .
that is , if we take two ample and symmetric divisors @xmath466 and @xmath467 , then we know that there exists an integer @xmath468 so that @xmath469 is ample ; similarly there exists @xmath470 so that @xmath471 is ample .
it follows from properties of the weil height machine that the heights @xmath472 and @xmath473 will then satisfy @xmath474 for real constants @xmath475 . upon passing to the canonical height ,
we conclude that @xmath476 on the abelian variety .
in particular , @xmath477 for some sequence in @xmath478 if and only if @xmath479 . now suppose we have a family of abelian varieties @xmath13 .
two line bundles @xmath480 and @xmath481 associated to relatively ample and symmetric divisors induce a canonical heights @xmath482 and @xmath483 on each fiber @xmath85 . but recalling that amplitude persists on zariski open sets ( * ? ? ?
* theorem 1.2.17 ) , there exist positive integers @xmath484 and @xmath485 so that the line bundles @xmath486 and @xmath487 are relatively ample on a zariski open subset of the base @xmath2 . passing to the canonical heights once again , we find that the relation holds uniformly over @xmath2 ( after possibly excluding finitely many points ) .
therefore , for any section @xmath87 , there exists a positive constant @xmath488 of theorem [ zhang language ] for height @xmath489 if and only if it there exists such a constant @xmath490 for @xmath491
. now let @xmath57 and @xmath58 be a pair of elliptic surfaces over the same base curve @xmath2 , defined over @xmath59 .
let @xmath12 be a section of the fiber product @xmath492 , so that @xmath12 can be expressed as @xmath61 with @xmath493 for @xmath494 .
let @xmath42 be the line bundle on @xmath495 associated to the divisor @xmath496 .
the canonical height @xmath497 on each fiber @xmath85 is easily seen to be the sum of canonical heights ( see , e.g. , @xcite for properties of the height functions ) , so that @xmath498 now assume that @xmath61 is not a special section .
define @xmath499 for @xmath500 and for all @xmath72 where both @xmath501 and @xmath502 are smooth elliptic curves .
it suffices to show that there is no infinite sequence @xmath503 for which @xmath504 as @xmath48 .
we have assumed that @xmath12 is not special , so neither @xmath505 nor @xmath506 is identically 0 . if @xmath63 is isotrivial , and if @xmath62 is constant non - torsion on @xmath63 , then @xmath507 for all @xmath24 and we are done .
thus , we may assume that the function - field heights @xmath508 are nonzero for @xmath500 .
let @xmath509 and @xmath510 be the adelically metrized line bundles on the base curve @xmath2 associated to the height functions @xmath505 and @xmath506 , from theorem [ good height ] .
these metrized line bundles @xmath511 are both equipped with continuous adelic metrics of non - negative curvature .
suppose there exists a sequence @xmath512 such that holds . in theorem [ good height
] , we showed that @xmath513 .
therefore , we may apply the observation of chambert - loir ( * ? ? ?
* proposition 3.4.2 ) , which builds upon on zhang s inequalities @xcite , to conclude that there exist integers @xmath514 and @xmath515 so that @xmath516 and @xmath517 are isomorphic as line bundles on @xmath2 and their metrics are scalar multiples of one another .
it follows that the height functions @xmath505 and @xmath506 are the same , up to scale , and in particular they have the same zero sets . in other words
, @xmath518 is a torsion point on @xmath501 if and only if @xmath519 is a torsion point on @xmath502 ( for all but finitely many @xmath24 in @xmath2 ) .
we have therefore reduced our problem to the main results of @xcite , and we conclude that @xmath61 must have been a special section . as this was ruled out by assumption , we deduce that there exists a constant @xmath520 so that the set @xmath521 is finite .
this concludes the proof of theorem [ zhang language ] .
in this final section , we provide a few illustrations of the distributions @xmath50 for an archimedean place @xmath51 , arising in corollary [ equidistribution on b ] . in proposition [ dense ] , we present a complex - dynamical proof that the archimedean measures @xmath50 will have support equal to all of @xmath2 .
given @xmath78 and section @xmath12 , we plot the parameters @xmath24 where @xmath8 is a torsion point on the fiber @xmath15 of specified order . as proved in corollary [ equidistribution on b ] , the local height function at each place @xmath522 determines the distribution of the torsion parameters ; it is a potential for the measure @xmath50 ( away from the singularities ) . recall that if we have two sections @xmath12 and @xmath523 that are linearly related on @xmath18 , then the distributions of their torsion parameters in @xmath2 will be the same .
figure [ example 1 ] , top , illustrates the example of silverman from ( * ? ? ?
* theorem i.0.3 ) . here
, we have @xmath524 with @xmath525 and @xmath526 in @xmath527-coordinates .
plotted are the torsion parameters of orders @xmath528 for all @xmath529 ; that is , the points @xmath24 in the base @xmath2 where @xmath8 is torsion of order @xmath528 on the fiber @xmath15 .
roughly , a smaller yellow dot corresponds to higher order of torsion .
figure [ example 1 ] , bottom , is another section of the same family , where the @xmath100-coordinate of @xmath8 is constant and equal to @xmath530 .
( strictly speaking , this second @xmath12 is not a section of our given @xmath531 , because the @xmath532-coordinate will not lie in @xmath533 but in an extension ; however , the property of being torsion and the determination of its order is independent of which point in the fiber we choose . )
observe the distinctly different pattern of yellow dots in the first and second pictures , especially in the left half of the two pictures , illustrating the linear independence in @xmath534 of the two sections .
figure [ example 2 ] illustrates the torsion parameters for two independent sections of the legendre family , @xmath535 over @xmath525 , studied in @xcite .
the chosen sections are @xmath75 , with constant @xmath100-coordinate equal to @xmath536 , and @xmath537 , with constant @xmath100-coordinate equal to 5 . as in figure
[ example 1 ] , we plot the torsion parameters of orders @xmath528 for all @xmath529 ; generally , a smaller yellow dot signifies higher order of torsion .
it was proved in @xcite that the limiting distributions for sections with constant @xmath100-coordinate satisfy @xmath538 ( at an archimedean place ) if and only if @xmath539 .
it was proved in @xcite and @xcite that there are no @xmath540 for which both @xmath541 and @xmath542 are torsion on @xmath15 .
again , observe the difference in the geometry of the yellow dots for the two independent sections .
figure [ example equidistribution ] illustrates our equidistribution result , corollary [ equidistribution on b ] , for the example of the legendre family with the section @xmath537 .
plotted are the torsion parameters of orders @xmath528 with ( a ) @xmath543 , ( b ) @xmath529 , and ( c ) @xmath544 .
observe how the yellow dots fill in the grid structure " in the base curve @xmath2 , exactly as do the torsion points for one elliptic curve .
[ h ] and @xmath526 , shown in the region @xmath545 .
the singular fibers occur at @xmath546 , and one sees the effects of numerical error in a small neighborhood of these parameters . at bottom ,
torsion parameters for section @xmath12 having @xmath100-coordinate @xmath547 for all @xmath24.,title="fig:",width=432 ] and @xmath526 , shown in the region @xmath545 .
the singular fibers occur at @xmath546 , and one sees the effects of numerical error in a small neighborhood of these parameters . at bottom , torsion parameters for section @xmath12 having @xmath100-coordinate @xmath547 for all @xmath24.,title="fig:",width=432 ] [ h ] and @xmath97 . at
left , the section @xmath75 with @xmath548 for all @xmath24 ; at right , the section @xmath537 with @xmath549 for all @xmath24
. both are shown in the region @xmath550 , title="fig:",width=288 ] and @xmath97 . at left ,
the section @xmath75 with @xmath548 for all @xmath24 ; at right , the section @xmath537 with @xmath549 for all @xmath24
. both are shown in the region @xmath550 , title="fig:",width=288 ] [ h ] and @xmath97 and @xmath537 as in figure [ example 2 ] . at top ,
torsion parameters of orders @xmath528 for @xmath551 ; bottom left , of orders @xmath528 for @xmath529 , and bottom right , of orders @xmath528 for @xmath544 .
, title="fig:",width=288 ] + and @xmath97 and @xmath537 as in figure [ example 2 ] . at top ,
torsion parameters of orders @xmath528 for @xmath551 ; bottom left , of orders @xmath528 for @xmath529 , and bottom right , of orders @xmath528 for @xmath544 .
, title="fig:",width=288 ] and @xmath97 and @xmath537 as in figure [ example 2 ] . at top ,
torsion parameters of orders @xmath528 for @xmath551 ; bottom left , of orders @xmath528 for @xmath529 , and bottom right , of orders @xmath528 for @xmath544 .
, title="fig:",width=288 ] as mentioned above , the smaller yellow dots in the illustrations correspond , roughly , to higher orders of torsion .
these images are produced with a standard escape - rate algorithm .
we use the dynamical system @xmath262 on @xmath125 , induced from multiplication by 2 on the elliptic curve @xmath15 from section [ dynamics ] , line . the coordinates on @xmath125
are chosen so that @xmath552 is the image of the 0 of @xmath15 .
we mark @xmath24 yellow if @xmath553 for some @xmath529 . in all of these examples
, the yellow dots will fill in the picture as the order of torsion grows , and the support of the measures @xmath50 is equal to @xmath9 .
in fact , this will always be the case , for any ( nontrivial ) section of a complex elliptic surface , as our final result , proposition [ dense ] , shows .
let @xmath78 be an elliptic surface over a projective curve @xmath2 , defined over @xmath554 , and let @xmath192 a section for which @xmath40 ( over the function field @xmath555 ) .
let @xmath556 be the measure on @xmath2 defined as in proposition [ measure description ] , as the pullback of the current @xmath557 that restricts to haar measure on each smooth fiber .
in other words , @xmath556 is locally defined as the laplacian of the function @xmath270 introduced in proposition [ escape rate subharmonic ] , which is well defined when working over @xmath554 .
[ dense ] let @xmath78 be an elliptic surface over a projective curve @xmath2 , defined over @xmath554 , and let @xmath192 be a section for which @xmath40 ( over the function field @xmath555 ) .
then the set @xmath558 is dense in @xmath9 and @xmath559 we give a complex - dynamical proof , viewing proposition [ dense ] as a consequence of the main result of @xcite .
( we do not use the equidistribution result , corollary [ equidistribution on b ] . ) an analytic proof is also presented in ( * ? ? ?
* notes to chapter 3 ) .
let @xmath560 be a finitely - punctured riemann surface such that the fiber @xmath15 is smooth for all @xmath561 .
let @xmath562 be the degree - two projection and @xmath563 be the rational map induced by multiplication - by-2 on @xmath15 , as defined in the introduction to section [ dynamics ] .
it is well known that the holomorphic family @xmath564 is structurally stable ; see , e.g. , ( * ? ? ?
* chapter 4 ) .
thus , over any simply - connected subset @xmath185 of @xmath565 , there is a holomorphic motion of the periodic points of @xmath262 which extends uniquely to a holomorphic motion of all of @xmath125 , conjugating the dynamics .
the key observation is that @xmath556 is precisely the bifurcation measure " of the pair @xmath566 on @xmath565 .
see @xcite and @xcite for definitions .
the support of @xmath556 is equal to the bifurcation locus of @xmath566 ; in particular , the parameters @xmath567 for which @xmath568 is preperiodic for @xmath262 are dense in @xmath569 .
therefore , it suffices to show that @xmath570 .
suppose to the contrary that there is an open disk @xmath571 for which @xmath572 .
then the pair @xmath566 is stable on @xmath185 , and therefore @xmath568 can not be a repelling periodic point for any @xmath285 . from the uniqueness of the holomorphic motion
, it follows that @xmath573 is part of the holomorphic motion on @xmath185 . by analytic continuity , then , we deduce that @xmath568 must follow the motion of a point over all of @xmath565 .
this implies that the pair @xmath566 is stable throughout @xmath565 and the measure @xmath556 is 0 .
but this is absurd by the assumption that @xmath40 ; see ( * ? ? ?
* theorem 1.1 ) .
antoine chambert - loir . .
motivic integration and its interactions with model theory and non - archimedean geometry . volume ii _
, volume 384 of _ london math .
lecture note ser .
_ , pages 150 .
cambridge univ .
press , cambridge , 2011 .
|
let @xmath0 be an elliptic surface defined over a number field @xmath1 , where @xmath2 is a smooth projective curve , and let @xmath3 be a section defined over @xmath1 with canonical height @xmath4 . in this article , we show that the function @xmath5 on @xmath6 is the height induced from an adelically metrized line bundle with non - negative curvature on @xmath2 .
applying theorems of thuillier and yuan , we obtain the equidistribution of points @xmath7 where @xmath8 is torsion , and we give an explicit description of the limiting distribution on @xmath9 .
finally , combined with results of masser and zannier , we show there is a positive lower bound on the height @xmath10 , after excluding finitely many points @xmath11 , for any non - special " section @xmath12 of a family of abelian varieties @xmath13 that split as a product of elliptic curves .
|
artificially engineered _ metamaterials _ have unique electromagnetic ( em ) properties , which are of great interest both from the basic physics viewpoint and for numerous applications @xcite . in such media ,
the effective permittivity @xmath0 and permeability @xmath1 is such that , in certain frequency bands , the metamaterial displays either a right - handed ( rh ) behavior ( @xmath2 , @xmath3 ) or a left - handed ( lh ) behavior ( @xmath4 , @xmath5 ) ; in other words , energy and wavefronts may travel in the same or opposite directions in the rh or the lh frequency region , respectively . metamaterials with a lh behavior , namely lh - metamaterials ( lhm ) , exhibit negative refraction at microwave @xcite or optical frequencies @xcite .
apart from the effective medium description , there has also been a large interest in studying equivalent transmission line ( tl ) models for lhms in the microwave frequency region .
in such a case , the effective permittivity @xmath6 and permeability @xmath1 are directly connected to the serial and shunt impedance of the tl model ; this way , so - called , _ composite right - left handed _ ( crlh ) tls @xcite may exhibit either a rh or a lh behavior depending on the frequency band . in practice , crlh - tls
may be implemented with the coplanar waveguide ( cpw ) @xcite or microstrip technology @xcite .
importantly , crlh - tls have led to many microwave applications and devices , including dual - band branch - line couplers , asymmetric backward - wave directional couplers , resonators , antennas , and so on @xcite . on the other hand , _ nonlinear _ metamaterials , namely structures
in which @xmath0 and @xmath1 ( or the serial and shunt impedance in the respective tl models ) depend on the em field intensities ( or voltages and currents in the tl models ) , have also been a subject of interest .
such structures may be implemented by embedding an array of wires and split ring resonators ( srrs ) into a nonlinear dielectric @xcite , or by inserting diodes into resonant conductive elements ( such as the srrs ) @xcite .
nonlinear metamaterials may prove useful in various applications , including `` switching '' the material properties from left- to right - handed and back , tunable structures with intensity - controlled transmission , negative refraction photonic crystals , etc .
furthermore , fundamental effects , such as harmonic generation , nonlinearity - induced localization of em waves and soliton formation , are possible . more specifically , relevant nonlinear phenomena @xcite and soliton formation
have already been predicted to occur in nonlinear metamaterials , using either the effective medium description ( see , e.g. , refs .
@xcite ) or the tl description ( see , e.g. , refs . @xcite ) . from the viewpoint of nonlinear tl experiments , pulse propagation @xcite and envelope soliton formation @xcite
were recently observed ( see also the review of ref .
@xcite and the relevant work of ref .
@xcite ) ; moreover , analytical approximations , based on a continuum nonlinear schrdinger ( nls ) equation , allowed the description of bright @xcite or dark @xcite envelope solitons observed in the experiments . in this work , we study _ quasi - discrete _ microwave solitons that may be formed in a _
planar _ lhm .
such structures have the advantage of being easily fabricated ( by means of standard mask / photoetching techniques ) , they are compact , and are compatible with monolithic microwave integrated circuits . at this point
we should mention that , generally , metamaterial tls , operating at microwave frequencies , are artificial lines consisting of a host line loaded with reactive elements .
such tls can be implemented by means of two main approaches : ( i ) the so - called `` cl - loaded approach '' , where rh - tls are loaded with series capacitances and shunt inductances @xcite ( see also ref .
@xcite for results in a nonlinear case ) , and ( ii ) the so - called `` resonant - type approach '' , where the tls are loaded with sub - wavelength resonators , such as srrs @xcite .
tls of the latter type exhibit controllable electrical characteristics , beyond what can be achieved in conventional tls , implemented , e.g. , in printed circuit boards ( pcbs ) : the size of such a tl is determined by the size of the resonators and , thus , they can easily be miniaturized .
note that the fabrication of a pertinent prototype device , based on a cpw with an array of srrs being etched at the bottom of the substrate , was first introduced in ref .
@xcite , while the corresponding tl model was presented in ref .
@xcite . here
, we consider a nonlinear counterpart of the tl model of ref .
@xcite , with the shunt capacitors of the model being nonlinear ; such a nonlinear tl model may be implemented by incorporating a nonlinear dielectric thin film in the structure , whose dielectric constant may be controlled by a proper bias voltage ( see , e.g. , refs .
@xcite ) . in our analysis , starting from the discrete lump element model of the crlh - tl under consideration , we derive a nonlinear lattice equation , which is then treated in the framework of the _ quasi - discrete _ ( alias _ quasi - continuum _ ) approximation ( see , e.g. , ref . @xcite and ref .
@xcite for a review ) : this way , seeking for envelope soliton solutions of the nonlinear lattice model , characterized by a _ discrete carrier _ and a _ continuum envelope _
, we employ a multi - scale perturbation method to derive an effective nls equation .
the coefficients of this equation , which determine the type ( bright or dark ) of the envelope soliton , are found and it is shown that bright nls solitons are supported by the srr - cpw structure in a relatively wide range of frequencies inside the lh frequency band .
our analytical predictions are corroborated by numerical simulations , which reveal ( apart from the basic properties ) the robustness of the predicted quasi - discrete microwave solitons . and width @xmath7 is linear , while the one of relative dielectric constant @xmath8 and width @xmath9 is nonlinear .
bottom panel : layout of the bottom plate of the configuration , where the srrs are placed . shown
are all relevant dimensions appearing in the equations of appendix a.,title="fig:",width=302 ] and width @xmath7 is linear , while the one of relative dielectric constant @xmath8 and width @xmath9 is nonlinear .
bottom panel : layout of the bottom plate of the configuration , where the srrs are placed . shown
are all relevant dimensions appearing in the equations of appendix a.,title="fig:",width=302 ] the paper is organized as follows : in sec .
ii we present the srr - cpw nonlinear structure and derive a nonlinear lattice model describing the evolution of the voltage across this structure ; then we employ the quasi - discrete approximation and derive the nls equation describing envelope solitons that can be supported in this setting . in sec .
iii we present numerical simulations in the framework of the nonlinear lattice model , reveal the propagation properties of the predicted quasi - discrete solitons , and compare the numerical results with the analytical predictions .
we also propose changes in the configuration to increase the domains of existence of solitons . finally , in sec .
iv we discuss our conclusions .
our presentation is complemented by three appendices : the first details the parameters of the srr - cpw model , the second analyzes the method of multiple scales used to obtain the relevant nls equation , while the third touches upon the continuum limit of the quasi - discrete approximation developed herein .
we start by considering a nonlinear version of the srr - cpw model introduced in refs .
@xcite , as shown in the top panel of fig .
[ fig : circuit model ] : here , the structure incorporates a nonlinear dielectric film , of relative dielectric constant @xmath8 and width @xmath9 .
the rest of the srr - cpw configuration is identical to the one of refs .
@xcite : one may observe the srrs ( of external radius @xmath10 ) at the bottom plate of the structure , which are aligned with the slots ( of width @xmath11 and separation distance @xmath12 ) at the top plate of the structure .
notice that below we will focus on the case where the nonlinear dielectric film is introduced as shown in the top panel of fig .
[ fig : circuit model ] ; according to this consideration , nonlinearity is only introduced in certain elements ( i.e. , shunt capacitors ) in the equivalent discrete unit - cell model of the system .
nevertheless , below we will first consider the case where the serial capacitors , associated with the srrs , are also nonlinear ; in practice , this can be done , e.g. , by inserting diodes in the srr slots @xcite . the discrete element model ( unit cell ) of the considered srr - cpw structure
is shown in fig .
[ fig : circuit model ] . here ,
@xmath13 and @xmath14 denote the equivalent per section inductance and capacitance of the line , respectively , @xmath15 and @xmath16 are the equivalent inductance and capacitance of the srr , which is coupled with the transmission line , while the inductance @xmath17 is the equivalent inductance of the shunt strips .
the above elements are directly connected with the physical parameters of the srr - cpw structure ( see , e.g. , ref .
@xcite and details in appendix a ) .
let us now consider kirchhoff s voltage and current laws for the srr ( see the @xmath15@xmath16 combination in fig .
[ fig : circuit model ] ) equivalent circuit , namely , @xmath18 the above equations , together with the auxiliary kirchhoff s current law @xmath19 , lead to the following system for the unknown voltages @xmath20 and @xmath21 : @xmath22 \nonumber \\ & & + \left(1+\frac{l_{r}}{{l_s}}\right ) ( u_{n}-u_{n+1 } ) + ( v_{n-1}-2v_{n}+v_{n+1})=0 , \nonumber \\ \label{eq : sys01 } \\ & & l_{r}\frac{d^{2}}{dt^{2}}[c(v_{n})v_{n}]-(v_{n-1}-2v_{n}+v_{n+1})\nonumber \\ & & + \frac{l_{r}}{l}v_{n}- ( u_{n}-u_{n+1})=0 , \label{eq : sys02}\end{aligned}\ ] ] where @xmath23 and @xmath24 . at this point , in order to further simplify eqs .
( [ eq : sys01])-([eq : sys02 ] ) , we need to make some considerations concerning the nonlinear ( voltage - dependent ) capacitance @xmath25 and @xmath26 .
as discussed above , @xmath27 is effectively filled by a nonlinear dielectric , whose dielectric constant may be controlled by a proper bias voltage , say @xmath28 ( see , e.g. , refs . @xcite and discussion below ) .
on the other hand , the effective nonlinearity of @xmath29 is due to the presence of a diode , biased at a constant voltage , say @xmath30 .
assuming that the voltages @xmath20 and @xmath21 , respectively applied in the effective capacitors @xmath27 and @xmath29 , do not change significantly from their relevant bias voltages @xmath31 and @xmath32 , we can taylor expand @xmath33 and @xmath26 taking into regard only the lowest order terms , namely , @xmath34 where @xmath35 and @xmath36 are constant capacitances corresponding to @xmath28 and @xmath30 , respectively , and similarly @xmath37 , @xmath38 , take constant values depending on the particular forms of @xmath39 and @xmath40 ( see also below ) .
next , substituting eqs .
( [ eq : taylor1])-([eq : taylor2 ] ) into eqs .
( [ eq : sys01])-([eq : sys02 ] ) , and introducing the scale transformations @xmath41 [ where @xmath42 and @xmath43\{v_n , u_n \}$ ] , we cast the system of eqs .
( [ eq : sys01])-([eq : sys02 ] ) into the following form : @xmath44 and @xmath45 where the constant parameters are given by : @xmath46 in the above expressions , @xmath47 and @xmath48 denote series and shunt frequencies , while @xmath49 and @xmath50 denote characteristic frequencies related to the srr and the rh part of the circuit , respectively ; the above frequencies read : @xmath51 the system of eqs .
( [ eq : mod01])-([eq : mod02 ] ) is a model of two coupled nonlinear lattice equations , for the unknown voltages @xmath52 and @xmath21 , describing the dynamics of the system .
the analysis of the above system by means of analytical ( and/or even numerical ) techniques is a far more involved task and will be deferred to a future study .
in fact , we choose to consider a variant of the above model , which is analytically ( and numerically ) more tractable , corresponding to the case where only the capacitance @xmath27 ( associated with the cpw structure ) is nonlinear , while the capacitance @xmath29 is linear ( i.e. , nonlinearity in the considered srr - cpw structure is only introduced by the insertion of the nonlinear dielectric film , as shown in fig . [ cpwsrr ] ) . in such a case ,
the parameter @xmath53 ( governing the srr nonlinearity ) becomes @xmath54 , and the system of eqs .
( [ eq : mod01])-([eq : mod02 ] ) can be decoupled , leading to a single nonlinear lattice equation for the function @xmath52 , namely : @xmath55 it is now useful , for the purposes of our analytical and numerical considerations , to adopt experimentally relevant parameter values .
first , following ref .
@xcite , the parameters related to the cpw structure are chosen as follows .
the waveguide has a width @xmath56 mm and a thickness @xmath57 mm , the central strip width is @xmath58 mm , the width of the slots is @xmath59 mm , while the main linear dielectric substrate ( namely a rogers ro3010 ) , of width @xmath60 mm , has a relative dielectric constant @xmath61 .
the values for the srr characteristics are also borrowed from ref .
@xcite and are assumed to take the following values .
the internal radius is @xmath62 mm , the distance between the rings is @xmath63 mm and the rings width is @xmath64 mm . as far as the nonlinear dielectric film is concerned , following ref .
@xcite , we have assumed a strontium barium titanate ( sbto ) paraelectric thin film , of width @xmath65 m , and relative dielectric constant @xmath66 . using the above parameter values , we may determine ( as per the relevant equations provided in appendix a ) the values of the effective capacitances and inductances involved in the srr - based cpw structure .
this way , we find that @xmath67 nh , @xmath68 nh , @xmath69 nh , @xmath70 pf , and @xmath71 pf . notice that the value of @xmath72 is also obtained consistently from the effective voltage - dependent capacitance of ref .
@xcite , namely : @xmath73 , \label{cainonlin}\ ] ] where @xmath74 pf , @xmath75 , @xmath76 v@xmath77 , while @xmath78 v is the constant ( dc ) bias voltage .
according to the above expression , the values of the constant coefficients in eq .
( [ eq : taylor1 ] ) are given by @xmath79 = 2.44 $ ] pf [ identical to the result of eq .
( [ cr ] ) ] , and @xmath80 pf / v . according to the above , the frequencies in eq .
( [ eq : freq ] ) , which are in the microwave regime , take the values @xmath81 ghz , @xmath82 ghz , @xmath83 ghz , and @xmath84 ghz ; accordingly , the values of the normalized parameters @xmath85 , @xmath86 and @xmath87 appearing in eq .
( [ eq : model ] ) take the following values : @xmath88 thus , in our simulations ( see sec .
iii.a ) , we will use the above values to investigate quasi - discrete microwave solitons in this setting , and also discuss possible modifications of the considered setup in order to study how relevant changes in parameter values affect the domains of existence of these nonlinear structures ( see sec .
iii.b ) . notice that for the above mentioned choice of the physical parameters , the time unit associated to eq .
( [ eq : model ] ) is @xmath89 picoseconds , while the voltage unit is @xmath90 volts . before proceeding further
, we should note the following . as indicated by eq .
( [ eq : taylor1 ] ) , in our considerations we take into regard only the first order approximation in the @xmath33 dependence , while this restriction may not be accurate enough for artificial nonlinearities ( induced , e.g. , by inserting nonlinear elements , such as diodes , in the srr slots see ref .
@xcite and @xcite for relevant theoretical and experimental studies , respectively ) .
nevertheless , in our case , where the nonlinearity is induced by the insertion of the nonlinear dielectric thin film ( see fig .
[ cpwsrr ] ) , the considered approximation is quite reasonable : indeed , in our numerical simulations ( see sec .
iii ) we use a value for the initial voltage equal to @xmath91 v ( in physical units ) .
for such a value of the voltage ( similar , and even smaller , values have also been used in relevant experimental works @xcite ) , it can be found that @xmath92 pf , i.e. , @xmath93 higher than the value of @xmath94 pf . if we had taken into regard the quadratic term in the taylor expansion in eq .
( [ eq : taylor1 ] ) , namely @xmath95 , this term would take the value @xmath96 pf , i.e. , only @xmath97 higher than the above mentioned value of @xmath98 pf ( that corresponds to the lowest order of approximation ) .
thus , according to the above arguments , and given that the initial voltage value @xmath31 can also be controlled by other additional parameters stemming from our analysis [ see parameters @xmath0 and @xmath99 in eq .
( [ v0 ] ) below ] , we will proceed by analyzing eq .
( [ eq : model ] ) that takes into account only the first order approximation in the @xmath33 dependence [ as per eq .
( [ eq : taylor1 ] ) ] . in this section
, we will employ the quasi - discrete approximation ( see , e.g. , refs .
@xcite and appendix b ) .
generally , this approach is a variant of the multi - scale perturbation method , which is a well - known powerful tool to derive effective evolution equations ( valid under certain conditions and for appropriate spatial and temporal scales ) that are much simpler than the original models @xcite . in our case , since our original model [ cf .
( [ eq : model ] ) ] is actually a nonlinear dynamical lattice , we adopt the quasi - discrete approximation due to the fact that it takes into regard the discreteness of the system : this approach , allows for the description of quasi - discrete envelope solitons ( satisfying an effective nls model ) , which are characterized by a discrete carrier and a slowly - varying continuum pulse - like envelope . notice that , alternatively , one could adopt a continuum approximation , i.e. , take the continuum limit of eq .
( [ eq : model ] ) and analyze the latter in the framework of a multi - scale perturbation scheme ( as in refs .
@xcite ) . however , as we will show below ( and as was the case in ref .
@xcite ) , the quasi - discrete approximation is more accurate than the continuum one on providing estimates for the domains of existence of envelope solitons . furthermore , the analytically determined soliton profile and characteristics ( such as the center of mass and width ) will be found to be in good agreement with direct simulations obtained in the framework of the original lattice model of eq .
( [ eq : model ] ) see sec .
iii.a below .
we seek for solutions of eq .
( [ eq : model ] ) in the form : @xmath100 where @xmath101 ( @xmath102 ) are unknown envelope functions depending on the slow scales @xmath103 ( where @xmath104 is the group velocity , to be determined in a self - consistent manner ) and @xmath105 ; here , @xmath106 is a formal small parameter related to the soliton amplitude ( see below ) .
additionally , the function @xmath107 , with @xmath108 ( with @xmath109 and @xmath110 denoting frequency and wavenumber , respectively ) describes the carrier . in the above ansatz ,
the envelope ( carrier ) is obviously continuous ( discrete ) in space ; the results obtained in the framework of the quasi - discrete approximation , may be directly viewed in the continuum limit ( and would correspond to the continuum approximation ) of @xmath111 ; see also appendix c. substituting eq .
( [ eq : ansatz ] ) into eq .
( [ eq : model ] ) we obtain the following results ( see more details in appendix b ) .
first , to order @xmath112 ) ( linear limit ) , we derive the following dispersion relation : @xmath113 ( in ghz ) as a function of the wavenumber @xmath110 ( in rad / cell ) . solid ( blue ) lines and dashed ( red ) lines correspond to the quasi - discrete and continuum approximations [ cf .
( [ eq : dispqc ] ) and ( [ eq : dispc ] ) , respectively ] .
there exist a high - frequency rh band , for @xmath114 ghz @xmath115 ghz , and a low - frequency lh band , for @xmath116 ghz @xmath117 ghz ; the inset shows a magnification of the dispersion relation in the lh frequency band.,width=264 ] in fig .
[ fig : dispersion ] we plot the frequency @xmath118 ( in ghz ) as a function of the wavenumber @xmath110 ( in rad / cell ) .
it is clear that , apart from the shown gaps ( where em wave propagation is not supported ) , there exist two different frequency bands ( where propagation of em waves is possible ) : a high - frequency ( hf ) and a low - frequency ( lf ) band , with @xmath114 ghz @xmath115 ghz and @xmath116 ghz @xmath117 ghz , respectively .
it should be noticed that the lower ( upper ) cutoff frequency of the hf ( lf ) frequency band is equal to @xmath48 ( @xmath47 ) see eq .
( [ eq : freq ] ) .
obviously , in the hf ( lf ) band the product of the wavenumber @xmath110 and the group velocity @xmath119 is positive ( negative ) and , thus , energy and wavefronts travel in the same ( opposite ) directions in the hf ( lf ) frequency region .
thus , in the lf band ( which is clearly a lh frequency band ) the considered srr - cpw structure apparently behaves as a lh transmission line ; the dispersion relation in this lh frequency band , is shown in the inset of fig .
[ fig : dispersion ] .
note that in the continuum limit ( see pertinent results , for @xmath120 , in appendix c ) the lh frequency band under consideration becomes slightly wider , i.e. , it extends in the interval @xmath121 ghz @xmath117 ghz ; a similar effect is also observed for the rh band ( @xmath122 ghz ) .
next , proceeding to the next order [ @xmath123 in the perturbation scheme , we obtain the group velocity , given by : @xmath124 finally , to order @xmath125 , we obtain a nonlinear evolution equation for the unknown voltage @xmath126 , namely the following nls equation , @xmath127 with dispersion and nonlinearity coefficients , @xmath128 and @xmath129 respectively , given by the following expressions : @xmath130 and @xmath131\mathcal{g } } , \label{eq : nlcoef}\ ] ] where the function @xmath132 is given by : @xmath133 [ cols="^,^ " , ] next , keeping fixed the parameter values of the cpw structure as in sec .
ii.a ( in this case @xmath59 mm ) , we have studied changes in the srr geometry , namely the radius @xmath134 and width @xmath135 of the srrs ( for fixed spacing between the srrs , @xmath63 mm ) .
first , in the case with fixed @xmath64 mm , the increase ( decrease ) of the srr radius @xmath134 in the interval @xmath136 mm to @xmath137 mm results in a decrease ( increase ) of the values of parameters @xmath85 and @xmath1 , while the parameter @xmath87 takes the approximately constant value of @xmath138 .
accordingly , it is found that the width of the lh frequency band is increased ( decreased ) , while the central frequency regime where bright solitons can be formed is increased ( decreased ) as well . as an example , for @xmath139 mm ( recall that @xmath62 mm for the `` regular '' srr - cpw structure ) , the lh frequency band extends from @xmath140 ghz up to @xmath141 ghz , i.e. , it has increased @xmath142 .
furthermore , in the same case , the central frequency region where bright solitons can be formed is also increased by @xmath143 , with the frequency dependence of coefficient @xmath129 being similar to the one shown in the bottom panel of fig .
[ pr2 ] . in this case
, the parameters @xmath85 , @xmath1 and @xmath87 take the following values : @xmath144 numerical simulations for bright soliton propagation in this regime ( e.g. , we have performed long evolution runs for the frequency @xmath145 ghz ) lead to results qualitatively similar to the ones presented above ( cf . figs .
[ fig : simul1 ] , [ fig : simullong ] and [ slong2 ] ) .
we have studied analytically and numerically quasi - discrete microwave bright solitons that can be formed in the left - handed frequency band of a split ring resonator ( srr ) based coplanar waveguide ( cpw ) structure .
we have used the nonlinear transmission line analogue of this structure to derive a nonlinear lattice equation governing the voltage across the fundamental ( unit cell ) element of the system .
this lattice equation was then treated analytically , by means of the quasi - discrete approximation .
the latter is a variant of the multi - scale perturbation method , which takes into regard the discreteness of the system by considering the carrier ( envelope ) of the wave as a discrete ( continuum ) object .
this approach allowed us to derive , in the small - amplitude approximation and for certain space- and time - scales , a nonlinear schrdinger ( nls ) model for the unknown voltage envelope function .
the nls model was then used to predict formation of bright solitons in certain sub - regions of the left - handed frequency band .
importantly , the conditions for soliton formation were found to depend on the _ discreteness _ of the system : in fact , if the continuum instead of the quasi - discrete approximation was used , then the allowable ( for soliton formation ) frequency bands would be significantly modified . furthermore , the quasi - discrete approximation predicts effects , such as the appearance of resonance frequencies in the coefficients of the effective nls model , which suggest optimum operating frequency bands for the observation of quasi - discrete solitons . generalizing these observations
, one should expect that the adopted analytical approach , based on the quasi - discrete approximation , should provide more reliable results concerning conditions for soliton formation in left - handed structures .
numerical simulations performed in the framework of the nonlinear lattice equation , with initial conditions borrowed from the effective nls equation , revealed that the ( bright ) solitons may indeed be formed in the frequency sub - intervals predicted in the analytics . furthermore , it was shown that if the carrier frequency is chosen to be sufficiently far from characteristic frequencies ( where the dispersion and nonlinearity coefficients of the effective nls model vanish or have resonance poles ) , the numerically found soliton profile and characteristics ( center of mass and width ) are in very good agreement with the analytical predictions . although the numerical simulations were performed for relatively large computation times , the corresponding physical times were small due to the fact that all the characteristic frequencies of the system were in the microwave regime : under any choice , the physical time unit is less than a nanosecond and , thus , simulations corresponding to a few seconds would correspond to computation times of order of @xmath146 .
this difficulty , particular to the microwave structure under consideration , did not allow us to be definitively conclusive as concerns the robustness of the solitons for realistic experimental times . nevertheless , very long simulations ( corresponding to computing times @xmath147 or @xmath148 millisecond in physical units ) have shown that the predicted quasi - discrete microwave solitons are quite robust and do not appear to be modified during their propagation .
this , in turn , allows us to conjecture that these nonlinear wave structures have a good chance to be observed experimentally in the near future .
we have also presented a study showing how certain physical parameters of the configuration may affect the results .
in particular , we have focused on two cases , namely the effects of decrease of the width of the slots in the cpw structure , and of the increase of the srr radius .
these choices led to modified values of the parameters of the model ( @xmath87 , @xmath85 , and @xmath1 ) , such that the width of the left - handed frequency band or / and the domain of existence of quasi - discrete bright solitons are increased .
this way , we have also proposed certain experimentally relevant changes in the considered configuration , to facilitate observation of solitons in future experiments .
there are many interesting directions for future studies . in that regard ,
first , we should mention that in the analysis of the considered srr - cpw structure we have actually excluded the srr from nonlinearity .
it would be interesting to see what happens if srr is also subject to nonlinear modulation : this is certainly a very challenging direction , in terms of the development of both analytical and numerical techniques , due to the form of the pertinent models , which are coupled nonlinear lattice equations [ see , e.g. , eqs .
( [ eq : mod01])-([eq : mod02 ] ) ] .
such studies would lead to potentially interesting and relevant results concerning nonlinear wave propagation in such settings , as well as the design of nonlinear left - handed transmission lines ( and related structures ) . on the other hand
, we note that following the analytical procedure adopted in this work , it would be possible to analyze soliton formation in relevant left - handed structures . furthermore , employing this analytical approach , one could in principle derive self - consistently [ at @xmath149 in the perturbation scheme ] a generalized nls model , incorporating higher - order dispersive and nonlinear effects ( see , e.g. , ref .
such a higher - order nls model could also predict soliton propagation but , in this case , the soliton characteristics would be modified by the presence of the higher - order effects .
thus , an interesting challenge would be the derivation of such a higher - order nls model , the study of its soliton solutions and a comparison of such findings to direct simulations .
finally , while the theory and simulations presented in this work assume ideal components , for which excellent performance can be observed , the reality to be met in possible experiments might deteriorate the performance by unavoidable dispersion in component parameters .
the effect of disorder is expected to be quite significant in nonlinear settings : in fact , its paramount importance has been demonstrated even in the simpler case of linear resonant systems ( see , e.g. , ref .
@xcite ) and , thus , the presence of disorder may drastically affect the results in nonlinear structures . in that regard , a study of how much the considered system ( or other relevant ones ) is sensitive to disorder is certainly a relevant and important direction for future studies .
j.c . acknowledges financial support from the micinn project fis2008 - 04848 .
p.g.k . gratefully acknowledges support from the nsf ( dms-0806762 and cmmi-1000337 ) , the alexander von humboldt foundation and the alexander s. onassis public benefit foundation . the work of d.j.f .
was partially supported by the special account for research grants of the university of athens .
the elements @xmath150 , @xmath151 and @xmath152 associated with the cpw structure are connected with the physical parameters of the system by means of the following equations ( see , e.g. , ref .
@xcite ) : @xmath153 in the above expressions , @xmath154 is the dielectric constant of vacuum , @xmath155 and @xmath156 are the complete elliptic integral of the first kind and its complementary function , respectively @xcite , the arguments of these functions are given by : @xmath157 } \,\,\,\ , ( j=1,2 ) , \label{kj } \ ] ] and , finally , @xmath158 and @xmath159 denote , respectively , the inductance of the shunt strip and the effective inductance of the cpw structure ( note that the former takes the approximate value @xmath160 nh @xcite ) .
as far as the values of the srr parameters , @xmath15 and @xmath16 , are concerned , they are given by the following expressions @xcite : @xmath161 in the above expressions , @xmath162 is the fractional area of the slots occupied by the rings @xcite , while @xmath163 and @xmath164 denote , respectively , the inductance and capacitance of the srrs , and are given by : @xmath165 \nonumber \\ & & + \left(r+d+\frac{3w}{2}\right)\ln\left [ \frac{8(2r+3w+2d)}{w}-2\right ] , \label{lsrr } \\ & & c_{\rm srr } = 4\epsilon_0 \frac{k(\kappa_3)}{k'(\kappa_3)}+2\epsilon_0 ( \epsilon_{r1}-1)\frac{k(\kappa_4)}{k'(\kappa_4 ) } , \label{csrr}\end{aligned}\ ] ] where @xmath166 is the permeability of vacuum , @xmath134 , @xmath167 and @xmath135 respectively denote the radius of the internal ring , the distance between the internal and external rings , and the width of the rings ( see fig . [ cpwsrr ] ) , while the arguments of the elliptic integrals @xmath155 and @xmath156 are now given by : @xmath168}. \label{k34 } \ ] ]
as mentioned in sec .
ii.b , the quasi - continuum approximation is a variant of the method of multiple scales @xcite .
we thus introduce , at first , a set of new independent temporal variables , @xmath169 ( @xmath170 ) , and acordingly expand the derivative operator @xmath171 as @xmath172 .
next , we seek solutions of eq .
( [ eq : model ] ) in the form : @xmath173 where @xmath174
. then , we substitute eq .
( [ eq : ansatz01 ] ) into eq .
( [ eq : model ] ) and employ a continuum approximation for the unknown envelope functions @xmath175 , namely @xmath176 , where @xmath177 and @xmath178 being the lattice spacing ( the latter parameter does not appear in the results below , as one may readily rescale @xmath179 as @xmath180 ) .
furthermore , similarly to the introduction of the temporal variables , we introduce the set of the spatial variables @xmath181 ( and , thus , @xmath182 ) . to this end , equating coefficients of like powers of @xmath0 , we obtain the following ( first three ) perturbation equations : @xmath183 , \label{eq : term3}\end{aligned}\ ] ] where the operators are given by @xmath184 it is clear that the first - order pertubation solution to eq .
( [ eq : term1 ] ) reads : @xmath185 where @xmath186 is an unknown complex function , @xmath187 , while @xmath109 and @xmath110 satisfy the dispersion relation of eq .
( [ eq : dispqc ] ) .
next , substituting eq .
( [ eq : u1 ] ) into eq .
( [ eq : term2 ] ) , we obtain the following results : first , the non - secularity condition : @xmath188\frac{\partial v_1}{\partial x_1}=0 , \label{eq : secular}\ ] ] suggests that @xmath189 , where @xmath190 and @xmath104 is the group velocity of eq .
( [ eq : gvelqc ] ) , now consistently determined ; second , we obtain a uniformly valid solution for the second - order perturbation equation , in the form : @xmath191 where @xmath192 is given by eq .
( [ eq : gparam1 ] ) and the dependence of @xmath193 on higher - order scales has been omitted . to this end , substituting eqs .
( [ eq : u1])-([eq : u2 ] ) into eq .
( [ eq : term2 ] ) , and using the variables @xmath194 and @xmath195 , we derive from the non - secularity condition at @xmath125 the nls eq .
( [ eq : nls ] ) .
in the continuum limit ( for @xmath120 ) , the dispersion relation [ cf . eq .
( [ eq : dispqc ] ) ] is reduced to the form , @xmath196 the group velocity , @xmath197 , is now given by @xmath198 finally , the expressions for the dispersion and nonlinearity coefficients @xmath128 and @xmath129 in the continuum approximation read : @xmath199 and @xmath200\mathcal{g } } , \label{eq : nlcoefc}\ ] ] where the function @xmath201 is now given by : @xmath202 f. falcone , t. lopetegi , m. a. g. laso , j. d. baena , j. bonache , m. beruete , r. marqus , f. martn , and m. sorolla , phys . rev . lett . * 93 * , 197401 ( 2004 ) ; f. falcone , f. martn , j. bonache , r. marqus , and m. sorolla , microwave opt . technol .
lett . * 40 * , 3 ( 2004 ) ; f. aznar , j. bonache , a. valcarcel , and f. martn , microwave opt .
technol . lett . *
51 * , 926 ( 2009 ) .
|
we study the propagation of quasi - discrete microwave solitons in a nonlinear left - handed coplanar waveguide coupled with split ring resonators . by considering the relevant transmission line analogue
, we derive a nonlinear lattice model which is studied analytically by means of a quasi - discrete approximation .
we derive a nonlinear schrdinger equation , and find that the system supports bright envelope soliton solutions in a relatively wide subinterval of the left - handed frequency band .
we perform systematic numerical simulations , in the framework of the nonlinear lattice model , to study the propagation properties of the quasi - discrete microwave solitons .
our numerical findings are in good agreement with the analytical predictions , and suggest that the predicted structures are quite robust and may be observed in experiments .
= 1
|
a control of processes of quantum tunneling through potential barriers by external signals is a part of the field called quantum control which is actively developed now , see , for example , ref .
@xcite and references therein .
excitation of molecules , when one should excite only particular chemical bonds @xcite , formation of programmable atomic wave packets @xcite , a control of electron states in heterostructures @xcite , and a control of photocurrent in semiconductors @xcite , are typical examples of control by laser pulses .
a control of quantum tunneling through potential barriers is also a matter of interest , since tunneling is a part of many processes in nature .
the computation of probability for a classically forbidden region has a certain peculiarity from the mathematical stand point : there necessarily arises here the concept of motion in imaginary time or along a complex trajectory @xcite .
the famous semiclassical approach of wentzel , kramers , and brillouin ( wkb ) @xcite for tunneling probability can be easily ( in the case of a static potential ) reformulated in terms of classical trajectories in complex time as a simple change of variables .
the method of complex trajectories can also be applicable to a nonstationary case @xcite which is not trivial .
the method has been further developed in papers @xcite , when singularities of the trajectories in the complex plane were accounted for an arbitrary potential barrier ( see also @xcite ) .
recent achievements in the semiclassical theory are present in refs.@xcite .
let us focus on the main aspects of tunneling under nonstationary conditions .
when the electric field @xmath0 acts on a tunneling particle of the initial energy @xmath1 , it can absorb the quantum @xmath2 ( with the probability proportional to the small parameter @xmath3 ) and tunnel after that in a more transparent part of the barrier with the higher energy @xmath4 .
the pay in the absorption probability may be compensated by the probability gain in tunneling . in this case
the system tends to absorb further quanta to increase the total probability of passing the barrier .
this mechanism of barrier penetration is called photon - assisted tunneling .
if @xmath2 is not big , the process of tunneling , with the simultaneous multi quanta absorption , can be described in a semiclassical way by the method of classical trajectories in the complex time @xcite .
when a tunneling particle of the energy @xmath1 is acted by a short - time pulse , the tunneling probability is associated with the particle density carrying away in the outgoing wave packet .
the particle energy after escape is @xmath5 , where the energy gain @xmath6 , should be extremized with respect to the number of absorbed quanta @xmath7 and the energy @xmath8 of each quantum @xcite .
this mechanism relates to semiclassical method for nonstationary potentials .
the semiclassical method in quantum mechanics is very elegant and constructive . on the other hand , this method is based on the delicate mathematical issue called stokes phenomenon when a solution is not expected to appear but it appears @xcite .
this makes a use of the semiclassical method to be not trivial even in the case of a static potential .
in this situation the role of exactly solvable problems in quantum mechanics is very important .
for example , in the problem of reflection of a particle from certain static potentials one can follow in details a formation of a semiclassical solution from the exact one @xcite .
is it possible to find an exactly solvable problem for a nonstationary potential ( excepting not interesting parabolic one , where the effect is trivial @xcite ) to see how the exact solution is reduced to complex classical trajectories under semiclassical conditions ?
an exactly solvable problems would be extremely useful since the method of classical trajectories in nonstationary problems is still challenged and an exact analytical solution would place classical trajectories in the rank of mathematical theorem .
such exactly solvable nonstationary problem exists .
this is an inter - band tunneling ( zener breakdown ) in a semiconductor @xcite in a nonstationary electric field which is a constant in space .
the latter condition makes the problem to be exactly solvable since in the momentum representation the schrdinger equation is reduced to the first order which can be solved by the method of characteristics . in this paper
it is studied how the exact solution turns over into one corresponding to classical trajectories and under which conditions this is possible .
the goal of the paper is not to investigate zener effect in real semiconductors but to use this situation to mathematically justify the method of classical trajectories .
another issue of this paper is that a new mechanism of _ enhanced _ tunneling through static non - one - dimensional barriers is proposed in addition to well known _
normal _ tunneling solely described by a trajectory in imaginary time . as shown in the paper , under certain conditions on the barrier shape and the particle energy , the probability of enhanced tunneling
is not exponentially small even for non - transparent barriers , in contrast to the case of normal tunneling .
a penetration of a particle through a potential barrier is forbidden in classical mechanics . only due to quantum effects
the probability of passing across a barrier becomes finite and it can be calculated on the basis of wkb approach , which is also called the semiclassical theory .
the transition probability through the barrier , shown in fig . [ fig1 ] , is @xmath9\ ] ] where @xmath10}\ ] ] is the classical action measured in units of @xmath11 .
the integration goes under the barrier between two classical turning points where @xmath12 .
one can use the general estimate @xmath13 , where @xmath14 is the barrier height and @xmath15 is the frequency of classical oscillations in the potential well .
a semiclassical barrier relates to a big value @xmath16 .
what happens when the static potential barrier @xmath17 is acted by a weak nonstationary electric field @xmath18 ?
in this case there are two possibilities for barrier penetration : ( i ) the conventional tunneling , which is not affected by @xmath18 , shown by the dashed line in fig .
[ fig2](a ) , and ( ii ) an absorption of the quantum @xmath2 of the field @xmath18 and subsequent tunneling with the new energy @xmath4
. the latter process is called photon - assisted tunneling .
the total probability of penetration across the barrier can be schematically written as a sum of two probabilities @xmath19 where @xmath20 is the fourier component of the field @xmath18 and the length @xmath21 is a typical barrier extension in space . the second term in eq .
( [ 3 ] ) relates to photon - assisted tunneling and it is a product of probabilities of two quantum mechanical processes : absorption of the quantum @xmath2 and tunneling through the reduced barrier @xmath22 . since in quantum mechanics
one should add amplitudes but not probabilities , eq . (
[ 3 ] ) is rather schematic and serves for general illustration only .
for example , the accurate perturbation theory starts with a linear @xmath23-term
. when the frequency is high @xmath24 , the second dominates at sufficiently small nonstationary field @xmath25 .
this is a feature of tunneling processes since , normally , a nonstationary field dominates at bigger amplitudes @xmath26 . when the second term in eq .
( [ 3 ] ) exceeds the first one , further orders of perturbation theory should be accounted which correspond to the multiple absorption , shown in fig .
[ 2](b ) .
let us specify a shape of a field pulse in the form @xmath27 with the fourier component @xmath28 . in this case , in addition to the almost steady flux from the barrier ,
an outgoing wave packet ( of the maximum amplitude @xmath29 ) is created which carries away a certain particle density .
the probability of the transition through the process of absorption of @xmath7 quanta and subsequent tunneling with the higher energy @xmath30 , shown in fig .
[ 2](b ) , can be estimated as @xmath31= \left(\frac{\pi}{\hbar}\hspace{0.05cm}\theta a{\cal e}\right)^{2n}\exp(-a)\ ] ] where @xmath32 here the total energy transfer @xmath33 is introduced ( @xmath34 ) .
the maximum squared amplitude @xmath35 of the outgoing packet is associated with a maximum value of @xmath36 and is determined by the extreme @xmath37 to get a minimum of @xmath38 .
this results in the condition @xmath39 .
an existence of such a minimum is possible if growing up of small @xmath37 reduces @xmath38 , that is , under the condition @xmath40 of sufficiently short pulses . in other words , sufficiently short and not very small pulses ( however , still much smaller than the static barrier field ) strongly enhance tunneling by photon assistance . in principal
, this type of treatment of nonstationary tunneling is reasonable since a maximum of the outgoing packet should relate to some extreme condition .
nevertheless , the approach ( [ 5 ] ) is schematic since a quantum interference between absorption and tunneling is neglected .
what is a `` scientific '' way to account precisely a nonstationary field in tunneling ?
we go towards this in [ sec : traj ] .
according to feynman @xcite , when the phase of a wave function is big , it can be expressed through classical trajectories of the particle .
but in our case there are no conventional trajectories since a classical motion is forbidden under a barrier .
suppose a classical particle to move in the region to the right of the classical turning point @xmath41 in fig .
[ fig1 ] and to reach this point at @xmath42 . then , close to the point @xmath41 , @xmath43 ( @xmath44 ) and there is no a barrier penetration as at all times @xmath45 . nevertheless , if @xmath46 is formally imaginary , @xmath47 , the penetration becomes possible since @xmath48 is less then @xmath41 .
therefore , one can use classical trajectories in imaginary time to apply feynman s method to tunneling . in the absence of a nonstationary field a classical trajectory
satisfies newton s equation in imaginary time @xmath49 where @xmath17 is the static barrier in fig .
the classical turning point @xmath41 in fig .
[ fig1 ] is reached at @xmath50 with the initial condition @xmath51 .
the classical trajectory @xmath52 can be considered as a change of variables @xmath53 in the wkb exponent ( [ 2 ] ) when it becomes of the form @xmath54\ ] ] so , in the absence of a non - stationary field , use of classical trajectories in imaginary time is obvious and it is simply reduced to a change of variable .
is it possible to extend the method of classical trajectories in tunneling to a non - stationary case ?
suppose some pulse of an external field , for example ( [ 4 ] ) , acts on a tunneling particle .
at the point @xmath41 , besides an almost constant background , there is a wave packet of the outgoing particles shown in fig .
if the method of classical trajectories is applicable , the maximum of the squared amplitude of the outgoing packet @xmath55 should correspond , with the exponential accuracy , to the classical action @xmath56\ ] ] since a classical trajectory relates to an extreme value of the action .
the classical trajectory should satisfy newton s equation @xmath57 with the conditions @xmath58 we suppose the potential well to be narrow and localized close to @xmath59 as in fig .
the under - barrier `` time '' @xmath60 is expressed through the particle energy @xmath1 in the well @xmath61 which is weakly violated by a small non - stationary field .
certain semiclassical conditions ( not extremely small and sufficiently slow varying @xmath18 ) should be fulfilled .
( [ 9]-[13 ] ) constitute the method of classical trajectories in tunneling under non - stationary conditions .
this method relates to calculation solely of the maximum in time @xmath62 of the probability @xmath63 of outgoing particles .
in contrast to a static barrier , the method of classical trajectories is not trivial in application to a nonstationary case .
this can be seen from that the non - stationary field of the type @xmath0 in imaginary time becomes proportional to @xmath64 and may be very big .
the same relates to the field ( [ 4 ] ) which is singular in imaginary time .
this means , formally , that arbitrary weak amplitude @xmath23 of a nonstationary field may produce a big effect .
the arguments in [ sec : ph - as ] also show an increase of an effective @xmath23 in tunneling , but it is clear physically that an effect of an extremely small nonstationary field is negligible and the method of classical trajectories should not work in this case . as one can see , the method of classical trajectories is delicate and requires an accurate justification . in the papers @xcite for the certain particular potential well and nonstationary pulses the wave function has been found in the form of expansion with respect to powers of @xmath11 @xmath65\exp\left[\frac{i}{\hbar}\hspace{0.1cm}s(x , t)\right]\ ] ] where @xmath66 is the classical action .
the semiclassical formalism , developed in @xcite , confirms the method of classical trajectories and coincides with the general scheme ( [ 9]-[13 ] ) .
nevertheless , it would be extremely instructive to find any exactly solvable case of tunneling under nonstationary conditions in order to see how the method of classical trajectories follows not from a semiclassical formalism but from an exact mathematical solution . in [ sec : zener ] this program is completed .
let us consider an one dimensional two band semiconductor in an external homogeneous electric field @xmath67 , where @xmath68 is the static component .
the schrdinger equation has the form @xmath69\psi\ ] ] where @xmath70 , @xmath71 is the energy gap , and @xmath72 is the velocity at big momentum . the square root in eq . ( [ 15 ] ) may have two signs according to two energy bands . in a static electric field
@xmath68 the inter - band quantum tunneling is possible which is called zener breakdown @xcite . the incident flux of particles from left to the right in fig . [ fig4 ]
is mainly reflected back from the tilted energy gap but a small fraction penetrates the other band and goes to @xmath73 with the probability @xcite @xmath74 where @xmath75 and @xmath76 is the tunneling length in fig .
( [ 16 ] ) holds under the semiclassical condition @xmath77 the peculiarity of this problem is that it can be solved exactly with a nonstationary field by making the fourier transformation with respect to @xmath78 . in the new variables , momentum @xmath79 and time @xmath46 , the schrdinger equation ( [ 15 ] )
is of the first order with respect to derivatives @xmath80 and @xmath81 and can be solved by the method of characteristics . if to measure the coordinate in units of @xmath21 and time in units of @xmath82 , the solution of eq .
( [ 15 ] ) has the form @xmath83\right\}\ ] ] where @xmath84^{2}}\ ] ] and @xmath85 .
below we consider the case when the nonstationary field is much less compared to the static value @xmath86 if @xmath18 is an analytical function of @xmath46 , @xmath87 is an analytical function of the complex argument @xmath79 and it has only two branch points at @xmath88 . according to this , eqs .
( [ 17]-[18 ] ) can be written through the complex paths @xmath89\right\};\hspace{1 cm } f_{t}(q)=\int_{c{q}}duf_{t}(u , q)\ ] ] where the contour @xmath90 on the complex plane of @xmath79 and the contour @xmath91 on the complex plane of @xmath92 are shown in fig .
two cuts are denoted in fig . [ 5](a ) by solid lines and the regular branch of the function @xmath87 is chosen in order to get the limit @xmath93 the integral ( [ 20 ] ) is divergent and has to be cut off at the lower limit by a big negative value which sets an irrelevant constant phase shift . under the semiclassical condition ( [ 16a ] ) the @xmath79-integration in eq .
( [ 19 ] ) goes mainly in the vicinity of the saddle point(s ) determined by the equation @xmath94 suppose the nonstationary field @xmath95 at @xmath96 when the particle energy is zero and the particle goes to the tilted energy gap in fig .
[ fig4 ] from the left . for @xmath78 to the left of the point @xmath97 in fig .
[ fig4 ] ( @xmath98 in the dimensionless units used ) an influence of the nonstationary field under the condition ( [ 18a ] ) is small and two saddle points @xmath99 are shown in fig .
[ fig6](a ) .
they correspond to the incident and the reflected de broglie waves .
since the tunneling probability is small , at @xmath98 the wave function can be shown to have the form @xmath100 in the classically allowed regions , to the left and to the right of the tilted gap , the classical trajectory @xmath101 ( the equation , it satisfies , is written below ) is weakly violated by the non - stationary field under the condition ( [ 18a ] ) @xmath102 excepting that @xmath103 now .
the energy of the incident particle is zero and the energy after exit from the tilted gap conserves and equals @xmath104 . for each @xmath78 and @xmath46 the saddle point condition ( [ 21 ] ) determines a certain @xmath79 . in the absence of a nonstationary field
the amplitude of the outgoing wave is almost stationary . under action of a pulse
there is an additional outgoing wave packet at @xmath105 which goes to the right and keeps the constant maximum amplitude ( if to neglect small quantum effects of smearing ) at the classical trajectory ( [ 23 ] ) .
we are interested to find this maximum amplitude of the outgoing packet since it defines , with the exponential accuracy , a total number of particles passing the barrier . with @xmath106 the saddle point condition ( [ 21 ] )
is satisfied by @xmath107 since @xmath108 is small for real @xmath46 . in order to reach the saddle point at positive @xmath78
, one should deform the contour @xmath90 as shown in fig .
[ fig6](b ) .
this could be done with no problems since @xmath87 has only the branch point at @xmath109 in the upper half plane .
the main contribution to the @xmath79-integration over the bent contour @xmath90 comes from the saddle point , shown in fig .
[ fig6](b ) . the saddle point condition ( [ 21 ] ) now reads @xmath110^{2 } } } \hspace{0.1cm}h(i\tau)\ ] ] the nonstationary field is supposed to be symmetric @xmath111 , as the field ( [ 4 ] ) , so that @xmath112 is real .
@xmath60 is a smallest positive root of the equation @xmath113^{2}=1\ ] ] and can be called the tunneling time . if to define the transition probability @xmath35 as a square of the ratio of the maximum amplitude of the outgoing packet and the amplitude of the incident wave , eq .
( [ 17 ] ) gives with the exponential accuracy @xmath114 where @xmath115^{2}}\ ] ] the action @xmath38 is collected due to the @xmath79-integration in the narrow vicinity of the right - hand - side saddle in fig .
[ fig6](b ) where @xmath79 is real and positive . for such @xmath79 the contour @xmath91 in fig . [ fig5](b ) goes around the branch point @xmath116 and leads to the relation ( [ 27 ] ) . without a non - stationary field eqs .
( [ 26 ] ) and ( [ 27 ] ) give the static result ( [ 16 ] ) .
using of the above method of saddle point means that the length @xmath117 of the steepest descent integration should be sufficiently short compared to the typical scale of the problem ( @xmath118 in dimension units ) which can be estimated from the relation @xmath119 . for the pulse ( [ 4 ] ) the condition reads @xmath120 in presence of a nonstationary field , the semiclassical condition ( [ 16a ] ) should be supplemented by the additional semiclassical condition ( [ 28 ] ) . according to it , the nonstationary field @xmath23 and its duration @xmath121
should not be too small .
note , that under the semiclassical condition ( [ 28 ] ) the nonstationary amplitude @xmath23 can be still smaller than the static field @xmath68 . in refs .
@xcite the method of classical trajectories was developed for the present problem of inter - band tunneling in presence of a nonstationary field . according to this method
, one has to minimize the classical action of a particle with the spectrum , as in ( [ 15 ] , ) and in a homogeneous electric field @xmath122\right\}\ ] ] to find a classical trajectory .
it satisfies the equation @xmath123^{2}}}\ ] ] the classical trajectory starts at @xmath42 at the point @xmath124 in fig .
[ fig4 ] . at the `` moment '' @xmath125 it reaches some point inside the gap where two branches merge and , turning on the opposite side of the cut , it comes at @xmath42 in the point @xmath126 . within the formalism , developed in refs .
@xcite , in order to obtain a tunneling probability , one should substitute the classical trajectory into the action ( [ 29 ] ) and to integrate over the branch point @xmath125 .
this procedure leads to the result ( [ 27 ] ) . for the nonstationary pulse ( [ 4 ] ) under the condition ( [ 18a ] ) , the tunneling time @xmath127 at @xmath128 and @xmath129 at @xmath130 .
the action ( [ 27 ] ) becomes of the form @xmath131 this expression weakly depends on @xmath23 under the semiclassical condition ( [ 28 ] ) since the role of the nonstationary field ( [ 4 ] ) , due to its singularity in imaginary time , is only to set the new tunneling time @xmath60 .
obtained results for inter - band tunneling under nonstationary conditions may be interpreted as positive ( positive energy transfer ) photon - assisted tunneling shown in fig .
[ fig4 ] . in this problem
can not be such phenomenon as euclidean resonance @xcite , associated with a negative energy transfer .
analogously one can consider the monochromatic @xmath0 @xcite and the gaussian @xmath132 nonstationary fields which become exponentially big in imaginary time and to establish how much they can grow within the formalism of classical trajectories .
one can conclude now , that for inter - band tunneling in a nonstationary field under semiclassical conditions the method of classical trajectories follows from the exact analytical solution .
the bridge between exact theory and classical trajectories enables in problems of tunneling to treat a nonstationary field in complex time .
tunneling through one - dimensional static barrier is a well known problem described in almost every text book on quantum mechanics and the tunneling rate , with exponential accuracy , is given by wkb formulas ( [ 1 ] ) and ( [ 2 ] ) .
tunneling through a two- or many - dimensional potential barrier was considered in a variety of publications @xcite . the condition @xmath133 , where @xmath134 is the total energy , determines certain surfaces in the coordinate space .
for a two - dimensional case it is shown in fig .
[ fig7 ] . tunneling probability from the classically allowed region @xmath38 to another such region @xmath135
is given by eq .
( [ 1 ] ) where the exponent @xmath136\ ] ] is expressed through the classical trajectory @xmath137 satisfying newton s equation in imaginary time @xmath138 the trajectory connects different surfaces , being perpendicular to them at the connected points , terminates at the border of @xmath135 at @xmath42 , and starts at the border of @xmath38 at @xmath139 .
the time @xmath60 is defined by the total energy @xmath134 .
the above general scheme is applicable to a calculation of tunneling rate regardless of dimensionality of the problem ( we do not consider a possibility of caustics between @xmath38 and @xmath135 ) . nevertheless ,
when the dimensionality is bigger than one , another way may exist to tunnel through a static potential barrier .
this way differs from the normal scheme and may result in an enhanced tunneling rate .
any tunneling process obeys quantum mechanical rules : a tunneling particle at each point emits partial de broglie waves which interfere and produce a final outgoing wave .
this wave is exponentially small for semiclassical one - dimensional barriers . in two- or multi - dimensional case
the tunneling motion can be influenced by other ( non - tunneling ) motions leading to different conditions of interference . as a result
, the outgoing wave may be proportional to the enhanced exponent , which is much bigger than normal one or even of the order of unity .
the phenomenon of enhanced tunneling is considered below .
the mechanism of enhanced tunneling through a static barrier can be shown in the case of a long and narrow quantum wire when the radial motion ( @xmath78-direction ) is strongly quantized but the motion along the axis ( @xmath140-direction ) is almost classical .
we do nod discuss a detailed link to real quantum wires since the goal is solely to demonstrate a new mechanism of tunneling through static barriers .
to model this situation , suppose the particle to move in the two - dimensional potential @xmath141 plotted in fig .
the particle can tunnel from the energy level @xmath1 in the radial direction . in two dimensions the potential energy
is shown in fig .
[ fig9 ] . the particle tunnel from the region of small @xmath78 to the region of big @xmath78 . when the total potential energy is simply a sum of @xmath78-dependent and @xmath140-dependent potentials , variables @xmath78 and @xmath140
are separated and tunneling along @xmath78-direction occurs as in a one - dimensional case . if to add some `` interaction '' potential @xmath142 to the system , the total potential @xmath143 is not reduced to a sum of @xmath78- and @xmath140-term and motions in the @xmath78- and @xmath140-direction are no more independent .
the both barriers in fig . [
8 ] are weakly penetrated and , in the first approximation , the particle occupies the region of @xmath144 and positive @xmath140 in fig .
[ 9 ] with incident and reflected fluxes along @xmath140-direction as shown in fig .
[ fig8](b ) .
this is equivalent to the problem of two one - dimensional particles with the interaction @xmath142 when one particle in a well is acted by a steady flux of incident particles .
this is analogous to the situation in nuclear physics when an incident steady flux of protons acts on the alpha particle occupied some energy level inside the nuclear potential well @xcite .
this nuclear problem of excitation of alpha particle can be considered by two methods : ( i ) by a solution of the schrdinger equation for alpha particle and proton in the static potential or ( ii ) by a solution of the schrdinger equation for alpha particle in a nonstationary potential created by the classically moved proton which reflects by nuclear coulomb forces .
the both ways lead to the same small exponent in the semiclassical probability .
the second method is more convenient since the excitation of a higher level of alpha particle can be considered as one driven by a nonstationary force , acting during a finite time @xcite .
so , with exponential accuracy , the steady problem of decay of a metastable state under an incident flux can be mapped onto dynamical pulses . under conditions ,
when the motion in the @xmath140-direction is weakly violated by the motion in the @xmath78-direction , the total process can be considered as one - dimensional tunneling in the nonstationary potential @xmath145 $ ] , where @xmath146 is a classical trajectory in the static potential @xmath147 , shown in fig .
according to this , besides normal tunneling `` 0 '' in fig . [ 8](a ) , corresponding to absence of the pulse , there is also the process `` 1 '' related ( in the language of dynamical pulses ) to the under - barrier photon - assisted tunneling which is similar to fig .
[ fig4 ] . since a photon - assistance increases tunneling probability
, the process `` 1 '' relates to the phenomenon of enhanced tunneling . in the language of static potential
, one can say that tunneling along the radius is assisted by the motion along the axis .
this mapping is mentioned only for interpretation of enhanced tunneling `` 1 '' by means the language of dynamical pulses . in [ subsec : newtunnel ] enhanced tunneling is considered on the basis of a static barrier . the condition @xmath148 gives the curve shown in fig .
[ fig10 ] . to calculate a probability of enhanced tunneling `` 1 '' , one should account the fact that in semiclassical limit the phase @xmath149 of the wave function @xmath150 $ ] is big .
the function @xmath151 can be found from scrdinger equation as the main term in the expansion of the type ( [ 14 ] ) . for normal tunneling `` 0 ''
, @xmath151 coincides with the classical action , varies smoothly between the regions @xmath38 and @xmath135 , and its extreme value relates to the classical trajectory denoted by the dashed curve in fig .
[ fig10 ] . for enhanced tunneling `` 1 ''
the behavior of @xmath151 is much more complicated , since there are reconnection among different branches of @xmath66 . on each branch
the function @xmath151 coincides with some classical action according to findings of refs .
@xcite .
fortunately , to find a tunneling probability it is not necessary to know @xmath152 at all @xmath78 and @xmath140 .
it is sufficient to connect phases of the wave function between certain points at the borders of @xmath38 and @xmath135 using some formal method .
such formal method for the process `` 1 '' ( enhanced tunneling ) consists of two steps : a connection @xmath153 and a connection @xmath154 ( fig . [ fig10 ] ) .
for the first step , one can consider the vertical line at @xmath155 connecting the border of @xmath38 and some point @xmath156 in fig .
[ fig10 ] , where @xmath157 is expressed through the certain branch @xmath158 of the classical action , which satisfies the equation of hamilton - jacobi @xmath159 at @xmath160 the potentials @xmath161 , @xmath147 , and @xmath142 are smooth . since the potential @xmath161 is abrupt at @xmath155 , existence of the quantum level @xmath1 in the well results in the following semiclassical condition inside the barrier in fig . [ 8](a ) , when @xmath78 tends to @xmath162 from the right , @xmath163 the total energy is @xmath164 , where @xmath165 is the energy of the classical motion along the well .
the @xmath140-dependence of the action at @xmath155 can be easily found from eqs .
( [ 35 ] ) and ( [ 36 ] ) in the wkb form of the one - dimensional motion along the well .
the phase at the point @xmath156 ( relative to @xmath38 ) in fig .
[ fig10 ] is @xmath166}\ ] ] where @xmath167 and @xmath168 are determined by zero of the square root in eq .
( [ 37 ] ) at @xmath156 and @xmath38 respectively .
the position of the point @xmath156 , defined by the conditions @xmath155 and @xmath169 , is not flexible and depends only on the energy @xmath165 , since the quantum level @xmath1 is strictly determined by the potential in fig . [ 8](a ) .
the same relates to the phase @xmath170 at the point @xmath156 ( relative to @xmath38 ) . the value of @xmath158 , given by eq .
( [ 37 ] ) , is extreme with respect to variation of positions of the ends of the line @xmath155 in fig .
[ fig10 ] . as the second step
, the point @xmath156 can be connected to the region @xmath135 by the classical trajectory which provides an extreme of the phase of @xmath135 ( relative to @xmath156 ) and is based on the expression @xmath171 , as eq .
( [ 32 ] ) , but the trajectory connects now the regions @xmath156 and @xmath135 . the rate of enhanced tunneling `` 1 '' @xmath172 is given by the total phase difference and reads @xmath173 where @xmath158 is defined by eq .
( [ 37 ] ) .
as soon as normal tunneling is described solely by a classical trajectory ( [ 32])-([33 ] ) , enhanced tunneling , according to eq . ( [ 38 ] ) , requires more complicated description .
the total tunneling probability should be estimated as @xmath174 to calculate @xmath175 ( normal ) and @xmath176 ( enhanced ) one should specify the potential energy in the form ( [ 34 ] ) and use eqs .
( [ 32 ] ) , ( [ 33 ] ) , ( [ 37 ] ) , and ( [ 38 ] ) . the second term in eq .
( [ 38 ] ) , which is real and negative , works versus the big first term representing the double tunneling through the barriers in figs . [ 8](a ) and [ 8](b )
. the second term in eq . ( [ 38 ] ) can play a crucial role in reduction of @xmath176 .
enhanced tunneling `` 1 '' may be interpreted as a cooperative motion along a trajectory in the @xmath78-direction and along a bounce ( forth and back ) in the @xmath140-direction .
one should emphasize again , that the above semiclassical method gives only the formal calculation of the tunneling probability with exponential accuracy , but it does not relate to a real behavior of the wave function @xmath152 . for example , the formal density at the point @xmath156 can be big , the real wave function drops exponentially in that region .
there are two issues in this paper .
the first one is that the problem of inter - band tunneling in a semiconductor ( zener breakdown ) in a nonstationary and homogeneous electric field is solved exactly . on the basis of the exact analytical solution the formation of an approximation of classical trajectories
is studied .
this semiclassical formalism is very effective in tunneling under nonstationary conditions since it allows to reduce the problem to a solution of the newton equation in complex time .
the semiclassical approach , obtained from the exact solution , is of the same type as in other problems of nonstationary tunneling which do not allow exact solutions , for example , decay of a metastable state , penetration through a potential barrier , and over - barrier reflection .
the second issue relates to different types of tunneling through a static potential barrier if dimensionality of the problem is bigger than one .
a multi - dimensional case has an essential feature which differs it from one - dimensional problem .
namely , a tunneling coordinate @xmath177 can be influenced by non - tunneling one @xmath178 resulting in a different condition of interference of emitted partial waves during tunneling through the potential @xmath179 .
a multi - dimensional tunneling can be interpreted as one - dimensional but in the nonstationary potential @xmath180 $ ] , where @xmath146 is the classical non - tunneling motion .
this mapping onto dynamical problem enables to establish different types of tunneling .
as well known , there is _ normal _ tunneling through a multi - dimensional barrier related simply to a classical trajectory in imaginary time and connected two classically allowed regions .
this mechanism was always employed by authors in consideration of multi - dimensional tunneling .
normal tunneling corresponds to the case when the tunneling particle does not absorb quanta of the nonstationary field .
there is also a process when particle absorbs quanta ( photons ) from the nonstationary potential @xmath180 $ ] .
this is a photon - assisted tunneling with an enhanced probability . in the language of the static potential barrier
, this process corresponds to _ enhanced _ tunneling .
this is a new mechanism of quantum tunneling through a static multi - dimensional barrier .
depending on shape , sign , and value of the potential @xmath179 , two types of enhanced tunneling are possible : ( i ) with absorption of energy by the tunneling motion ( positive assistance ) or ( ii ) with emisssion of energy ( negative assistance ) .
the positive assistance was considered in ref .
@xcite and in this paper for zener breakdown .
the negative assistance was considered in ref .
it is remarkable that the method of calculation of probability of enhanced tunneling , proposed in this paper , is applicable equally to both positive and negative assistance . as it was found in ref .
@xcite for tunneling through a nonstationary barrier , the enhanced tunneling probability may be not exponentially small under conditions of negative assistance .
this phenomenon is called euclidean resonance which results from influence of nonstationary field on interference of emitted partial waves during tunneling .
euclidean resonance can also take place in tunneling through a static non - one - dimensional barrier . to get this phenomenon
, one should adapt barrier parameters and the particle energy to reach the condition @xmath181 in eq .
( [ 38 ] ) , as in the case of a nonstationary barrier @xcite .
the above theory is valid if @xmath182 and when this condition breaks down one should use a formalism generic with the multi - instanton approach .
nevertheless , the presented theory , as any other approach , can be approximately extended up to the limit of its validity @xmath183 wich relates to euclidean resonance .
enhanced tunneling may result in a dramatic increase of a tunneling rate .
for example , for an almost classical barrier , according to the normal mechanism , the tunneling probability is calculated to be @xmath184 , but enhanced tunneling , under conditions of euclidean resonance , results in the probability of @xmath185 .
euclidean resonance is a phase phenomenon and any effect of dephasing ( friction , for example ) may destroy it . to avoid this destruction ,
the attenuation time due to friction should be bigger compared to the under - barrier time . in conclusion , a new mechanism of enhanced tunneling through static non - one - dimensional barriers
is established in addition to well known normal tunneling . under certain conditions on shape of an almost classical barrier and particle energy , the probability of quantum tunneling through it
may be not exponentially small .
i am grateful to v. gudkov for valuable discussions .
|
the problem of inter - band tunneling in a semiconductor ( zener breakdown ) in a nonstationary and homogeneous electric field is solved exactly . using the exact analytical solution ,
the approximation based on classical trajectories is studied .
a new mechanism of enhanced tunneling through static non - one - dimensional barriers is proposed in addition to well known normal tunneling solely described by a trajectory in imaginary time . under certain conditions on the barrier shape and the particle energy , the probability of enhanced tunneling
is not exponentially small even for non - transparent barriers , in contrast to the case of normal tunneling
. 1.0 cm
|
consider the following scenario : we have a simple network consisting of two routes connecting @xmath0 to @xmath1 , see figure [ fig:4roads ] .
each route consists of two roads .
roads @xmath2 and @xmath3 are identical , as are roads @xmath4 and @xmath5 .
traffic is unidirectional in the direction from @xmath0 to @xmath1 .
travel time along roads @xmath2 and @xmath3 are given by @xmath6 , where @xmath7 is the number of vehicles on that road , while the travel time is @xmath8 for each of roads @xmath4 and @xmath5 , irrespective of the number of vehicles on that road . in equilibrium , vehicles will distribute evenly between the two routes connecting @xmath0 and @xmath1 , i.e. , roads @xmath2 & @xmath3 and @xmath4 & @xmath5 . assuming that initially @xmath9 vehicles start from @xmath0
, we find a travel time of @xmath10 along each of the two routes .
add a road @xmath11 as given in figure [ fig:5roads ] , and assume that the travel time is zero along this road .
drivers will start using the new road , reducing their travel time from @xmath10 to @xmath12 .
however , as more and more drivers use the new road , their travel time will increase to @xmath13 .
now , no driver will have an incentive to use the old roads , i.e. , avoiding road @xmath11 , as the travel time along those roads will be @xmath14 .
thus all drivers are worse off than before , in spite of having a new road .
this is the braess paradox in a nutshell : _ adding a new road to a network may make travel times worse for all .
_ in both cases the equilibrium is a _ wardrop equilibrium _
( i.e. , all routes used have the same travel time , and all unused routes have longer travel times ) as well as a _
nash equilibrium_. this is the simplest example of the braess paradox , introduced ( with a different example ) by braess in 1968 @xcite , see also @xcite .
this example and some generalizations have been studied in , e.g. , @xcite . in spite of the unrealistic assumptions in the prevalent example above
, the paradox has turned out to be ubiquitous and intrinsic to dynamical networks .
the paradox also appears in other situations not modeling traffic flow @xcite , see , e.g. , @xcite for an example involving mesoscopic electron systems , and @xcite for an example with mechanical springs .
furthermore , the paradox can be reformulated in the context of game theory .
in addition , there are well documented examples of the paradox occurring in real - life traffic situations , e.g. , in seoul @xcite and stuttgart @xcite , see also @xcite .
not surprisingly , the paradox has been well described also in general media , see , e.g. , @xcite and on wikipedia as well as youtube .
the extensive discussion about the braess paradox makes a complete reference list impossible , see , however , @xcite . in this paper
we only refer to articles directly related to the research at hand . here
we want to study the braess paradox with a more realistic nonlinear dynamics .
more specifically , we want to model unidirectional traffic along roads by a macroscopic model where only densities of vehicles are considered .
we believe this to be novel . in this class of models , introduced by lighthill
whitham @xcite and richards @xcite ( hereafter denoted the lwr model ) , vehicles , described by a density @xmath7 rather than individually , drive with a velocity determined by the density alone ; higher density yields slower speed while low density lets vehicles approach the speed limit . at a maximum density with bumper - to - bumper vehicles , traffic comes to a halt .
the dynamics is well described by the nonlinear partial differential equation @xmath15 see , e.g. , @xcite .
the function @xmath16 is denoted the flux function , or , in the context of traffic flow , the fundamental diagram .
it is in general a concave function that equals zero when @xmath7 vanishes and when @xmath7 equals the maximum possible road density .
hyperbolic conservation laws , as equations of the type are called , have been used to study traffic on a network , starting with holden and risebro @xcite , see , e.g. , the book by garavello and piccoli @xcite . related results on a game theoretic approach to network traffic through the lwr model ,
see @xcite . for general theory concerning hyperbolic conservation laws
we refer to @xcite .
however , the braess paradox describes an equilibrium situation , and it is not relevant to include time variation .
rather , we want to study stationary solutions where the velocity is a given function of the density of vehicles on the road . at a junction
, the differential equation will in general , if the two roads have different properties , establish a complicated wave pattern , creating waves that emanate from the junction in both directions .
however , in the equilibrium situation , this can not happen , as it would create time - dependent waves .
thus , we will set up the example in such a way that no waves are created at junctions . in this paper
we analyze the same simple network as described above , but with much more realistic dynamics .
more general examples are of course possible using the same methods .
however , calculations become more cumbersome and less transparent , and we here focus on presenting the ideas of the model , exemplified on the simple network in figures [ fig:4roads ] and [ fig:5roads ] . for another approach to the braess paradox , see , e.g. , @xcite .
the prevalence of the braess paradox is unwanted , and one would like to take measures to prevent its occurrence . in the example in the present paper , we use the velocity of the road @xmath11 as a control parameter . by properly adjusting the speed limit on road @xmath11 , one can force the braess paradox to disappear , and make the social optimum coincide with the nash equilibrium
. this can be illustrated in the simple example in the beginning of the introduction . given a `` benevolent dictator '' who wants to reduce the total travel time and reach the social optimum
, a short calculation shows that , with @xmath9 , 1750 vehicles should follow each of the routes @xmath2 & @xmath4 and @xmath5 & @xmath3 , and the remaining @xmath17 vehicles should follow the route @xmath2 , @xmath11 , and @xmath3 .
although a social optimum , this situation is neither a wardrop nor a nash equilibrium . this paper offers a framework applicable to general networks .
the input is , in addition to the network itself , the length and velocity fields of each road as well as the influx .
we assume that traffic is in the uncongested , or free , phase .
this will prevent waves from emanating from the junctions .
below , we denote @xmath18 and @xmath19^{n } \mid \sum_j { \vartheta}_j \leq 1\}$ ] is the standard simplex in @xmath20 . the sphere centered at @xmath21 with radius @xmath22
is denoted by @xmath23 .
two points @xmath0 and @xmath1 are connected through a network of roads . along each road
, traffic is described through the lwr model . at each junction
, the total flow exiting the junction equals the incoming one , so that the total quantity of vehicles is conserved .
the macroscopic description obtained solving along each road also provides the full microscopic portrait of the network .
indeed , once @xmath24 is known along the road @xmath22 connecting , say , the junction at @xmath0 to that at @xmath1 , the single vehicle leaving from @xmath0 at time @xmath25 travels along @xmath22 according to @xmath26 the travel time @xmath27 along the road @xmath2 is then implicitly defined by @xmath28 to compute @xmath27 , in general , one has first to provide with initial and boundary data , then solve the resulting initial - boundary value problem to obtain @xmath29 , use this latter expression to solve the ordinary differential equation and finally solve the equation .
observe that the right - hand side in the ordinary differential equation in is in general discontinuous , nevertheless in the present setting it is well - posed , see @xcite . in the present stationary framework ,
this procedure can be pursued explicitly , as we detail below in example [ ex : simple ] .
remark that , in a stationary regime , all travel times are independent of the starting time @xmath25 . for the above travel times to be a reliable measure of the network efficiency , it is necessary that they are independent from any particular initial data .
also the standard initial - boundary value problem for with zero initial density on the whole network is unsatisfactory , since it would give results that depend on the transient period necessary to fill the network .
we are thus bound to select _ stationary _ solutions , assigning a constant inflow at @xmath0 for all times @xmath30 . moreover , to allow for stationary solutions , we also assume that the total flow incoming at any junction never exceeds the total capacity of the roads exiting that junction . in the general lwr model , the flux function @xmath31 is a concave function that vanishes at zero density and at @xmath32 , the maximum density .
the flux has a unique maximum for some value @xmath33 . as usual
, we refer to densities below @xmath34 as the uncongested , or free , phase , and for densities above @xmath34 as the congested phase . in the remaining part of the paper , to obtain stationary solutions ,
we need to remain in the free phase only , so that @xmath35 $ ] throughout the network . in order to simplify the notation we will use the normalization @xmath36 for all roads .
we will not make any assumptions on , or reference to , @xmath37 above this value .
hence , on the flow function we pose the following assumption : ( q ) : : @xmath38 ; { { \mathbb{r}}}^+)$ ] , @xmath39 , @xmath40 and @xmath41 .
clearly , if @xmath37 satisfies * ( q ) * , then the speed law @xmath42 is well - defined , continuous , strictly positive and weakly decreasing , see lemma [ lem : speed ] . as a result ,
the travel along a road segment is a convex and increasing function of the inflow .
[ lem : traveltime ] let @xmath37 satisfy * ( q ) * with @xmath43 and call @xmath44 .
then , the travel time @xmath45 , which is defined by @xmath46 where @xmath47 is of class @xmath48 ; { { \mathbb{r}}}^+)$ ] , weakly increasing and convex .
the proof follows directly from lemma [ lem : propertiesv ] .
when @xmath49 is a route consisting of the adjacent roads @xmath50 , the travel time @xmath51 along @xmath49 is then defined as the sum @xmath52 of the travel times of all roads .
a network consists of several routes connecting @xmath0 to @xmath1 .
to describe it , we enumerate each single road ( or edge ) and construct the matrix @xmath53 setting @xmath54 we now assign a constant total inflow @xmath55 at @xmath0 and call @xmath56 the fraction of the drivers that reach @xmath1 along the route @xmath57
. a single road may well belong to more than one route , so that the flow along the road @xmath58 is @xmath59 and the travel time along that road results to be @xmath60 .
the total travel time @xmath61 along the @xmath62th route is in general a function of all partition parameters , more precisely @xmath63 from a global point of view , it is natural to evaluate the quality of a network through the mean global travel time @xmath64 or , using matrix notation @xmath65 $ ] , we find @xmath66 we call _ globally optimal _ a state @xmath67 that minimizes @xmath68 over @xmath69 , i.e. , @xmath70 .
this _ social optimum _ state conforms to wardrop s _ second principle _ , see @xcite .
[ prop : convexity ] let all road travel times @xmath71 be of class @xmath72 ; { { \mathbb{r}}}^+)$ ] , weakly increasing and convex .
then , the map @xmath68 is in @xmath72;{{\mathbb{r}}}^+)$ ] is convex . the proof is deferred to the appendix . for brevity , we call _ relevant _ those travel times @xmath61 such that @xmath73 .
[ def : eq ] a state @xmath74 is an _ equilibrium state _
if all relevant travel times coincide , i.e. , for all @xmath75 @xmath76 the common value @xmath77 of the travel times being the _ equilibrium time_. in other words , at equilibrium all drivers need the same time to go from @xmath0 to @xmath1 . a common criterion for optimality goes back to pareto .
[ def : pareto ] an equilibrium state @xmath78 is a _ local pareto point _ if there exists a positive @xmath79 such that for all @xmath80 if there exists a @xmath81 such that @xmath82 , then there exists also a @xmath83 such that @xmath84 .
in other words , no ( small ) perturbation of a pareto point may reduce all travel times . however , from a _
`` selfish '' _ point of view , each driver aims at reducing his / her own travel time .
it is then natural to introduce the following definition .
[ def : nash ] an equilibrium state @xmath85 is a _ local nash point _ if there exists a positive @xmath79 such that for all @xmath86 $ ] and all @xmath87 , @xmath88 where @xmath89 is the unit vector directed along the @xmath81th axis . in other words , it is not convenient for @xmath90 drivers to change from route @xmath83 to route @xmath81 , for any @xmath91 .
[ ex : simple ] consider the simple case of the network in figure [ fig : simple ] , and assume that its dynamics is described as follows : [ cols="^,^,^,^,^ " , ] the maximal inflow @xmath55 at @xmath0 that , for any @xmath92 $ ] , can be partitioned in @xmath93 along @xmath2 and @xmath94 along @xmath4 is @xmath95 . with this constant inflow as left boundary data in , the resulting ( stationary ) densities
are @xmath96 the corresponding constant traffic speeds @xmath97 inserted in , lead to the following travel times on the two roads : @xmath98 finally , the mean global travel time defined at is @xmath99 according to definition [ def : nash ] , we have a unique nash point at @xmath100 and a unique globally optimal state at @xmath101 , where @xmath102 , \end{cases } \qquad { \vartheta}_g = \begin{cases } 0 , & { \varphi}\in [ 0 , 1/12 ) , \\ \frac{1 + 4\ , { \varphi}}{16\ , { \varphi } } , & { \varphi}\in [ 1/12 , \sqrt{2}-1 ] . \end{cases}\ ] ] clearly , @xmath100 is also a pareto point according to definition [ def : pareto ] . note that the globally optimal state may well differ from the nash optimal one and both depend on the total inflow @xmath55 , see figure [ fig : simplenum ] . with @xmath103 , so that @xmath104 and @xmath105.,scaledwidth=75.0% ] consider the network in figure [ fig:4roads ] .
the network is given by two routes , denoted @xmath106 and @xmath107 , connecting @xmath0 and @xmath1 .
the route @xmath106 consists of roads @xmath2 and @xmath4 , the route @xmath107 consists of roads @xmath5 and @xmath3 .
roads @xmath2 and @xmath3 have the same length @xmath108 and the same fundamental diagram @xmath37 .
similarly , roads @xmath4 and @xmath5 share the same length @xmath109 and the same flow density relation .
traffic is always assumed to be unidirectional from @xmath0 to @xmath1 , and no obstructions , e.g. , traffic lights , are encountered at the junctions . along each road , the dynamics of traffic is described by the lwr model with flux functions that lead to the travel times @xmath110 so that the travel time @xmath111 along the route @xmath106 and @xmath112 along the route @xmath107 , are @xmath113 then , @xmath114 is ( weakly ) increasing , while @xmath115 is ( weakly ) decreasing . since @xmath116 , we have that @xmath117 is a nash ( and also pareto ) point for this system .
it is easy to verify that @xmath118 is also globally optimal , since it is the argument that minimizes @xmath119 over the simplex @xmath120 .
we now introduce a new road in figure [ fig:4roads ] , passing to the network described in figure [ fig:5roads ] . the new road @xmath11 , which has the direction from @xmath2 to @xmath3 , has length @xmath121 and its dynamics is characterized by a flow function @xmath122 satisfying * ( q)*. the presence of the road @xmath11 allows us to consider the route @xmath49 connecting @xmath0 to @xmath1 consisting of the roads @xmath2 , @xmath11 , and @xmath3 .
for all @xmath123 $ ] such that @xmath124 , we now let the inflow @xmath125 enter @xmath106 , @xmath126 enter @xmath107 and the remaining @xmath127 enter @xmath49 .
the travel times along the three routes are then : @xmath128 observe that @xmath129 .
the mean global travel time is @xmath130 we now compare the travel times obtained in the two cases described by figures [ fig:4roads ] and [ fig:5roads ] . to this end , observe that the travel times @xmath131 and @xmath132 in the case of four roads , and referring to figure [ fig:4roads ] , are obtained from those in the @xmath133 roads case setting @xmath134 [ thm:4roads ] let the travel times @xmath135 ; { { \mathbb{r}}}^+)$ ] be non decreasing and assume that @xmath136 or @xmath137 are not constant . if the travel times defined in satisfy @xmath138 then : * @xmath139 is the unique local nash point for the network with five roads in figure [ fig:5roads ] ; * the corresponding equilibrium time @xmath140 is worse than the globally optimal configuration for the network with four roads in figure [ fig:4roads ] . under the above conditions we have the occurrence of the braess paradox
observe that the point @xmath141 is the unique pareto point for the five roads networks .
condition allows us to construct several examples illustrating the braess paradox .
[ ex : braess ] with the notation in figure [ fig:5roads ] , choose @xmath142 condition then becomes @xmath143 and , for any @xmath144 $ ] , it can easily be met for suitable @xmath145 , @xmath146 , see figure [ fig : ex : braess ] . with @xmath147 , @xmath148 , @xmath149 , @xmath150 . above ,
@xmath151 and @xmath152 ; below @xmath153 and the global travel time @xmath68 . the color scales to the right are the same in all figures and display the maximal and minimal values of the diagrams to their left.,title="fig:",scaledwidth=49.0% ] + with @xmath147 , @xmath148 , @xmath149 , @xmath150 . above ,
@xmath151 and @xmath152 ; below @xmath153 and the global travel time @xmath68 .
the color scales to the right are the same in all figures and display the maximal and minimal values of the diagrams to their left.,title="fig:",scaledwidth=49.0% ] +
our next aim is proving that in the case of the network in figure [ fig:5roads ] , a carefully chosen speed limit imposed on the novel road @xmath49 makes the nash optimal state coincide with the globally optimal one .
we use the same notation as in section [ subs : braess ] , but we use the travel time @xmath154 along the @xmath11 road as control parameter .
equivalently , we impose that the speed along the road @xmath49 is @xmath146 , so that @xmath155 the next theorem says that there exists an optimal control .
[ thm : control ] let the travel time @xmath156 ; { { \mathbb{r}}}^+)$ ] be non decreasing and convex , one of the two being strictly convex .
then , there exists a constant travel time @xmath157 such that the network in figure [ fig:5roads ] admits a partition @xmath158 which is a nash optimal state and also globally minimizes the mean global travel time .
thus , by carefully selecting the travel time , or , equivalently , adjusting the maximum speed , one can avoid the occurrence of the braess paradox .
moreover , the nash equilibrium is steered to become globally optimal .
[ lem : speed ] let @xmath37 satisfy * ( q)*. then , the speed @xmath159 defined by @xmath160 is well - defined , continuous in @xmath161 $ ] , strictly positive and weakly decreasing .
we find @xmath177 moreover , using the explicit expressions above , @xmath178
\frac{d^2}{d{\vartheta}^2 } \left(\frac{1}{v\left(\rho ( { \vartheta})\right)}\right ) & = - 2 \ , \frac{\rho ' ( { \vartheta } ) \ , { \varphi}}{\left(q\left(\rho ( { \vartheta})\right)\right)^3 } \\ & \quad \times \left [ \frac{1}{2 } \left(\frac{q\left(\rho ( { \vartheta})\right)}{q'\left(\rho ( { \vartheta})\right)}\right)^2 q''\left(\rho ( { \vartheta})\right ) + q\left(\rho ( { \vartheta})\right ) - \rho ( { \vartheta } ) \ , q'\left(\rho ( { \vartheta})\right ) \right ] .
\end{aligned}\ ] ] call @xmath179
. observe that @xmath180 and @xmath181 thereby completing the proof .
observe that if @xmath182 is convex and increasing , then also the map @xmath183 is convex and increasing . by lemma [ lem : propertiesrho ] , for all @xmath184 ,
the map @xmath185 is convex for @xmath186 $ ] .
hence , also the map @xmath187 is convex for @xmath188^n$ ] . since @xmath189 , also the map @xmath190 is convex . by definition [ def : nash ] ,
the configuration @xmath100 with @xmath191 is clearly an equilibrium , the only relevant time being the equilibrium @xmath192 by , it is also a nash point , since @xmath193 and , by continuity , the same inequality holds in a neighborhood of @xmath100 .
assume there exists an other equilibrium point @xmath194 in the interior of @xmath120 .
then , by symmetry , @xmath195 and , by definition [ def : nash ] , @xmath196 by assumption , the left - hand side above is a strictly increasing function of @xmath197 , while the right - hand side is weakly decreasing , so that @xmath198 which contradicts . to complete the proof of the uniqueness of the nash points ,
consider the configuration @xmath199 . in this case , the only relevant time is @xmath200 and @xmath201 proving that @xmath202 is not a nash point .
the case of @xmath199 is entirely analogous .
[ lem : theta ] let the travel time @xmath156 ; { { \mathbb{r}}}^+)$ ] be non decreasing and convex , at least one of the two being strictly convex .
then , there exists a map @xmath204)$ ] such that the partition @xmath205 is the point of global minimum of the mean travel time @xmath68 defined in , , over @xmath69 .
the travel time @xmath68 is convex by proposition [ prop : convexity ] . by symmetry ,
its minimum is attained at a point @xmath206 and if @xmath207 , then this point satisfies @xmath208 .
straightforward we find @xmath209 hence @xmath210 , which shows that the map @xmath211 is strictly convex .
hence it admits a unique point of minimum @xmath212 in @xmath213 .
the standard implicit function theorem ensures that @xmath214 is continuous .
[ lem : tt ] let the travel time @xmath156 ; { { \mathbb{r}}}^+)$ ] be non decreasing and convex , at least one of the two being strictly convex .
then , there exists a map @xmath215 ; { { \mathbb{r}}}^+)$ ] such that assigning the travel time @xmath216 on road @xmath11 makes the configuration @xmath206 the unique local nash point in the sense of definition [ def : nash ] . given @xmath217 $ ] , we seek a @xmath154 such that @xmath206 is an equilibrium point . to this aim , we solve @xmath218 by symmetry consideration , to former equality is certainly satisfied for any @xmath219 $ ] .
the latter is equivalent to : @xmath220 therefore , we set @xmath221 by construction , @xmath206 is an equilibrium configuration in the sense of definition [ def : eq ] , once the travel time @xmath154 along the road @xmath11 is set equal end @xmath222 . when @xmath207 , to prove that @xmath206 is a local nash point , thanks to the present symmetries , it is sufficient to check that for all small @xmath223 we have @xmath224 or , equivalently , @xmath225 and all these inequalities hold by the monotonicity of the travel times .
let @xmath214 and @xmath226 be the maps defined in lemma [ lem : theta ] and lemma [ lem : tt ] , respectively .
define @xmath227 \to [ 0,1/2 ] \qquad \mbox { by } \qquad \upsilon = \theta \circ \tilde t,\ ] ] and call @xmath228 a fixed point for @xmath229 . by construction
, @xmath158 is a local nash point , once @xmath230 is fixed as the travel time along road @xmath11 .
pala , s. baltazar , p. liu , h. sellier , b. hackens , f. martins , v. bayot , x. wallart , l. desplanque , and s. huant .
transport inefficiency in branched - out mesoscopic networks : an analog of the braess paradox .
2012(108 ) 076802 .
|
we show the existence of the braess paradox for a traffic network with nonlinear dynamics described by the lighthill whitham - richards model for traffic flow . furthermore
, we show how one can employ control theory to avoid the paradox .
the paper offers a general framework applicable to time - independent , uncongested flow on networks .
these ideas are illustrated through examples .
|
the hot and dense matter expected to be formed in the partonic phase after ultra - relativistic heavy ion collisions dynamically evolve in space and time due to high internal pressure . consequently the system cools and reverts back to hadronic matter from the partonic phase . just after the formation , the entire energy of the system is thermal in nature and with progress of time some part of the thermal energy gets converted to the collective ( flow ) energy . in other words , during the expansion stage the total energy of the system is shared by the thermal as well as the collective degrees of freedom .
the evolution of the collectivity within the system is sensitive to the equation of state ( eos ) .
therefore , the study of the collectivity in the system formed in the quark gluon plasma ( qgp ) phase will be useful to shed light on the eos @xcite ( see @xcite for review ) and on the nature of the transition that may take place during the evolution process .
it is well known that the average magnitude of radial flow at the freeze - out surface can be extracted from the transverse momentum ( @xmath0 ) spectra of the hadrons .
however , hadrons being strongly interacting objects can bring the information of the state of the system when it is too dilute to support collectivity _
i.e. _ the parameters of collectivity extracted from the hadronic spectra are limited to the evolution stage where the collectivity ceases to exist .
these collective parameters have hardly any information about the interior of the matter . on the other hand electromagnetic ( em ) probes , _
i.e. _ photons and dileptons are produced and emitted @xcite ( see @xcite for review ) from each space time points .
therefore , estimating radial flow from the em probes will shed light on the time evolution of the collectivity in the system @xcite .
the invariant momentum distribution of photons produced from a thermal source depends on the temperature ( @xmath1 ) of the source through the thermal phase space distributions of the participants of the reaction that produces the photon @xcite . as a result
the @xmath0 spectra of photon reflects the temperature of the source .
hence ideally the photons with intermediate @xmath0 values ( @xmath2 gev , depending on the value of initial temperature ) reflect the properties of qgp ( realized when @xmath3 , @xmath4 is the transition temperature ) .
therefore , one should look into the @xmath0 spectra for these values of @xmath0 for the detection of qgp .
however , for an expanding system the situation is far more complex .
the thermal phase space factor changes - several factors _
e.g. _ the transverse kick received by low @xmath0 photons due to flow originating from the low temperature hadronic phase ( realized when @xmath5 ) populates the high @xmath0 part of the spectra @xcite . as a consequence
the intermediate or the high @xmath0 part of the spectra contains contributions from both qgp and hadrons .
for dilepton the situation is , however , different because in this case we have two kinematic variables - out of these two , the @xmath0 spectra is affected by the flow , however , the @xmath0 integrated invariant mass ( @xmath6 ) spectra is unaltered by the flow in the system .
it should be mentioned here that for @xmath6 below @xmath7 peak and above @xmath8 peak dileptons from qgp dominates over its hadronic counterpart ( assuming the contributions from hadronic cocktails are subtracted out ) within the framework of the present model . however
, the spectral function of low mass vector mesons ( mainly @xmath7 ) may shift toward lower invariant mass region due to non - zero temperature and density effects . as a consequence the contributions from the decays of such vector mesons to lepton pairs could populate the low @xmath6 window and may dominate over the contributions from the qgp phase @xcite ( and also @xcite for review ) . in the present work such thermal effects are not considered .
all these suggests that a judicious choice of @xmath0 and @xmath6 windows will be very useful to characterize the flow in qgp and hadronic phase . however , there are still some difficulties
. the calculations of em probes from thermal sources depend on the parameters like initial temperature ( @xmath9 ) , thermalization time ( @xmath10 ) , chemical freeze - out temperature ( @xmath11 ) , kinetic freeze - out temperature ( @xmath12 ) etc , which are not known uniquely . to minimize the dependence of thermal sources on these parameters the importance of the ratio of the transverse momentum spectra of photon
to dilepton has been emphasized previously @xcite in order to overcome the above mentioned uncertainties .
it may be mentioned here that in the limit of @xmath13 the lepton pairs ( virtual photons ) emerge as real photons .
therefore , the evaluation of the ratio of the @xmath0 spectra of photons to dileptons for various invariant mass bins along with a judicious choice of the @xmath0 and @xmath6 windows will be very useful to extract the properties of qgp as well as that of hadronic phase .
this will be demonstrated in the present work by analyzing wa98 and phenix photons and na60 and phenix dilepton spectra .
the paper is organized as follows . in the next section , the production of thermal photons and dileptons
are briefly outlined . in section iii
the expansion dynamics of the system is discussed .
section iv is devoted to results and discussions and finally in section v , we present summary and conclusions .
the @xmath0 and @xmath6 spectra of em probes measured in the experiments are mingled with all the sources of productions - broadly categorized as : ( i ) prompt productions resulting from the interactions of the partons of the colliding nuclei , ( ii ) thermal productions from the interactions of thermal partons as well as from thermal hadrons and ( iii ) finally from the decays of the long ( compared to strong interaction time scale ) lived mesons .
the contributions from @xmath14 collisions at a given collision energy can be used as a bench - mark to estimate the hard contributions . to estimate the thermal contributions we adopt the following procedure : _ thermal contributions = contributions from heavy ion collision minus @xmath15contributions from @xmath14 collisions _ where @xmath16 is the number of nucleon - nucleon interactions in the nuclear collisions at a given centrality .
instead of evaluating the hard contributions by applying pqcd we use the experimental data both for photons and dileptons from the @xmath14 collisions wherever available to minimize the uncertainties in the contributions of category ( i ) . for collisions with large nuclei _
@xmath17 or @xmath18 , the number of valence @xmath19 quarks are more than the number of valence @xmath20 quark because in these nuclei there are more neutrons than protons .
the magnitude of electric charge of the @xmath19 quark is half of that of @xmath20 quarks .
consequently , in the production of em probes from @xmath17 collisions - a larger number @xmath19 quarks with lesser charge and comparatively a smaller number of @xmath20 quarks with larger charge are involved .
therefore , for the photon production from @xmath17 interaction for the category ( i ) is not a mere superposition of the yield from @xmath21 interaction .
however , in the present work we concentrate in the kinematic region of central rapidity where the number of valence quarks are small . in fact , this is negligible for rhic energy .
it is worth mentioning here that the role of em probes produced from other mechanisms - like fragmentation of high energy quarks and the interaction of high energy partons with thermal qgp medium @xcite are ignored here .
it is expected that at the @xmath0 domain of our interest omission of these mechanism will not change the final results significantly . for the present work photons and dileptons from thermalized matter of partons and hadrons play the most crucial role .
the rate of thermal dilepton production per unit space - time volume per unit four momentum volume is given by @xcite @xmath22 where @xmath23 is the em coupling constant , @xmath24 is the imaginary part of the retarded photon self energy and @xmath25 is the thermal phase space distribution for bosons .
@xmath26=@xmath27 , arises from the final state leptonic current involving dirac spinors of mass @xmath28 . as mentioned before , in the limit of vanishing @xmath6 a lepton pair _
i.e. _ a virtual photon appears as a real photon .
therefore , the real photon production rate can be obtained from the dilepton emission rate by replacing the product of em vertex @xmath29 @xmath30 @xmath31 , the term involving final state leptonic current and the square of the ( virtual ) photon propagator by the polarization sum for the real photon . for an expanding system
@xmath32 should be replaced by @xmath33 , where @xmath34 and @xmath35 are the four momentum and the hydrodynamic four velocity respectively .
the hard thermal loop @xcite approximations has been used by several authors @xcite to evaluate the photon spectra originating from a thermal source of quarks and gluons .
the complete calculation of emission rate of photons from qgp to order o@xmath36 has been done by resuming ladder diagrams in the effective theory @xcite , which has been used in the present work .
a set of hadronic reactions with all possible isospin combinations have been considered for the production of photons @xcite from hadronic matter .
the effect of hadronic dipole form factors has been taken into account in the present work as in @xcite .
the lowest order process producing lepton pairs is @xmath37 and @xmath38 annihilation .
the correction of order @xmath39 to the lowest order rate of dilepton production from qgp has been calculated in @xcite , which is considered in the present work .
for the low @xmath6 dilepton production from the hadronic phase the decays of the light vector mesons @xmath40 and @xmath8 has been considered in @xcite .
the continuum part of the vector mesons spectral functions constrained by experimental data @xcite have been included here . as mentioned
before the contributions from the qgp phase dominates the @xmath6 spectra of the lepton pairs below @xmath7-peak and above the @xmath8-peak if no thermal effects of the spectral functions of the vector mesons ( see @xcite for review ) are considered .
since the continuum part of the vector meson spectral functions are included in the current work the processes like four pions annihilations @xcite are excluded to avoid double counting .
the space time evolution of the system formed in heavy ion collisions has been studied by using relativistic hydrodynamics with longitudinal boost invariance @xcite and cylindrical symmetry @xcite .
we assume that the system reaches equilibration at a time @xmath10 after the collision . the initial temperature , @xmath9 can be related to the measured hadronic multiplicity ( @xmath41 ) by the following relation : @xmath42 where @xmath43 is the radius of the system , @xmath44 is the riemann zeta function and @xmath45 where @xmath46 ( @xmath47 ) is the degeneracy of quarks and gluons in qgp , @xmath48=number of flavours .
the value of @xmath41 for various beam energies and centralities are calculated from the following equation @xcite : @xmath49 and tabulated in table i. lccccr @xmath50 & centrality & @xmath51&@xmath52&@xmath9(mev ) + 17.3 gev & 0 - 06%&700&1.0&200 + 200 gev & 0 - 20%&496&0.6&227 + & 20 - 40%&226&0.6&203 + & min .
bias&184&0.6&200 + @xmath53 is the number of collisions and contribute to @xmath54 fraction to the multiplicity @xmath55 measured in @xmath14 collision .
the number of participants , @xmath56 contributes fraction @xmath57 of @xmath55 .
the values of @xmath56 and @xmath53 are estimated by using glauber model and the results are in agreement with @xcite .
we have used @xmath58 and @xmath59 at @xmath60 gev .
it should be mentioned here that the values of @xmath41 ( through @xmath61 and @xmath16 in eq .
[ eq5 ] ) and hence the @xmath9 ( through @xmath41 in eq .
[ eq4 ] ) depend on the centrality of the collisions .
the values of @xmath43 for different centralities have been evaluated by using the equation @xmath62 .
some comments are in order here regarding the flow for non - central collisions which results in non - cylindrical geometrical shape of the system formed after the collisions . in the present work the non - centrality of the nuclear collisions is reflected in the initial temperature through hadronic multiplicity .
the geometry of the system due to non - centrality , should in principle be treated by ( 2 + 1 ) dimensional @xcite or for more rigorous treatment ( 3 + 1 ) dimensional hydrodynamical @xcite evolution .
however , we expect that the results obtained in the present work will not be affected substantially due to non - cylindrical geometric shape of the system . because the flow has been extracted from the ratio of photon to dilepton spectra for a given centrality .
we use the eos obtained from the lattice qcd calculations by the milc collaboration @xcite .
we consider kinetic freeze out temperature , @xmath12=140 mev for all the hadrons .
the ratios of various hadrons measured experimentally at different @xmath63 indicate that the system formed in heavy ion collisions chemically decouple at @xmath64 which is higher than @xmath12 which can be determined by the transverse spectra of hadrons @xcite .
therefore , the system remains out of chemical equilibrium from @xmath64 to @xmath12 .
the deviation of the system from the chemical equilibrium is taken in to account by introducing chemical potential for each hadronic species .
the chemical non - equilibration affects the yields through the phase space factors of the hadrons which in turn affects the productions of the em probes .
the value of the chemical potential has been taken in to account following ref .
it is expected that the chemical potentials do not change much for the inclusion of resonances above @xmath65 .
the prompt photons and dileptons ( drell - yan ) are normally estimated by using perturbative qcd . however , to minimize the theoretical model dependence here we use the available experimental data from @xmath14 collisions to estimate the hard photon and dilepton contributions in heavy ion collisions .
the wa98 photon spectra from pb+pb collisions is measured at @xmath66 gev .
however , no data at this collision energy is available for pp interactions . therefore , prompt photons for p+p collision at @xmath67 gev has been used @xcite to estimate the hard contributions for nuclear collisions at @xmath68 gev .
appropriate scaling ( see @xcite for details ) has been used to obtain the results at @xmath66 gev . for the pb+pb collisions
the result has been appropriately scaled by the number of collisions at this energy ( this is shown in fig .
[ fig1 ] as prompt photons ) .
the high @xmath0 part of the wa98 data is reproduced by the prompt contributions reasonably well . at
low @xmath0 the hard contributions under estimate the data indicating the presence of a thermal source .
the thermal photons with initial temperature @xmath69 mev along with the prompt contributions explain the wa98 data well ( fig .
[ fig1 ] ) , with the inclusion of non - zero chemical potentials for all hadronic species considered @xcite(see also @xcite ) . in some of the previous works @xcite the effect of chemical freeze - out
is ignored . as a result
either a higher value of @xmath9 or a substantial reduction of hadronic masses in the medium was required @xcite . in the present work
, the data has been reproduced without any such effects . following a similar procedure ,
the data @xcite from au - au collisions at @xmath60 gev has been reproduced well by adding the prompt contributions ( which is constrained by pp data at the same energy ) to the thermal photons .
the reproduction of data is satisfactory ( fig .
[ fig2 ] ) for all the centralities with the initial temperature shown in table i ( see also @xcite . the transverse mass distribution of dimuons produced in in+in collisions at @xmath66 gev has been evaluated for different invariant mass ranges ( see @xcite for details ) .
the quantity @xmath70 has been obtained by integrating the production rates over invariant mass windows @xmath71 to @xmath72 and @xmath73 is defined as @xmath74 where @xmath75 .
the results are compared with the data obtained by na60 collaborations @xcite at sps energy ( fig .
[ fig3 ] ) .
theoretical results contain contributions from the thermal decays of light vector mesons ( @xmath7 , @xmath76 and @xmath8 ) and also from the decays of vector mesons at the freeze - out @xcite of the system has also been considered .
the non - monotonic variation of the effective slope parameter extracted from the @xmath73 spectra of the lepton pair with @xmath77 evaluated within the ambit of the present model @xcite reproduces the na60 @xcite results reasonably well . for au+au collisions at @xmath50=200 gev
, we have evaluated the dilepton spectra for different invariant mass bins with the initial condition ( min bias ) shown in table i and lattice qcd equation of state .
the results are displayed in fig .
the slopes of the experimental data on @xmath0 distribution of lepton pairs for different invariant mass windows measured by the phenix collaboration @xcite could be reproduced well with the same initial condition that reproduces photon spectra @xcite .
in fact , the reproduction of data for the higher mass windows @xmath78m(gev)@xmath79 and @xmath80m(gev)@xmath81 do not need any normalization factors ( fig .
[ fig4 ] ) . for lower mass windows slopes
are reproduced well but fail to reproduce the absolute normalization .
therefore , it should be clarified here that the theoretical results shown in fig .
[ fig4 ] for lower mass windows ( to be precise for @xmath82m(gev)@xmath83 , @xmath84m(gev)@xmath85 and @xmath86m(gev)@xmath87 ) contain arbitrary normalization constant .
however , it should also be mentioned at this point that for the extraction of the flow within the present approach the absolute normalization is not essential , which is essential is the slope ( eq . [ radvel ] ) .
therefore , the non - reproduction of the absolute normalization of the @xmath0 spectra of lepton pairs for the lower mass windows does not affect the extraction of the magnitude of the radial flow . with @xmath0 for different invariant mass windows at sps energy
( see text ) . ] with @xmath0 for different invariant mass windows at rhic energy ( see text ) . ]
as mentioned before some of the uncertainties prevailing in the individual spectra may be removed by taking the ratio , @xmath88 of the @xmath0 distribution of thermal photon to dileptons . in the absence of experimental data for both photon and dilepton from the same colliding system for sps energies , we have calculated the ratio @xmath88 for pb+pb system , where the initial condition , the freeze - out condition and the eos are constrained by the measured wa98 photon spectra .
the results are displayed in fig .
[ fig5 ] .
next we evaluate the ratio of the thermal photon to dilepton spectra constrained by the experimental data from au+au collisions measured by phenix collaboration .
the results for the thermal ratio , @xmath88 displayed in fig .
[ fig6 ] is constrained by the experimental data .
the behaviour of @xmath88 with @xmath0 for different invariant mass windows which is extracted from the available data is similar to the theoretical results obtained in ref .
@xcite .
gev and 1.2 gev at sps energy . ]
= 0.625 gev and 0.9 gev at rhic energy . ] the ratio , @xmath89 for different @xmath6 windows ( figs . [ fig5 ] and [ fig6 ] ) can be parametrized as follows : @xmath90 with different values of @xmath91 for different invariant mass windows .
the argument of the exponential in eq .
[ ratpat ] can be written as @xcite ; @xmath92 where , @xmath93 is the blue shifted effective temperature for massless photons and @xmath94 , is the effective temperature for massive dileptons .
@xmath95 is the average temperature and @xmath96 is the average radial flow of the system . for a given @xmath0 and
@xmath6 eq . [ radvel ] can be written as @xmath97 .
the @xmath91 obtained from the parametrization of ratio at sps energy are 263 mev and 243 mev for m=0.75 and 1.2 gev respectively .
the average flow velocity @xmath96 versus @xmath95 have been displayed for m=0.75 gev and 1.2 gev in fig .
the hadronic matter ( qgp ) dominates the @xmath98 gev region .
therefore , these two mass windows are selected to extract the flow parameters for the respective phases . the @xmath96 increases with decreasing @xmath99 ( increase in time ) and reaches its maximum when the temperature of the system is minimum , i.e. , when the system attains @xmath12 , the freeze - out temperature .
therefore , the variation of @xmath96 with @xmath99 may be treated as to show how the flow develops in the system .
the @xmath96 is larger in the hadronic phase because the velocity of sound in this phase is smaller , which makes the expansion slower as a consequence system lives longer - allowing the flow to fully develop . on the other hand
, @xmath96 is smaller in the qgp phase because it has smaller life time where the flow is only partially developed . in fig .
[ fig8 ] the variation of average transverse velocity with average temperature for rhic initial conditions is depicted .
the magnitude of the flow is larger in case of rhic than sps because of the higher initial pressure . because of the larger initial pressure and qgp life time the radial velocity for qgp at rhic is larger compared to sps .
obtaining @xmath100 and @xmath101 from the individual spectra and eliminating @xmath99 one gets the variation of @xmath96 with @xmath6 .
[ fig9 ] ( left panel ) shows the variation of @xmath96 with @xmath6 for sps conditions .
the radial flow velocity increases with invariant mass m up to @xmath102 then drops .
how can we understand this behaviour ?
from the invariant mass spectra it is well known that the low @xmath6 ( below @xmath7 mass ) and high m ( above @xmath8 peak ) pairs originate from a partonic source @xcite .
the collectivity ( or flow ) does not develop fully in the qgp because of the small life time of this phase . which means that the radial velocity in qgp will be smaller for both low and high @xmath6 .
whereas the lepton pairs with mass around @xmath7-peak mainly originate from a hadronic source ( at a late stage of the evolution of system ) are largely affected by the flow resulting in higher values of flow velocity . in summary ,
the value of @xmath96 for @xmath6 below and above the @xmath7-peak is small but around the @xmath7 peak is large - with the resulting behaviour displayed in fig .
similar non - monotonic behaviour is observed in case of elliptic flow of photon as a function of @xmath0 @xcite .
the variation of @xmath96 with @xmath6 in rhic ( fig .
[ fig9 ] right panel ) is similar to sps , though the values of @xmath96 at rhic is larger than that of sps as expected due to higher initial pressure .
the photon and dilepton spectra measured at sps and rhic energies by different experimental collaborations have been studied .
the initial conditions have been constrained to reproduce the measured multiplicity in these collisions .
the eos , the other crucial input to the calculations has been taken from lattice qcd calculations .
the deviation of the hadronic phase from chemical equilibrium is taken in to account by introducing non - zero chemical potential for each hadronic species .
it is shown that simultaneous measurements of photon and dilepton spectra in heavy ion collisions will enable us to quantify the evolution of the average radial flow velocity for the system and the nature of the variation of radial flow with invariant mass indicate the formation of partonic phase at sps and rhic energy .
* acknowledgment : * we are grateful to tetsufumi hirano for providing us the hadronic chemical potentials .
this work is supported by dae - brns project sanction no .
2005/21/5-brns/2455 .
j. kapusta , p. lichard , and d. seibert , phys . rev .
d * 44 * ( 1991 ) 2774 ; r. bair , h. nakkagawa , a. niegawa , and k. redlich , z. phys .
c * 53 * ( 1992)433 ; p. aurenche , f. gelis , r. kobes , and h. zaraket , phys .
d * 58 * ( 1998 ) 085003 .
p. arnold , g.d .
moore , and l.g .
yaffe , j. high energy phys .
* 0111 * ( 2001 ) 057 ; p. arnold , g.d .
moore , and l.g .
yaffe , j. high energy phys .
* 0112 * ( 2001 ) 009 ; p. arnold , g.d .
moore , and l.g .
yaffe , j. high energy phys .
* 0206 * ( 2002 ) 030 .
i. arsene _
( brahms collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 1 ; b. b. back _ et al . _
( phobos collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 28 ; j. adams _ et al . _ ( star collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 102 ; k. adcox _
( phenix collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 184 .
|
an attempt has been made to extract the evolution of radial flow from the analysis of the experimental data on electromagnetic probes measured at sps and rhic energies .
the transverse momentum ( @xmath0 ) spectra of photons and dileptons measured by wa98 and na60 collaborations respectively at the cern - sps and the photon and dilepton spectra obtained by phenix collaboration at bnl - rhic have been used to constrain the theoretical models .
we use the ratio of photon to dilepton spectra to extract the flow , where some model dependence are canceled out . within the ambit of the present analysis
we argue that the variation of the radial velocity with invariant mass is indicative of a phase transition from initially produced partons to hadrons at sps and rhic energies .
|
we study geometry of 3-dimensional spaces which are isometric gluing of ( ideal and hyper - ideal ) tetrahedra in hyperbolic spaces .
our main focus is on the rigidity of these spaces .
the metrics of these spaces are given by the lengths of edges of tetrahedra ( in the underlying triangulation ) .
the curvatures of the spaces are @xmath1 less the cone angles at the edges .
our main results state that for a fixed triangulation , the curvature determines the edge lengths and hence these hyperbolic polyhedral metrics .
the tool used in the proof is a variational principle associated to the schlaefli formula and its legendre transformation .
the infinitesimal rigidity of hyperbolic cone metrics follows from the strict convexity of the volume of the ideal and hyper - idea tetrahedra in terms of the dihedral angles . in the dual setting , one considers the co - volume which is the dual of the volume of the tetrahedra and has the edge lengths as the variables .
the main difficult comes from the fact that the space of hyperbolic tetrahedra parametrized by the edge lengths is not convex .
we overcome the difficulty by showing that the co - volume function ( of the edge lengths ) can be extended to a @xmath2 smooth convex function defined on a convex open set .
this is very similar to the results established in @xcite . by establishing the convex extensions of the co - volume functions , we are able to prove several results on the volume optimization program of casson and rivin . for instance
, we show that the maximum volume angle structures are exactly those coming from the generalized polyhedral metrics ( see theorem 1.3 ) .
we now state our results more precisely .
suppose @xmath3 is a triangulated compact pseudo 3-manifold with a triangulation @xmath4 and the set of edges @xmath5 a _ decorated hyperbolic polyhedral metric _
( respectively _ hyper - ideal polyhedral metric ) _ on @xmath3 is obtained by replacing each tetrahedra in @xmath4 by an decorated ideal tetrahedron ( respectively hyper - ideal tetrahedron ) and replacing the affine gluing homeomorphisms by isometries preserving the decoration .
the _ curvature _ of the metric assign each edge @xmath6 @xmath1 less the cone angle at @xmath6 for interior edge @xmath6 and @xmath7 less the cone angle for boundary edge @xmath8 by the construction , these polyhedral metrics are determined by the lengths of the edges .
[ main ] suppose @xmath3 is a triangulated compact pseudo 3-manifold @xmath9 a. a decorated hyperbolic polyhedral metric on @xmath3 is determined up to isometry and change of decorations by its curvature .
b. a hyper - ideal hyperbolic polyhedral metric on @xmath3 is determined up to isometry by its curvature .
the rigidity results are closely related to the volume optimization of angle structures initiated by casson and rivin .
the program tries to find complete hyperbolic metrics on ideal triangulated 3-manifolds @xmath10 using _ angle structures ( see @xcite , @xcite ) .
recall that a non - negative ( respectively positive ) angle structure @xmath11 on @xmath10 assigns each edge in a tetrahedron a non - negative ( respectively positive ) number called the angle so that the sum of angles around each edge is @xmath1 and the sum of angles at three edges from each vertex of each tetrahedron is @xmath12 the _ volume of an angle structure @xmath11 is well defined using the lobachevsky function ( see 2.1 ) .
it is well known there are maximum volume non - negative angle structures .
if the maximum volume angle structure is positive , then it is known @xcite that there exists a geometric triangulation of a complete hyperbolic metric on the manifold @xmath13 realizing the angle structure .
our result gives a characterization of maximum angle structures in the case some angles are 0 .
_ _ [ 1.3 ] suppose @xmath10 is a triangulated closed pseudo 3-manifold which supports a positive angle structure and @xmath11 is a non - negative angle structure which maximizes volume in the space of all non - negative angle structures on @xmath14 then there exists an assignment of real number @xmath15 to each edge @xmath6 so that for each tetrahedron @xmath16 \(1 ) if all angles of @xmath17 in @xmath11 are positive , then @xmath11 are the dihedral angles of the decorated ideal tetrahedron of edge lengths given by @xmath18 and , \(2 ) if one angle of @xmath17 in @xmath11 is 0 , then all angles of @xmath17 in @xmath11 are @xmath19 and their edge lengths in @xmath18 satisfy @xmath20 where @xmath21 is the length of the edge of angle @xmath7 and @xmath22 and @xmath23 are lengths of opposite edges .
conversely , if @xmath24 is any function so that ( 1 ) and ( 2 ) hold , then the corresponding angle @xmath11 of @xmath18 defined by ( 1 ) and ( 2 ) maximizes volume . in section 6
, we introduce the corresponding notion of non - negative and positive angle structures of hyper - ideal type ( see definition [ angle - hyper ] ) , and prove the following counterpart of theorem 1.3 . [ 1.4 ] suppose @xmath10 is a triangulated closed pseudo 3-manifold which supports a positive angle structure of hyper - ideal type and @xmath11 maximizes the volume in the space of all non - negative angle structures of hyper - ideal type on @xmath14 then there exists an assignment of positive number @xmath15 to each edge @xmath6 so that for each tetrahedron @xmath16 \(1 ) if all angles of @xmath17 in @xmath11 are positive , then @xmath11 are the dihedral angles of the hyper - ideal tetrahedron of edge lengths given by @xmath18 and , \(2 ) if one angle of @xmath17 in @xmath11 is 0 , then all angles of @xmath17 in @xmath11 are @xmath19 and the numbers assigned by @xmath18 to the edges of @xmath17 are not the edge lengths of any hyper - ideal tetrahedron .
conversely , if @xmath25 is any function so that ( 1 ) and ( 2 ) holds , then the corresponding angle @xmath11 of @xmath18 defined by ( 1 ) and ( 2 ) maximizes volume .
there have been many important work on rigidity of hyperbolic cone metrics on 3-manifolds .
see work of hodgeson - kerckhoff@xcite , weiss@xcite , mazzeo - montcouquiol@xcite and fillastre - izmestiev@xcite and others .
the difference between their work and ours is that we consider the case where the singularity consists of complete geodesics from cusp to cusp or geodesics orthogonal to the totally geodesic boundary with possible cone singularities .
the paper is organized as follows . in sections 2 ,
we collect preliminary materials including decorated hyperbolic tetrahedra , angle structures , volume functions and the fenchel dual . in section 3
, we define the co - volume function and reveal its relationship with the fenchel dual of the volume function . as a consequence , we prove theorem [ main ] ( a ) and theorem [ 1.3 ] .
the second part of the paper focuses on polyhedral metrics and angle structures of hyper - ideal type .
the corresponding preliminary materials are included in section 4 , and theorem [ main ] ( b ) and theorem [ 1.4 ] are respectively proved in section 5 and section 6 .
the work is partially supported by the nsf dms 1105808 and nsf dms 1222663 .
we would like to thank steven kerckhoff for discussions .
since this paper involves topological triangulations , geometry of tetrahedra in hyperbolic space @xmath26 and convex optimization , we will briefly recall the related material in this section .
take a finite disjoint collection @xmath27 of euclidean tetrahedra and identify some of the codimension-1 faces in @xmath27 in pairs by affine homeomorphisms .
the quotient space @xmath3 is a compact pseudo 3-manifold @xmath28 together with a triangulation @xmath4 whose simplexes are the quotients of simplexes in @xmath29 if each codimension-1 face of @xmath27 are identified with another codimension-1 face , then @xmath28 is a closed pseudo 3-manifold .
otherwise , @xmath28 is a compact pseudo 3-manifold with non - empty boundary which is the quotient of the union of un - identified codimension-1 faces in @xmath29 two edges of tetrahedra in @xmath27 are called equivalent if they are mapped to the same set in @xmath30 we define _ edges in the triangulation @xmath4 to be equivalence classes of edges in tetrahedra in @xmath29 we use @xmath31 and @xmath32 to denote the sets of all edges and tetrahedra in @xmath4 respectively .
since a tetrahedron in the triangulation @xmath4 is the same as a tetrahedron in the original set @xmath33 we will identify @xmath32 with the set @xmath29 a _ quad in the triangulation @xmath4 is a pair of opposite edges in @xmath29 thus each tetrahedron contains three quads
. we use @xmath34 to denote the set of all quads in @xmath35 if @xmath36 @xmath37 and @xmath38 we use @xmath39 to denote that the quad @xmath40 is contained in the tetrahedron @xmath17 and use @xmath41 or @xmath42 to denote that @xmath43 and there exists @xmath44 with @xmath45 and @xmath46 . using these notations and the fact that angles at opposite edges in a tetrahedron are the same for any angle structure , a _
non - negative angle structure on a closed pseudo 3-manifold @xmath3 is a map @xmath47 so that ( 1 ) @xmath48 @xmath49 and ( 2 ) @xmath50 @xmath51 see for instance @xcite for more details . _ _ _ an _ ideal n - simplex @xmath52 in the hyperbolic n - space @xmath53 is the convex hull of n+1 points @xmath54 in @xmath55 so that @xmath56 are not in a round @xmath57-sphere .
any two ideal triangles are isometric .
an ideal tetrahedron in @xmath26 is determined up to isometry by its six dihedral angles .
these angles satisfy the condition that angles at opposite edges are the same and the sum of all angles is @xmath58 thus the space of all ideal tetrahedra modulo isometry can be identified with @xmath59 _ following penner @xcite , a _ decorated ideal n - simplex ( or simply decorated simplex ) is a pair @xmath60@xmath61 .@xmath62.,@xmath63 where @xmath52 is an ideal n - simplex and @xmath64 is an @xmath57-horosphere centered at the i - th vertex @xmath65 we call @xmath66 the _ decoration .
two decorated simplexes are _ equivalent if there is a decoration preserving isometry between the underlying ideal simplexes .
each edge @xmath67 in a decorated simplex has the signed _ length @xmath68 defined as follows .
the absolute value @xmath69 of the length is the distance between @xmath70 and @xmath71 so that @xmath72
if @xmath64 and @xmath73 are disjoint and @xmath74 if @xmath75 faces of decorated ideal simplexes are decorated ideal simplexes . also @xmath76 is isometric to a euclidean @xmath57-simplex . _ _ _ _ for a decorated ideal triangle
@xmath77 we call the length @xmath78 of the horocyclic arc in @xmath64 bounded by the two edges of @xmath52 from @xmath79 the _ angle at @xmath65 penner s cosine law says that for @xmath80 @xmath81given any three real numbers @xmath82 there exists a unique decorated ideal triangle whose lengths are @xmath83 see @xcite .
_ the characterization of the lengths of decorated ideal tetrahedron is well known ( see for instance @xcite lemma 2.5 , or @xcite lemma 4.2.3 ) .
[ l1 ] suppose @xmath84 are the edge lengths of a decorated ideal tetrahedron @xmath85 then all four euclidean triangles @xmath86 are similar to the euclidean triangle @xmath87 of edge lengths @xmath88 so that @xmath89 for @xmath90 the dihedral angle @xmath91 of @xmath17 at the edge @xmath92 is equal to the inner angle of @xmath87 opposite to the edge of length @xmath93 conversely , if @xmath94 satisfies the triangular inequalities above , then there is a unique decorated ideal tetrahedron having @xmath68 as the length at the edge @xmath95 one consequence of the lemma is that dihedral angles @xmath96 i.e. , dihedral angles at opposite edges are the same . thus , we can talk about the dihedral angle of a quad in a decorated simplex .
a _ generalized decorated tetrahedron is a ( topological ) 3-simplex of vertices @xmath97 so that each edge @xmath92 is assigned a real number @xmath98 called the _ length . a decorated ideal tetrahedron ( with the signed edge lengths ) is a generalized decorated tetrahedron .
the space of all generalized decorated tetrahedra parameterized by the length vectors @xmath99 is @xmath100 the subspace of all ( equivalence classes ) of decorated ideal tetrahedra is given by @xmath101 _ _ to define _ dihedral angles and _ volume of a generalized decorated tetrahedron @xmath102
let us begin with the notion of _ generalized euclidean triangles and their _ angles .
generalized euclidean triangle @xmath103 is a ( topological ) triangle of vertices @xmath104 so that each edge is assigned a positive number , called edge length .
let @xmath105 be the assigned length of the edge @xmath106 where @xmath107 the _ inner angle @xmath108 at the vertex @xmath79 is defined as follows .
if @xmath109 satisfy the triangle inequalities that @xmath110 for @xmath111 then @xmath78 is the inner angle of the euclidean triangle of edge lengths @xmath112 opposite to the edge of length @xmath105 ; if @xmath113 then @xmath114 @xmath115 it is known ( see for instance @xcite ) that _ _ _ _ _ _ [ lemma2.2 ] the angle function @xmath116 $ ] is continuous so that @xmath117 and the @xmath118-smooth differential 1-form @xmath119 is closed on @xmath120 furthermore , for @xmath121 the integral @xmath122 is a @xmath2-smooth convex function in @xmath123 on @xmath124 so that @xmath125 is strictly convex when restricted to @xmath126 and @xmath127 for all @xmath128 . here
a @xmath118-smooth 1-form is defined to be closed if its integration over any @xmath2-smooth null homotopic loop is zero .
for a generalized decorated tetrahedron of length vector @xmath129 the _ dihedral angle @xmath130 at the edge @xmath92 is defined to be the inner angle of the generalized euclidean triangle of edge lengths @xmath131 @xmath132 and @xmath133 so that @xmath91 is opposite to the edge of length @xmath134 for @xmath135 distinct .
in particular , the dihedral angles at opposite edges are the same and the total sum of all six dihedral angles are @xmath58 if @xmath17 is a decorated ideal tetrahedron , then the two definitions of dihedral angles coincide .
_ recall that the lobachevsky function is defined by @xmath136 it is a continuous function of period @xmath7 so that @xmath137 the _ volume of a generalized decorated simplex @xmath17 of lengths @xmath68 s , denoted by @xmath138 is defined to be @xmath139 if the generalized decorated tetrahedron @xmath17 is a decorated ideal tetrahedron , then lobachevsky showed that @xmath140 is the hyperbolic volume of the underlying ideal tetrahedron . if @xmath17 is not a decorated ideal tetrahedron , then by definition , @xmath141 _ since the space of all ideal tetrahedra can be parameterized by @xmath142 the volume function @xmath143 defined on @xmath144 is given by @xmath145 by lemma [ lemma2.2 ] , we have [ pp2 ] \(1 ) the function @xmath146 is continuous .
( 2)(rivin ) the volume function @xmath147 is smooth strictly concave and extends continuously to the closure @xmath148 so that @xmath149 if one of @xmath150 is 0 .
\(3 ) the volume function on the space of generalized decorated tetrahedra @xmath151 @xmath152 is continuous .
recall that a _ proper convex function @xmath153 $ ] is a convex function so that @xmath154 for some @xmath155 a function @xmath156 $ ] is called _ lower semi - continuous if for all @xmath157 @xmath158 if @xmath159 is a non - empty closed convex set and @xmath160 is a convex lower semi - continuous function , then the new function @xmath161 $ ] defined by @xmath162 and @xmath163 for @xmath164 is a proper lower semi - continuous convex function . see for instance the classical book @xcite for details . _ _
if @xmath165 we use @xmath166 or @xmath167 to denote the standard inner product of @xmath168 and @xmath169 the fenchel dual @xmath170 $ ] of a proper function @xmath171 $ ] is @xmath172 it is known that if @xmath173 is a proper convex function then @xmath174 is a proper convex lower semi - continuous function .
a fundamental fact about @xmath174 is the following ( see @xcite ) , if @xmath171 $ ] is a proper lower semi continuous convex function , then @xmath175 since the volume function @xmath176 is concave and continuous , we obtain a proper lower semi continuous convex function @xmath177 $ ] defined by @xmath178 if @xmath179 and @xmath180 if @xmath181 [ prop2.6 ] the fenchel dual @xmath182 of @xmath183 is the @xmath2-smooth convex function defined by @xmath184 where @xmath185 are inner angles of the generalized euclidean triangle of edge lengths @xmath186 @xmath187 and @xmath188 so that @xmath78 is opposite to the edge of length @xmath189 furthermore , \(1 ) @xmath190 and \(2 ) if @xmath191 so that @xmath192 then the convex function @xmath193 satisfies that @xmath194 for all @xmath195 and @xmath196 for @xmath197 define @xmath198 by definition , @xmath199 let @xmath200 if @xmath201 then @xmath202 has a critical point at @xmath203 in @xmath204 indeed , for the tangent vector @xmath205 to @xmath206 @xmath207 at @xmath208 due to the sine law .
since @xmath202 is a strictly concave function in @xmath209 it follows that @xmath210 is the unique maximum point of @xmath202 in @xmath206 i.e. , @xmath211 if @xmath212 say @xmath213 then @xmath114 @xmath115 the same calculation above shows that @xmath214 has no critical points in @xmath204 therefore @xmath215 on the other hand @xmath216 we obtain @xmath217 but for @xmath218 @xmath219 so that equality holds for @xmath220 with @xmath221 and @xmath222 therefore , @xmath223 due to @xmath114 @xmath115 in particular , we see that @xmath182 is convex and hence continuous on @xmath224 to show @xmath2-smoothness of @xmath225 we first note that by the sine law for euclidean triangles of angles @xmath226 and lengths @xmath227 @xmath228 see for instance @xcite
. therefore , if @xmath229 @xmath230 i.e. , @xmath231 in the open set @xmath232 we have @xmath233 hence we also have @xmath234 on @xmath235 on the other hand , the @xmath2-smooth function @xmath236 defined in lemma [ lemma2.2 ] satisfies @xmath237 on @xmath224 therefore these two functions @xmath182 and @xmath125 have the same gradient on the open dense subset @xmath238 since @xmath125 and @xmath182 are continuous , these two functions defer by a constant . hence @xmath182 is @xmath2-smooth . to prove the last statement , note that @xmath194 and convexity of @xmath239 follow from the definition .
furthermore by lemma [ lemma2.2 ] , @xmath239 is strictly convex when restricted to @xmath240 let @xmath241 be the vector so that the euclidean triangle of edge lengths @xmath242 has inner angles @xmath243 and @xmath244 then @xmath245 consider the plane latexmath:[$p=\{y \in { \mathbb r}^3 is a convex function with a minimal point @xmath18 so that @xmath239 is strictly convex near @xmath247 therefore @xmath248 due to @xmath249 this shows , by projecting @xmath124 to @xmath250 that @xmath251 we remark that the function @xmath252 $ ] has appeared before in the work of cohn - kenyon - propp @xcite and bobenko - pinkahl - springborn @xcite .
the proposition is similar to the work of colin de verdire @xcite .
suppose @xmath3 is a triangulated compact pseudo 3-manifold with the sets of edges @xmath31 and quads @xmath253 an angle assignment on @xmath3 is a map @xmath254 so that for each tetrahedron @xmath255 @xmath256 an angle assignment is called positive if @xmath257 for all @xmath258 the _ cone angle _ of @xmath11 is defined to be @xmath259 where @xmath260 the _ volume _ of @xmath261 denoted by @xmath262 is defined to be @xmath263 given @xmath264 the space of all angle assignments of cone angle @xmath265 is denoted by @xmath266 and the space of all positive angle assignments of cone angle @xmath265 is denoted by @xmath267 by definition , @xmath268 and @xmath269 note that if @xmath270 then the closure of @xmath271 in @xmath272 is @xmath273 however , it is possible that @xmath274 and @xmath275 the usual angle structures on closed pseudo 3-manifolds @xmath3 are positive angle assignments of cone angle @xmath1 at each edge . as a consequence of lemma [ pp2 ]
, we have the volume function @xmath276 is continuous , concave and is smooth strictly concave when restricted to @xmath267 a generalized decorated metric on @xmath3 is given by a map @xmath277 called the _ edge length function_. for each @xmath278 by replacing each tetrahedron @xmath17 in @xmath27 by a generalized decorated tetrahedron whose edge lengths are given by @xmath279 we define the _ dihedral angle of @xmath18 at an edge @xmath6 in a tetrahedron @xmath280 to be the corresponding dihedral angles in the generalized decorated tetrahedron whose edge lengths are given by @xmath247 since dihedral angles are the same at two opposite edges in a generalized decorated tetrahedron , the dihedral angle @xmath281 of the generalized decorated metric @xmath279 @xmath282 is an angle assignment , i.e. , @xmath283 @xmath284 the cone angle of @xmath11 is called the _ cone angle of the metric @xmath247 we denote it by @xmath285 i.e. , @xmath286 the _ volume of a generalized decorated metric @xmath287 is defined to be the volume of its dihedral angle assignment , i.e. , @xmath288 the _ covolume of @xmath278 denoted by @xmath289 is defined to be @xmath290 where @xmath291 is the cone angle of @xmath18 and @xmath166 is the standard inner product of @xmath292 by the definition of cone angles , we have @xmath293.\ ] ] _ _ _ _ [ prop2.10 ] the covolume function defined on the space of all generalized decorated tetrahedra @xmath294 is given by @xmath295 where @xmath296 are the edge lengths so that @xmath105 and @xmath297 are lengths of opposite edges .
in particular , @xmath298 is a @xmath2-smooth convex function so that @xmath299 when @xmath300 is the dihedral angle at the i - th edge . for @xmath301 let @xmath302 and @xmath303 be the angles of the generalized euclidean triangle of edge lengths @xmath304 then by proposition [ prop2.6 ] , @xmath305 $ ] and @xmath306 now by definition , the dihedral angles of generalized tetrahedron of lengths @xmath105 s are @xmath300 and @xmath307 for @xmath308 therefore by ( [ cov ] ) , @xmath309 = 2\phi^*(y).$ ] furthermore , for @xmath301 @xmath310 the result also works for @xmath311 due to @xmath312
a _ decorated ideal hyperbolic polyhedral metric on @xmath313 is a generalized metric @xmath287 so that each tetrahedron @xmath44 becomes a decorated ideal hyperbolic tetrahedron in the length @xmath247 these metrics are the same as complete finite volume hyperbolic cone metrics on @xmath314 which are obtained as isometric gluing of ideal tetrahedra along codimension-1 faces together with a horosphere centered at each cusp .
the collection of the horospheres is called a _ decoration . _
_ we will establish the main results for decorated ideal hyperbolic polyhedral metrics in this section .
these results imply theorems stated in 1 .
[ thm31 ] suppose @xmath3 is a compact triangulated pseudo 3-manifold . 1 .
if @xmath315 is a generalized decorated metric on @xmath3 with dihedral angle @xmath316 of cone angle @xmath317 then @xmath11 is a maximum volume point for @xmath143 on @xmath273 2 .
if @xmath318 so that @xmath319 and @xmath320 is a maximum volume angle assignment , then there exists a generalized decorated metric @xmath321 so that its dihedral angle function is @xmath322 furthermore , the maximum volume angle point @xmath11 is unique .
[ thm32 ] a decorated ideal hyperbolic polyhedral metric on @xmath313 is determined up to isometry and change of decoration by its cone angles at edges .
theorems [ thm31 ] and [ thm32 ] are consequences of a duality result for volume and covolume to be proved below . for a triangulation @xmath4 of edges
@xmath323 and vertices @xmath324 the vector space @xmath325 acts linearly on @xmath326 by @xmath327 where @xmath328 @xmath329 and the edge @xmath330 has vertices @xmath331 we will identify @xmath325 with the linear subspace @xmath332 of @xmath333 for each tetrahedron @xmath38 let @xmath334 and @xmath335 be the sets of edges and vertices in @xmath17 so that @xmath336 acts on @xmath337 according to ( [ action ] ) .
let @xmath338 be the composition of the restriction map and quotient map .
the following lemma was proved in @xcite , also see @xcite page 1354 .
[ choi ] the kernel of the linear map @xmath339 is @xmath340 in particular , the induced linear map @xmath341 is injective .
indeed , if @xmath342 is in the kernel , then for each tetrahedron @xmath102 we can find a function @xmath343 so that @xmath344 the goal is to show that if @xmath17 and @xmath345 are two tetrahedra sharing the same vertex @xmath346 then @xmath347 now if @xmath17 and @xmath345 have a common triangle face @xmath348 then the result follows by considering the three equations ( [ triangeq ] ) at three vertices of @xmath349 the geneneral case follows by producing a sequence of tetrahedra @xmath350 @xmath351 so that @xmath352 and @xmath353 have a common triangular face @xmath354 for a triangulated pseudo 3-manifold @xmath355 define @xmath356 to be the space of all cone angles of angle assignments .
for a tetrahedron @xmath17 with sets of edges @xmath334 and vertices @xmath357 the covolume function @xmath358 sends @xmath359 to @xmath360 $ ] is @xmath2-smooth and convex . by proposition [ prop2.10 ] , we have @xmath361 furthermore , by lemma [ lemma2.2 ] , for any @xmath362 and @xmath363 the function @xmath364 satisfies @xmath365 in particular , @xmath366 is a convex function defined on the quotient space @xmath367 furthermore , if @xmath368 so that @xmath369 for all @xmath370 , then @xmath371=+\infty.\ ] ] this follows from corollary 2.10 and lemma 2.2 .
for @xmath372 ( i.e. , @xmath373 for an angle assignment @xmath374 ) , consider the function @xmath375 [ prop3.4 ] \(1 ) for any @xmath376 @xmath377 \(2 ) if @xmath378 for all @xmath379 then @xmath380 = + \infty.\ ] ] in particular , @xmath381 has a minimal point in @xmath333 we can rewrite the function as @xmath382.\ ] ] therefore , statement ( 1 ) follows from that of covolume function for single tetrahedron ( [ 11234 ] ) .
part ( 2 ) follows from the condition that @xmath378 for all @xmath379 lemma [ choi ] and ( [ limit1 ] ) .
since the function @xmath383 is convex on @xmath384 and is invariant under the linear action of @xmath385 it is a convex function on the euclidean space @xmath386 which tends to infinity as @xmath220 tends to infinity .
therefore , @xmath383 has a minimal point in @xmath386 which implies that it has a minimal point in @xmath387 the space of all cone angles @xmath388 defined by ( [ cone ] ) is a compact convex polytope in @xmath387 indeed , it is the image of the compact convex polytope of all angle assignments in @xmath272 under a linear map . define @xmath389 to be the function @xmath390 the function @xmath391 encodes the volume optimization program .
for instance , @xmath392 is the casson - rivin s program of finding complete hyperbolic metrics of finite volume .
[ continuous ] the function @xmath393 is convex and lower semi - continuous in the compact convex set @xmath394 this proposition follows from the lemma [ 4561 ] below by taking @xmath395@xmath396@xmath397 @xmath398 and @xmath399 to be @xmath400 [ 4561 ] suppose @xmath401 is a compact convex set , @xmath402 is a continuous convex function and @xmath403 is linear . then @xmath404 is convex and lower semi - continuous on @xmath405 take @xmath406 and choose @xmath407 so that @xmath408 and @xmath409 for @xmath410 and @xmath411.$ ] then @xmath412 therefore , @xmath413 i.e. , @xmath202 is convex . to see lower semi continuity of @xmath414
suppose @xmath415 so that @xmath416 by the compactness of @xmath417 after selecting a subsequence , we may assume that @xmath418 so that @xmath419 in @xmath420 and @xmath421 clearly due to continuity @xmath422 and @xmath423 this shows that @xmath424 one of the two main technical results of the paper is the following , [ dual ] the fenchel dual of the @xmath2 smooth convex covolume @xmath425 is the lower semi - continuous convex function function @xmath426 $ ] defined by ( [ vol1 ] ) . by the fenchel duality theorem for convex function , it suffices to show the dual @xmath427 of @xmath428 is the function @xmath429 take @xmath430
we will show that if @xmath431 then @xmath432 and if @xmath433 then @xmath434 * case 1 . * @xmath435 suppose otherwise that @xmath436 i.e. , there exists @xmath437 so that @xmath438 for all @xmath439 since @xmath440 is uniformly bounded for all @xmath220 and @xmath441 there exists @xmath442
so that for all @xmath443 @xmath444 where @xmath445 is the cone angle of @xmath220 considered as a generalized decorated metric . consider the linear map @xmath446 defined by @xmath447 by the assumption that @xmath448 @xmath449 since @xmath450 is a closed convex cone in @xmath451 by the separation theorem applied to @xmath452 and @xmath453 there exists @xmath454 so that @xmath455 and for all @xmath456 @xmath457 where @xmath167 is the standard inner product in @xmath458 we can rewrite above two inequalities as @xmath459 and , @xmath460 + \sum_{\sigma \in t } [ h(\sigma ) \sum_{q \subset \sigma } t(q ) ] \leqslant 0.\ ] ] equation ( [ eqn:44 ] ) can be written as , @xmath461 since it holds for all @xmath462 hence for all @xmath370 @xmath463 now since ( [ eqn:221 ] ) holds for all @xmath443 it holds for the projection of @xmath464 to @xmath387 taking this projection as @xmath220 in ( [ eqn:221 ] ) , letting the diehedral angle of @xmath220 be @xmath465 and using ( [ eqn:31 ] ) and ( [ eqn:44 ] ) , we obtain @xmath466 here we have used ( [ eqn:31 ] ) and ( [ eqn:51 ] ) in steps 2 and 4 in the derivation above .
this is a contradiction .
therefore , @xmath467 * case 2 .
* @xmath433 say @xmath468 for an angle assignment @xmath469 where @xmath470 for this choice of @xmath471 and for any @xmath472 with dihedral angles @xmath473 we can write @xmath474 @xmath475.\ ] ] [ 3227 ] if @xmath476 and @xmath477 then for any @xmath478 so that @xmath303 are inner angles of the generalized euclidean triangle of edge lengths @xmath479 we have @xmath480 the lemma follows from proposition [ prop2.6 ] on convexity and @xmath2-smoothness of the covolume function @xmath481 ( for single tetrahedron ) . for any @xmath2-smooth convex function @xmath125 on @xmath482 we have
@xmath483 now using the fact that @xmath484 and take @xmath485 so that the angles of the generalized euclidean triangle of edge lengths @xmath242 are @xmath486 s , the result follows . summing up the inequalities in lemma [ 3227 ] over all tetrahedra in @xmath33
we obtain @xmath487 this implies , for any angle assignment @xmath488 we have @xmath489 and hence @xmath490 taking the supremum in @xmath220 of the inequality above , we obtain @xmath491 we claim that the equality holds .
there are two steps involved in the proof . in the first step ,
we assume that @xmath492 where @xmath493 in this case , by proposition [ prop3.4 ] , the function @xmath494 of variable @xmath220 has a minimal point @xmath495 in @xmath387 let @xmath496 be the dihedral angles of @xmath497 then by @xmath498 we obtain @xmath499 ( due to @xmath500 for @xmath501 ) .
therefore , at this point @xmath502 @xmath503 this shows @xmath504 combining with ( [ 432 ] ) , we conclude @xmath505 next , both @xmath506 and @xmath507 are convex and semi - continuous on the closed convex set @xmath508 so that they coincide in the subset @xmath509 of @xmath508 which contains the relative interior of @xmath510 therefore , @xmath511 on @xmath508 is a consequence of the lemma below .
[ relativeint ]
suppose @xmath159 is a closed convex set and @xmath512 are convex semi - continuous functions .
if @xmath173 and @xmath202 coincide on the relative interior of @xmath417 then @xmath513 we begin with the following , [ uniq123 ] suppose @xmath514 so that @xmath515 then there is a unique maximum point of the volume function @xmath516 the maximum point exists since the @xmath143 is continuous on the compact set @xmath273 suppose otherwise that @xmath517 are two distinct maximum points of volume in @xmath273 then due to convexity , all points in the line segment @xmath518 for @xmath519 $ ] are maximum point .
on the other hand , the work of rivin @xcite shows that , due to @xmath270 for any maximum volume point @xmath520 if @xmath521 for some @xmath522 then for the other two quads @xmath523 @xmath524 this shows if @xmath525 then @xmath526 for otherwise , the values of the maximum point @xmath527 at the three quads @xmath528 would be @xmath529 for some @xmath530 now due to @xmath270 @xmath531 therefore , there is a tetrahedron @xmath44 so that @xmath532 for all @xmath533 hence the function @xmath534 is a sum of concave functions so that one of it is strictly concave in @xmath349 this implies @xmath535 is not a constant function which contradicts that @xmath535 s are the maximum value @xmath536 to prove part ( 1 ) of theorem [ thm31 ] , take any @xmath537 whose cone angle is @xmath538 by definition @xmath539 but by the duality theorem [ dual ] , @xmath540 thus @xmath541 i.e. , @xmath11 is a maximum volume point . to prove part ( 2 ) of theorem [ thm31 ] , since @xmath542 by proposition [ prop3.4 ] , the function @xmath543 has a minimal point @xmath544 let @xmath545 be the dihedral angle of @xmath247 then by @xmath546 at @xmath547 we see that the cone angle of @xmath548 is @xmath317 i.e. , @xmath549 by part ( 1 ) , @xmath548 is a maximum volume point in @xmath273 therefore , both @xmath548 and @xmath11 are maximum volume points in @xmath273 by proposition [ uniq123 ] , @xmath550 thus the result follows . to prove theorem [ thm32 ] ,
suppose that @xmath551 are two decorated ideal hyperbolic polyhedral metrics on @xmath3 so that their cone angles @xmath445 and @xmath552 are the same .
we will show that @xmath553 for some @xmath376 i.e. , @xmath554 differ by a change of decoration .
let @xmath555 by the assumption that @xmath220 is a decorated ideal hyperbolic polyhedral metric , @xmath556 by theorem [ thm31 ] , the dihedral angles @xmath557 and @xmath558 of @xmath220 and @xmath559 are the maximum volume points in @xmath560 on the other hand , proposition [ uniq123 ] shows the maximum volume point is unique
. therefore @xmath561 this implies that the underlying hyperbolic metrics for @xmath220 and @xmath559 are isometric .
therefore , @xmath220 and @xmath559 differ by a change of decoration , i.e. , @xmath562 for some @xmath563 note that we have proved a slightly stronger statement that we only need to assume @xmath220 is decorated hyperbolic
metric and @xmath559 is a generalized decorated metric of the same cone angle .
we recall some of the basic results on hyper - ideal tetrahedra in this subsection . following @xcite and @xcite , a _ hyper - ideal tetrahedron _
@xmath17 in @xmath564 is a compact convex polyhedron that is diffeomorphic to a truncated tetrahedron in @xmath565 with four hexagonal faces right - angled hyperbolic hexagons ( see figure [ hyper - ideal ] ( a ) ) .
the four triangular faces isometric to hyperbolic triangles are called _
vertex triangles_. an _ edge _ in a hyper - ideal tetrahedron is the intersection of two hexagonal faces , and a _ vertex edge _ is the intersection of a hexagonal face and a vertex triangle .
the _ dihedral angle _ at an edge is the angle between the two hexagonal faces adjacent to it .
the dihedral angle between a hexagonal face and a vertex triangle is always @xmath566 let @xmath567 @xmath568 be the four vertex triangles of @xmath569 we use @xmath570 to denote the edge joining @xmath571 to @xmath572 and use @xmath573 to denote the hexagonal face adjacent to @xmath574 @xmath575 and @xmath576 ( see figure [ hyper - ideal ] ( a ) . ) the length of @xmath570 is denoted by @xmath68 and the dihedral angle at @xmath570 is denoted by @xmath577 the length of the vertex edge @xmath578 is denoted by @xmath579 as a convention , we always assume @xmath580 and @xmath581 + [ bb - fu ] ( @xcite , @xcite ) suppose @xmath17 is a hyper - ideal tetrahedron in @xmath582 a. the isometry class of @xmath17 is determined by its dihedral angle vector @xmath583 which satisfies the condition that @xmath584 and @xmath585 for each fixed @xmath586 b. conversely , given @xmath587 so that @xmath588 for each @xmath589 where @xmath590 there exists a hyper - ideal tetrahedron having @xmath591 as its dihedral angle at the @xmath592-th edge . c. the isometry class of @xmath17 is determined by its edge length vector @xmath593 thus , the space of isometry classes of hyper - ideal tetrahedra parametrized by dihedral angles is the open convex polytope @xmath594 let @xmath595 be the hyperbolic volume of the hyper - ideal tetrahedra considered as a function in the dihedral angles
. then the schlaefli formula says @xmath596 see @xcite for the schlaefli formula in a more general setting .
[ vol ] the volume function has the following properties . a. ( @xcite ) the volume function @xmath597 is smooth and has positive definite hessian matrix at each point in @xmath598 b. ( @xcite ) the function @xmath143 can be extended continuously to the compact closure @xmath599 of @xmath600 in @xmath601 where @xmath602 let @xmath603 be the set of vectors @xmath604 such that there exists a hyper - ideal tetrahedron having @xmath68 as the length of the @xmath592-th edge .
it follows from proposition [ bb - fu ] that @xmath603 is a simply connected open subset of @xmath605 the volume function can be considered as defined on @xmath606 the legendre transform of @xmath607 to be called the _ co - volume _ function , is @xmath608 given by @xmath609 it is known ( see @xcite ) that @xmath428 has a positive definite hessian matrix at each @xmath610 and hence @xmath428 is locally strictly convex . however , the open subset @xmath611 is not convex .
one of the main technical results in this paper is that @xmath428 can be extended to a @xmath2-smooth and convex function on @xmath612 by studying the flat hyper - ideal tetrahedra .
recall that a _ flat hyper - ideal tetrahedron _ is defined as follows .
take a right - angled hyperbolic octagon @xmath613 with eight edges cyclically labelled as @xmath614@xmath615@xmath616@xmath617@xmath618@xmath619@xmath620 let @xmath621 ( and @xmath622 ) be the shortest geodesic arc in @xmath613 joining @xmath623 to @xmath624 ( and @xmath625 and @xmath626 ) .
we call @xmath627 a _ flat hyper - ideal tetrahedron with six edges @xmath628 see figure [ hyper - ideal ] ( b ) .
the dihedral angles at @xmath621 and @xmath622 are @xmath7 and are @xmath629 at all other edges .
the volume of a flat hyper - ideal tetrahedron is defined to be zero .
_ in this subsection , we investigate the space of hyper - ideal tetrahedra parametrized by the edge lengths and their degenerations .
one of the goals is to extend the locally convex function @xmath428 to a convex function defined on @xmath630 to this end , let us define _ a generalized hyper - ideal tetrahedron to be a topological truncated tetrahedron so that each edge is assigned a positive number , called the edge length .
we will define the _ dihedral angles and _ volume and _ covolume of generalized hyper - ideal tetrahedra in this section .
_ _ _ _ suppose @xmath17 is a generalized hyper - ideal tetrahedron with edges @xmath570 joining the i - th and the j - th vertices and @xmath580 is the edge length of @xmath628 to define the dihedral angle @xmath591 at @xmath574 we need the following compatibility property , which is a special case of proposition 3.1 of @xcite .
[ comp ] for @xmath631 and @xmath632 let @xmath633 for @xmath634 and let @xmath635 and @xmath636 then @xmath637 let @xmath638 @xmath639 @xmath640 and @xmath641 for @xmath642 by definition , we have @xmath643 and similarly @xmath644 to see @xmath645 it suffices to show that the two denominators in ( [ 43 ] ) and ( [ 44 ] ) are the same .
to this end , we have @xmath646 therefore , @xmath647 note that if @xmath17 is a hyper - ideal tetrahedron of edge lengths @xmath648 then by the cosine law , @xmath649 and @xmath650 in lemma [ comp ] are the lengths of the vertex edge @xmath578 and the dihedral angle at @xmath651 in particular , the conclusion of the lemma is obvious for @xmath569 for a generalized hyper - ideal tetrahedron @xmath652 using lemma [ comp ] we call @xmath649 the _ length of the vertex edge and define the function @xmath653 by @xmath654 then we have _ [ lengths ]
the space of all hyper - ideal tetrahedra parametrized by the edge lengths is @xmath655 if @xmath18 is the edge length vector of a hyperbolic tetrahedra , then each of the dihedral angles @xmath656 by the cosine law for the hyperbolic triangle @xmath657 and right - angled hexagon @xmath658 we have @xmath659 this shows @xmath660 conversely , for each @xmath661 with @xmath662 for all @xmath663 by ( [ eq5.2 ] ) , @xmath664 @xmath665 and @xmath666 satisfy the triangular inequality .
then there exists a unique hyperbolic triangle @xmath667 having them as edge lengths .
taking @xmath668 by lemma [ comp ] , we see that @xmath669 @xmath670 and @xmath671 are the angles of @xmath672 hence they satisfy @xmath673 by proposition [ bb - fu ] , there is a unique hyper - ideal tetrahedron @xmath17 with dihedral angles @xmath674 applying the cosine law to the vertex triangles @xmath675 and hexagons @xmath676 we see that @xmath18 is the edge lengths of @xmath569 [ degeneration ] let @xmath677 be the frontier of @xmath603 in @xmath678 then @xmath679 where each @xmath680 @xmath301 is a real analytic codimension-@xmath681 submanifold of @xmath678 the complement @xmath682 is a disjoint union of three manifolds @xmath683 with boundary so that @xmath684 @xmath308 for @xmath663 let @xmath685 and @xmath686 then by proposition [ lengths ] , we have @xmath687 [ tech ] for @xmath632 we have 1 .
@xmath688 2 .
@xmath689 and @xmath690 3 .
@xmath691 and @xmath692 for each @xmath693 we have @xmath694 and @xmath695 which implies @xmath696 and @xmath697 therefore , ( 1 ) holds and @xmath698 on the other hand , for each @xmath699 we have @xmath700 and @xmath701 which implies that @xmath702 and @xmath703 as a consequence , @xmath704 and @xmath705 therefore , we have @xmath706 hence @xmath707 by symmetry , we have @xmath708 from which we see @xmath709 now for @xmath710 we have @xmath711 which implies that @xmath712 if @xmath713 then @xmath714 and @xmath715 if @xmath716 then @xmath717 and @xmath718 therefore , we have @xmath719 on the other hand , since @xmath720 and @xmath721 we have @xmath722 from which ( 2 ) follows . ( 3 ) follows from the same argument with the inequalities replaced by equalities .
let @xmath723 @xmath724 and @xmath725 for each @xmath726 and @xmath727 we have @xmath728 which implies @xmath729 therefore , @xmath730 is a regular values of @xmath731 by the implicit function theorem , @xmath732 is a smooth codimension-@xmath681 submanifold of @xmath678 since each @xmath733 is real analytic in @xmath734 the submanifold @xmath735 is real analytic .
let @xmath736 @xmath737 and @xmath738 and similarly let @xmath739 @xmath740 and @xmath741 as a consequence of lemma [ tech ] , we have @xmath742 is a disjoint union of three @xmath743-dimensional submanifolds with boundary . by lemma [ tech ] , @xmath744 and @xmath745
we claim that @xmath746 indeed , for each @xmath747 say , we construct a sequence @xmath748 convergent to @xmath18 as follows .
we let @xmath749 and define @xmath750 then for @xmath751 large enough , by ( [ xijk ] ) @xmath752 and @xmath753 satisfy the triangular inequalities , and by ( [ eq5.2 ] ) each @xmath754 by proposition [ lengths ] , @xmath755 for @xmath751 large enough . therefore , we have @xmath756 which completes the proof . [ comp0 ]
the function @xmath733 extends continuously to @xmath757 and @xmath758 when @xmath759 since the denominator of @xmath733 is never equal to @xmath760 the function continuously extends @xmath761 furthermore a direct calculation show if @xmath762 i.e. , @xmath763 then @xmath764 indeed , both numerator and denominator in ( [ eq5.4 ] ) are @xmath765 [ degeneration2 ] for each subset @xmath766 of the edges of a tetrahedron , let @xmath767 and let @xmath768 @xmath301 be the closure of @xmath769 in @xmath770 if @xmath771 then @xmath772 is a real analytic codimension-@xmath681 submanifold of @xmath773 let @xmath774 be the closure of @xmath775 in @xmath770 we first observe that if @xmath776 then @xmath777 indeed , by lemma [ comp0 ] , if @xmath778 then @xmath779 and if @xmath780 then @xmath764 therefore , if @xmath781 then @xmath782 and hence @xmath783 by lemma [ tech ] , @xmath784 which implies @xmath785 hence @xmath786 letting @xmath787 @xmath788 and @xmath789 be as before , we have @xmath728 which implies that the projection of @xmath790 to the tangent space of @xmath791 at @xmath792 is non - vanishing , i.e. , @xmath791 and @xmath774 transversely intersect . by the implicit function theorem , the intersection @xmath793 is a smooth codimension-@xmath681 submanifold of @xmath773 since each @xmath733 is real analytic , the submanifold is real analytic in @xmath773 let @xmath794 be the closure of @xmath603 in @xmath795 and let @xmath796 for @xmath797 by definding @xmath798 and @xmath799 we have [ dihedral ] the dihedral angle function @xmath800 can be extended continuously to @xmath801 so that its extension , still denoted by @xmath802 is a constant on each component of @xmath803 for each subset @xmath766 of the edges .
we call @xmath591 in corollary [ dihedral ] the _ dihedral angle of the generalized hyper - ideal tetrahedron @xmath804 _
the locally convex function covolume function @xmath805 defined by ( [ 2.1 ] ) satisfies the schlaefli identity that @xmath806 for @xmath797 where @xmath800 is the dihedral angle function at the @xmath592-th edge .
in particular , the differential @xmath681-form @xmath807 is closed in @xmath808 and we can recover covolume @xmath428 by the integration @xmath809 for each @xmath810 we let @xmath811 where @xmath812 by corollary [ dihedral ] , we can extend the function @xmath800 to a continuous function @xmath813 by @xmath814 and call @xmath815 the _ dihedral angle vector _ of @xmath816 we define a new continuous @xmath681-form @xmath817 on @xmath612 by @xmath818 [ close ] the continuous differential @xmath681-form @xmath819 is closed in @xmath601 i.e. , for any euclidean triangle @xmath103 in @xmath601 @xmath820 we prove it in two steps . in the first step ,
we prove that @xmath817 is closed in @xmath821 next , we show @xmath817 is closed in @xmath605 by corollary [ dihedral ] , the differential @xmath681-form @xmath819 is continuous in @xmath605 the restriction @xmath822 is closed . by corollary [ dihedral ] , @xmath591 is a constant in each connected component of @xmath823 proposition [ degeneration ] shows that the subset @xmath824 in @xmath825 is open and bounded by a smooth codimension-@xmath681 submanifold .
now we use the following lemma .
[ l3](propositions 2.4 and 2.5 , @xcite ) suppose @xmath826 is an open set and @xmath827 is a continuous 1-form on @xmath828 1 .
if @xmath829 is an open subset bounded by a smooth codimension-@xmath681 submanifold of @xmath830 and @xmath831 and @xmath832 are closed , then @xmath817 is closed in @xmath828 2 . if @xmath833 is simply connected , then @xmath834 is a @xmath2-smooth function such that @xmath835 3 .
if @xmath833 is convex and @xmath836 is an open subset of @xmath833 bounded by a codimension-@xmath681 real analytic submanifold of @xmath833 so that @xmath837 and @xmath838 are locally convex , then @xmath125 is convex in @xmath828 thus , by lemma [ l3 ] ( 1 ) , the differential @xmath681-form @xmath817 is closed in @xmath630 for each subset @xmath766 of the edges , by equation ( [ eq5.4 ] ) , lemma [ comp0 ] and a direct calculation , @xmath839 is closed in @xmath840 by definition , @xmath839 is constant in each connected component of @xmath841 now proposition [ degeneration2 ] shows that the subset @xmath842 in @xmath791 is open and bounded by a smooth codimension-@xmath681 submanifold .
thus , by lemma [ l3 ] ( 1 ) , the differential @xmath681-form @xmath839 is closed in @xmath773 for each euclidean triangle @xmath103 in a quadrant @xmath613 of @xmath843 let @xmath766 be set of edges @xmath6 so that @xmath844 for all @xmath845 and let @xmath846 be the projection of @xmath103 to @xmath773 by definition and lemma [ comp0 ] , @xmath847 if @xmath848 which implies @xmath849 as a consequence , @xmath817 is closed in each of the quadrants of @xmath605 repeating applying lemma [ l3 ] ( 1 ) , we conclude that @xmath817 is closed in @xmath605 [ extended ] the function @xmath850 defined by the integral @xmath851 is a @xmath2-smooth convex function . by lemma [ l3 ] ( 2 ) and proposition [ close ] , @xmath428 is a @xmath2-smooth function in @xmath852 by lemma [ l3 ] ( 3 ) and lemma [ dihedral ] , @xmath428 is convex in @xmath853 by the continuity , for each subset @xmath766 of the edges , @xmath854 is convex in @xmath855 since @xmath847 if @xmath848 we have @xmath856 and @xmath857 for some @xmath858 as a consequence , @xmath428 is convex in each quadrant of @xmath605 repeat using lemma [ l3 ] ( 3 ) , we conclude that @xmath428 is convex in @xmath852 by the work of ushijima @xcite , @xmath859 where @xmath860 is the lobachevsky function defined by @xmath861 and @xmath862 is the maximal volume amongst the generalized hyperbolic tetrahedra .
in this section , we prove theorem [ main ] ( b ) using the convex extension of @xmath428 in corollary [ extended ] .
let @xmath863 be triangulated closed pseudo @xmath0-manifold and let @xmath864 @xmath865 and @xmath866 respectively be the sets of edges , vertices and tetrahedra in @xmath867 replacing each 3-simplex in @xmath4 by a hyper - ideal tetrahedron and gluing them along codimension-1 by isometries , we obtain an _ hyper - ideal polyhedral metric on @xmath9 this metric is the same as assigning a positive number to each edge @xmath868 so that each tetrahedron @xmath17 becomes a hyper - ideal tetrahedron with assigned numbers as edge lengths .
they are the same as hyperbolic cone metrics on @xmath869 with singularity consisting of geodesic arcs between totally geodesic boundary . here
@xmath870 is an open regular neighborhood of @xmath871 in @xmath30 we denote by @xmath872 the space of all hyper - ideal polyhedral metrics on @xmath873 parametrized by the edge length vector @xmath874 if @xmath875 its _ curvature is a map @xmath876 sending each edge to @xmath1 less the sum of dihedral angles at the edge .
the _ curvature map @xmath877 sends @xmath18 to @xmath878 _ _ _ the rigidity theorem [ main ] ( b ) can be rephrased as [ hyper - idealrigid ] for any closed triangulated pseudo 3-manifold @xmath355 a hyper - ideal polyhedral metric on @xmath3 is determined by its curvature , i.e. , the curvature map @xmath879 is injective . for each @xmath880 and each tetrahedron @xmath255 let @xmath881 be the edge length vector of @xmath17 in the hyperbolic polyhedral metric @xmath247
define the _ co - volume _ function @xmath882 on @xmath883 by proposition [ vol ] ( a ) , the hessian matrix of the function @xmath884 is positive definite at each point in @xmath606 therefore , the hessian matrix of @xmath428 is positive definite . in particular
, @xmath428 is locally strictly convex . on the other hand ,
it is shown in @xcite that the gradient of @xmath428 is @xmath1 minus the curvature map @xmath885 i.e. , @xmath886 therefore , it suffices to show that @xmath887 is injective .
we extend the co - volume function @xmath888 defined by ( [ 3.1 ] ) to a @xmath2-smooth convex function , still denoted by @xmath889 by @xmath890 where @xmath884 is the extended convex function given by corollary [ extended ] .
the convexity of @xmath428 follows from the fact that each summand @xmath891 is convex .
now suppose otherwise that there exist @xmath892 so that @xmath893 joint @xmath21 and @xmath894 in @xmath895 by the line segment @xmath896 @xmath897,$ ] and consider the convex function @xmath898 @xmath897.$ ] by the construction , @xmath899\rightarrow\mathbb{r}$ ] is a @xmath2-smooth convex function so that @xmath900 now @xmath901 and @xmath902 it follows that @xmath903 since @xmath904 is convex , @xmath905 mush be a linear function in @xmath349 however , @xmath428 is strictly convex near @xmath21 and @xmath906 therefore , @xmath904 is strictly convex in @xmath907 near @xmath629 and @xmath908 this is a contradiction .
let @xmath909 be closed triangulated pseudo 3-manifold with set of edges @xmath910 and set of tetrahedra @xmath29 [ angle - hyper ] an _ angle assignment _
( respectively _ positive angle assignment _ ) of hyper - ideal type on @xmath3 assigns each edge @xmath6 in each tetrahedra @xmath17 a non - negative ( respectively positive ) number @xmath911 call the dihedral angle of @xmath6 in @xmath102 so that sum of the dihedral angles at three edges in the same tetrahedron @xmath17 adjacent to each vertex is less than or equal to ( respectively strictly less than ) @xmath12 the _ cone angle _ of an angle assignment is the function @xmath912 sending each edge @xmath6 to the sum of dihedral angles at @xmath8 for any @xmath913 we denote respectively by @xmath914 and @xmath915 the spaces of angle assignments and positive angle assignments with cone angles @xmath538 note that if @xmath916 then @xmath917 is the compact closure of @xmath918 when @xmath919 we denote respectively by @xmath920 and @xmath921 the space of the corresponding angle assignments and positive angle assignments of cone angles @xmath58 we note that @xmath921 coincides with the space of _ linear hyperbolic structures _ on @xmath922 defined in @xcite . by the work of @xcite and @xcite , a positive angle assignment @xmath923 on @xmath3
is the same as making each tetrahedra @xmath44 a hyper - ideal tetrahedron so that its dihedral angles are given by @xmath924 in particular , we can define the volume of a positive angle assignment as the sum of the hyperbolic volume of the hyper - ideal tetrahedra , i.e. , the _ volume function _
@xmath925 is @xmath926 where @xmath927 is the hyper - ideal tetrahedron with dihedral angles given by @xmath924 by the work of schlenker @xcite , @xmath143 is smooth and strictly convex on @xmath928 a theorem of rivin@xcite shows that @xmath143 can be extended continuously to the compact closure @xmath914 of @xmath929 for each @xmath932 by replacing each tetrahedron @xmath17 in @xmath27 by a generalized hyper - ideal tetrahedron whose edge lengths are given by @xmath279 we define the _ dihedral angle of @xmath18 at an edge @xmath6 in a tetrahedron @xmath933 to be the corresponding dihedral angles in the generalized hyper - ideal tetrahedron whose edge lengths are given by @xmath247 the _ cone angle of @xmath18 at @xmath6 is the sum of dihedral angles of @xmath18 at @xmath6 in all the tetrahedra @xmath17 that contain @xmath8 the goal of this section is to prove the following _ _
a. there is a unique @xmath935 that achieves the maximum volume , and b. for each generalized hyper - ideal metric @xmath936 with cone angles @xmath317 the dihedral angles @xmath937 of @xmath18 is the maximum volume on @xmath938 c. if @xmath935 achieves the maximum volume , then there exists a generalized hyperbolic metric @xmath939 whose dihedral angles @xmath940 theorem [ unique2 ] implies theorem [ 1.4 ] in 1 .
it is known to kojima@xcite that every compact hyperbolic @xmath0-manifold with totally geodesic boundary admits a geometric ideal triangulation so that each tetrahedron is either hyper - ideal or flat hyper - ideal whose dihedral angles are @xmath629 and @xmath12 one consequence of theorem [ unique2 ] ( b ) is [ main3 ] suppose @xmath28 is a compact hyperbolic @xmath0-manifold with totally geodesic boundary and @xmath941 is a geometric ideal triangulation of @xmath28 so that each tetrahedron is either hyper - ideal or flat hyper - ideal . if @xmath942 then the maximum volume on @xmath943 is equal to the hyperbolic volume of @xmath30 recall that @xmath944 is the space of dihedral angle vectors of hyper - ideal tetrahedra defined by ( [ equation ] ) and let @xmath599 be its closure , i.e. , @xmath945 by @xcite , the volume function @xmath143 on @xmath946 is continuous and convex . to study the volume optimization
, we need to classify points in @xmath947 we denote by @xmath954 @xmath955 and @xmath956 respectively the set of generalized dihedral angles of types i , ii and iii .
note that @xmath957 for each @xmath663 we let @xmath958 be the function defined by @xmath959 where @xmath960 and @xmath961 by the cosine law , if @xmath962 is the dihedral angles of a hyper - ideal tetrahedron @xmath17 with @xmath963 the edge lengths , then @xmath964 for @xmath973 we let @xmath974 and call @xmath975 the _ associated edge lengths _ of @xmath924 given a tetrahedron @xmath102 let @xmath766 be a set of edges in @xmath17 and define @xmath976 the set @xmath977 is an open convex polytope in the smallest affine space containing it since @xmath977 is defined by strict linear inequalities and linear equalities . by lemma [ length ] , the edge lengths continuously extend to @xmath977 with @xmath980 for @xmath981 from the definition , each @xmath982 for @xmath983 is smooth in @xmath979 as a consequence , the function @xmath984 is smooth and the following differential @xmath681-form @xmath985 is smooth in @xmath979 from the definition and a direct calculation , we have @xmath986 for @xmath261 @xmath987 i.e.
, the differential @xmath681-form @xmath988 is closed in @xmath979 for each @xmath989 take @xmath990 close to @xmath991 and define a function @xmath992 by @xmath993 where the path in the integral is any smooth path in @xmath977 connecting @xmath994 and @xmath924 since @xmath988 is smooth and closed , @xmath202 is well defined and smooth in @xmath977 and @xmath995 for each corner @xmath996 we claim that @xmath997 the claim implies that the volume function @xmath143 is smooth in @xmath977 and @xmath998 for each @xmath996 now since @xmath994 is close to @xmath991 we can take a vector @xmath999 so that @xmath1000 @xmath1001 let @xmath1002 and let @xmath1003 where the summation is over all the dihedral angles . by the continuity of @xmath1004 we have @xmath1005 one the other hand ,
since @xmath1006 continuously extends to @xmath1007 it is uniformly continuous on the compact parallelogram determined by @xmath1008 @xmath991 @xmath1009 and @xmath1010 as a consequence , @xmath173 is uniformly continuous on the compact square @xmath1011\times[0,1]$ ] and @xmath1012 again , since @xmath173 is uniformly continuous , we can switch the order of taking limit and integrating , and we have @xmath997 now we show the strict concavity of @xmath143 in @xmath1013 let @xmath1014 be map defined by the restriction of the associated edge lengths and let @xmath1015 from the definition , each @xmath1016 for @xmath983 is smooth in @xmath1017 hence the function @xmath1018 is smooth in @xmath840 now we have two smooth maps @xmath1019 and @xmath1020 which are , by the cosine law , inverses of each other . as a consequence , the map @xmath1021 is a local diffeomorphism and the jocobi matrix @xmath1022_{\alpha,\beta\in s}$ ] is non - singular . since @xmath1023 is connected , the signature of the hessian matrix @xmath1024_{\alpha,\beta\in s}$ ] of the volume function @xmath143 on @xmath977 is independent of the choice of @xmath1025 by a direct calculation , using the formula in guo(@xcite , theorem 1 ) , the matrix @xmath1026_{\alpha,\beta\in s}$ ] is negative definite at @xmath1027 for each @xmath996 this implies that @xmath143 is locally strictly concave in @xmath979 since @xmath977 is convex , the volume function @xmath143 is strictly concave in @xmath979 let @xmath1033\rightarrow\mathbb r$ ] be the function defined by @xmath1034 for ( 1 ) , by the concavity of @xmath143 and the mean value theorem , we have @xmath1035 for all @xmath1036 since @xmath1037 @xmath1038 and @xmath1039 by proposition [ degeneration ] , @xmath824 is open in @xmath825 and @xmath1040 is a smooth codimension@xmath681 submanifold , hence for each @xmath1041 there exists a sequence @xmath1042 converging to @xmath18 and the corresponding dihedral angles @xmath1043 converging to @xmath1044 since the volume function @xmath143 is strictly concave in @xmath1045 we have for all @xmath751 and for all @xmath1046 that @xmath1047 by schlaefli formula , @xmath1048 since @xmath143 is continuous on @xmath599 and @xmath1049 as @xmath751 approaches @xmath1050 @xmath1051 in particular , when @xmath1052 we have @xmath1053 by ( [ label ] ) and ( [ label2 ] ) , we have @xmath1054 hence @xmath1055 for @xmath1056 by proposition [ lengths ] , there exists an @xmath1057 such that @xmath1058 by the previous case , we have @xmath1059 without loss of generality , we may assume that @xmath1060 by cosine law , @xmath1061 since @xmath1062 we have @xmath1063 hence @xmath1064 as a consequence , @xmath1065 combined with ( [ label3 ] ) , we have @xmath1066 we let @xmath1071 @xmath1072 and @xmath1073 for @xmath663 and let @xmath1074 for @xmath1075 then @xmath1076 let @xmath1077 and @xmath1078 respectively be the asymptotic orders of @xmath1079 and @xmath1080 as @xmath907 approaches @xmath760 i.e. , @xmath1081 and @xmath1082 for some constants @xmath1083 and @xmath1084 as @xmath907 approaches @xmath760 @xmath1085 and by schlaefli formula , @xmath1086 is a finite number .
therefore , to prove the result , it suffices to prove that the @xmath1087 is strictly negative .
first we consider the case that , up to a permutation of the vertices , @xmath1088 for some @xmath1089 and @xmath1090 for some @xmath1091 in this case , if @xmath1092 for some @xmath663 then @xmath1093 which was ruled out by the assumption .
if @xmath1094 for some @xmath663 then we claim that @xmath1095 i.e. , @xmath1096 indeed , letting @xmath1097 and @xmath1098 we have @xmath1099 and also @xmath1100 since @xmath1101 we have @xmath1102 @xmath1103 and the equality holds if and only if @xmath1104 and one of @xmath1105 and @xmath1106 equals @xmath1107 therefore , since @xmath1108 @xmath1109 only if @xmath1110 @xmath1111 and one of @xmath1112 and @xmath1113 equals @xmath1107 if , say , @xmath1114 then from the last equation , we have @xmath1115 and @xmath1116 with @xmath1117 we have @xmath1118 and @xmath1119 from @xmath1120 and @xmath1121 we have @xmath1122 as a consequence , up to a permutation of indices , @xmath1123 with @xmath1124 which was ruled out by the assumption . hence the claim is true . now for @xmath970 since @xmath1069 we have @xmath1125 and @xmath1126 by the definition of @xmath1127 we have @xmath1128 if and only if @xmath1129 since @xmath923 is of type iii , there exists at least one @xmath1130 with @xmath1129 for such @xmath589 since @xmath1131 we have @xmath1132 thus , @xmath1133 the two sporadic cases are verified by a direct calculation . in the case that @xmath1134 with @xmath1135 we have @xmath1136 for each @xmath1137 @xmath1138 and @xmath1139 since @xmath1140 and @xmath1141 we have @xmath1142 @xmath1143 and @xmath1144 therefore , @xmath1145 in the case that @xmath1146 with @xmath1147 we have @xmath1148 @xmath1149 @xmath1150 and @xmath1151 since @xmath1152 we have @xmath1153 and @xmath1154 let @xmath863 be closed triangulated pseudo 3-manifold .
the space of all generalized hyper - ideal polyhedral metrics on @xmath1155 parameterized by the edge length vectors , is @xmath1156 the _ co - volume _ function @xmath1157 is defined by @xmath1158 where each summand is the function defined by ( [ 3.2 ] ) . by corollary
[ extended ] , @xmath428 is @xmath2-smooth and convex , hence its fenchel dual function @xmath1159 is well - defined , convex and lower semicontinuous in @xmath1160 the goal of this subsection is to show that the fenchel dual function @xmath1161 optimizes the volume function on the space of non - positive angle assignments of given cone angles .
we will use the method developed by colin de verdire s @xcite . take any @xmath1174 we rewrite @xmath1172 as @xmath1175 for each @xmath255
we let @xmath1176 be the edge length vector of the hyper - ideal tetrahedron whose dihedral angles are @xmath1177 then @xmath1178 is the unique critical point of the convex function @xmath1179 defined by @xmath1180 since @xmath1181 is strictly convex near @xmath1182 the function @xmath1181 is _ closed in @xmath601 i.e. , @xmath1183 as a consequence , the function @xmath1172 is closed and convex in @xmath1160 this shows that @xmath1172 has a critical point @xmath18 in @xmath1160 moreover , since @xmath1184 @xmath1185 for each edge @xmath8 this implies that @xmath844 for each @xmath1186 i.e. , @xmath1187 indeed , if otherwise @xmath1188 for some @xmath1186 then by lemma [ comp0 ] , all the dihedral angles of @xmath18 at @xmath6 are zero , hence @xmath1189 which is a contradiction . _
denoting by @xmath1194 the convex open polytope @xmath1195 we claim that @xmath1196 indeed , since @xmath1197 we have that @xmath1198 on the other hand , by proposition [ image ] , for each @xmath1199 there exists an @xmath1200 such that @xmath1201 it now suffices to show that @xmath1202 since @xmath1203 the cone angle of @xmath18 at each edge @xmath6 is in the range @xmath1204 as a consequence , all the dihedral angles of @xmath941 in @xmath18 are in the range @xmath1204 thus , all the tetrahedra of @xmath941 are hyper - ideal in @xmath279 and @xmath18 is in @xmath1205 we first show that if @xmath1208 then @xmath1209 for all @xmath1210 since the space @xmath1211 of all possible dihedral angles is compact and @xmath143 is continuous , there exists a constant @xmath437 so that @xmath1212 for all @xmath1213 since @xmath1214 and @xmath1215 are compact and convex in @xmath1216 by the separation theorem of convex sets , there exists an @xmath1217 so that @xmath1218 for all @xmath1219 in particular , letting @xmath1220 be the cone angle vector of @xmath1221 we have @xmath1222 therefore , @xmath1223 now we prove that @xmath1165 in @xmath1224 by proposition [ image2 ] , if @xmath1225 the function @xmath1172 has a critical point @xmath1226 and @xmath1227 to show that @xmath1228 i.e. , @xmath937 achieves the maximum volume , it suffices to show that the sub - derivative @xmath1229 for each @xmath1230 let @xmath1231 and let @xmath1232 for each @xmath1233 we have @xmath1234 hence @xmath1235 for each edge @xmath8 from this , we see @xmath1236 let @xmath1237 be the subset of @xmath1238 consisting of hyper - ideal tetrahedra in @xmath18 and let @xmath1239 we have @xmath1240 then by schlaefli formula , @xmath1241 where the inequality is from lemma [ lemma6.7 ] . hence @xmath1242 and @xmath833 coincide on the subset @xmath1243 of @xmath1244 which contains the relative interior of @xmath1224 since @xmath1242 and @xmath833 are convex and lower semicontinuous , by lemma [ relativeint ] , @xmath1165 on @xmath1224 for ( a ) , suppose otherwise that there exist @xmath1245 that achieve the maximum volume . connect @xmath1246 and @xmath1247 by the line segment @xmath1248 @xmath897,$ ] and consider the concave function @xmath1249 by the maximality of @xmath1250 the function @xmath143 is constant in @xmath1011.$ ] on the other hand , we let @xmath1251 be the restriction of @xmath1252 to @xmath17 and let @xmath1253 for each @xmath1233 then by the maximality and lemma [ lemma ] , each @xmath1254 is not of type iii . as a consequence ,
the interior of the line segment @xmath1255 lies in @xmath1256 or @xmath1257 since @xmath955 is discrete in @xmath599 and @xmath1258 there is at least one @xmath1259 such that @xmath1260 lies in @xmath1261 we claim that the interior of @xmath1260 lies in @xmath1262 for some subset @xmath1263 of the edges of @xmath1264 indeed , if @xmath1265 and @xmath1266 for some @xmath1267 @xmath1268 then the interior of @xmath1260 lies in @xmath1269 by proposition [ strict ] , the function @xmath1270 is strictly concave , hence @xmath1271 is strictly concave in @xmath1011,$ ] which is a contradiction . for ( b ) , by the assumption , @xmath18 is a critical point of the function @xmath1272 and @xmath1273 by theorem [ fd3 ] , @xmath1274 for ( c ) , by proposition [ image ] , there exists an @xmath1190 such that @xmath1275 by ( b ) , @xmath937 achieves the maximum volume , and by ( a ) , @xmath940 let @xmath1190 be the edge length vector in the hyperbolic metric of @xmath922 for which @xmath4 is geometric .
then the dihedral angles @xmath1276 by theorem [ unique2 ] ( b ) , @xmath937 achieves the maximum volume on @xmath1277 since the triangulation @xmath941 is geometric , @xmath1278 equals the hyperbolic volume of @xmath30 kojima , sadayoshi , _ polyhedral decomposition of hyperbolic 3-manifolds with totally geodesic boundary_. aspects of low - dimensional manifolds , 93112 , adv .
pure math .
, 20 , kinokuniya , tokyo , 1992 .
neumann , walter , _ combinatorics of triangulations and the chern - simons invariant for hyperbolic 3-manifolds _ , topology 90 ( columbus , oh , 1990 ) , 243271 , ohio state univ
publ . , 1 , de gruyter , berlin , 1992 .
ushijima , akira , _ a unified viewpoint about geometric objects in hyperbolic space and the generalized tilt formula_. hyperbolic spaces and related topics , ii ( japanese ) ( kyoto , 1999 ) .
surikaisekikenkyusho kokyuroku no .
1163 ( 2000 ) , 8598 .
feng luo + department of mathematics , rutgers university + new brunswick , nj 08854 , usa + ( [email protected] ) + tian yang + department of mathematics , stanford university + stanford , ca 94305 , usa + ( [email protected] )
|
we investigate the rigidity of hyperbolic cone metrics on @xmath0-manifolds which are isometric gluing of ideal and hyper - ideal tetrahedra in hyperbolic spaces .
these metrics will be called ideal and hyper - ideal hyperbolic polyhedral metrics .
it is shown that a hyper - ideal hyperbolic polyhedral metric is determined up to isometry by its curvature and a decorated ideal hyperbolic polyhedral metric is determined up to isometry and change of decorations by its curvature .
the main tool used in the proof is the fenchel dual of the volume function .
|
the epoch of reionization marks the second major change in the ionization state of the universe .
reionization begins when the first sources of neutral hydrogen ( ) ionizing photons form within dark matter potential wells and start building an ionization region around themselves , the so - called strmgren sphere .
however , the reionization history and the redshift at which it ends still remain the subject of much discussion .
this is because the reionization process depends on a number of parameters including the initial mass function ( imf ) of the first sources , their star formation rates ( sfr ) , their stellar metallicity and age , the escape fraction of ionizing photons produced by each source and the clumping of the intergalactic medium ( igm ) , to name a few .
given the large number of free parameters that inevitably enter into the construction of theoretical reionization models , it is imperative to compare and update the models as fresh data sets are acquired . in this sense , it has been suggested ( malhotra & rhoads 2004 , 2005 ; santos 2004 ; mesinger , haiman & cen 2004 ; haiman & cen 2005 ; dijkstra , lidz & wyithe 2007 ; mesinger & furlanetto 2008 ; dayal , ferrara & gallerani 2008 ; dayal et al .
2009a , 2009b ) that a class of high redshift galaxies called lyman alpha emitters ( laes ) could be an important addition to complementary data sets to constrain the reionization history .
laes , galaxies identified by means of their very strong ly@xmath0 line ( 1216 ) emission , have been becoming increasingly popular as probes of reionization for three primary reasons .
firstly , specific signatures like the strength , width and the continuum break bluewards of the ly@xmath0 line make the detection of laes unambiguous to a large degree .
secondly , since ly@xmath0 photons have a large absorption cross - section against , their attenuation can be used to put constraints on the ionization state of the igm .
thirdly , there are hundreds of confirmed laes at @xmath5 ( finkelstein et al .
2007 ) , @xmath2 ( malhotra et al . 2005 ; shimasaku et al . 2006 ) and @xmath6 ( taniguchi et al . 2005 ; kashikawa et al . 2006 ) , which exactly probe the redshift range around which reionization is supposed to have ended . the data accumulated on laes shows some very surprising features ; while the apparent ly@xmath0 luminosity function ( lf ) does not show any evolution between @xmath7 - @xmath8 ( ouchi et al .
2008 ) , it changes appreciably between @xmath9 and @xmath10 , with @xmath11 ( the luminosity of the break , after which the number density decreases rapidly ) at @xmath12 ( kashikawa et al .
2006 ) being about 50% of the value at @xmath9 ( shimasaku et al .
unexpectedly , however , the ultraviolet ( uv ) lf does not show any evolution between these same redshifts .
although kashikawa et al .
( 2006 ) have proposed this evolution in the ly@xmath0 lf to be indicative of a sudden change in the ionization state of the igm , the problem of why reionization would affect the high luminosity tail of the ly@xmath0 lf rather than the faint end , as expected , remains . in spite of a number of diverse approaches ,
the effects of the ionization state of the the igm on the visibility of laes still remain poorly understood .
this is primarily because understanding/ constraining the ionization state of the igm using laes requires : ( a ) a detailed knowledge of the physical properties of each galaxy , including the sfr , stellar age and stellar metallicity , needed to calculate the intrinsic ly@xmath0 and continuum luminosity produced by stellar sources , ( b ) the intrinsic ly@xmath0 and continuum luminosity produced by the cooling of collisionally excited in the interstellar medium ( ism ) , ( c ) an understanding of the dust formation , enrichment and distribution in each galaxy , necessary to calculate the fractions of escaping ly@xmath0 and continuum photons , and ( d ) a full radiative transfer ( rt ) calculation to obtain the fraction of ly@xmath0 luminosity transmitted through the igm for each galaxy .
although a number of simulations have been used in the past to study laes , they lack one or more of the aforementioned ingredients . since they used an n - body simulation , mcquinn et al .
( 2007 ) were forced to neglect the intrinsic properties of the galaxies and their dust enrichment , although they carried out analytic rt .
iliev et al .
( 2008 ) used a simulation that only followed dark matter ; they also could not use the intrinsic galaxy properties or calculate the dust enrichment , although they carried out a complete rt calculation .
nagamine et al .
( 2008 ) used an sph ( smoothed particle hydrodynamics ) simulation to obtain the intrinsic sfr for each galaxy ; however , they did not calculate the dust enrichment and assumed a fixed value of the igm transmission , ignoring rt .
dayal et al .
( 2009a , 2009b ) used an sph simulation and the intrinsic galaxy properties to calculate the luminosity from both stellar sources and cooling of , calculated the dust enrichment and the igm transmission ; the only missing ingredient in their work was the rt calculation .
most recently , zheng et al .
( 2009 ) have carried out a rt calculation on an sph simulation ; however , they have not included any dust calculation or the luminosity contribution from the cooling of .
our aim in this work is to build a lae model containing all these ingredients , so as to determine the relative importance of dust , peculiar velocity fields and patchy reionization , in shaping the observed ly@xmath0 and uv lfs .
we start with an introduction to the sph and rt simulations used in sec .
[ simulations ] .
we then present the main ingredients of our lae model , including the calculation of the intrinsic luminosity from stellar sources / cooling of , the dust enrichment and the igm transmission , in sec .
[ model ] .
once the model is laid down , we present the results obtained with it and quantify the relative importance of dust , peculiar velocities and reionization on the ly@xmath0 and uv lfs in sec .
[ reio effects ] .
the physical properties of the galaxies identified as laes are shown in sec .
[ phy prop ] .
we conclude by mentioning the caveats and shortcomings of our model in sec .
the simulations used in this work are based on a @xmath13cdm cosmological model with @xmath14 , @xmath15 , @xmath16 , @xmath17 , where @xmath18 and a scale invariant power spectrum of the initial density perturbations is normalized to @xmath19 .
the ly@xmath0 and uv lfs presented in this work are based on the results from combined runs of sph and rt simulations , carried out using gadget-2 ( springel 2005 ) and crash ( maselli , ferrara & ciardi 2003 ; maselli , ciardi & kanekar 2009 ) respectively , which are coupled to a previously developed lae model ( dayal et al .
2008 ; dayal et al .
2009a , 2009b ) .
gadget-2 generates the redshift evolution of the density field , the baryonic density distribution and the velocity fields in a 100@xmath20 mpc comoving volume ; we obtain a snapshot of the simulation at @xmath21 . the specific simulation used in this work
is the g5 run described in springel and hernquist ( 2003 ) , which is part of an accurate study focused on modelling the star formation history of the universe .
we have chosen the g5 run since it contains several physical ingredients necessary for our investigation .
firstly , the volume is large enough so that cosmic variance is minimized in the determination of the lfs .
secondly , the sfr ( @xmath22 ) is self - consistently inferred from physically motivated prescriptions that convert gas particles into stellar particles and that properly take account of the mechanical / chemical feedback associated to supernovae and galactic winds .
this point is particularly relevant since the intrinsic luminosity of the galaxies , both in the ly@xmath0 and uv , depends sensitively on @xmath22 .
also , since the physics governing star formation and galaxy evolution is modelled accurately , the simulation gives us a reliable representation of the galaxy population .
as expected , the large simulation volume naturally leads to a relatively coarse mass resolution with the resolution mass being @xmath23 @xmath24 ( @xmath25 @xmath24 ) for dark matter ( gas particles ) . running a friends - of - friends ( fof ) algorithm on the sph particle distribution
, we identify galaxies and obtain their intrinsic properties , including @xmath22 , the mass weighted stellar metallicity ( @xmath26 ) , the total gas and stellar masses .
these are then used to calculate the intrinsic ly@xmath0/continuum luminosity , the dust enrichment and the escape fraction of continuum / ly@xmath0 photons from the galaxy .
all these are presented and discussed in greater detail in sec .
[ model ] and [ reio effects ] .
the rt calculations have been carried out in post - processing mode using the 3d crash code .
the main assumptions made for running crash are the following : ( a ) we initially assume , at @xmath27 , the gas to be in photoionization equilibrium with a uniform ultraviolet background ( uvb ) produced by sources below the simulation resolution scale , corresponding to an average neutral hydrogen fraction , @xmath28 , ( b ) we assume that galaxies have formed stars in the past at an average rate @xmath22 , equal to the one derived at @xmath27 and ( c ) since the rt simulation is carried out on a sph snapshot which contains about 2500 galaxies , to reduce the computational running time , we run the rt calculation in the monochromatic mode , with @xmath29 ev .
this is a reasonable assumption since the ly@xmath0 lfs depend mostly on the average ionization fraction and on the 3d topology of the fully ionized regions ; we are therefore not interested in the detailed profiles of the ionization front or a highly accurate estimate of the igm temperature , to probe which the ionizing radiation of each galaxy would have to be computed from its spectrum .
the specific photoionization rate is calculated using the expression given in eq .
12 of dayal , ferrara & gallerani ( 2008 ) and then averaging this value over @xmath26 and stellar ages ( @xmath30 ) of the galaxies in the the sph simulation .
the contribution from all the galaxies determined under the above - mentioned assumptions is then super - imposed on the average uvb value .
finally , we use @xmath31 ( gnedin et al .
2008 ) , as a characteristic value of the escape fraction of ionizing photons from each galaxy . the total ionizing radiation field , which is the sum of that produced by all galaxies identified in the simulation and the uvb , is then evolved with crash up to the hubble time corresponding to @xmath21 . however , as we are interested in assessing the visibility of lae as a function of the mean ionization fraction which could correspond to different epochs along any assigned reionization history , no particular meaning should be attached to the redshift parameter . in fig .
[ rt_maps ] we show maps of the fraction across a 2d cut through the rt simulation box , at different simulation times .
this time sequence then represents a time - line of the reionization process , due to the galaxy population present within the box . though an approximation , this description catches many of the important features of the reionization process , including the complex topology produced by the inhomogeneities in the density field , by the galaxy properties and by their spatial distribution .
[ rt_maps ] clearly shows the reionization process in its main three phases .
first , isolated ionized hydrogen ( ) regions start growing around point sources which are preferentially located along the over - dense filaments of matter . in this stage
the size of the isolated regions grows differentially according to the production rate of ionizing photons from the specific source ; the few galaxies with large @xmath22 ( e.g. the largest galaxy located at the center having @xmath32 ) can build an ionized region of about 20 mpc , in a time as short as 10 myr ( see upper - left panel ) , which grows at a very fast rate due to the vigorous ionizing photon output .
if observed at the same evolutionary timescale , due to the smaller ionizing photon budget , the regions of smaller galaxies are confined to sizes of the order of few mpc ( @xmath33 ) , which grow slowly .
after 10 myr of continuous star formation activity , the average neutral hydrogen fraction decreases only to @xmath34 . at this early stage ,
galaxies with @xmath35 , transmit only about 20% of their ly@xmath0 luminosity through the igm .
consequently the low luminosity end of the lf is depressed with respect to the intrinsic emissivity of these sources . on the other hand ,
galaxies with larger @xmath22 are able to transmit a larger amount ( @xmath36% ) of their luminosity through the igm .
as the ionizing photon production from galaxies continues , the sizes of the regions increase with time . at about @xmath37 myr ( upper right panel ) , the isolated regions start overlapping , resulting in an enhancement of the local photo - ionization rate close to the smaller sources .
this allows the ionized regions associated to the latter to grow faster , and results in the build - up of large regions also around smaller galaxies .
this results in an enhancement in the transmission of the ly@xmath0 luminosity for these sources , such that this value increases to @xmath38% , consequently leading to a boost of the low - luminosity end of the ly@xmath0 lf .
as expected , the overlap phase speeds - up the reionization process ; for a star formation time @xmath39 myr , @xmath40 decreases to @xmath41 , respectively . finally at about 300 myr ( lower center panel ) ,
the simulation volume is almost completely ionized .
this configuration corresponds to the post - overlap phase which leads to a further acceleration of the reionization process . for about 400 myr of star formation activity
, @xmath42 drops to a value of about @xmath43 .
this is a consequence of the fact that , as the igm becomes more ionized , even the ionizing radiation from the numerous smaller galaxies contributes to the overall radiation field .
it is very important to point out that , even in this final stage , the ionization field is highly inhomogeneous , with regions of higher ionization fraction corresponding to the environment of point ionizing sources ( green regions in the maps ) .
this topology is of fundamental importance in the estimation of the ly@xmath0 lfs , as the ly@xmath0 transmissivity associated to each source depends on the extent of the highly ionized regions as well as on the residual neutral hydrogen within them ; a very tiny fraction of residual neutral hydrogen is sufficient to suppress the intrinsic ly@xmath0 luminosity significantly .
even for an ionization fraction , @xmath44 , which results after 500 myr of star formation , the transmission value ranges between @xmath45% for galaxies with increasing @xmath22 ; the blue part of the line can be significantly attenuated even by the tiny residual fraction of about @xmath46 , close to the source .
as mentioned in sec . [ simulations ] , for each galaxy in the sph simulation snapshot , we obtain the sfr , @xmath22 , the mass weighted stellar metallicity , @xmath26 , the total gas mass , @xmath47 , and the stellar mass , @xmath48 ; the stellar age is then calculated as @xmath49 , i.e. assuming a constant sfr .
this assumption has been made for consistency with the rt calculation ( see sec . [ simulations ] ) .
these properties are then used to calculate the intrinsic ly@xmath0 ( @xmath50 ) and continuum luminosity ( @xmath51 ) , considering the contribution from both stellar sources and from the cooling of collisionally excited in the ism of the galaxy .
the intrinsic values of the ly@xmath0 and continuum luminosities so obtained can then be used to calculate the observed luminosity values .
it is comparatively easy to calculate the observed continuum luminosity , @xmath52 , since the wavelength range ( 1250 - 1500 ) of the continuum luminosity band is chosen so as to be unaffected by ; @xmath52 in fact only depends on the fraction of the continuum photons , @xmath53 , that escape the galactic environment undamped by dust , i.e. @xmath54 on the other hand , since ly@xmath0 photons have a large absorption cross - section against , the observed ly@xmath0 luminosity , @xmath55 , depends both on : ( a ) the fraction of ly@xmath0 photons , @xmath56 , that escape out of the galaxy into the igm , undamped by dust and ( b ) the fraction , @xmath57 , of these escaped " photons that are transmitted through the igm , undamped by .
the observed ly@xmath0 luminosity can therefore be expressed as @xmath58 all galaxies with @xmath59 and an observed equivalent width , @xmath60 are identified as laes and used to construct the cumulative ly@xmath0 and uv lfs which can then be compared to the observations . in principle , our results can be compared to both the observed lfs at @xmath2 and @xmath10 .
however , there exists a huge uncertainty between the complete photometric and confirmed spectroscopic sample at @xmath6 ; we limit the comparison between the theoretical model and observations to the data accumulated at @xmath2 by shimasaku et al .
( 2006 ) , in this work .
the calculations of @xmath50 , @xmath51 , @xmath56 , @xmath53 and @xmath57 are explained in more detail in what follows .
star formation produces photons with energies @xmath61 ryd , which ionize the in the ism . due to the high density of the ism ,
recombinations take place on a short time scale and this produces a large number of ly@xmath0 photons .
the stellar luminosity component is calculated by using starburst99 ( leitherer et al .
1999 ) , a population synthesis code , to obtain the spectra and hence , the ionizing photon rate ( @xmath62 ) of each galaxy , using a salpeter imf , @xmath26 , @xmath30 and @xmath22 as input parameters . then the intrinsic ly@xmath0 luminosity produced by stellar sources @xmath63 , is calculated as @xmath64 where @xmath65 , for consistency with the rt calculation ( see sec . [ simulations ] ) , @xmath66 is the planck constant , @xmath67 is the frequency of the ly@xmath0 photons and the factor of two - thirds enters assuming case b recombination ( osterbrock 1989 ) . the intrinsic continuum luminosity produced by stellar sources , @xmath68 , is calculated at 1375 , at the middle of the continuum band between 1250 - 1500 .
the ly@xmath0 and continuum luminosity from the cooling of collisionally excited in each galaxy ( @xmath69 respectively ) is calculated using the number density of electrons and , which depend on the temperature distribution of the ism gas mass and the gas distribution scale of the galaxy .
the temperature dependence comes from the fact that the recombination coefficient , which determines the neutral hydrogen fraction in the ism , decreases sensitively with higher temperatures .
the number density is calculated by assuming a primordial ism composition ( @xmath70 hydrogen and @xmath71 helium ) and calculating the gas distribution scale assuming the to be concentrated in a radius , @xmath72 . here
the spin parameter , @xmath73 ( ferrara , pettini & shchekinov 2000 ) and @xmath74 is the virial radius , assuming the collapsed region has an overdensity of 200 times the critical density at the redshift considered .
the calculation of the electron number density requires a knowledge of the temperature distribution of the igm gas .
however , since we do not have this information from the sph simulation used in this work , we use the temperature distribution averaged over distinct halo mass ranges shown in tab . 1 of dayal et al .
( 2009b ) and again use the gas distribution scale to calculate the electron number density .
complete details of this calculation can be found in sec .
3.2 of dayal et al .
( 2009b ) .
it is important to model the dust in the ism of galaxies since it absorbs both ly@xmath0 and continuum photons , thereby , strongly influencing the observed luminosity values .
dust is produced both by supernovae and evolved stars in a galaxy .
however , several authors ( todini & ferrara 2001 ; dwek et al .
2007 ) have shown that the contribution of agb stars can be neglected for @xmath75 , since the typical evolutionary time - scale of these stars ( @xmath76 gyr ) becomes longer than the age of the universe above that redshift .
hence , assuming snii to be the primary dust factories at @xmath21 , we calculate the dust enrichment for each galaxy , taking into account three processes : ( a ) snii produce dust in the expanding ejecta ; the average dust mass produced per snii is taken to be @xmath77 ( todini & ferrara 2001 ; nozawa et al .
2003 , 2007 ; bianchi & schneider 2007 ) , ( b ) snii destroy dust in the ism they shock to velocities @xmath78 , for which we use a destruction efficiency value @xmath79 ( mckee 1989 ) , and ( c ) a homogeneous mixture of gas and dust is assimilated into star formation ( astration ) .
once the final dust mass , @xmath80 , is calculated for each galaxy in the simulation , we can translate this into an optical depth , @xmath81 , for continuum photons as @xmath82 where @xmath83 is the dust surface mass density , @xmath84 is the radius of dust distribution , @xmath85 and @xmath86 are the radius and material density of graphite / carbonaceous grains , respectively ( @xmath87 , @xmath88 ; todini & ferrara 2001 ; nozawa et al .
2003 ) . this can be used to obtain the escape fraction of continuum photons from the galaxy as @xmath89 .
the determination of @xmath84 follows in sec .
[ reio effects ] and complete details of the calculations mentioned here can be found in dayal et al .
( 2009b ) . different extinction curves , including that for the milky way , small magellanic cloud and for supernovae give different relations between the extinction of the ly@xmath0 and continuum photons , for homogeneously distributed dust . in this spirit , and
to get a hint of the dust inhomogeneity , we combine @xmath56 and @xmath53 , and use the escape fraction of ly@xmath0 photons relative to the continuum ones , @xmath90 , to be the free parameter to calculate @xmath55 in our model .
a more detailed explanation of the determination of @xmath53 and @xmath56 , required to reproduce the observations , follows later in sec .
[ reio effects ] .
the calculation of @xmath57 is now explained in better detail .
if @xmath91 and @xmath92 are the redshifts of the emitter and the observer respectively , we calculate the total optical depth ( @xmath93 ) to the ly@xmath0 photons along a line of sight ( los ) as @xmath94 where @xmath95^{-1}$ ] is the rest - frame velocity of a photon with wavelength @xmath96 , relative to the line centre ( rest - frame wavelength @xmath97 , velocity @xmath98 ) , @xmath99 is the voigt profile , @xmath100 is the number density of along the los , @xmath101^{-1}$ ] and @xmath102 represents the light speed .
we use @xmath103^{-1}$ ] , where @xmath104 , @xmath105 are the electron charge and mass respectively and @xmath106 is the oscillator strength ( 0.4162 ) . for regions of low density , the natural line broadening is not very important and the voigt profile can be approximated by the gaussian core : @xmath107 where @xmath108 is the velocity of a photon of initial velocity @xmath109 at a redshift @xmath110 along the los , @xmath111 is the peculiar velocity at @xmath110 and @xmath112 is the velocity of the ly@xmath0 photons at @xmath110 , which is 0 in this expression .
the doppler width parameter is expressed as @xmath113 , where @xmath114 is the hydrogen mass , @xmath115 is the boltzmann constant and @xmath116 is the igm temperature .
while we use a @xmath117 for each cell with an ionization fraction @xmath118 , we use @xmath119 k if the igm is pristine , i.e. has an ionization fraction @xmath120 .
, only 8 cells out of @xmath121 have @xmath120 , with the complete rt simulation as described in sec .
[ simulations ] . hence , to simplify the calculations , we use @xmath122 k for all the cells for all the crash outputs . ] in more dense regions the lorentzian damping wing of the voigt profile becomes important . according to peebles ( 1993 ) ,
this can be approximated as @xmath123 } \label{lorentz}\ ] ] where @xmath124^{-1}$ ] and @xmath125 s@xmath126 is the decay constant for the ly@xmath0 resonance .
although computationally more expensive than the above approximations , using the voigt profile to compute the absorption cross - section gives precise results , and therefore we have implemented it in our code to obtain all the results presented below .
the total ly@xmath0 optical depth is calculated for each galaxy identified in the crash outputs corresponding to @xmath127 as mentioned in sec .
[ simulations ] .
the transmission along any los is calculated as @xmath128 , integrating from the position of each galaxy to the edge of the box . to get a statistical estimate of the transmission , we construct 6 los starting from the position of each galaxy , one along each of the box axes , for each of the mentioned ionization states .
the average value of the transmission for each galaxy is then obtained by averaging the transmission over the 6 los .
this is done for all the ionization state configurations resulting from the rt calculation .
once the above model is in place , the only _ two free parameters _ of our model to match the theoretical ly@xmath0 and uv lfs to the observed ones are : the dust distribution radius , @xmath84 , and the relative escape fraction of ly@xmath0 photons as compared to the continuum photons , @xmath90 .
both these parameters remain quite poorly understood due to a lack of observational data about the dust distribution / topology in high - redshift galaxies ; they must therefore be inferred by comparing the theoretical lfs to the observed ones .
details on how these parameters are determined follow in sec . [ reio effects ] .
once the combined sph+rt calculations are carried out and the lae model implemented , we are in a position to quantify the importance of reionization , peculiar velocities and the dust enrichment on the observed lfs .
physically , the reionization process leads to a decrease in @xmath42 , thereby increasing @xmath57 ; on the other hand , peculiar velocities caused by galactic scale outflows ( inflows ) , redshift ( blueshift ) the ly@xmath0 photons , thereby leading to a higher ( lower ) value of @xmath57 .
a handle on the dust enrichment is necessary since dust grains absorb both ly@xmath0 and continuum photons , thereby affecting their escape fractions from the galaxy .
we begin our study by ignoring the effects of the peculiar velocities and assuming all the galaxies to be dust free , to quantify how each of these two parameters shapes the observed ly@xmath0 and uv lfs .
the former assumption means that @xmath129 in eqs .
[ gauss ] , [ lorentz ] ; the latter implies that all photons produced inside the galaxy escape into the igm , i.e. @xmath130 in eqs .
[ lc_int ] , [ la_int ] .
we then use the prescriptions detailed in sec . [ model ] to identify the galaxies that would be visible as laes in each crash output and build their ly@xmath0 lf , as shown in fig .
[ lya_novel_nodust ] .
since both peculiar velocities and dust are neglected , in this case , while the uv lf is simply the intrinsic lf , the ly@xmath0 lf is shaped solely by the transmission through the igm .
as mentioned in sec . [ simulations ] , after a star formation time scale as short as 10 myr ( @xmath131 ) , the galaxies in the simulation snapshot are able to form regions , with the region size and the ionization fraction inside it increasing with @xmath22 ; this results in a @xmath57 which increases with @xmath132 . as the galaxies continue to form stars , the sizes of the regions built by each source increase with time , leading @xmath133 to decrease to @xmath134 at @xmath135 myr respectively .
this leads to an increase in the transmission of the red part of the ly@xmath0 line for all the sources , in turn yielding a corresponding increase of the ly@xmath0 luminosity of each source , as depicted in fig .
[ lya_novel_nodust ] .
however , after about 200 myr , @xmath42 reduces to @xmath136 ; the regions built by each source are large enough so that almost all of the red part of the ly@xmath0 line is transmitted and the value of @xmath57 saturates for all laes .
this results in very similar lfs for @xmath137 , as seen from the same figure .
the key point here is that , if peculiar velocities and dust effects are neglected , at no stage of the reionization history , either the slope or the amplitude of the observed ly@xmath0 lf can be reproduced ; analogous problems arise also when the uv lfs are considered .
the above discussion implies the need of one or more physical effects attenuating the ly@xmath0 and continuum photons , the most obvious of which is the presence of dust in the ism of these galaxies , which would absorb both ly@xmath0 and continuum photons , simultaneously reducing @xmath56 and @xmath53 .
we then include the dust model described in sec .
[ dust_model ] into our calculations .
since the uv is unaffected by the in the igm as mentioned in sec . [ model ] , the same value of @xmath138 ( @xmath139 is the ism gas distribution scale , sec . [ dust_model ] ) reproduces the uv lf for all of the ionization states of the igm , ranging from @xmath131 to @xmath140 .
this assumption of the dust distribution scale leads to an average escape fraction of continuum photons , @xmath141 ( refer sec .
[ dust_model ] for details ) for the galaxies we identify as laes .
we now momentarily digress to discuss the effect of reionization on @xmath57 ( defined by eq .
[ la_int ] ) for different stellar mass ranges .
however , since @xmath142 scales with @xmath48 ( details in sec .
[ phy prop ] ) , we discuss @xmath57 in terms of @xmath22 . as mentioned in sec .
[ simulations ] , we initialize the crash runs with @xmath143 . in a star formation timescale of 10 myr , by virtue of the regions that already start growing
, the total photoionization rate ( sum of contributions from the uvb and all the galaxies in the simulation ) has a value of @xmath144 , the total photoionization rate , is calculated assuming ionization - recombination equilibrium over average values of density and ionization fraction in the simulation volume , so that @xmath145 here , @xmath146 is the average hydrogen number density in the simulation volume and @xmath147 is the case b recombination co - efficient . ]
@xmath148 and @xmath42 decreases slightly to 0.295 . at this point , due to their smaller ionizing photon output ,
galaxies with @xmath35 , have @xmath149 ; galaxies with larger @xmath22 , have @xmath150 , as shown in panel ( a ) of fig .
[ fractx_ms ] . as
the star formation continues , for 50 ( 100 ) myr of star formation , the region sizes increase and @xmath144 increases slightly to @xmath151 @xmath152 ; @xmath57 increases and ranges between @xmath153 ( 0.34 - 0.44 ) for @xmath154 , panel b ( c ) , fig .
[ fractx_ms ] . finally for @xmath155 , corresponding to @xmath156 , the transmission settles to @xmath157 for @xmath154 ( panels d - f ) ; in about 200 myr from the ignition of star formation , the strmgren spheres built by these laes are large enough so that the redshifted ly@xmath0 photons are no longer affected by the outside this region .
however , the residual inside this ionized region leads to an absorption of the photons blueward of the ly@xmath0 line , and hence , about half of the line is transmitted .
the values of @xmath144 , @xmath42 and the @xmath57 values averaged for all laes in each crash output are shown in tab .
[ tab1 ] . using the above @xmath57 values and dust model fixed by the uv lf , the only free parameter we are left with , to match the theoretical and observed ly@xmath0 lfs is @xmath90
; this only scales the ly@xmath0 lf without affecting its shape . as mentioned above , the @xmath57 value for galaxies identified as laes settles for @xmath158 , which leads to a corresponding saturation in the ly@xmath0 lf .
we find that @xmath159 reproduces the slope and the magnitude of the observed ly@xmath0 lf quite well for all the crash outputs where @xmath158 , as shown in fig .
[ lya_novel_incdust ] .
however , due to decreasing values of @xmath57 , the ly@xmath0 lf is progressively under - estimated for increasing values of @xmath42 , as seen from the same figure .
once the above framework is in place , we include and study the effect of the final component that can affect the observed luminosity , the presence of peculiar velocities .
once this is done by consistently deriving the peculiar velocity field from the simulations , we again reproduce both the magnitude and slope of the observed uv lf as shown in fig .
[ uv_incvel_incdust ] . of course
this does not come as a surprise since velocity fields do not affect uv photons .
however , galactic scale outflows ( inflows ) from a galaxy , redshift ( blueshift ) the ly@xmath0 photons , thereby leading to a higher ( lower ) @xmath57 value .
it is quite interesting to see that we can again match the theoretical ly@xmath0 lfs to the observed ones by a simple scaling between @xmath56 and @xmath53 for @xmath158 , as seen from fig .
[ lya_incvel_incdust ] .
however , we require a much higher value of @xmath160 ( models @xmath161 , tab .
[ tab1 ] ) , compared to the value of @xmath162 excluding velocity fields ( models @xmath163 , tab .
[ tab1 ] ) .
again , @xmath57 , and hence the ly@xmath0 lf , get progressively more damped with an increase in @xmath42 , as seen from fig .
[ lya_incvel_incdust ] , ( models @xmath164 , tab .
[ tab1 ] ) .
we now discuss the reason for the higher @xmath160 value required to reproduce the ly@xmath0 lf when velocity fields are considered , as compared to the ratio of @xmath162 , when they are not , for @xmath158 .
the galaxies we identify as laes have halo masses @xmath165 , which correspond to @xmath166 fluctuations at @xmath21 . these laes are therefore subject to strong inflows since they lie in dense regions .
as these inflows blue - shift the ly@xmath0 photons , even photons in the red part of the line are attenuated , thereby reducing @xmath57 .
this can be seen clearly from fig .
[ fractx_ms ] , where the solid lines , which represent the outcomes from the models including velocities , are always significantly below the dotted ones , which represent the model with no velocity field included .
when velocity fields are included we find @xmath167 ( @xmath168 ) for @xmath131 ( @xmath169 ) for @xmath154 .
averaging over all the laes for @xmath170 , we find the value of @xmath57 @xmath171 ( @xmath172 ) including ( excluding ) the effects of peculiar velocity fields .
this requires that when peculiar velocities are included , a correspondingly larger fraction ( @xmath173 ) of ly@xmath0 photons must escape the galaxy , undamped by dust to bring the observed luminosity up to the levels it would reach in the absence of these inflows .
although many galaxies do show outflows , powered by supernova explosions , these dominate at very small scales ( @xmath174 physical mpc ) . however , the small redshift boost imparted by these is negligible compared to the blue - shifting of the ly@xmath0 line because of large scale inflows ; eventually , it is these dominant inflows that determine the value of @xmath57 .
further , including velocity fields changes the slope of the lf ; when velocity fields are not included , @xmath57 basically scales with @xmath22 ; when these are included , @xmath57 is the most damped for the largest masses , since these see the strongest inflow velocities by virtue of their largest potential wells .
this can be seen clearly from fig .
[ fractx_ms ] where the slope of @xmath57 is visibly shallower for the solid curves ( including velocities ) than for the dotted ones ( neglecting velocities ) .
this has the effect of _ flattening _ the slope of the ly@xmath0 lf as can be seen in fig .
[ lya_incvel_incdust ] ) . [ cols="^,^,^,^,^,^,^,^,^ " , ] as an important result , the above analysis shows that there exists a degeneracy between the ionization state of the igm and dust clumping ( or grain properties ) inside high - redshift galaxies , i.e a high ( low ) @xmath57 can be compensated by a low ( high ) @xmath56 .
this is shown in fig .
[ prob_contours ] , where we find that within a @xmath175 error , for @xmath176 , we can not distinguish an igm with @xmath177 from one where @xmath178 . within the area under the @xmath3 error , @xmath179 can also fit the ly@xmath0 lf for @xmath180 .
this leads to the very interesting conclusion that the ionization state of the igm can not be tightly constrained unless the relative escape fraction of ly@xmath0 compared to the continuum photons is reasonably well understood .
we now discuss the dusty nature of the laes we identify from the sph simulation .
we require that for the complete lae model ( @xmath181 , tab .
[ tab1 ] ) , the value of @xmath90 must range between @xmath182 ( @xmath183 ) for an average neutral hydrogen fraction of @xmath184 ( @xmath185 ) .
however , no single extinction curve gives a value of @xmath186 .
one of the simplest ways of explaining this large relative escape fraction is to invoke the multiphase ism model as proposed by neufeld ( 1991 ) , wherein the ism is multiphase and consists of a warm gas with cold dust clumps embedded in it .
this inhomogeneity of the dust distribution can then lead to a larger attenuation of the continuum photons relative to the ly@xmath0 .
we also translate @xmath53 into the color excess , @xmath187 , for models @xmath181 ( tab .
[ tab1 ] ) and compare this value to other high redshift lae observations . at high redshifts , the observed properties of the most distant quasars ( maiolino et al .
2006 ) and gamma - ray bursts ( stratta et al . 2007 ) can be successfully interpreted using a sn extinction curve ( todini & ferrara 2001 ; bianchi & schneider 2007 ) . using the same curve , we find the average value of the color excess , @xmath188 , corresponding to an average @xmath189 .
this inferred color excess value is in good agreement with recent experimental determinations : by fitting the seds of 3 laes at @xmath190 , lai et al .
( 2007 ) have inferred @xmath191 ; in a sample of 12 laes at @xmath192 , finkelstein et al .
( 2009 ) have found @xmath193 ; finally , pirzkal et al . ( 2007 ) have found @xmath194 for 3 galaxies at @xmath195 .
we now summarize our two main results : ( a ) we find that the ly@xmath0 lf can be well reproduced ( to within a @xmath3 error ) by an average neutral hydrogen fraction as high as 0.24 ( an almost neutral igm ) , to a value as low as @xmath140 , corresponding to an ionized igm , provided that the increase in the transmission is compensated by a decrease in the ly@xmath0 escape fraction from the galaxy , ( b ) we find that to reproduce the ly@xmath0 lf , for any ionization state of the igm , we require @xmath196 , a value that can not be obtained using any existing extinction curve ; this raises the need to invoke a multiphase ism model , in which dust clumps are embedded in a highly ionized ism , to facilitate the ly@xmath0 photon escape relative to that of continuum photons .
we are now in a position to discuss the physical properties of the galaxies we identify as laes , including the stellar and gas mass , metallicity , stellar ages , dust mass and escape fraction of ly@xmath0 photons from the ism . as mentioned in sec .
[ reio effects ] , while the uv lf is independent of the ionization state of the igm , the ly@xmath0 lf can be well reproduced if a low ( high ) ly@xmath0 escape fraction from the galaxy is compensated by a high ( low ) transmission through the igm .
hence , by scaling up @xmath90 for increasing values of @xmath197 ( models m1-m3 , tab . [ tab1 ] ) , we would broadly always identify the same galaxy population as laes .
we now show the physical properties for the laes identified using model m7 ( tab .
[ tab1 ] ) , which includes the full rt calculation , density / velocity fields and dust , with @xmath198 .
we find that there is the direct correlation between @xmath47 and @xmath48 ; laes with a larger @xmath48 are also more gas rich .
however , the ratio @xmath199 is the largest ( smallest ) for laes with the smallest ( largest ) @xmath48 , as one can deduce from panel ( a ) of fig .
[ phy_prop ] .
if we assume the halo mass to scale with the total baryonic mass ( @xmath200 ) , according to the cosmological ratio , this implies that more massive galaxies are more efficient in turning their gas into stars .
this trend is as expected , since even a small amount of star formation activity in low mass galaxies can lead to large outflows of gas , thereby suppressing further star formation .
this situation however , does not occur in galaxies with larger masses , which do not witness large , galactic scale outflows , by virtue of their much larger potential wells ( mac low & ferrara 1999 ) .
laes with @xmath201 have @xmath202 , i.e. most of the laes have a sustained but not exceptionally large star formation activity .
however , galaxies with larger stellar masses are much more efficient in converting their gas into stars , leading to sfr as large as @xmath203 , as seen from panel ( b ) , fig .
[ phy_prop ] , which is just a positive feedback of the @xmath204 relation mentioned above .
although the average age of the galaxies are calculated as @xmath205 , it is interesting to see that all the galaxies we identify as laes have @xmath206 myr , as shown in panel ( c ) of fig .
[ phy_prop ] ; further the standard deviation on the smallest ages are the smallest . since we calculate the ages assuming @xmath22 to have been constant over the entire star formation history , it is entirely possible that many of the laes are actually older ( younger ) if @xmath22 in the past was smaller ( larger ) than the final value at @xmath21 .
however , we are unable to comment on this further in absence of the complete star formation history for each galaxy . using these average ages
, we find that laes are intermediate age galaxies , instead of being extremely young ( @xmath207 10 myr ) or extremely old ( @xmath208 gyr ) objects .
this confirms an earlier result obtained in dayal et al .
( 2009a ) , based on an accurate modelling of the star formation history , that it is unlikely that young ages could be responsible for the large equivalent widths ( @xmath209 ) observed for a number of laes at @xmath210 by dawson et al .
( 2007 ) and shimasaku et al .
( 2006 ) respectively .
a complete discussion on the ew distribution is deferred to sec .
[ conc ] of this paper .
the mass weighted stellar metallicity of laes scales with @xmath48 as shown in panel ( d ) of fig .
[ phy_prop ] ; laes with a higher @xmath22 are more dust enriched , which is only to be expected since the metals have a stellar origin .
the metallicity values for the smallest halos show the largest dispersion , possibly arising due to the differing values of feedback in these low mass halos .
compared to the analogous mass - metallicity relation observed at lower redshifts ( tremonti et al .
2004 ; panter et al . 2008 ; maiolino et al .
2008 ) , we do not see the sign of a flattening of metallicity towards larger masses which might imply that at high redshift galaxies are not massive enough to retain their metals and behave as close - boxes . in our model
, we have assumed snii to be the primary dust factories , with the snii rate , where @xmath211 for a salpeter imf between the lower and upper mass limits of @xmath212 and @xmath213 respectively ( ferrara , pettini and shchekinov 2000 ) . ] being proportional to @xmath22 . in our model , the dust amount is regulated solely by stellar processes : dust is produced by snii , destroyed in the ism shocked by snii and astrated into stars , as mentioned before in sec .
[ dust_model ] . the dust amount , therefore , scales with @xmath22 , with the most star forming galaxies being most dust enriched , as shown in panel ( e ) of fig . [ phy_prop ] .
a caveat that must be mentioned here is that as outflows tend to occur on smaller scales with respect to inflows , they are only marginally resolved by our rt simulations . lacking this information
, we have not included their destructive impact into the computation of the dust mass , which could therefore be somewhat overestimated ( refer dayal et al .
2009b for details ) . as expected , we find that @xmath56 decreases with increasing dust enrichment of the galaxy ; galaxies with larger @xmath48 , and hence , @xmath22 , have a smaller ly@xmath0 escape fraction .
the value of @xmath56 decreases by a factor of about 3 , going from 0.9 to 0.3 as @xmath48 runs from @xmath214 , as shown in panel ( f ) , fig .
[ phy_prop ] .
although a number of simulations have been used in the past few years ( mc quinn et al . 2007 ;
iliev et al .
2008 , nagamine et al . 2008 ; dayal et al .
2009a , 2009b ; zheng et al .
2009 ) , the relative importance of reionization , velocity fields and dust , on shaping the observed ly@xmath0 and uv lfs has remained rather obscure so far . to this end
, we build a coherent model for laes , which includes : ( a ) cosmological sph simulations run using gadget-2 , to obtain the galaxy properties ( @xmath22 , @xmath26 , @xmath48 ) , ( b ) the ly@xmath0 and continuum luminosities produced , calculated using the intrinsic properties of each galaxy , accounting for both the contribution from stellar sources and cooling of collisionally excited in their ism , ( c ) a model for the dust enrichment , influencing the escape fraction of ly@xmath0 and continuum photons , and ( d ) a complete rt calculation , carried out using crash , including the effects of density and velocity fields , used to calculate the transmission of ly@xmath0 photons through the igm . in spite of this wealth of physical effects ,
we are still left with two free parameters : the radius of dust distribution , @xmath84 , which determines the optical depth to continuum photons ( see eq .
[ tau_dust ] ) and , the escape fraction of ly@xmath0 photons relative to the continuum photons , @xmath90 .
these then must be constrained by comparing the theoretical model to the observations .
starting our analysis by assuming all galaxies to be dust - free and ignoring gas peculiar velocities , we find that with such a scheme , none of the ionization states of the igm with @xmath215 to @xmath216 , can reproduce either the slope or the galaxy number density of the observed uv or ly@xmath0 lf .
we then include the dust model described in sec .
[ dust_model ] into our calculations , such that each snii produces @xmath217 of dust , snii shocks destroy dust with an efficiency of about @xmath218% and the dust distribution radius is about half of the gas distribution radius , @xmath219 .
since the uv is unaffected by the in the igm as mentioned in sec . [ model ] , the same dust model parameters reproduce the uv lf for all of the ionization states of the igm , ranging from @xmath131 to @xmath140 . we find that we can reproduce the ly@xmath0 lf using @xmath159 ( 3.7 ) for all ionization states such that @xmath158 excluding ( including ) peculiar velocity fields , since the ly@xmath0 lf settles to a constant value at this point due to a saturation in @xmath57 , as explained in sec .
[ reio effects ] .
higher values of @xmath42 naturally lead to a lower transmission , thus leading to a progressive underestimation of the ly@xmath0 lf .
the higher @xmath90 value required to reproduce the observations when velocity fields are included , arise because laes reside in @xmath220 fluctuations ; laes are therefore subject to strong inflows which blue - shift the ly@xmath0 photons , thereby reducing @xmath57 .
this decrease in @xmath57 must therefore be compensated by a larger escape fraction from the galaxy to fit the observations .
further , due to the larger masses , @xmath57 is most damped for the largest galaxies , when velocity fields are included , which has the effect of _ flattening _ the slope of the ly@xmath0 lf .
the above _ degeneracy _ between the ionization state of the igm and the dust distribution / clumping inside high - redshift galaxies has been quantified ( see fig . [ prob_contours ] ) ; the ly@xmath0 lf can be well reproduced ( to within a @xmath3 error ) by @xmath221 , corresponding to a highly neutral igm , to a value as low as @xmath140 , corresponding to an ionized igm , provided that the increase in @xmath57 is compensated by a decrease in the ly@xmath0 escape fraction from the galaxy .
this leads to the very interesting conclusion that the ionization state of the igm can not be constrained unless the escape fraction of ly@xmath0 versus continuum photons is reasonably well understood .
this is possible only through simulations that model the physics of the ism , including the clumping of dust , and include a rt calculation for both ly@xmath0 and continuum photons on such ism topologies .
however , the clumping of dust depends on the turbulence in the ism , which itself remains only poorly understood .
therefore , attempts to break the mentioned degeneracy must be deferred to future works . as for the dusty nature of laes , we find that @xmath222 , averaged over all the laes in any snapshot , which corresponds to a color excess , @xmath223 , using a sn extinction curve .
this value is in very good accordance with the observed values of @xmath191 ( lai et al .
2007 ) , @xmath193 ( finkelstein et al . 2009 ) and @xmath194 ( pirzkal et al .
2007 ) . secondly , since no single extinction curve ( milky way
, small magellanic cloud , supernova ) gives a value of @xmath186 ; this larger escape fraction of ly@xmath0 photons relative to the continuum can be understood as an indication of a multi - phase ism model ( neufeld 1991 ) where dust clumps are embedded in a more diffuse ionized ism component .
laes with a larger @xmath48 are also more gas rich . however ,
the ratio @xmath199 is the largest ( smallest ) for laes with the smallest ( largest ) stellar mass , which implies that more massive galaxies ( @xmath224 ) are more efficient in turning their gas into stars ( @xmath225 ) , i.e. , star formation is suppressed ( @xmath226 ) in low mass galaxies ( @xmath227 ) due to mechanical feedback .
all the galaxies we identify as laes have @xmath228 myr ; laes are intermediate age galaxies , instead of being extremely young ( @xmath207 10 myr ) or extremely old ( @xmath208 gyr ) objects , as shown in dayal et al .
( 2009a ) .
the mass weighted stellar metallicity of laes scales with @xmath48 but the metallicity of the smallest halos has a large dispersion , possibly arising due to the increasing importance of feedback towards low mass halos .
since in our model , we have assumed snii to be the primary dust factories , the dust amounts are regulated solely by stellar processes .
the dust amounts , therefore , scale with @xmath22 , with the most star forming galaxies being most dust enriched . as expected , we find that @xmath56 decreases with increasing dust enrichment of the galaxy ; galaxies with larger @xmath48 , and hence , @xmath22 , have a smaller ly@xmath0 escape fraction .
the value of @xmath56 decreases by a factor of about 3 , going from 0.9 to 0.3 as @xmath48 increases from @xmath229 .
we finally comment on the ew distribution obtained from our model and compare it to the observations ( fig .
there is a trend of increasing ew in more luminous objects which can not be yet solidly identified in the available ( and uncertain ) observational data .
in addition , the observed ews show a large dispersion , varying by as much as a factor of 3 for the same @xmath55 value ; this suggests that the observed ew depends on several structural parameters of laes , namely @xmath22 , peculiar velocities and dust clumping , to name a few .
by virtue of using a lae model which depends on the intrinsic galaxy properties , includes a dust calculation and takes peculiar velocities/ density inhomogeneities into account , we also obtain a large spread ( 20 - 222 ) in the observed ews .
this is a clear improvement on our own previous work ( dayal et al .
2009 ) in which some of these ingredients were not yet accounted for . at the very last ,
we discuss a few caveats .
first , we have used a constant value of the escape fraction of ionizing photons , @xmath31 in all our calculations . however , the value of @xmath230 remains poorly constrained both observationally and theoretically .
while @xmath231 galaxies have been used to obtain values of @xmath232 ( fernandez - soto et al .
2003 ) to @xmath233 ( steidel et al .
2001 ) , observations in the local universe range between @xmath234 ( deharveng et al . 1997 ) to @xmath235 ( catellanos et al .
. a larger ( smaller ) value of @xmath230 would lead to larger ( smaller ) regions , thereby affecting the progress of the reionization process and hence @xmath57 ; a change in @xmath230 would also affect @xmath50 , sec .
[ intlum_model ] .
secondly , the average age of the galaxies are calculated as @xmath205 .
since we calculate the ages assuming constant @xmath22 , it is entirely possible that some of the laes could be slightly older ( younger ) if @xmath22 in the past was smaller ( larger ) than the final value at @xmath21 .
galaxies younger than @xmath236 myr would have higher ews as compared to the values shown here ; however , this could be possible only for the the smallest galaxies , which could assemble due to mergers and undergo star formation in a time as short as 10 myr .
third , in the dust model explained in sec .
[ dust_model ] , we have considered dust destruction by forward sweeping snii shocks . however , bianchi & schneider ( 2007 ) have shown that reverse shocks from the ism can also lead to dust destruction , with only about 7% of the dust mass surviving the reverse shock , for an ism density of @xmath237 .
since we neglect this effect , we might be over - predicting the dust enrichment in laes .
this , however , does not affect our results because the dust optical depth depends on the surface density of the dust distribution as shown in eq .
[ tau_dust ] ; a large ( small ) dust mass can be distributed in a large ( small ) volume to obtain identical values of the optical depth .
we thank c. evoli , s. gallerani , r. schneider and r. valiante for useful comments . discussions and the stimulating environment at david iv , held at oaarcetri ,
florence is kindly acknowledged .
|
we present the first lyman alpha emitter ( lae ) study that combines : ( i ) cosmological sph simulations run using gadget-2 , ( ii ) radiative transfer simulations ( crash ) , and ( iii ) a previously developed lae model .
this complete lae model accounts for the intrinsic lae ly@xmath0/continuum luminosity , dust enrichment and ly@xmath0 transmission through the intergalactic medium ( igm ) , to quantify the effects of reionization , dust and velocity fields on the ly@xmath0 and uv luminosity functions ( lf ) .
we find that a model neglecting dust sorely fails to reproduce either the slope or the magnitude of the observed ly@xmath0 and uv lfs .
clumped dust is required to simultaneously fit the observed uv and ly@xmath0 lfs , such that the intrinsic ly@xmath0-to - continuum luminosity is enhanced by a factor @xmath1 ( 3.7 ) excluding ( including ) peculiar velocities .
the higher value including velocity fields arises since laes reside in large potential wells and inflows decrease their ly@xmath0 transmission . for the first time , a _
degeneracy _ is found between the the ionization state of the igm and the clumping of dust inside high - redshift galaxies .
the ly@xmath0 lf @xmath2 can be well reproduced ( to within a @xmath3 error ) by a wide range of igm average neutral hydrogen fraction , @xmath4 , provided that the increase in the ly@xmath0 transmission through a more ionized igm is compensated by a decrease in the ly@xmath0 escape fraction from the galaxy due to dust absorption .
the physical properties of laes are presented , along with a discussion of the assumptions adopted .
[ firstpage ] methods : numerical - radiative transfer - galaxies : high redshift - luminosity function - ism : dust - cosmology : theory
|
if @xmath0 is a set of matroids , then let @xmath1 be the set of matroids such that @xmath6 if and only if @xmath7 has no minor isomorphic to a member of @xmath0 .
thus , if @xmath8 , then @xmath9 is the set of graphic matroids of planar graphs . hall s
classical theorem on the graphs without a @xmath10-minor @xcite can be interpreted as saying that @xmath11 contains only a single @xmath5-connected matroid , namely @xmath12 itself .
this motivates the following definition : if @xmath0 is a set of matroids , then @xmath2 is a _
superfluous _ subset of @xmath0 if @xmath4 contains only finitely many @xmath5-connected matroids .
thus @xmath13 is a superfluous subset of @xmath14 .
obviously every subset of a superfluous subset is itself superfluous . in this article
we characterize the superfluous subsets of six well - known collections of excluded minors .
we will concentrate on the excluded minors for classes of matroids representable over partial fields .
partial fields were introduced by semple and whittle @xcite , prompted by whittle s investigation of classes of ternary matroids @xcite .
a _ partial field _ is a pair @xmath15 , where @xmath16 is a commutative ring with identity , and @xmath17 is a subgroup of the multiplicative group of @xmath16 , such that @xmath18 .
note that every field , @xmath19 , can be seen as a partial field , @xmath20 . for more information on partial fields , and matroid representations over them ,
we refer to @xcite . to date
, the class of matroids representable over a partial field has been characterized via excluded minors in only six cases .
those cases are : the fields @xmath21 , @xmath22 , and @xmath23 , the regular partial field , and two of the partial fields discovered by whittle , namely the sixth - roots - of - unity partial field , and the near - regular partial field .
we will characterize the superfluous subsets of all these collections of excluded minors .
first of all , tutte @xcite showed that the only excluded minor for the class of @xmath21-representable matroids is @xmath24 .
it is clear that the only superfluous subset in this case is the empty set . for a more interesting example
, we examine the _ regular _ partial field , @xmath25 .
tutte also proved that the set of excluded minors for @xmath26-representable matroids is @xmath27 .
it is a well - known application of seymour s splitter theorem @xcite that @xmath28 is a splitter for the class @xmath29 .
the next theorem follows easily from this fact .
[ thm : regsuper ] the only non - empty superfluous subsets of @xmath27 are @xmath30 and @xmath31 .
the only @xmath5-connected matroid in @xmath32 is @xmath28 .
next we consider the excluded - minor characterization of @xmath22-representable matroids , due to bixby and seymour @xcite .
[ thm : gf3excluded ] the set of excluded minors for @xmath22-representable matroids is @xmath33 .
[ thm : gf3super ] the only non - empty superfluous subsets of @xmath33 are @xmath30 and @xmath31 .
the only @xmath5-connected matroid in @xmath34 is @xmath28 .
the set of excluded minors for @xmath23-representable matroids was characterized by geelen , gerards , and kapoor @xcite .
[ thm : gf4excluded ] the set of excluded minors for the class of @xmath23-representable matroids is @xmath35 . let @xmath36 be the set of excluded minors in theorem [ thm : gf4excluded ] .
geelen , oxley , vertigan , and whittle showed the following : [ thm : govw02 ] let @xmath7 be a @xmath5-connected matroid .
then one of the following holds : 1 .
@xmath7 is @xmath23-representable ; 2 .
@xmath7 has a minor isomorphic to one of @xmath37 ; 3 .
@xmath7 is isomorphic to @xmath38 ; 4 .
@xmath7 is isomorphic to a minor of @xmath39 .
this implies that @xmath40 is a superfluous subset of @xmath36 .
we complement this theorem by showing that it is best possible : [ thm : govwcomplement ] the only superfluous subsets of @xmath36 are the subsets of @xmath40 . the only @xmath5-connected matroids in @xmath41 are isomorphic to @xmath42 , or minors of @xmath39 .
let @xmath43 be the _ sixth - roots - of - unity _ partial field , so that a matroid is @xmath44-representable if and only if it is both @xmath22- and @xmath23-representable . by combining theorems [ thm : gf3excluded ] and [ thm : gf4excluded ] , geelen , gerards , and kapoor derived the following result ( * ? ? ?
* corollary 1.4 ) .
[ thm : sruexcluded ] the set of excluded minors for the class of @xmath44-representable matroids is @xmath45 . let @xmath46 be the set of excluded minors in theorem [ thm : sruexcluded ] .
[ thm : srusuper ] the only superfluous subsets of @xmath46 are the subsets of @xmath47 and @xmath48 .
the only @xmath5-connected matroids in @xmath49 are isomorphic to @xmath28 , or minors of @xmath39 .
let @xmath50 be the _ near - regular _ partial field .
a matroid is @xmath51-representable if and only if it is representable over @xmath22 , @xmath23 , and @xmath52 .
the next theorem is proved in @xcite .
[ thm : nregexcluded ] the set of excluded minors for the class of @xmath51-representable matroids is @xmath53 let @xmath54 be the set featured in theorem [ thm : nregexcluded ] .
[ thm : nregsuper ] the only superfluous subsets of @xmath54 are the subsets of @xmath55 and @xmath56 .
the only @xmath5-connected matroids in @xmath57 are isomorphic to @xmath28 , @xmath58 , @xmath59 , @xmath60 , or @xmath61 .
we note here that all undefined matroids appearing in the paper can be found in the appendix of oxley @xcite .
we assume that the reader is familiar with the terminology and notation from that source .
we use the terms _ line _ and _ plane _ to refer to rank-@xmath62 and rank-@xmath5 subsets of the ground set . by performing a @xmath63 exchange on @xmath58 ,
we obtain @xmath64 , which is represented over @xmath22 by @xmath65 $ ] , where @xmath66 is the following matrix .
@xmath67 the paper is built up as follows . in section [ sec : splitters ] we use seymour s splitter theorem to prove that certain subsets are superfluous . to prove that a subset @xmath68 is not superfluous , we need to generate an infinite number of @xmath5-connected matroids in @xmath69 .
we do so by the simple expedient of growing arbitrarily long fans .
section [ sec : growfans ] proves the technical lemmas that allow us to do so . in section [ sec : infinites ]
we introduce several matroids to which our method of growing fans will be applied , and in section [ sec : final ] we will round up the results .
note that the proofs in sections [ sec : splitters ] and [ sec : infinites ] are finite case - checks that could be replaced by computer checks .
however , at the moment of writing no sufficiently reliable software for this existed .
the following result is very well - known ( * ? ? ?
* proposition 12.2.3 ) . [ prop : f7splitter2 ] the matroid @xmath28 is a splitter for the class @xmath29 . our next result , which seems not to be in the literature , proves a generalization of proposition [ prop : f7splitter2 ] . [ thm : f7splitter ] the matroid @xmath70 is a splitter for the class @xmath71 . by seymour s splitter theorem
we only have to check that @xmath70 has no @xmath5-connected single - element extensions and coextensions in @xmath71 . if @xmath7 is a @xmath5-connected matroid such that @xmath72 , then either @xmath73 is on exactly one line of @xmath70 , or @xmath73 is on no line of @xmath70 . in either case
@xmath74 contains a @xmath75-minor .
therefore we will assume that @xmath7 is a @xmath5-connected matroid such that @xmath76 and @xmath7 belongs to @xmath71 .
let @xmath77 be the class of matroids that are either binary or ternary .
now @xmath77 is a minor - closed class , and its excluded minors are characterized in @xcite .
certainly @xmath7 is not binary , since that would lead to a contradiction to proposition [ prop : f7splitter2 ] .
moreover , @xmath7 is not ternary , as it contains an @xmath28-minor .
therefore @xmath7 is not contained in @xmath77 .
hence ( * ? ? ?
* theorem 4.1 ) implies that @xmath7 contains a @xmath5-connected excluded minor for @xmath77 .
there are only @xmath78 such excluded minors , and as @xmath7 does not contain @xmath75 or @xmath79 as a minor , @xmath7 must contain one of the matroids obtained from the affine geometry @xmath80 or from @xmath81 by relaxing a circuit - hyperplane . as @xmath7 contains only @xmath82 elements
, @xmath7 must be isomorphic to the unique relaxation of @xmath80 .
but this matroid has an @xmath83-minor ( ( * ? ? ?
* page 646 ) ) .
this contradiction completes the proof .
we can make short work of the case in which we do not exclude @xmath84 .
geelen et al .
* theorem 1.5 ) proved the following result : if @xmath7 is a @xmath5-connected matroid in @xmath85 , and @xmath7 has a @xmath84-minor , then @xmath7 is a minor of @xmath39 .
since each of @xmath86 , @xmath87 , @xmath88 has a minor in @xmath89 , we immediately have [ cor : p8splitter ] if @xmath7 is a @xmath5-connected matroid in @xmath90 , and @xmath7 has a @xmath84-minor , then @xmath7 is a minor of @xmath39 .
next , we determine what happens if we do nt exclude @xmath58 .
our starting point is the automorphism group of @xmath58 .
note that it is transitive on elements of the ground set ( ( * ? ? ?
* page 653 ) ) . for each element @xmath91 in @xmath58
, there is a unique element @xmath92 such that @xmath91 and @xmath92 are not on a @xmath5-point line of @xmath58 .
any automorphism will map @xmath93 to another such pair , so specifying the image of @xmath91 also specifies the image of @xmath92 .
consider automorphisms of the diagram in figure [ fig : agdefig ] that pointwise fix @xmath94 and @xmath82 .
it is easy to confirm that the permutations below ( presented in cyclic notation ) , @xmath95 and @xmath96 are two such automorphisms .
the next result follows easily from this discussion . .
] [ lem : transitivefixed ] let @xmath91 and @xmath92 be points in @xmath58 such that there is no @xmath5-point line containing @xmath91 and @xmath92 .
the subgroup of the automorphism group of @xmath58 that pointwise fixes @xmath91 and @xmath92 is transitive on @xmath97 .
[ lem : transitivewhirlbasis ] let @xmath98 and @xmath99 be bases of @xmath58 such that every pair @xmath100 , and every pair @xmath101 spans a @xmath5-point line .
there is an automorphism of @xmath58 mapping @xmath98 to @xmath99 .
if @xmath102 is any element of @xmath58 , then let @xmath103 be the point that is in no @xmath5-point line with @xmath102 .
let @xmath104 .
the hypotheses of the lemma imply that @xmath105 .
let @xmath106 be the unique point such that @xmath107 is a circuit .
define @xmath108 and @xmath109 in the same way . then @xmath110 .
as @xmath58 contains only @xmath82 points , we can relabel as necessary , and assume @xmath109 is in @xmath111 . since @xmath109 is in a non - trivial line with @xmath112 and @xmath113 , it follows that @xmath114 , so that @xmath115 is a circuit .
let @xmath116 . by relabeling and using the same arguments
, we can assume that @xmath117 is a @xmath5-point line of @xmath58 .
consider the automorphism that maps @xmath118 to @xmath91 .
it must map @xmath119 to @xmath92 . by composing this automorphism with an automorphism that fixes @xmath91 and @xmath92 , and referring to lemma [ lem : transitivefixed ]
, we can assume that @xmath120 is mapped to @xmath112 .
but an automorphism maps lines to lines , so then @xmath121 must be mapped to @xmath113 , and the result follows . in the proof of the next lemma
we will show several times that a matroid @xmath122 $ ] is isomorphic to one of @xmath64 , @xmath84 , @xmath123 , or @xmath124 . unless the isomorphism is obvious ( i.e. one merely needs to permute rows and columns )
, we will specify which isomorphism we use .
for this we use the representation of @xmath64 with elements labeled as in .
moreover , we will label the elements of @xmath84 , @xmath123 , @xmath124 so that @xmath125 $ ] , @xmath126 $ ] , and @xmath127 $ ] , where @xmath128 and @xmath129 are the following matrices over @xmath22 .
@xmath130 [ lem : ag23-del - e ] let @xmath7 be a @xmath5-connected @xmath44-representable matroid such that @xmath131 for some @xmath132 .
then @xmath7 has @xmath133 as minor .
suppose that @xmath7 is a counterexample .
let @xmath134 .
there exists a set @xmath135 such that @xmath136 and @xmath137 .
suppose @xmath138 has no 5-point planes .
first we show that @xmath138 has no @xmath5-point lines .
observe that each line of @xmath138 is a line of @xmath58 , so @xmath138 has no @xmath78-point lines .
suppose @xmath139 is a line of @xmath138 .
if @xmath102 is on another 3-point line , then the union of those lines would be a 5-point plane , a contradiction .
it follows that @xmath140 is simple .
furthermore , @xmath141 is in no @xmath5-point line in @xmath140 , or else the union of this line with @xmath142 is a @xmath143-point plane in @xmath138 .
therefore @xmath144 is simple , has rank @xmath62 , and contains @xmath143 points .
therefore @xmath138 has a @xmath75-minor , which is impossible since it is @xmath44-representable . hence @xmath138 has no @xmath5-point lines .
let @xmath73 be an arbitrary point in @xmath145 .
then @xmath146 is a simple rank-@xmath5 matroid with @xmath147 points .
since @xmath138 has no 5-point planes , @xmath146 has no 4-point lines .
hence @xmath146 can not be the union of two lines , so it is @xmath5-connected .
then @xmath146 is isomorphic to one of the matroids @xmath28 , @xmath148 , @xmath149 , or @xmath150 ( see ( * ? ? ?
* page 292 ) ) . since @xmath146 is @xmath44-representable , it is not isomorphic to @xmath28 or @xmath148 .
furthermore , @xmath150 contains a @xmath78-point line , so @xmath146 must be isomorphic to @xmath149 . by the uniqueness of representation over @xmath22
, we can assume that the following @xmath22-matrix @xmath151 is such that @xmath152 $ ] .
@xmath153 as @xmath138 has no @xmath5-point lines , all of @xmath154 , @xmath155 , and @xmath156 are non - zero . by scaling the row labeled @xmath73
, we assume that @xmath157 . if @xmath158 then @xmath159 is a triangle .
it follows that @xmath160 . if @xmath161 , then @xmath162 , or else @xmath163 . therefore @xmath164 . if @xmath165 , then @xmath151 represents @xmath166 , which is impossible as @xmath7 is @xmath23-representable
. therefore @xmath167 . by the discussion above , @xmath168 .
but in @xmath169 , the sets @xmath170 , @xmath171 , and @xmath172 are triangles containing @xmath73 , whereas @xmath173 , and @xmath174 are triangles containing @xmath143 .
this is a contradiction , since @xmath175 has only one element that is on three lines .
therefore @xmath176 .
it follows that @xmath177 , or else @xmath178 is a triangle of @xmath138 .
assume that @xmath164 , from which it follows that @xmath179 .
then we find that @xmath180 , with isomorphism @xmath181 therefore we must have @xmath182 , and hence @xmath183 .
but then again @xmath180 , with isomorphism @xmath184 from this final contradiction we conclude that the claim holds .
let @xmath185 be a set of 5 points of a plane of @xmath138 , and @xmath186 .
note that @xmath187 , as @xmath188 contains no rank-2 flat with 5 elements . since @xmath188 is isomorphic to @xmath58 , we can distinguish three cases .
either @xmath189 is a 3-point line of @xmath188 , or @xmath189 is a basis of @xmath188 , and every pair of elements of @xmath189 spans a @xmath5-point line in @xmath188 , or @xmath189 is a basis of @xmath188 , and there is exactly one pair of elements in @xmath189 that does not span a @xmath5-point line of @xmath188 .
we can use lemmas [ lem : transitivefixed ] and [ lem : transitivewhirlbasis ] , and the fact that the automorphism group of @xmath58 is transitive on @xmath5-point lines ( ( * ? ? ?
* page 653 ) ) , and thereby assume that either @xmath190 or @xmath191 or @xmath192 , where the elements of @xmath193 are labeled as in figure [ fig : agdefig ] .
[ [ case - i ] ] case i + + + + + + suppose @xmath190 , so that @xmath194 . since @xmath195 is not a coloop and not in a series pair , there are two elements in @xmath189 that are not spanned by @xmath185 in @xmath138 .
let @xmath196 be the automorphism in equation , so that @xmath189 is an orbit of @xmath196 .
there is some @xmath197 such that @xmath198 takes the two elements in @xmath199 to @xmath200 .
now @xmath198 induces a relabeling of the elements of @xmath138 that set - wise fixes @xmath185 . after applying this relabeling
, @xmath188 is still equal to @xmath58 , as labeled in figure [ fig : agdefig ] .
moreover , @xmath185 is a @xmath143-point plane of @xmath138 that does not span @xmath78 or @xmath201 . by the uniqueness of representations over @xmath22
we can assume that @xmath122 $ ] for some @xmath22-matrix of the form @xmath202 with @xmath203 .
if @xmath204 then @xmath205 , with isomorphism @xmath206 hence @xmath207 .
but now @xmath208 .
this completes the analysis in case i. from now on , we assume that @xmath189 is not a triangle of @xmath209
. we will also assume that if @xmath185 spans an element @xmath210 , then there is no triangle @xmath211 of @xmath209 that contains @xmath212 . to justify this assumption , note that if @xmath213 , then @xmath214 must be a triad of @xmath7 , so that @xmath215 .
furthermore , @xmath189 is not a triangle in @xmath209 , so @xmath211 contains exactly one element of @xmath185 .
therefore , if @xmath211 exists , we can replace @xmath185 with @xmath216 , and replace @xmath189 with @xmath211 , and reduce to case i. [ [ case - ii ] ] case ii + + + + + + + suppose @xmath191 . since any pair of elements from @xmath217 is in a triangle of @xmath209 , we can assume that @xmath185 spans no element of @xmath189 , by the argument in the previous paragraph
. hence we have @xmath122 $ ] for some @xmath22-matrix of the form @xmath218 where @xmath154 and @xmath155 are non - zero . if @xmath219 , then @xmath220 , with isomorphism @xmath221 if @xmath222 , then @xmath223 , with isomorphism @xmath224 contradicting @xmath23-representability of @xmath7 . if @xmath225 , then @xmath226 , with isomorphism @xmath227 if @xmath228 , then @xmath220 , with isomorphism @xmath229 thus @xmath7 has a @xmath64-minor .
[ [ case - iii ] ] case iii + + + + + + + + suppose @xmath192 .
since @xmath230 and @xmath231 are triangles of @xmath209 , we assume that neither @xmath78 nor @xmath143 is in the span of @xmath185 , by the argument immediately preceding case ii . hence @xmath122 $ ] for some @xmath22-matrix of the form @xmath232 where @xmath233 . if @xmath204 then @xmath234 , with isomorphism @xmath235 therefore @xmath236 .
but now @xmath237 , with isomorphism @xmath238 the result follows .
we must now study coextensions of @xmath60 .
luckily our previous analysis can be used for this .
[ lem : delelemag23 ] let @xmath7 be a @xmath5-connected @xmath44-representable matroid such that @xmath239 for some @xmath132 .
then @xmath7 has an element @xmath240 such that @xmath241 is @xmath5-connected .
let @xmath7 be as stated , and suppose the result is false , so for each element @xmath242 , @xmath241 is not 3-connected .
since @xmath243 is 3-connected , @xmath244 must be in a triad with @xmath195 .
two distinct triads @xmath245 and @xmath246 , both containing @xmath195 , intersect only in @xmath195 , or else @xmath247 contains a triad . from this
we find that @xmath248 can be partitioned into series pairs .
however , this matroid has an odd number of elements , a contradiction .
[ cor : agde ] let @xmath7 be a @xmath5-connected @xmath44-representable matroid such that @xmath239 for some @xmath132 .
then @xmath7 has @xmath64 as minor .
let @xmath244 be an element as in lemma [ lem : delelemag23 ] .
then @xmath241 is a matroid satisfying all conditions of lemma [ lem : ag23-del - e ] , and the result follows .
now we combine the previous results and the splitter theorem to prove the following theorem .
[ thm : agde ] let @xmath7 be a @xmath5-connected matroid in @xmath249 then either @xmath7 is near - regular , or one of @xmath7 and @xmath250 is isomorphic to a member of @xmath251 . by the excluded - minor characterization of @xmath44-representable matroids ( theorem [ thm : sruexcluded ] ) , it follows that @xmath7 is @xmath44-representable .
we assume that @xmath7 is not @xmath51-representable .
then theorem [ thm : nregexcluded ] implies that @xmath7 contains a minor isomorphic to @xmath58 or its dual . by duality
, we assume that @xmath7 has an @xmath58-minor . if @xmath252 , we are done , so we assume otherwise . by seymour
s splitter theorem , @xmath7 contains a @xmath5-connected minor @xmath138 , such that @xmath138 is a single - element extension or coextension of @xmath58 .
lemma [ lem : ag23-del - e ] implies that @xmath138 is a single - element extension of @xmath58 .
thus @xmath138 is simple and @xmath253 .
moreover @xmath254 , so ( * ? ? ?
* theorem 2.1 ) implies that @xmath255 . if @xmath256 , we are done , so we assume that @xmath7 has a @xmath5-connected minor @xmath257 , such that @xmath257 is a single - element extension or coextension of @xmath60 .
but @xmath258 , or else we have contradicted ( * ? ? ?
* theorem 2.1 ) .
therefore @xmath259 , for some element @xmath195 .
corollary [ cor : agde ] implies that @xmath257 has a @xmath64-minor , a contradiction .
in this section we prove two results that allow us to replace a fan by a bigger fan while keeping a certain minor @xmath260 , without losing 3-connectivity , and without introducing an undesired minor @xmath261 ( subject to the conditions that @xmath261 is 3-connected and has no 4-element fans ) .
we will use brylawski s generalized parallel connection @xcite for this .
we refer the reader to oxley ( * ? ? ?
* section 11.4 ) for definitions and elementary properties , including the following : [ lem : genparconflats ] let @xmath7 and @xmath260 be matroids having a common restriction @xmath211 , which is moreover a modular flat of @xmath260 .
let @xmath262 . 1 .
a subset @xmath263 is a flat of @xmath138 if and only if @xmath264 is a flat of @xmath260 and @xmath265 is a flat of @xmath7 ; 2 .
@xmath266 and @xmath267 ; 3 .
if @xmath268 then @xmath269 ; 4 . if @xmath270 then @xmath271 ; 5 . if @xmath272 then @xmath273 ; 6 . if @xmath274 then @xmath275 .
let @xmath7 be a matroid on the ground set @xmath276 .
a subset of @xmath276 is _ fully - closed _ if it is closed in @xmath7 and @xmath250 .
if @xmath277 , then @xmath278 is the intersection of all fully - closed sets that contain @xmath185 .
we can obtain @xmath278 by applying the closure operator to @xmath185 , applying the coclosure operator to the result , and so on , until we cease to gain any new elements .
[ lem : simplecosimple2sep ] let @xmath7 be a simple , cosimple , connected matroid , and let @xmath279 be a @xmath62-separation of @xmath7 . then @xmath280 is a @xmath62-separation .
it is simple to verify that @xmath281 . if @xmath280 is not a @xmath62-separation , then @xmath282 .
this means that we can order the elements of @xmath98 as @xmath283 , so that @xmath284 is in @xmath285 or @xmath286 , for all @xmath287 .
hence @xmath288 , so @xmath289 .
thus @xmath290 is either dependent or codependent . in either case
we have a contradiction .
[ def : fan ] let @xmath7 be a matroid , and @xmath291 an ordered subset of @xmath292 , with @xmath293 .
we say @xmath294 is a _ fan _ of @xmath7 if , for all @xmath295 , @xmath296 is either a triangle or a triad , and if @xmath297 is a triad , then @xmath298 is a triangle ; if @xmath297 is a triangle then @xmath298 is a triad .
assume that @xmath299 is a fan .
then @xmath294 is a fan of @xmath300 .
we say that @xmath294 is a _ maximal _
fan if there is no fan @xmath301 such that @xmath302 and @xmath303 .
we say @xmath304 is a _ rim element _ if @xmath305 and @xmath306 is contained in exactly one triangle that is contained in @xmath307 , or if @xmath308 and @xmath306 is contained in no such triangle .
we say @xmath304 is a _ spoke element _ if it is not a rim element .
[ lem : fan2sep ] let @xmath7 be a simple , cosimple , connected matroid , let @xmath309 be a fan of @xmath7 , and let @xmath279 be a @xmath62-separation of @xmath7
. then @xmath7 has a @xmath62-separation @xmath310 with @xmath311 .
let @xmath7 , @xmath294 , and @xmath279 be as stated .
assume , by dualizing @xmath7 if necessary , that @xmath312 is a triangle .
clearly one of @xmath313 and @xmath314 has size at least two ; by relabeling assume @xmath315 . by lemma [ lem : simplecosimple2sep ]
, we can replace @xmath279 with @xmath280 .
if @xmath316 then we are done , so assume that @xmath317 is the smallest index such that @xmath318 .
certainly @xmath211 is contained in @xmath319 , so @xmath320 .
but then either @xmath321 or @xmath322 , which leads to a contradiction . in
what follows , the elements of the wheel @xmath323 and whirl @xmath324 are labeled @xmath325 where , for all indices @xmath326 ( interpreted modulo @xmath327 ) , @xmath328 is a triangle and @xmath329 is a triad .
hence , @xmath330 is the set of spokes and @xmath331 is the set of rim elements .
[ thm : growfan ] let @xmath7 be a @xmath5-connected matroid , and let @xmath332 be a fan of @xmath7 with @xmath312 a triangle .
let @xmath333 be an integer , and relabel the elements @xmath334 , @xmath335 , @xmath336 of @xmath323 by @xmath337 , @xmath338 , @xmath339 , in that order .
let @xmath340 , and @xmath341 .
then @xmath257 has the following properties : 1 .
[ eq : growfan1]@xmath342 is a fan of @xmath257 ; 2 .
[ eq : growfan2 ] @xmath7 is isomorphic to a minor of @xmath257 , with the isomorphism fixing all elements but @xmath338 ; 3 .
[ eq : growfan3 ] @xmath257 is @xmath5-connected ; let @xmath7 , @xmath294 , @xmath211 , @xmath327 , @xmath138 , and @xmath257 be as stated , and define @xmath343 .
it follows immediately from ( * ? ? ?
* corollary 6.9.10 ) that @xmath211 is a modular flat of @xmath260 , so @xmath344 is defined .
it follows from lemma [ lem : genparconflats ] that @xmath345 is a fan of @xmath138 and of @xmath257 , since the complement in @xmath138 of each triad of @xmath260 is a hyperplane of @xmath138 , and each triangle of @xmath260 other than @xmath211 is a triangle of @xmath138 . if @xmath346 , then holds .
hence we assume that @xmath347 .
we only need to show that @xmath348 is a triad of @xmath257 .
consider @xmath349 .
since @xmath350 and @xmath351 are hyperplanes of their respective matroids , @xmath352 is a flat of @xmath138 . since @xmath353
, it follows that @xmath354 is a cocircuit of @xmath138 .
but then @xmath348 is a cocircuit of @xmath257 , as desired .
statement is a straightforward consequence of lemma [ lem : genparconflats ] and the observation that @xmath323 has a minor in which @xmath355 is a triangle and some element is in parallel with @xmath335 .
statement follows immediately from ( * ? ? ?
* corollary 2.8 ) .
we will denote the matroid @xmath257 , as described in the statement of theorem [ thm : growfan ] , by @xmath356 .
theorem [ thm : growfan ] shows that we can make a fan arbitrarily long while keeping 3-connectivity .
our next task is to show that we can do so without introducing certain minors .
the following lemma , whose elementary proof we omit , will be useful : [ lem:4eltfan : no ]
let @xmath260 be a @xmath5-connected matroid without @xmath78-element fans .
let @xmath7 be a @xmath5-connected matroid having @xmath260 as minor , and let @xmath294 be a @xmath78-element fan of @xmath7 .
then @xmath357 .
recall that if @xmath211 is a coindependent triangle of the matroid @xmath7 , then @xmath358 is the matroid obtained from @xmath7 by a @xmath63 exchange ( see ( * ? ? ?
* section 11.5 ) ) .
[ thm : nofanfan ] let @xmath260 be a @xmath5-connected matroid with no @xmath78-element fan .
let @xmath7 be a @xmath5-connected matroid with at least @xmath143 elements that does not have an @xmath260-minor .
let @xmath359 be a fan of @xmath7 , where @xmath360 is a triangle , and let @xmath361 be an integer . if @xmath356 has an @xmath260-minor , then so does @xmath358 .
we will assume that @xmath361 has been chosen so that it is as small as possible , subject to the condition that @xmath356 has an @xmath260-minor .
let @xmath261 be a minor of @xmath356 that is isomorphic to @xmath260 .
first assume that @xmath362 .
since @xmath363 is a 4-element fan of @xmath356 , it follows from lemma [ lem:4eltfan : no ] that this set is not contained in @xmath364 .
we claim that @xmath365 has an @xmath260-minor .
assume this is not the case .
if @xmath366 has an @xmath260-minor , then , as @xmath367 is a parallel pair , @xmath365 has an @xmath260-minor .
therefore @xmath366 does not have an @xmath260-minor .
similarly , @xmath368 is a series pair in @xmath369 , so we assume that @xmath369 has no @xmath260-minor .
as @xmath370 is a parallel pair in @xmath371 , this means that @xmath371 has no @xmath260-minor .
moreover , @xmath372 is a series pair in @xmath373 , so this matroid does not have an @xmath260-minor . as @xmath374 is a series pair in @xmath375 , and we concluded that @xmath371 has no @xmath260-minor , neither does @xmath375 . since @xmath376 is a parallel pair in @xmath377 , and deleting @xmath378 destroys all @xmath260-minors , @xmath377 has no @xmath260-minor .
deleting @xmath379 creates the series pair @xmath380 , and contracting @xmath378 destroys all @xmath260-minors , so @xmath381 does not have an @xmath260-minor .
lastly , contracting @xmath382 creates the parallel pair @xmath374 , so @xmath383 does not have an @xmath260-minor , or else @xmath369 does . from this discussion , we conclude that @xmath384 , contradicting our earlier conclusion .
therefore @xmath365 has an @xmath260-minor .
since contracting @xmath378 and deleting @xmath385 from @xmath323 produces a copy of @xmath386 , it follows easily from lemma [ lem : genparconflats ] that @xmath365 is isomorphic to @xmath387 .
thus our assumption on the minimality of @xmath327 is contradicted .
now we must assume that @xmath388 .
if @xmath389 , then as this set is a triad in @xmath356 , obtaining @xmath261 involves contracting one of @xmath390 .
contracting any of these elements in @xmath391 produces a matroid consisting of the triangle @xmath392 with parallel points added to two distinct elements .
now we can use lemma [ lem : genparconflats ] to show that contracting an element in @xmath390 from @xmath356 produces a matroid that is isomorphic to @xmath7 or @xmath393 , up to the addition of parallel elements .
therefore @xmath7 has an @xmath260-minor , contrary to hypothesis .
it follows that @xmath394 .
since @xmath395 is a triangle in @xmath356 , we deduce that @xmath396 , or else @xmath397 is a @xmath78-element fan in @xmath261 .
since @xmath380 is a parallel pair in @xmath398 , and @xmath261 contains no parallel pairs , @xmath261 is a minor of @xmath399 . as @xmath400 is also a triangle in @xmath356
, we can use exactly the same arguments to show that @xmath261 is a minor of @xmath373 .
so @xmath261 is a minor of @xmath401 .
since @xmath402 , it is easy to prove that any triangle of @xmath7 is coindependent ( ( * ? ? ?
* lemma 8.7.5 ) ) . therefore @xmath401 is isomorphic to @xmath358 , and we are done .
in this section we describe a collection of matroids to which we can apply our operation of growing fans . recall that @xmath36 , @xmath46 , and @xmath54 , respectively , denote the sets of excluded minors for @xmath23-representable , @xmath44-representable , and @xmath51-representable matroids , as listed in theorems [ thm : gf4excluded ] , [ thm : sruexcluded ] , and [ thm : nregexcluded ] .
let @xmath403 be the rank-@xmath5 matroid shown in figure [ fig : m9 ] .
then @xmath403 is represented over @xmath22 by @xmath404 $ ] , where @xmath66 is the following matrix .
@xmath405 .
] [ lem : m9 ] let @xmath211 be the triangle @xmath406 of @xmath403 . if @xmath361 is an integer , then @xmath407 is @xmath5-connected , and contains an @xmath148-minor but no minor in @xmath408 .
clearly @xmath403 is @xmath5-connected , and @xmath409 is a fan , so we can apply theorem [ thm : growfan ] .
thus @xmath407 is @xmath5-connected , by statement .
since @xmath410 is isomorphic to @xmath148 , it follows from statement that @xmath407 has an @xmath148-minor for any @xmath361 .
now assume that @xmath407 has a minor in @xmath408 .
therefore either @xmath403 or @xmath411 has such a minor , by theorem [ thm : nofanfan ] .
since @xmath403 and @xmath411 are both ternary ( ( * ? ?
* lemma 11.5.13 ) ) , neither has a minor isomorphic to @xmath86 , @xmath87 , @xmath88 , @xmath38 , @xmath75 , @xmath79 , @xmath70 , or @xmath412 . as @xmath413 , and @xmath414 ,
neither contains @xmath415 .
since @xmath413 , and @xmath411 contains the triangle @xmath174 and has @xmath82 elements , neither contains @xmath166 . as @xmath411 has rank @xmath78 , and @xmath82 elements , it does not contain @xmath58 . as @xmath403 has @xmath82 elements and a @xmath78-point line
, it does not contain @xmath58 .
similarly , @xmath403 has rank @xmath5 , so it does not contain @xmath64 .
also @xmath411 has two triangles , @xmath174 and @xmath416 , so it does not contain @xmath64 either .
the only matroid left to check is @xmath124 .
obviously @xmath403 does not contain an @xmath124-minor .
assume that @xmath411 does .
as @xmath124 has no triangles , @xmath417 must be isomorphic to @xmath124 .
now @xmath406 is a triad of this matroid , and performing a @xmath418 exchange on this triad should produce a copy of @xmath148 .
instead it produces a copy of @xmath419 , which contains disjoint triangles , and is therefore not isomorphic to @xmath148 .
let @xmath420 be the matroid represented by @xmath65 $ ] over @xmath22 , where @xmath66 is the following matrix .
@xmath421 then @xmath420 is represented by the geometric diagram in figure [ fig : m11 ] . . ]
[ lem : m11 ] let @xmath211 be the triangle @xmath422 of @xmath420 . if @xmath361 is an integer , then @xmath423 is @xmath5-connected , and contains an @xmath64-minor , but no minor in @xmath424 .
it is clear that we can apply theorem [ thm : growfan ] .
thus @xmath423 is @xmath5-connected .
since @xmath425 is isomorphic to @xmath64 , it follows that @xmath423 has a @xmath64-minor for any @xmath361 .
if the lemma is false , then by theorem [ thm : nofanfan ] , either @xmath420 or @xmath426 contains as a minor a ternary member of @xmath424 , which is to say , one of @xmath148 , @xmath124 , @xmath166 , @xmath58 , or @xmath59 .
we start by noting that in @xmath427 , the sets @xmath428 and @xmath429 are 4-point lines .
therefore any @xmath147-element restriction of @xmath430 has either a @xmath78-point line or two disjoint triangles .
it follows that @xmath430 has no minor in @xmath54 .
as @xmath431 and @xmath432 are @xmath78-point lines of @xmath433 , we can also see that @xmath433 has no minor in @xmath54 .
the triangles of @xmath420 are @xmath434 , @xmath422 , and @xmath435 .
it follows easily that every @xmath82-element restriction of @xmath420 contains at least one triangle , so @xmath436 does not have @xmath84 as minor .
the rank of @xmath436 is too low to have @xmath415 as minor .
if @xmath436 has @xmath58 as minor , then this minor must be obtained by a single contraction .
since @xmath58 is simple , we can not contract an element from a 3-point line .
this leaves only elements @xmath147 and @xmath82 , and we have already decided that contracting either of these does not produce a minor in @xmath54 .
suppose @xmath420 has a @xmath124-minor . to obtain this minor we must delete two elements in such a way that no triangles remain .
since deleting @xmath437 gives us @xmath64 again , which has no @xmath124-minor , we must delete @xmath5 and one of @xmath438 .
but @xmath439 has disjoint triads @xmath440 and @xmath441 , whereas @xmath442 has disjoint triads @xmath443 and @xmath444 .
hence neither is isomorphic to @xmath124 .
therefore we assume that @xmath420 has an @xmath148-minor .
we must contract a single element from @xmath420 , and then delete a single element to obtain a copy of @xmath148 .
if we contract either @xmath5 or @xmath437 , then we produce two disjoint parallel pairs , which can not be rectified with a single deletion .
if we contract one of @xmath94 , @xmath62 , @xmath78 , or @xmath201 then we create a single parallel pair , so up to isomorphism we must delete , respectively , @xmath62 , @xmath94 , @xmath201 , or @xmath78 to obtain a copy of @xmath148 .
but in these minors , the triangle @xmath422 is disjoint from , respectively , the triangles @xmath445 , @xmath217 , @xmath446 , and @xmath443 .
therefore we do not contract @xmath94 , @xmath62 , @xmath5 , @xmath78 , @xmath201 , @xmath147 , @xmath82 , or @xmath437 .
if we contract @xmath143 , then up to isomorphism we must delete @xmath5 to obtain a copy of @xmath148 , but in this minor @xmath447 and @xmath448 are disjoint triangles .
thus @xmath420 does not contain a minor in @xmath424 .
assume that @xmath426 contains a minor @xmath261 that is isomorphic to a ternary member of @xmath424 .
if @xmath449 , then an element @xmath450 is contracted to obtain @xmath261 . but
@xmath451 , by ( * ? ? ?
* lemma 2.13 ) , so @xmath261 is isomorphic to a minor of @xmath420 .
since this contradicts the conclusion of the previous paragraph , it follows that @xmath211 is a triad of @xmath261 .
therefore @xmath261 is isomorphic to @xmath124 , or @xmath59 .
it follows easily from ( * ? ? ?
* corollary 2.17 ) and seymour s splitter theorem , that @xmath452 is a minor of @xmath453 .
if @xmath454 , then @xmath455 , and this leads to a contradiction . therefore @xmath456 .
the definition of @xmath418 exchange implies that @xmath457 .
but @xmath64 is a self - dual matroid , so @xmath420 has a minor isomorphic to @xmath64 that contains @xmath422 in its ground set . to obtain this minor
, we must delete a single element , but in each case the result has two triangles , namely @xmath422 and at least one of @xmath434 and @xmath435 .
this is a contradiction as @xmath58 has only one triangle . for a third infinite class , consider the following matrix , @xmath66 , over @xmath458 . here
@xmath154 is an element that satisfies @xmath459 .
let @xmath460 be @xmath404 $ ] .
a geometric representation of @xmath460 can be found in figure [ fig : m8 ] .
@xmath461 . ]
[ lem : m8 ] let @xmath211 be the triangle @xmath416 of @xmath460 .
if @xmath361 is an integer , then @xmath462 is @xmath5-connected , and contains a @xmath463-minor , but no minor in @xmath464 . once again , the only way the lemma can fail is if @xmath460 or @xmath465 contains a minor in @xmath464 .
( note that @xmath466 . ) as @xmath460 and @xmath465 have only @xmath147 elements , any such minor must be isomorphic to @xmath86 , @xmath87 , @xmath148 , or @xmath124 .
but @xmath460 and @xmath465 are @xmath458-representable , and therefore contain no @xmath148-minors or @xmath124-minors .
obviously @xmath460 does not contain @xmath87 .
it does not contain @xmath86 , as every element is on at least one triangle .
if @xmath465 contains a @xmath86 minor , then we must contract an element from @xmath211 to obtain this minor , as @xmath86 has no triads .
but contracting an element from @xmath211 produces a matroid isomorphic to a minor of @xmath460 , implying that @xmath460 has a @xmath86-minor .
assume @xmath465 has a @xmath87-minor .
by the previous argument , @xmath211 must be a triad of this minor .
then @xmath467 has a minor isomorphic to @xmath468 that contains @xmath211 .
but it is easy to see that the only way to obtain a @xmath463-minor from @xmath460 is to delete the element on three triangles , namely @xmath94 .
as @xmath469 , this is a contradiction .
if @xmath470 is @xmath5-connected , then @xmath7 has an @xmath28-minor , and proposition [ prop : f7splitter2 ] implies that @xmath7 is isomorphic to @xmath28 .
therefore @xmath30 is certainly superfluous .
as @xmath471 consists of the duals of the matroids in @xmath32 , it follows that @xmath31 is also superfluous . since @xmath472 contains all non - binary rank-@xmath62 uniform matroids , @xmath473
is contained in no superfluous subset .
similarly , @xmath474 contains all binary projective geometries . therefore @xmath475 is contained in no superfluous subset .
the result follows .
theorem [ thm : f7splitter ] implies that the only @xmath5-connected matroid in @xmath34 is @xmath28 itself . by duality , @xmath83 is the only @xmath5-connected matroid in @xmath476 .
thus @xmath30 and @xmath31 are superfluous subsets . on the other hand
, @xmath477 contains all the non - ternary rank-@xmath62 uniform matroids , so @xmath478 is not contained in any superfluous subset .
similarly , @xmath479 contains all the non - ternary corank-@xmath62 uniform matroids .
finally , @xmath480 contains all binary projective geometries , so @xmath475 is not superfluous .
theorem [ thm : govw02 ] implies that if @xmath7 is a @xmath5-connected matroid in @xmath41 , then @xmath7 is isomorphic to @xmath42 or a minor of @xmath39 .
thus @xmath40 is superfluous . as @xmath481 contains all rank-@xmath62 uniform matroids with at least @xmath201 elements , @xmath482 , and by duality @xmath483 ,
is not contained in any superfluous subset . by lemma [ lem : m9 ] ,
the set @xmath484 contains all matroids of the form @xmath407 , so @xmath485 , and by duality @xmath486 , is not contained in any superfluous subset .
finally , lemma [ lem : m8 ] shows that @xmath487 contains an infinite number of @xmath5-connected matroids , so @xmath488 is not contained in any superfluous subset .
let @xmath7 be a @xmath5-connected matroid in @xmath49 .
if @xmath7 has an @xmath28-minor , then theorem [ thm : f7splitter ] implies that @xmath489 .
hence we assume that @xmath7 does not have an @xmath28-minor , so that @xmath7 has a @xmath166-minor .
corollary [ cor : p8splitter ] says that @xmath7 is a minor of @xmath39 .
therefore @xmath47 is superfluous .
duality implies that the only @xmath5-connected matroids in @xmath490 are @xmath83 , and minors of @xmath491 , so @xmath48 is superfluous .
however , @xmath492 contains infinitely many uniform matroids , and @xmath493 contains all matroids of the form @xmath407 .
duality implies that none of @xmath478 , @xmath494 , @xmath485 , @xmath486 is contained in a superfluous subset .
finally , @xmath495 contains all binary projective geometries , so @xmath475 is contained in no superfluous subset .
let @xmath7 be a @xmath5-connected matroid in @xmath496 if @xmath7 contains an @xmath28-minor , then theorem [ thm : f7splitter ] implies that @xmath489 .
we assume that @xmath7 has no @xmath28-minor .
then theorem [ thm : agde ] says that @xmath7 is isomorphic to @xmath58 , @xmath60 , or the dual of one of these matroids .
therefore @xmath55 is superfluous . by duality
, @xmath56 is superfluous .
as @xmath497 contains infinitely many uniform matroids , and @xmath498 contains all matroids of the form @xmath407 , none of @xmath478 , @xmath494 , @xmath485 , @xmath486 is contained in a superfluous subset .
moreover , @xmath499 contains all matroids of the form @xmath423 , by lemma [ lem : m11 ] .
therefore @xmath500 is contained in no superfluous subset .
again , we observe that @xmath501 contains infinitely many binary matroids , so the proof is complete .
we conclude with the remark that , although our characterizations of @xmath1 are strong when @xmath0 contains all non - superfluous excluded minors in our class , we have made no attempt to characterize the infinite families . clearly some of these families are highly structured .
for instance , it is known that every rank-3 matroid with a @xmath75-minor also has a @xmath79-minor . before writing our proofs we experimented to uncover the truth .
these experiments were done using the macek software by hlien @xcite , and occasionally we queried mayhew and royle s database of small matroids @xcite . c. semple ,
g. whittle , on representable matroids having neither @xmath504- nor @xmath505-minors , in matroid theory ( seattle , wa , 1995 ) , volume 197 of contemp .
math . , pp .
soc . , providence , ri ( 1996 ) .
|
if @xmath0 is a set of matroids , then @xmath1 denotes the set of matroids that have no minor isomorphic to a member of @xmath0 . if @xmath2 , we say that @xmath3 is _ superfluous _
if @xmath4 contains only finitely many @xmath5-connected matroids .
we characterize the superfluous subsets of six well - known collections of excluded minors .
|
shear bands are narrow zones of intense shear observed during the dynamic deformation of many metals at high strain rates .
shear localization forms a striking instance of material instability , often preceding rupture , and its study has attracted considerable attention in the mechanics _ e.g.
_ @xcite , numerical @xcite , or mathematical literature @xcite . in experimental investigations of high strain - rate deformations of steels , observations of shear bands
are typically associated with strain softening response past a critical strain of the measured stress - strain curve @xcite .
it was proposed by zener and hollomon @xcite , and further precised by clifton et al @xcite , that the effect of the deformation speed is twofold : an increase in the deformation speed changes the deformation conditions from isothermal to nearly adiabatic , and the combined effect of thermal softening and strain hardening of metals may produce a net softening response .
on the other hand , strain - rate hardening has an effect _ per se _ , inducing momentum diffusion and playing a stabilizing role .
strain softening has a destabilizing effect ; it is known at the level of simple models ( _ e.g. _ wu and freund @xcite ) to induce _ hadamard instability _ - an ill - posedness of a linearized problem .
it should however be remarked that , while hadamard instability indicates the catastrophic growth of oscillations around a mean state , what is observed in localization is the orderly albeit extremely fast development of coherent structures , the shear bands . despite considerable attention to the problem of localization ,
little is known about the initial formation of shear bands , due to the dominance of nonlinear effects from the early instances of localization .
our aim is to study the onset of localization for a simple model , lying at the core of various theories for shear band formation , @xmath0 which will serve to assess the effects of strain softening @xmath1 and strain - rate sensitivity @xmath2 and analyze the emergence of a shear band out of the competition of hadamard instability and strain - rate hardening .
the model describes shear deformations of a viscoplastic material in the @xmath3-plane , with @xmath4 the velocity in the @xmath5-direction , @xmath6 the plastic shear strain ( elastic effects are neglected ) , and the material is obeying a viscoplastic constitutive law of power law type , @xmath7 the constitutive law can be thought as describing a plastic flow rule on the yield surface .
the model captures the bare essentials of the localization mechanism proposed in @xcite .
early studies of appear in hutchinson and neal @xcite ( in connection to necking ) , wu and freund @xcite ( for linear rate - sensitivity ) and tzavaras @xcite . the uniform shearing solutions , @xmath8 form a universal class of solutions to for any @xmath9 .
when @xmath10 and @xmath11 the system is elliptic in the @xmath12-direction and the initial value problem for the associated linearized equation presents hadamard instability ; nevertheless it admits as a special solution .
rate sensitivity @xmath13 offers a regularizing mechanism , and the associated system belongs to the class of hyperbolic - parabolic systems ( _ e.g. _ @xcite ) .
the linearized stability analysis of has been studied in @xcite for ( even more complicated models including ) - . as is time dependent
, the problem of linearized stability leads to the study of non - autonomous linearized problems .
this was addressed by fressengeas - molinari @xcite and molinari - clifton @xcite who introduced the study of relative perturbations , namely to assess the stability of the ratios of the perturbation relative to the base time - dependent solution , and provided linearized stability results . such linearized stability and instability results compared well with studies of nonlinear stability ( _ e.g _ @xcite or @xcite for a survey ) . here , we restrict to the constitutive function @xmath14 in but retain the dependence in @xmath15 , and study @xmath16 this model has a special and quite appealing property : after considering a transformation to relative perturbations and a rescaling of variables , @xmath17 the problem of stability of the time - dependent uniform shearing solution is transformed into the problem of stability of the equilibrium @xmath18 for the nonlinear but _ autonomous _ parabolic system @xmath19 the following heuristic argument leads to a conjecture regarding the effect of rate sensitivity @xmath15 on the dynamics : as time proceeds the second equation in , which is of relaxation type , relaxes to the equilibrium manifold @xmath20 . accordingly ,
the stability of is determined by the equation describing the effective equation @xmath21 the latter is parabolic for @xmath22 and backward parabolic for @xmath23 , what suggests instability in the range @xmath23 .
the argument is proposed in @xcite in connection to the development of an asymptotic criterion for the quantitative assessment of shear band formation and can be quantified by means of an asymptotic expansion ( see section [ sec : coarse_scale ] ) . in this article
we study the dynamics of the system .
first , we provide a complete analysis of linearized stability . the linearized system around the equilibrium @xmath24 takes the form @xmath25 and is simple to analyze via fourier analysis .
a complete picture emerges : * for @xmath26 , high - frequency modes grow exponentially fast and indicate catastrophic growth and hadamard instability .
* for @xmath27 , the modes still grow and are unstable but at a tame growth rate . * for @xmath28 strain - rate dependence
is strong and stabilizes the motion .
the instability occurring in rate - dependent localization resembles at the linearized level to the _ turing instability _ familiar from problems of morphogenesis @xcite ( _ cf . _
remark [ remarkturing ] ) .
next , we turn to the nonlinear system and proceed to analyze the competition between hadamard instability and strain - rate dependence in the nonlinear regime in the parameter range @xmath27 . exploiting scaling properties of we construct a class of _ focusing _ self - similar solutions of the form @xmath29 for @xmath30 and as opposed with the usual self - similar solutions of diffusion equations the information will propagate on the lines @xmath31 and focus towards the center @xmath32 .
the solution depends on four - parameters : @xmath15 , the growth - rate @xmath30 , and the initial data @xmath33 standing for the sizes of the initial nonuniformities : @xmath34 in the strain and @xmath35 in the strain rate .
the profiles @xmath36 , @xmath37 , solve the singular system and are constructed numerically . for the construction
, it turns out that the parameters need to satisfy the constraints @xmath38 the uniform shear solution corresponds to the choice @xmath39 associated to no growth @xmath40 .
the solution shows that rate - sensitivity suppresses , at the nonlinear level , the oscillations resulting from hadamard instability and that the combined process leads into a single runaway " of concentrated strain that appears like the shear bands observed in experiments .
it is sketched at various time instances in figure [ fig : gamma ] for the strain @xmath6 , in figure [ fig : u ] for the strain rate @xmath41 , in figure [ fig : v ] for the velocity @xmath4 , and in figure [ fig : sigma ] for the stress @xmath42 .
the figure for the stress provides an analytical justification of the phenomenon of stress collapse across the shear band , predicted in theoretical results of @xcite and in numerical computations of wright and walter @xcite .
our analysis validates for the model the onset of localization predicted by the asymptotic analysis in @xcite along the lines of the chapman - enskog expansion of kinetic theory .
the result complements an earlier study of similar behavior for a thermally softening temperature dependent non - newtonian fluid @xcite . in a companion article @xcite
, we provide an existence proof for the heteroclinic orbit ( computed numerically here ) ; this is accomplished even at the level of the plastic flow rule in the parameter range @xmath43 .
( this region is optimal as it is known that for @xmath44 the uniform shear is asymptotically stable @xcite . )
the proof of existence for the heteroclinic employs the geometric singular perturbation theory for dynamical systems and is outside the scope of the present work .
the article is organized as follows : in section [ sec : model ] we introduce the mathematical model along with its basic properties .
the idea of relative perturbations for the stability of the time - dependent uniform shearing solutions is reviewed in section [ sec : relative ] and the complete linearized stability analysis is presented in section [ sec : linear ] .
a dichotomy of stability for the linearized problem appears , depending on the strain - rate sensitivity , in accord with @xcite and results on nonlinear stability @xcite . from section [ sec : coarse_scale ] onwards , we study nonlinear effects .
first , an effective equation is derived via an asymptotic analysis following @xcite that postulates instability in the parameter regime @xmath27 .
the emergence of localization is studied in section [ sec : localization ] : we introduce an ansatz of _ focusing _ self - similar solutions @xmath45 where @xmath46 , @xmath47 and @xmath48 is a parameter .
their existence is based on constructing a solution @xmath49 for the nonlinear system of singular ordinary differential equations @xmath50 such singular systems may ( or may not ) have solutions and this is determined by a case - by - case analysis .
one can remarkably de - singularize and convert the problem of existence of localizing profiles to that of constructing a suitable heteroclinic orbit for a system of ordinary differential equations ( see - ) . using a combination of dynamical systems ideas and numerical computation
, we numerically construct the heteroclinic connection and show that it gives rise to a coherent localizing structure .
the heteroclinic orbit is represented by the red dotted line in figure [ fig:2manifold ] .
the properties of the localizing solutions are summarized in section [ sec : localiz ] .
the simplest model for analyzing the dynamics of shear band formation is the one - dimensional shear deformation of a viscoplastic material that exhibits strain softening and strain - rate sensitivity .
the motion of the specimen occurs in the @xmath5-direction , with shear direction that of the @xmath51-axis , and is described by the _ velocity _
@xmath52 , _ plastic strain _ @xmath53 and _ stress _ @xmath54 .
these field variables satisfy the balance of linear momentum and the kinematic compatibility equation @xmath55 respectively . in the simplest situation ,
the elastic effects are neglected and one focusses on a viscoplastic model , where the stress depends only on the ( plastic ) strain @xmath6 and the strain rate @xmath41 , @xmath56 the material exhibits strain softening when @xmath57 .
a simple constitutive law of that form is @xmath58 where @xmath59 for strain softening , while the strain - rate sensitivity parameter @xmath13 is thought as very small @xmath60 .
* uniform shearing solutions*. the system - with admits a special class of solutions describing uniform shearing , @xmath61 due to the strain - softening assumption @xmath59 , the system - , , with @xmath10 is an elliptic initial - value problem which is ill - posed exhibiting hadamard instability .
nevertheless , both the system with @xmath10 and its regularized version with @xmath13 admit the class of the uniform shearing solutions . our goal is to study the stability of the uniform shearing solutions in both cases @xmath10 and @xmath13 .
for the remainder of this work we focus on the particular choice of @xmath62 , namely @xmath63 the reason for this restriction is the following : the uniform shearing solutions are time - dependent and their analysis ( linearization and nonlinear analysis ) leads very quickly to issues with non - autonomous problems .
the special choice of the constitutive relation has the property that it leads to autonomous problems for its relative perturbation ( see system - ) and appears to be indicative of the general response in the unstable regime
. with the choice system - reads @xmath64 an initial - boundary problem for - is considered in @xmath65 \times \mathbb{r}^+$ ] with the following initial and boundary conditions @xmath66 the boundary condition reflects imposed boundary shear . as a consequence of the boundary conditions we have @xmath67 an equivalent formulation of - and is obtained by considering the strain rate as the primary field variable ( replacing the velocity ) .
introducing the strain rate @xmath68 the system becomes @xmath69 the corresponding initial conditions are @xmath70 note that the boundary conditions on @xmath4 and the compatibility condition imply that @xmath71 the resulting initial - boundary problem consists of - subject to - .
the constraint is inherited from the initial data due to the conservation of @xmath72 .
it is apparent from that there is a diffusion mechanism in the strain - rate , manifesting the effect of strain - rate sensitivity .
motivated by the form of the uniform shearing solutions and @xcite , one may introduce a rescaling of the dependent variables and time in the following form : @xmath73 in we select @xmath74 which normalizes the initial relative perturbation @xmath75 from - we see that the new field variables satisfy , @xmath76 the boundary condition together with , and the equations , , imply the restrictions @xmath77 note that under the transformation the uniform shearing motion is transformed to an equilibrium @xmath78 for the transformed problem - .
despite the fact the solution is time dependent , the transformed problem is still autonomous .
in fact , the latter property is the reason for restricting to the constitutive class , so that spectral analysis can be used to obtain information for the linearized problem . in order to assess the growth or decay of perturbations of the uniform shearing solutions , we linearize the system - around .
let @xmath79 be a small parameter ( describing the size of the perturbation ) and consider the asymptotic expansions @xmath80 by neglecting @xmath81-terms , we obtain from - the linearized system @xmath82 the boundary condition and the constraint imply @xmath83 growth and decay modes of the linearized system - can be captured via spectral analysis .
let us assume the even extension of @xmath84 in @xmath85 $ ] , which is compatible with and consider a cosine series expansion of @xmath86 , @xmath87 in view of , we have also @xmath88 then the coefficients @xmath89 satisfy @xmath90 the characteristic polynomial of the coefficient matrix is @xmath91 with discriminant @xmath92 thus two eigenvalues are real and @xmath93 hence , for @xmath94 , @xmath95 there is always one negative and one positive eigenvalue . the eigenvalues can be computed explicitly @xmath96 the mode @xmath97 needs special attention . in this case
the eigenvalues are @xmath98 and @xmath99 . due to the constraint
, we have @xmath100 thus the zero - th mode decays exponentially to zero .
we summarize the result : * case 1 .
@xmath26 : hadamard instability * the eigenvalues are @xmath101 and satisfy @xmath102 for @xmath103 and @xmath104 for @xmath105 . using the taylor series expansion , @xmath106 , the leading order terms of , are @xmath107 so @xmath108 increases linearly as @xmath109 and is thus unbounded .
this kind of catastrophic instability , called _ hadamard _ instability , is associated with the strain softening behavior and is typical in initial value problems of elliptic equations . * case 2 .
@xmath110 : turing instability * strain - rate dependence provides a diffusive mechanism , which moderates but does not entirely suppress the instability , as can be seen by the following lemma . the behavior in this regime
is that of turing instability [ lem : turing ] for @xmath111 , the eigenvalues are @xmath112 and satisfy the properties a. @xmath108 is increasing in @xmath113 , b. @xmath114 , c. @xmath115 as@xmath109 .
let @xmath116 and write @xmath117 since @xmath118 for @xmath119 , we have @xmath120 and hence @xmath121 which proves @xmath122 .
the eigenvalue @xmath123 satisfies @xmath124 differentiation with respect to @xmath51 gives @xmath125 and the upper bound of @xmath123 proves @xmath126 .
further , the asymptotic expansion @xmath127 yields @xmath128 . [ remarkturing ] ( i ) in view of , positive eigenvalues imply linearized instability and negative linearized stability . invoking that @xmath129 , lemma [ lem : turing ] implies that the rates are of polynomial order and the precise rate of growth or decay can be easily computed .
the upper bound of growth rate indicates that the rate of growth is bounded and the bound is proportional to @xmath130 . by ,
the perturbations exhibit oscillatory response .
\(ii ) the instability arising in the case @xmath131 is a _
turing_-type instability , namely , the combined effect of two different stabilizing mechanisms leads to an instability : note that the coefficient matrix of can be decomposed as @xmath132 taking note of both matrices on the right - hand - side give rise to marginally stable systems for the eigenmodes ; however , the matrix obtained as the sum of the two is equiped with a family of strictly positive eigenvalues .
this decomposition corresponds to visualizing the linearized problem - as the sum of two problems @xmath133 which , subject to the restriction , are both marginally stable .
the stabilizing mechanism in the first one is viscosity , while in the second one are the inertial effects ( stemming from the fact that excess growth is needed to overcome the uniform shear solution ) .
this perspective clarifies the stabilizing mechanisms present in this problem ; still , the prime driver of instability is the softening response of the system that effects the coupling of the two systems .
the analysis in section [ sec : linear ] concerns the behavior of the linearized problem and captures the onset of instability .
focusing next in the nonlinear regime , we devise an asymptotic criterion , in the spirit of @xcite , for the onset of localization .
the goal is to derive an effective equation for the evolutions of @xmath134 and @xmath135 valid at a coarse space and time scale . to this end
, we consider the rescaling of independent variables @xmath136 and rewrite - as @xmath137 for small values of @xmath138 , we can view as a moment equation and equation as a relaxation process towards the equilibrium curve @xmath139 we are interested in calculating the equation describing the effective response of , for @xmath138 sufficiently small .
we consider a chapman - enskog type expansion with @xmath140 for the field variables @xmath141 then we have @xmath142 from and , collecting together the @xmath143-terms , we have @xmath144 this implies that the equation describing the effective dynamics at the order @xmath145 is . since @xmath111 , equation is a backward parabolic equation . on the one hand , this indicates instability , on the other the asymptotic procedure will cease to be a good approximation at this order @xmath145 .
we thus proceed to calculate the effective equation at the order @xmath146 .
collecting together the @xmath147-terms , we obtain @xmath148 hence , @xmath149 we thus obtain an effective equation for @xmath134 which captures the effective response of the system up to order @xmath150 , @xmath151 the leading term in is backward parabolic while the first order correction is a fourth order .
the fourth order term introduces a stabilizing mechanism to the equation , in the sense that the linearized equation around the equilibrium @xmath152 is stable . to see that , write @xmath153 , and compute the linearized equation satisfied by @xmath154
this has the form @xmath155 to study the stability properties of we take the fourier transform and obtain @xmath156 we can see that the right - hand - side becomes negative with high frequency @xmath157 due to the fourth order term .
the fourth order term acts as a stabilizing mechanism .
experimental observations of shear bands indicate that the development of shear localization proceeds in a fast but organized and coherent fashion .
it is not associated with oscillatory response , rather a coherent structure emerges from the process leading to strain localization . the reader should note that the unstable modes of the linearized problem - for @xmath158 contain oscillatory modes ; such individual oscillatory modes are not observed in experiments .
the conjecture is that the combined effect of instability and nonlinearity suppresses the oscillations and results in a runaway " of the strain at a point . in this section
, we will construct a class of self - similar solutions that captures this process for the system - .
the goal is to exploit the invariance properties of the system - in order to construct self - similar focusing solutions that depict the initial stage of localization .
we focus on values of the parameter space @xmath111 and carry out the following steps : * in section [ sec : emergence ] , we consider a variant of - and show that its invariance properties suggest a similarity class of solutions of the form with the profile @xmath159 in satisfying - . * in section [ sec : assds ] , the system - is de - singularized and transformed to a system of three autonomous differential equations - . the profile @xmath159 is transformed to a heteroclinic connection for the orbit @xmath160 of - . * in section [ sec : equilibria ] , we identify the equilibria @xmath161 , @xmath162 , of the system - and calculate their corresponding stable and unstable manifolds @xmath163 and @xmath164 . * in section [ sec : hetero ] , we probe the possibilities for heteroclinic connections between the equilibria , and characterize our target orbit as a connection joining @xmath165 and @xmath166 ( following @xmath167 near @xmath165 ) . *
the heteroclinic connection is computed numerically in section [ sec : numerical_const ] as an orbit between @xmath165 and @xmath166 .
the implications on localizing solutions are summarized in section [ sec : localiz ] .
we return to the system - , restrict henceforth to @xmath111 , and consider it in the entire domain @xmath168 , with the objective to construct a special class of solutions . upon introducing the change of variables ,
we obtain @xmath169 where @xmath170 , and the last system is considered for @xmath171 .
equation may be expressed as @xmath172 the system - is scale invariant .
indeed , given @xmath173 , one verifies that @xmath174 defined by @xmath175 satisfies again - .
note that the time @xmath176 is not rescaled here .
the scaling invariance property motivates to seek for self - similar solutions of - in the form @xmath177 functions of the similarity variable @xmath178 where @xmath179 is specified later .
( the format of is suggested by setting @xmath180 in thus exploiting the property that rescalings of @xmath176 do not enter in the scale invariance property ) . with this form of solutions ,
the problem of localization is transformed to the problem of construction of an appropriate self - similar profile " functions @xmath159 .
the reason is the following : suppose a smooth profile @xmath159 is constructed associated to an increasing function @xmath179 satisfying @xmath181 as @xmath182 .
then will describe a coherent localizing structure where @xmath159 will depict the profile of that structure .
indeed , if for example @xmath183 is a bell - shaped even profile , the transformation of variables will have the effect of narrowing the width and increasing the height of the profile as time increases .
eventually it forms a singularity at @xmath184 as @xmath185 .
a simple calculation shows that the functions @xmath159 satisfy a system of ordinary differential equations , @xmath186 where we have selected @xmath187 independently of @xmath176 , or equivalently @xmath188 in this ansatz the parameter @xmath189 parametrizes the focusing rates of growth and narrowing during the developing localizing structure . invoking @xmath129
, we see that @xmath190 and thus we are seeking a family of solutions growing at a polynomial order . the system - is solved subject to initial data @xmath191 because the problem is singular the values of the data can not be independent and - imply the compatibility conditions @xmath192 therefore , among the five parameters @xmath15 , @xmath189 , @xmath193 , @xmath194 and @xmath195 , involved in determining these solutions , only three are independent .
we adopt the view that @xmath15 and the rate @xmath189 precise the system - , while one parameter ( say ) @xmath193 reflects choice of the initial data .
further inspection of - shows that if @xmath196 is a solution , then @xmath197 is also a solution .
this symmetry implies @xmath198 is odd , @xmath199 and @xmath200 are even , and suggests to impose @xmath201 in summary , we seek a family of solutions to - in the half space @xmath202 subject to the initial conditions and .
the scale invariance property of - is inherited by the system - : @xmath203 is a solution if @xmath204 , @xmath205 is a solution .
note that conditions persist under rescaling but the initial value does not .
in other words , once we have a solution with the initial value @xmath206 , then we obtain solutions with different initial values by rescaling . in the previous section
the emergence of a localizing structure was reduced to the existence of an orbit for the initial value problem and satisfying the initial conditions - .
the system - is non - autonomous and singular at @xmath207 . in this section
, we introduce a series of transformations in order to de - singularize the problem and obtain an autonomous system of ordinary differential equations .
the solution associated with a localizing structure rises as a heteroclinic orbit of the new system .
define @xmath208 , @xmath209 , @xmath210 , @xmath211 and a new independent variable @xmath212 by @xmath213 where @xmath214 \ , $ ] and @xmath215 is such that @xmath216 then @xmath208 , @xmath209 , @xmath217 satisfy the autonomous system @xmath218 in the new independent variable @xmath212 , where we have denoted @xmath219 by @xmath220 .
we observe that even if @xmath221 as @xmath222 for some @xmath223 , then @xmath224 and thus @xmath225 and @xmath226 diverge to infinity .
this causes analytical difficulties , and in order to overcome this difficulty we introduce a second non - linear transformation to obtain an equivalent system with all equilibria lying in a finite region .
we define new variables @xmath227 a cumbersome but straightforward computation shows that @xmath228 satisfies the autonomous system @xmath229 in the following section we study the equilibria of system - and the local behavior around them , with the objective of understanding the existence of heteroclinic orbits between the equilibria . system - has four equilibria , for @xmath230 , which are @xmath231 when @xmath232 the equilibria @xmath165 , @xmath166 , @xmath233 persist , but @xmath234 is replaced by equilibria distributed on the entire @xmath235-axis .
since , we will be interested in the heteroclinic from @xmath165 to @xmath166 we do not consider this case separately .
furthermore , we restrict attention to the placement that @xmath166 lies above the plane @xmath236 , which implies the restriction @xmath237 the placement of equilibria is depicted in figure [ fig : equilibria0 ] .
the local behavior of the dynamical system around each equilibrium is determined ( for hyperbolic equilibria ) from the eigenstructure of the linearized problem .
we denote the three eigenvalues of the linearization at each @xmath161 by @xmath238 and the associated eigenvectors by @xmath239 , @xmath240 .
@xmath241 @xmath165 _ is as unstable node _ : all eigenvalues are real and positive .
the first and the last eigenvectors lie on the @xmath242-plane , the first one is pointing towards the origin on the @xmath243-axis and the last one is pointing towards right and down on the @xmath242-plane . the second eigenvector is going off the @xmath242-plane .
( see figure [ fig : equilibria0 ] . )
@xmath244 @xmath241 @xmath166 _ is a saddle _ : all eigenvalues are real , two are negative and one is positive .
the first and the last eigenvector lie on the @xmath242-plane , the first one is an unstable direction , pointing toward the @xmath233 and the last one is a stable direction , pointing toward @xmath166 in right bottom direction .
the second eigenvector is pointing off the plane .
@xmath245 @xmath241 @xmath233 _ is a stable node _ : all eigenvalues are real and negative .
the eigenvectors are the coordinate basis vectors , @xmath246 @xmath241 @xmath234 _ is at the origin and is a saddle _ : all eigenvalues are real , the second eigenvalue is positive and the last eigenvalue is negative . the sign of the first eigenvalue bifurcates by @xmath247 ; it is negative if @xmath248 and is positive if @xmath249 .
the eigenvectors are the coordinate basis vectors , @xmath250 in this section , we study the heteroclinic connections between the equilibrium points .
since there are four equilibria , the unstable and stable manifolds of the equilibria might intersect in various ways and could conceivably produce multiple connections .
we aim to identify the one that is physically relevant and to construct it in phase space .
we assert that the relevant orbit joins @xmath165 to @xmath166 as @xmath212 runs from @xmath251 to @xmath252 .
a schematic sketch of this orbit is depicted in figure [ fig : hetero ] .
let us explain how that is singled out
. we will be interested in the orbit converging to the equilibrium @xmath166 .
the reason is the following : from the perspective of shear band formation , the strain of a localizing solution should grow with time , as the material is loading .
note that if @xmath53 grows polynomially as @xmath253 , then the ratio @xmath254 as @xmath12 increases .
on the other hand , the transformations we imposed imply @xmath255 thus we are interested in @xmath256 with @xmath257 as @xmath258 , which suggests to restrict attention to orbits converging to @xmath166 .
the behavior of the nonlinear problem near the saddle point @xmath166 is determined by the corresponding linearization and the orbit @xmath259 in a neighbourhood of @xmath166 is expressed as @xmath260 we are interested in orbits that have a nontrivial component out of the plane @xmath261 .
this is achieved by requiring the coefficient @xmath262 in .
note that when @xmath263 the term @xmath264 determines the asymptotic response when @xmath262 .
the reader should note that the plane @xmath265 is invariant for the flow of the dynamical system - .
in fact the flow can be explicitly computed in this case and provides the heteroclinic connecting @xmath165 to @xmath166 on the plane @xmath266 .
as this orbit is known we focus on the ones that venture out of the plane @xmath265 , hence imposing @xmath262 .
figure [ fig : hetero ] illustrates schematically the orbit approaching @xmath166 in @xmath267 .
the orbit emanating from @xmath165 ( as @xmath269 ) has the asymptotic expansion @xmath270 \rightarrow \kappa\vec{x}_{02 } , \quad \text{as $ \eta \rightarrow -\infty$ for some constant $ \kappa>0$.}\ ] ] this claim follows from the initial conditions and , via an asymptotic analysis of the behavior of the solution @xmath196 of - near the singular point @xmath271 , which is presented in the appendix .
figure [ fig : hetero ] depicts the orbit exhibiting the above asymptotics ; the orbit emerges in the direction of @xmath167 of the second unstable eigenspace .
the numerical calculations performed in section [ sec : numerical_const ] ( see fig [ fig:2manifold ] ) suggest there is a two - parameter family of heteroclinic orbits connecting @xmath165 and @xmath166 .
this conjecture is also supported by the following arguments : @xmath166 has a two dimensional stable manifold while @xmath165 is an unstable node ( with a three - dimensional unstable manifold ) .
one heteroclinic connection between @xmath165 and @xmath166 lies in the plane @xmath266 and can be explicitly computed .
its computation results from noticing that the plane @xmath265 is invariant under the flow of the dynamical system - and integrating the resulting system of ( two ) differential equations .
this provides one connection on the plane @xmath265 . since
, the numerical computations suggest there is an out - of - plane heteroclinic connection this indicates that there is a two parameter family of such connections .
indeed , they are systematically computed in section [ sec : numerical_const ] .
given the two parameter family , we proceed to show how to select the desired heteroclinic that satisfies and .
the objective is to select the heteroclinic connection that satisfies the asymptotic relation .
note that using and we can compute from the data the coefficient @xmath272 in .
the general heteroclinic orbit @xmath259 satisfies , in a sufficiently small neighbourhood of @xmath165 , the expansion @xmath273 as @xmath268 . for @xmath15 small , @xmath274 ,
the first term in the right hand side of dominates the remaining terms .
therefore , dictates that the coefficient @xmath275 and thus fixes one curve and reduces by one the degrees of freedom .
let us denote by @xmath276 a heteroclinic orbit that emanates in the direction of the eigenvector @xmath167 .
there is one degree of freedom at our disposal provided by the translation invariance of the selected orbit .
having selected the curve @xmath277 , we proceed to select the translation factor @xmath278 as follows : let @xmath279 be the constant associated to the asymptotic behavior of @xmath280 in .
we set our target heteroclinic @xmath281 and compute @xmath282 = e^{-2\eta}\left[\begin{pmatrix } p(\eta+\eta_0 ) \\
q(\eta+\eta_0 ) \\
r(\eta+\eta_0 ) \end{pmatrix } - m_0\right]\rightarrow { \kappa_2}e^{2\eta_0}\vec{x}_{02 } , \quad \text{as $ \eta \rightarrow -\infty$.}\end{aligned}\ ] ] therefore , given @xmath283 in , we compute @xmath278 by @xmath284 in summary , there exists a two parameter family of heteroclinic orbits joining @xmath165 and @xmath166 , whose existence is supported by the numerical calculations , and singles out one orbit among them . among the parameters @xmath15 , @xmath189 and the data @xmath193 , @xmath194 and @xmath195 , only three are independent due to and . given @xmath15 and two parameters ,
say @xmath285 and @xmath35 with @xmath286 , we select @xmath287 and the translaton factor @xmath278 by : @xmath288 the inequality restricts the data @xmath289 now , we pick one heteroclinic @xmath290 having the desired asymptotic behavior by the procedure described in section [ sec : singling ] . the solution is reconstructed by inverting to obtain @xmath291 , @xmath292 , @xmath293 and @xmath294 , @xmath295 it is instructive to relate @xmath296 to the original variables @xmath297 , @xmath53 , @xmath54 , @xmath298 .
using we recover @xmath299 which are smooth functions of @xmath300 .
moreover , using and , the original variables @xmath4 , @xmath6 , @xmath42 and @xmath301 are reconstructed by the formulas @xmath302 the focusing behavior of the profiles is readily deduced from the above formulas and is also illustrated in the figures [ fig : gamma]-[fig : sigma ] , obtained from the computed heteroclinic orbit .
the asymptotic behavior of the profiles @xmath303 as @xmath304 and @xmath305 captures the behavior of the localizing solution and is listed in the following proposition : [ prop : asymptotic ] for the orbit @xmath306 , the corresponding profile @xmath303 defined in @xmath307 by satisfies the properties : a. at @xmath271 it satisfies the boundary conditions @xmath308 b. the asymptotic behavior as @xmath309 is given by @xmath310 c. the asymptotic behavior as @xmath311 is given by @xmath312 indeed the behavior in ( i ) and ( ii ) follows readily from the analysis in section and the appendix . to prove ( iii ) : note that the heteroclinic orbit @xmath313 must lie in the intersection @xmath314 and thus satisfies as @xmath258 the asymptotic expansion .
the term multiplying @xmath315 decays much faster than the term @xmath316 and can be ignored .
using , together with , , we compute the asymptotic behavior @xmath317 in turn , via , @xmath318 and ( iii ) follows from and @xmath319 .
having identified the asymptotic behavior of @xmath320 associated to the selected heteroclinic orbit , we can identify the behavior of @xmath321 . using and proposition [ prop : asymptotic](iii ) ,
we obtain : \(a ) for the strain : @xmath322 the tip of the strain @xmath323 grows at superlinear order and @xmath53 , @xmath324 , grows in sublinear order as @xmath325 .
recalling that the growth of the uniform shear motion is linear , this observation points to localization .
\(b ) for the strain rate @xmath326 @xmath327 decays to @xmath328 for @xmath329 in the presence of @xmath110 as @xmath325 .
\(c ) for the stress @xmath330 as time proceeds , the stress collapses at the origin quicker to @xmath328 than at points @xmath331 .
in this section we construct numerically the heteroclinic orbit proposed in section [ sec : hetero ] and schematically presented in figure [ fig : hetero ] . to this end
we numerically compute solutions of the dynamical system - using appropriate routines from the odesuite of matlab .
care has to be exercised to deal with the known difficulties in numerically computing heteroclinic orbits .
the first task is to compute the collection of heteroclinic orbits joining @xmath165 and @xmath166 .
such heteroclinics appear as intersections of @xmath332 , the unstable manifold of the equilibrium @xmath165 , and @xmath333 , the stable manifold of the equilibrium @xmath166 .
we capture this intersection numerically by using a shooting argument .
we start with initial conditions near @xmath166 at the directions spanned by the stable eigenvectors of the equilibrium @xmath166 .
such data lie in the stable manifold @xmath333 ; we compute numerically the solution of - solving backwards in @xmath212 .
the orbits that tend near the equilibrium @xmath165 are retained and form a two dimensional manifold of heteroclinics , which comprises the intersection @xmath334 .
the result of the computation is depicted in fig .
[ fig:2manifold ] .
it is expected from the asymptotic behavior in formula that the heteroclinics in this manifold will be separated to those approaching @xmath165 along the eigenvector @xmath335 ( the blue curves ) and one single orbit approaching in the direction @xmath167 . to capture computationally the latter , we proceed as follows .
starting now near @xmath165 we compute the solutions of - with initial data close to @xmath165 in the tangent space spanned by @xmath167 and @xmath336 .
we note that the unstable manifold @xmath337 is the whole space and splits into the blue curves ( forming the heteroclinic surface ) and the orange curves in the transversal directions .
this second computation captures the orange curves .
when the orbits reverse directionality the computation captures the ( red ) heteroclinic emanating in the direction @xmath167 ; see figure [ fig : intersection ] for a zoom of this computation .
having computed the heteroclinic orbit @xmath338 of the dynamical system - , our next task is to introduce the computed solution to the transformation to produce the numerically computed form of the associated self - similar solution .
the computed solution is presented in figures [ fig : gamma]-[fig : sigma ] .
the graph is presented in terms of the original variables @xmath339 , @xmath53 , @xmath327 and @xmath54 depicting the solution at various instances of time . from figures [ fig : gamma ] - [ fig :
sigma ] , one sees that initially all the variables are small perturbations of the corresponding uniform shear solutions .
these nonuniformities then are amplified and eventually develop a shear band .
the strain and strain rate exhibit sharp peaks localized around the origin and the velocity field becomes a step like function , with the transition again occurring around the origin .
the stress field initially assumes finite values , but as the shear band forms it collapses to computable zero at the center of the band .
we have considered a model for shear deformation of a viscoplastic material , with a yield relation depending the plastic strain @xmath6 and the strain rate @xmath41 , modeling a viscoplastic material exhibiting strain softening and strain rate sensitivity , in the form of inverse and power law respectively .
the effects of those two mechanisms were first assessed at the linearized level .
the uniform shearing solutions are stable for @xmath28 , see @xcite , and we considered the range @xmath111 . in this parameter range ,
strain rate sensitivity alters the nature of instability from _ hadamard instability _ to _ turing instability_. then we turned into explaining the emergence of a coherent structure in the nonlinear regime .
an effective equation is derived which captures the combined effect of the competition between strain - rate hardening and strain - softening .
we analyzed a class of focusing self - similar solutions exploiting the invariance properties of the underlying model .
this offers a quantitative way for describing the development of singularity beyond the linearized level .
the associated localized solution arises as a heteroclinic orbit of an induced autonomous dynamical system . for a certain range of the parameters ,
the existence of such a heteroclinic orbit was shown numerically .
the associated profiles exhibit coherent localizing structures that have the morphology of shear bands .
the result hinges on a computational construction of the heteroclinic orbit .
the proof of existence of the heteroclinic orbit requires advanced results from dynamical systems , the geometric singular perturbation theory , and is presented in @xcite . to establish the claim ,
the starting point is the boundary conditions and .
we compute a taylor expansion of the variables @xmath340 , @xmath341 and @xmath342 near @xmath207 .
initial conditions imply an asymptotic behavior near @xmath207 of the form @xmath343 the values of the variables and their derivatives evaluated at @xmath207 can be inferred by differentiating system - repeatedly .
after a straightforward calculation , we find @xmath344 and @xmath345 then , the taylor expansion of @xmath340 near @xmath207 gives @xmath346 and the taylor expansion of @xmath341 is @xmath347\\ & = n\bar\gamma(0)\bu(0)^{1-n}\xi^2 + o(\xi^2 ) .
\end{aligned}\ ] ] finally , notice first that @xmath348 the taylor expansion of @xmath349 near @xmath207 yields @xmath350 + o(\xi^2)\\ & = -\big(\bar\gamma(0)\bu(0)^{1-n}\big)\frac { \lambda } { 2-n } \frac { \lambda } { 1-n } \bigg(\frac{1 } { \frac{1}{2 } - \frac{n}{1-n } \frac { \lambda } { r_0}}\bigg)\xi^2 + o(\xi^2 ) \ , .\end{aligned}\ ] ] from @xmath351 , we conclude that @xmath352 \rightarrow { \bar\gamma(0)\bu(0)^{1-n}}\vec{x}_{02 } , \quad \text{as $ \eta \rightarrow -\infty$. } \end{aligned}\ ] ] now specifies @xmath353 and in view of , @xmath354 notice that @xmath355 has the role of determining a translation factor of the curve .
let @xmath356 be a solution of - with , and @xmath306 be the corresponding heteroclinic orbit of - .
further , let @xmath357 be the rescaled solution as in and @xmath358 be the corresponding heteroclinic orbit
. then @xmath359 with @xmath360 .
|
metals deformed at high strain rates can exhibit failure through formation of shear bands , a phenomenon often attributed to hadamard instability and localization of the strain into an emerging coherent structure .
we verify formation of shear bands for a nonlinear model exhibiting strain softening and strain rate sensitivity .
the effects of strain softening and strain rate sensitivity are first assessed by linearized analysis , indicating that the combined effect leads to turing instability . for the nonlinear model a class of self - similar solutions
is constructed , that depicts a coherent localizing structure and the formation of a shear band .
this solution is associated to a heteroclinic orbit of a dynamical system .
the orbit is constructed numerically and yields explicit shear localizing solutions .
|
understanding the physical properties of baryonic matter above nuclear saturation density is one of the fundamental challenges in modern nuclear astrophysics .
many aspects of neutron stars ( e.g. , the mass - radius relation , cooling and rotational dynamics , glitches and pulsar kicks ) depend on these properties .
for example , the equation of state of supranuclear baryonic matter plays the key role in determining the maximum possible mass of neutron stars .
the harder ( softer ) equation of state is , the larger ( smaller ) maximum mass can be .
the equation of state of dense baryonic matter is also one of the essential ingredients that determines the dynamics of core - collapse supernovae and , in turn , the mass distribution of black holes in the universe @xcite .
several characteristics of the gravitational wave emission from merging neutron stars @xcite , which may be observed , e.g. , by advanced laser interferometer gravitational wave observatory ( ligo ) , are sensitive to the details of the equation of state . from the theoretical viewpoint
, currently there is no consensus even regarding the qualitative state of matter at the highest densities reached in stellar cores .
the most conservative possibility is that such matter is made of only nucleonic degrees of freedom .
the study of several neutron stars in ref .
@xcite , for example , does not exclude such a possibility , although a phase transition may be in agreement with their analysis , provided no extreme softening of the equation of state occurs . another astrophysical determination of masses and radii of three neutron stars in ref .
@xcite suggests , however , that the actual equation of state is too soft to be pure nucleonic .
such contradictory interpretations are representative and show that the current knowledge is too limited to settle the issue
. theoretically , it may be also appropriate to mention that the observables associated with the equation of state alone have limited power to probe the actual nature of dense matter @xcite . in fact , a true insight regarding the stellar interior may require a comprehensive understanding not only of the thermodynamical , but also transport properties and neutrino emission rates of various possible states of superdense matter . in this paper , we assume that baryonic matter at the highest stellar densities is deconfined quark matter .
the possible formation of quark matter in stars is an old hypothesis @xcite that dates back to the time when the concept of quarks was first introduced @xcite .
this is also supported by general considerations @xcite based on the property of asymptotic freedom in quantum chromodynamics ( qcd ) @xcite .
the main uncertainties of this scenario are ( i ) the value of the critical density , at which the deconfinement transition occurs , and ( ii ) the actual highest density reached in stars . if quark matter is formed , as we assume here , it is also likely to be a color - superconductor @xcite .
( for reviews on color superconductivity and its general effects on stellar properties see , for example , refs . @xcite . ) in this paper , in particular , we concentrate on the scenario , in which color superconductivity is due to same - flavor , spin - one cooper pairing @xcite .
the fact of liberation of quark degrees of freedom and the formation of a color - superconducting state of matter is likely to be revealed through a detailed study of the observational features of neutron stars .
for example , one promising class of observables is related to the rates of weak processes .
such processes are known to be responsible for the cooling rates @xcite and damping of the rotational ( r - mode ) instabilities @xcite in stars .
the cooling is primarily determined by the neutrino emission rate , while the damping of r - modes is controlled by the viscosity of dense matter @xcite . the bulk viscosity in the normal phase of three - flavor quark matter is usually dominated by the nonleptonic weak processes @xcite .
it was argued in ref .
@xcite , however , that the interplay between the semi - leptonic and nonleptonic processes may be rather involved even in the normal phase of quark matter . indeed , because of the resonance - like dynamics responsible for the bulk viscosity and because of a subtle interference between the two types of the weak processes
, a larger rate of the nonleptonic processes may not automatically mean its dominant role .
in fact , it was shown that the contributions of the two types of weak processes are generally not separable and that , for a range of parameters , taking into account the semi - leptonic processes may substantially modify the nonleptonic result @xcite . in this paper
, we extend the analysis of ref .
@xcite and study the effect that spin - one color superconductivity has on the bulk viscosity when the interplay between the two types of weak processes is carefully taken into account .
( for calculation of the bulk viscosity in other color superconducting phases see refs .
the necessary ingredients for the calculation of the bulk viscosity are the rates of semi - leptonic ( urca ) and nonleptonic weak processes . while the needed rates for the semi - leptonic processes in several spin - one color superconducting phases were obtained several years ago in ref .
@xcite , the corresponding rates of the nonleptonic processes remained unknown until very recently @xcite . here
we utilize both to obtain the bulk viscosity .
the rest of the paper is organized as follows .
the general formalism for the calculation of the bulk viscosity in strange quark matter with several active weak processes is reviewed in sec . [ formalism ] .
this formalism is then used in sec .
[ calculation ] to obtain our main results for the bulk viscosity as a function of temperature and the frequency of density oscillations .
there we also study the enhancement effect of color superconductivity on the bulk viscosity , as well as the interplay of semi - leptonic and nonleptonic processes . in sec .
[ discussion ] , we discuss the results and their potential implications for the physics of compact stars .
two appendices at the end of the paper contain our fits for the numerical suppression factors of the semi - leptonic and nonleptonic rates in spin - one color superconducting strange quark matter .
in this study , in order to calculate the bulk viscosity in the presence of several types of active weak processes , we follow the general formalism of ref . @xcite .
we assume that small oscillations of the quark matter density are described by @xmath1 where @xmath2 is the magnitude of the oscillations .
for such a periodic process , the bulk viscosity @xmath3 is defined as the coefficient in the expression for the energy - density dissipation averaged over one period , @xmath4 , @xmath5 where @xmath6 is the hydrodynamic velocity associated with the density oscillations . by making use of the continuity equation , @xmath7 , we derive @xmath8 such an energy - density dissipation of a pulsating hydrodynamic flow is the outcome of a net work done on a macroscopic volume over a period of the oscillation , @xmath9 where @xmath10 is the specific volume . by matching the hydrodynamic definition in eq .
( [ zeta - def ] ) with the relation in eq .
( [ diss - energy ] ) , we derive the expression for the bulk viscosity , @xmath11 the dominant mechanism behind the bulk viscosity is related to weak processes @xcite .
a periodic oscillation of the density is responsible for an instantaneous departure from @xmath12-equilibrium in the system . as a result ,
the forward and backward weak processes ( e.g. , @xmath13 and @xmath14 ) , which have equal rates in equilibrium , become unbalanced .
their net effect is to restore the equilibrium composition .
however , since the weak rates are relatively slow , a substantial time lag between the oscillations of the fermion number density ( and , thus , the specific volume ) and the chemical composition ( and , thus , the pressure ) develops .
if the resulting relative phase shift of the two oscillations is @xmath15 , one finds from eqs .
( [ diss - energy ] ) and ( [ zeta - pressure ] ) that the corresponding energy dissipation and the bulk viscosity are proportional to @xmath16 .
( note that the departure from the thermal equilibrium is negligible because it is restored by strong forces on much shorter time scales . )
it should be clear that the instantaneous flavor composition in oscillating quark matter and the rate difference of the forward and backward weak processes in fig .
[ fig - weak ] are related to each other .
the difference of the rates changes the composition , while the composition in turn influences the difference of rates .
the corresponding dynamics can be conveniently described in terms of the time dependent deviations of the chemical potentials from their equilibrium values . in @xmath12 equilibrium , the chemical potentials of the three lightest quarks are related as follows : @xmath17 . here
@xmath18 , @xmath19 and @xmath20 are the chemical potentials of up , down and strange quarks , while @xmath21 is the electron chemical potential . in pulsating matter , the instantaneous departure from equilibrium is described by the following two independent parameters : @xmath22 where @xmath23 denotes the deviation of the chemical potential @xmath24 from its equilibrium value .
( note that @xmath25 is not independent . ) when @xmath23 are non - zero , the corresponding pairs of forward and backward weak processes in fig .
[ fig - weak ] have different rates .
to leading order , the rate differences are linear in @xmath23 , @xmath26 [ ratediff0 ] the corresponding @xmath27-rates have been calculated for the normal phase @xcite as well as several color superconducting phases of quark matter @xcite .
the results for the normal phase , in particular , read @xmath28 these will be used below as a benchmark for the rates in spin - one color - superconducting phases .
processes that contribute to the bulk viscosity of dense quark matter.,title="fig:",scaledwidth=35.0% ] + processes that contribute to the bulk viscosity of dense quark matter.,title="fig:",scaledwidth=35.0% ] + processes that contribute to the bulk viscosity of dense quark matter.,title="fig:",scaledwidth=35.0% ] the semi - leptonic rate @xmath29 is determined by the urca processes @xmath30 and @xmath31 , shown in diagrams @xmath32 and @xmath33 in fig .
[ fig - weak ] .
it was calculated in ref .
@xcite for four different spin - one color - superconducting phases of quark matter .
the result has a form of the product of the rate in the normal phase @xmath34 and a phase - specific suppression factor , @xmath35 , \label{lambda_3_general}\ ] ] where @xmath36 is the spin - one color - superconducting gap parameter , and @xmath37 is a suppression factor for the processes involving gapped quasiparticles .
( the first term in square brackets is the contribution of ungapped quasiparticles . )
when @xmath38 , the suppression factor @xmath37 approaches @xmath39 and the normal phase result is restored .
a simple fit to the numerical data of ref .
@xcite for @xmath37 is presented in appendix [ urca - rates ] .
because of similar kinematics and phase space constraints for the other pair of semi - leptonic processes , @xmath40 and @xmath41 , shown in diagrams @xmath42 and @xmath43 in fig .
[ fig - weak ] , the dependence of the rate @xmath44 on the color - superconducting gap should take the same form as @xmath45 in eq .
( [ lambda_3_general ] ) , i.e. , @xmath46 . \label{lambda_2_general}\ ] ] in contrast , the rate @xmath47 is determined by the nonleptonic processes @xmath13 and @xmath48 , see diagrams @xmath49 and @xmath50 in fig . [ fig - weak ] , which have a qualitatively different kinematics . in spin - one color - superconducting phases of quark matter ,
this was recently calculated in ref .
the numerical result can be conveniently summarized by the following expression : @xmath51 , \label{lambda_1_general}\ ] ] where , in addition to the suppression factor @xmath52 , we also introduced a constant @xmath53 , which determines a relative contribution of the ungapped quasiparticles to the corresponding rate . in the four spin - one phases studied in ref .
@xcite , the constant takes the following values : @xmath54 , @xmath55 , and @xmath56 . a simple fit to the numerical data for @xmath52
is given in appendix [ nonleptonic - rates ] . when the rates ( [ lambda_3_general ] ) , ( [ lambda_2_general ] ) and ( [ lambda_1_general ] ) are known , the calculation of the instantaneous pressure and , thus , the bulk viscosity from eq .
( [ zeta - pressure ] ) is straightforward @xcite .
here we quote only the final expression for the viscosity , @xmath57 where @xmath58 , \label{zeta11 } \\
\zeta_{2 } & = & \frac{n}{\omega}\frac{\alpha_{1}\alpha_{3}}{g_{1}^2+g_{2}^2 } \big [ \alpha_{1}\alpha_{2}\alpha_{3 } c_{2}^2\nonumber\\ & + & \left(\alpha_{1}+\alpha_{2}+\alpha_{3}\right ) \left[\left(a_{2}-b_{2 } \right ) c_{1 } -a_{2}c_{2}\right]^2\big ] , \label{zeta22 } \\
\zeta_{3 } & = & \frac{n}{\omega}\frac{\alpha_{1}\alpha_{2}}{g_{1}^2+g_{2}^2 } \big [ \alpha_{1}\alpha_{2}\alpha_{3 } \left(c_{1}-c_{2}\right)^2\nonumber\\ & + & \left(\alpha_{1}+\alpha_{2}+\alpha_{3}\right ) \left(b_{1}c_{2}-b_{2}c_{1}\right)^2\big ] , \label{zeta33}\end{aligned}\ ] ] [ zeta1 - 3 ] and @xmath59 here @xmath60 ( @xmath61 ) and @xmath62 is the baryon density of quark matter .
the quantities @xmath63 , @xmath64 and @xmath65 are susceptibility - like functions , see ref .
@xcite for the definition . to leading order in @xmath66 , they are the same as in the normal phase . for comparison ,
let us also note that the bulk viscosity in the limit of the vanishing semi - leptonic rates reads @xmath67
in our calculation of the bulk viscosity in spin - one color - superconducting quark matter below , we choose the same two representative sets of model parameters as in ref .
@xcite : [ tableab ] [ cols="<,<",options="header " , ] + for model parameters in set a , the numerical results are presented in fig .
[ results - set - a ] . as we can see
, the value of @xmath68 determines the point where the bulk viscosity starts to deviate from the benchmark result in the normal phase ( shown by the gray solid line ) .
the upper panels show the dependence of the bulk viscosity @xmath3 on temperature for two representative values of the oscillation frequency , @xmath69 and @xmath70 .
the lower panels in the same figure show the temperature dependence of the ratio @xmath71 , where @xmath3 is the bulk viscosity that takes into account all weak processes , while @xmath72 is an approximate result , see eq .
( [ zeta_non ] ) , in which only the nonleptonic processes are included and the semi - leptonic processes are not . when the ratio @xmath71 is substantially larger than 1
, it is an indication that the semi - leptonic processes play an important role and , thus , can not be neglected .
compared to the normal phase result , the main features of the temperature dependences in spin - one color - superconducting phases ( see the upper panels in fig . [ results - set - a ] ) are ( i ) a smoothed shape of the semi - leptonic hump " and ( ii ) an overall enhancement of the bulk viscosity due to color superconductivity for a substantial range of temperatures below @xmath68 . as in the case of the normal phase ,
the semi - leptonic processes are responsible for an increase ( hump " ) of the bulk viscosity in a region of temperatures around @xmath73 , where @xmath74 [ t_hump ] are the approximate positions of the peak of the hump in the normal phase in the case of the model parameters in set a and set b , respectively . in order to derive these results
, we used an approximate expression for the bulk viscosity in eq .
( 24 ) of ref .
@xcite , which is valid when the nonleptonic rate is infinitely large while the semi - leptonic rates are finite .
the maximum of that expression corresponds to @xmath75 , whose solution determines an approximate value for @xmath73 .
two remarks are in order here : ( i ) the scaling law @xmath76 follows from the power - law temperature dependence of the semi - leptonic rates @xmath77 and ( ii ) the overall value in eq .
( [ t_hump ] ) is slightly corrected to match the actual numerical results in the case of a finite nonleptonic rate . when @xmath78 the semi - leptonic hump is partially washed out by the presence of color superconductivity .
this is most clearly seen from the ratio of the bulk viscosities @xmath71 in the lower panels in fig .
[ results - set - a ] .
while the inclusion of the semi - leptonic processes leads to an increase of the viscosity , the effect is not as large as in the normal phase .
of course , this conclusion is sensitive to the choice of the color - superconducting critical temperature @xmath68 . in general ,
two qualitatively different regimes can be realized .
when @xmath79 , the hump occurs in the normal phase and , therefore , its shape is almost unaffected by color superconductivity . in the opposite case ,
@xmath80 , the effect is present and gets stronger as @xmath68 increases relative to @xmath73 .
now , let us turn to an overall enhancement of the bulk viscosity due to color superconductivity below @xmath68 .
this is observed almost for the whole range of temperatures @xmath81 , where @xmath82 is the temperature at which the bulk viscosity of the normal phase has a global maximum .
the value of @xmath82 can be easily estimated by considering an approximate expression for the bulk viscosity ( [ zeta_non ] ) when only the nonleptonic processes are taken into account . the maximum of eq .
( [ zeta_non ] ) corresponds to @xmath83 . after solving this for the temperature ,
we obtain @xmath84 [ t0max ] where @xmath85 is the period of oscillations measured in milliseconds .
notably , the location of the maximum is almost the same for both sets of model parameters . because of the superconductivity ,
the location of the maximum is shifted to a higher temperature , @xmath86 , where @xmath53 is the same parameter that appears in eq .
( [ lambda_1_general ] ) . taking the shift of the maximum into account
, we find that the enhancement relative to the normal phase is observed for @xmath87 with @xmath88 being the point between @xmath82 and @xmath89 , at which the bulk viscosities for the normal and superconducting phases cross . at lower temperatures , @xmath90 ,
the effect of color superconductivity is opposite : it reduces the bulk viscosity .
+ the range of temperatures , in which the bulk viscosity increases relative to the normal phase of quark matter , depends on the value of the critical temperature @xmath68 and the frequency of oscillations .
while the actual enhancement of the viscosity also depends on the specific pattern of spin - one pairing , the qualitative features in all four phases studied here are similar . as an example , let us consider the csl phase in more detail . in fig .
[ diagram - t - vs - tc ] , we show the contour plot for the bulk viscosity enhancement factor due to color superconductivity . the ratio @xmath91 is larger than @xmath39 only in the colored regions in fig . [ diagram - t - vs - tc ] . in white regions
, it is either @xmath39 ( when @xmath92 ) or less than @xmath39 ( otherwise ) . as evident from fig .
[ diagram - t - vs - tc ] , the enhancement of the bulk viscosity by spin - one color superconductivity occurs in a rather wide range of temperatures , especially when the frequency of density oscillations is not too large and the value of @xmath68 is not too small . at @xmath69 , for example , it extends over an order of magnitude or more in temperature , provided @xmath93 . at @xmath70 , in contrast , an order of magnitude or wider temperature range for the enhancement is seen only if @xmath94 .
( it should be noted that , in the case @xmath69 shown in the left panel in fig .
[ diagram - t - vs - tc ] , the ratio @xmath91 is truly less than @xmath39 in a small white region just below the @xmath95 line .
this abnormality " is due to a subtle interplay between the semi - leptonic and nonleptonic processes when the value of @xmath68 is fine - tuned to be near @xmath73 . ) by ignoring the subtle complications due to the semi - leptonic hump around @xmath73 , we find that the enhancement of the bulk viscosity in the window of temperatures @xmath87 ( as well as the suppression at lower temperatures , @xmath90 ) is primarily due to the reduction of the nonleptonic rate @xmath96 in color - superconducting phases . at temperatures below @xmath68 ,
when all gapped quasiparticles effectively cease to contribute , the corresponding reduction factor for the rate is approximately given by the value of @xmath53 .
this means that the enhancement factor for the viscosity approaches its inverse value , @xmath97 . by making use of the numerical results for @xmath53 , we find that the enhancement factor for the bulk viscosity reaches up to about @xmath98 in the @xmath99- and polar phases , @xmath100 in the planar phase and @xmath101 in the csl phase .
( the suppression factors at @xmath90 approach the same values . ) in the region of the hump , of course , the behavior is more complicated , but the overall effect of superconductivity is still mainly to increase the bulk viscosity .
the numerical results in the case of the model parameters in set b are shown in fig .
[ results - set - b ] .
the qualitative features are similar to those obtained for set a. however , the effect of the semi - leptonic processes is less pronounced : the corresponding hump is almost non - existent and the ratio @xmath71 does not much deviate from 1 . at the same time , the effect of color superconductivity is very well pronounced .
compared to the normal phase result , an enhancement of the bulk viscosity by a factor of about @xmath97 is seen in a relatively wide window of temperatures from @xmath102 to @xmath68 .
in this study , we calculated the bulk viscosity in spin - one color - superconducting strange quark matter by carefully taking into account the interplay between the nonleptonic and semi - leptonic week processes .
as expected , the nonleptonic processes give the dominate contribution to the viscosity in a wide range of parameters . yet , as in the normal phase @xcite , the semi - leptonic processes may also lead to a substantial correction in a window of temperatures around @xmath73 , see eq .
( [ t_hump ] ) . the value of @xmath73 scales as @xmath103 and happens to be of order @xmath104 for millisecond pulsars .
the size and the relative importance of the hump can be conveniently measured by the ratio @xmath71 when it is noticeably larger than @xmath39 . for millisecond pulsars ,
however , this ratio remains close to @xmath39 .
the effect is more pronounced when the period is a few orders of magnitude longer .
we also find that the corresponding hump in the temperature dependence of the bulk viscosity of color superconductors is partially washed out compared to the normal phase .
the higher is @xmath68 relative to @xmath73 , the larger wash out of the hump is seen . at sufficiently low @xmath68 ,
i.e. , @xmath105 , the hump occurs in the normal phase and , therefore , its shape is unaffected by color superconductivity .
if the critical temperature of the spin - one color - superconducting phase transition @xmath68 is larger than @xmath82 , see eq .
( [ t0max ] ) , the main effect of color superconductivity is an overall increase of the bulk viscosity in a range of subcritical temperatures , @xmath87 , see fig .
[ diagram - t - vs - tc ] .
the corresponding range of temperatures widens with increasing the value of @xmath68 and with decreasing the frequency of oscillations .
the increase of the viscosity is primarily due to the suppression of the nonleptonic rate by color superconductivity . at almost all temperatures below @xmath68 ,
the rate is dominated by the ungapped quasiparticles , whose relative contribution is scaled by the factor @xmath53 with respect to the normal phase ( note that @xmath106 ) .
it is the inverse value @xmath97 that determines the maximal enhancement of the bulk viscosity at subcritical temperatures .
the corresponding enhancement factor is equal to @xmath98 in the @xmath99- and polar phases , about @xmath100 in the planar phase and about @xmath101 in the csl phase .
( at temperatures below @xmath102 , color superconductivity leads to a suppression of the bulk viscosity , and the maximal suppression will approach the same value of @xmath97 . ) in relation to this result , it might be appropriate to note that a similar enhancement mechanism was previously observed for spin - zero color superconductors @xcite .
a special feature of spin - one color superconductivity is that the maximum enhancement factor can be much larger . in our analysis
, we utilized the same spin - one pairing pattern as in refs .
@xcite . in the case of zero quark masses ,
the main signature of the corresponding phases is the presence of ungapped quasiparticles .
when quarks have small masses , the gaps of the corresponding modes are of order @xmath107
. these may be still too small to significantly affect our main results .
however , if the spin - one gaps are larger , as some studies suggest @xcite , the suppression of the nonleptonic rates and , therefore , the enhancement of the bulk viscosity in color superconducting matter may turn out to be even stronger . in application
to compact stars , we may speculate that the transition to a spin - one color superconducting phase in a stellar core can have a stabilizing effect against the r - modes driven by the gravitational radiation @xcite .
if the critical temperature of the corresponding phase transition is on the order of or above @xmath104 , the corresponding dynamics can affect even relatively young stars .
the study of the actual quantitative effect that spin - one color superconductivity has on the reduction of the instability window in the pulsar frequency and temperature plane can be done along the lines of refs .
@xcite . however , this problem is beyond the scope of the present paper .
the authors would like thank mark alford , thomas schfer , andreas schmitt and kai schwenzer for useful comments . the work of x.w
. is also supported in part by the arizona state university graduate fellowship .
the work of i.a.s . is supported in part by the start - up funds from the arizona state university and by the u.s .
national science foundation under grant no .
the rates of the semi - leptonic processes in spin - one color superconducting quark matter were calculated in ref .
the general expression for the rate takes the following form : @xmath108 \quad\mbox{for}\quad i=2,3,\ ] ] where @xmath109 is the corresponding rate in the normal phase of quark matter and @xmath110 is a phase - specific suppression factor . by construction , it satisfies the constraint @xmath111 , which corresponds to the case of the normal phase .
we used the numerical data of ref .
@xcite to obtain the following fits for the suppression factors as functions of the dimensionless ratio @xmath112 in the four spin - one color superconducting phases of quark matter : @xmath113 where @xmath114 , @xmath115 , @xmath116 , @xmath117 , @xmath118 and @xmath119 , @xmath120 where @xmath121 , @xmath122 , @xmath123 , @xmath124 , @xmath125 and @xmath126 , @xmath127 where @xmath128 , @xmath129 , @xmath130 , @xmath131 and @xmath132 , @xmath133 where @xmath134 , @xmath135 , @xmath136 , @xmath137 and @xmath138 .
the @xmath27-rate of the nonleptonic processes in spin - one color superconducting quark matter was calculated in ref .
the general expression for the rate takes the following form : @xmath139 , \ ] ] where @xmath140 is the corresponding rate in the normal phase of quark matter , @xmath53 is a constant that determines the relative contribution of ungapped quasiparticles to the rate , and @xmath141 is a phase - specific suppression factor due to gapped quasiparticles .
the normal phase corresponds to @xmath142 , in which case there is no suppression and @xmath143 .
the value of @xmath53 for each phase reads for this study we used the numerical data of ref .
@xcite to obtain the following fits for the suppression factors as functions of the dimensionless ratio @xmath112 : @xmath145 where @xmath146 , @xmath147 and @xmath148 , @xmath149 where @xmath150 , @xmath151 and @xmath152 , @xmath153 where @xmath154 , @xmath155 , @xmath156 and @xmath157 , @xmath158 where @xmath159 , @xmath160 , @xmath161 and @xmath162 .
a. w. steiner , j. m. lattimer and e. f. brown , arxiv:1005.0811 .
f. zel , g. baym and t. gver , arxiv:1002.3153 [ astro-ph.he ] . f. zel , nature * 441 * , 1115 ( 2006 )
. m. alford , d. blaschke , a. drago , t. klahn , g. pagliara and j. schaffner - bielich , nature * 445 * , e7 ( 2007 ) [ arxiv : astro - ph/0606524 ] .
j. c. collins and m. j. perry , phys .
* 34 * , 1353 ( 1975 ) .
h. d. politzer , phys .
lett . * 30 * , 1346 ( 1973 ) ; d. j. gross and f. wilczek , phys .
d * 8 * , 3633 ( 1973 ) ; phys .
d * 9 * , 980 ( 1974 ) .
b. c. barrois , nucl .
b * 129 * , 390 ( 1977 ) .
d. bailin and a. love , phys .
b * 137 * , 348 ( 1984 ) .
k. rajagopal and f. wilczek , arxiv : hep - ph/0011333 .
m. g. alford , ann .
nucl . part .
* 51 * ( 2001 ) 131 .
s. reddy , acta phys .
b * 33 * , 4101 ( 2002 ) .
m. buballa , phys .
rept . * 407 * , 205 ( 2005 ) .
i. a. shovkovy , found .
phys . * 35 * , 1309 ( 2005 ) .
m. g. alford , a. schmitt , k. rajagopal , and t. schfer , rev .
phys . * 80 * , 1455 ( 2008 ) .
t. schfer , phys .
d * 62 * , 094007 ( 2000 ) .
a. schmitt , q. wang and d. h. rischke , phys .
d * 66 * , 114010 ( 2002 ) ; phys .
* 91 * , 242301 ( 2003 ) .
a. schmitt , phys .
d * 71 * , 054016 ( 2005 ) .
m. g. alford , j. a. bowers , j. m. cheyne , and g. a. cowan , phys .
d * 67 * , 054018 ( 2003 ) .
n. iwamoto , phys .
* 44 * , 1637 ( 1980 ) .
n. andersson , astrophys .
j. * 502 * , 708 ( 1998 ) .
j. madsen , phys .
lett . * 81 * , 3311 ( 1998 ) .
q. d. wang and t. lu , phys .
* 148b * , 211 ( 1984 ) .
h. dong , n. su , and q. wang , phys . rev .
d * 75 * , 074016 ( 2007 ) .
m. g. alford , s. mahmoodifar and k. schwenzer , arxiv:1005.3769 [ nucl - th ] .
b. a. sad , i. a. shovkovy , and d. h. rischke , phys .
d * 75 * , 125004 ( 2007 ) .
m. g. alford , m. braby , s. reddy , and t. schfer , phys .
c * 75 * , 055209 ( 2007 ) ; m. g. alford , m. braby , and a. schmitt , j. phys .
g * 35 * , 115007 ( 2008 ) . m. g. alford and a. schmitt , j. phys .
g * 34 * , 67 ( 2007 ) .
b. a. sad , i. a. shovkovy , and d. h. rischke , phys .
d * 75 * , 065016 ( 2007 ) .
b. a. sad , arxiv:0806.3359 [ astro - ph ] .
m. mannarelli and c. manuel , phys .
d * 81 * , 043002 ( 2010 ) .
x. wang , h. malekzadeh and i. a. shovkovy , phys .
d * 81 * , 045021 ( 2010 ) .
j. madsen , phys .
d * 47 * , 325 ( 1993 ) .
anand , a. goyal , v.k .
gupta , and s. singh , astrophys .
j. * 481 * , 954 ( 1997 ) .
j. r. ellis , e. gardi , m. karliner and m. a. samuel , phys .
d * 54 * , 6986 ( 1996 ) .
f. marhauser , d. nickel , m. buballa and j. wambach , phys .
d * 75 * , 054022 ( 2007 ) .
j. madsen , phys .
lett . * 85 * , 10 ( 2000 ) .
n. andersson , d. i. jones and k. d. kokkotas , mon . not .
astron . soc .
* 337 * , 1224 ( 2002 ) .
p. jaikumar , g. rupak and a. w. steiner , phys .
d * 78 * , 123007 ( 2008 ) .
|
the bulk viscosity in spin - one color - superconducting strange quark matter is calculated by taking into account the interplay between the nonleptonic and semi - leptonic week processes . in agreement with previous studies
, it is found that the inclusion of the semi - leptonic processes may result in non - negligible corrections to the bulk viscosity in a narrow window of temperatures .
the effect is generally more pronounced for pulsars with longer periods .
compared to the normal phase , however , this effect due to the semi - leptonic processes is less pronounced in spin - one color superconductors . assuming that the critical temperature of the phase transition is much larger than @xmath0 , the main effect of spin - one color superconductivity in a wide range of temperatures is an overall increase of the bulk viscosity with respect to the normal phase .
the corresponding enhancement factor reaches up to about 9 in the polar and _ a_-phases , about 25 in the planar phase and about 29 in the csl phase .
this factor is determined by the suppression of the nonleptonic rate in color - superconducting matter and , therefore , may be even larger if all quark quasiparticles happen to be gapped .
|
hysteresis is the history dependence of the properties of a material when driven by an applied external field @xcite .
it is observed in many different systems in non - equilibrium conditions . in general
it is a dynamical effect which must be treated within a time - dependent formalism . here
we will focus on a special class of systems that display , under certain conditions , time independent hysteresis properties .
the prototypes are ferromagnets , at low temperature , in which magnetization versus magnetic field paths are not influenced by the applied field rate , at least for low enough rates . in such materials ,
when these paths are studied in detail , it is commonly observed that they are not continuous but rather composed of small bursts or avalanches : this is the so - called barkhausen noise @xcite .
this abrupt and stochastic response of the magnetization to a continuously varying external field has been related to the disorder existing in the system @xcite .
actually , domain walls are pinned by impurities and therefore a magnetization change can only occur if the corresponding energetic barriers are overcome ( depinning ) . since these barriers are large , thermal activation is not effective , and the evolution of the sytem can only proceed by modifying the external field ( athermal behavior ) . in addition to ferromagnetic materials , a wide variety of complex systems like martensitic materials@xcite , superconducting films@xcite , or capillary condensation systems@xcite , display similar phenomenology .
all these systems can be described in the framework of fluctuationless first - order ( flfo ) phase transitions , i.e. they exhibit , when driven by an external field , a first - order phase transition influenced by disorder with irrelevant thermal fluctuations .
recently , different versions of spin lattice models with disorder , as the random bond @xcite and random field ising models @xcite , have been used to model hysteresis and avalanches in systems undergoing flfo phase transitions .
deterministic algorithms with synchronous dynamics corresponding to local energy relaxation have been employed to study hysteresis paths at zero temperature .
it has been found that the shape of hysteresis loops depends on the amount of disorder @xcite . for low disorder ,
when sweeping the field from -@xmath1 to + @xmath2 , the system reverses sharply the magnetization ( infinite avalanche ) at a given field , while for a high amount of disorder the path is smoother and extends over a wide range of fields .
interestingly the change in the hysteresis cycle shape has been attributed to the existence of a critical point at a given degree of disorder , where the infinite avalanche disappears .
for such a degree of disorder , avalanches of all sizes are detected indicating that the system behaves critically : the avalanche size distribution follows a power law characterized by a universal exponent depending only on the space dimensionality @xcite .
actually , in agreement with the models , it has been experimentally observed that the amplitude of barkhausen pulses in ferromagnetic materials , and the amplitude of the acoustic signals generated during martensitic transitions , extend over several decades indicating the possibility of a universal critical behavior @xcite .
a number of attempts to make a comparison between models and experiments have been reported .
@xcite nevertheless , the existence of criticality induced by changing the amount of disorder has not been confirmed .
the difficulty arises from the possibility of controlling the amount of disorder in a given system . in any case
, it must be mentioned that the studied models appear as very simple idealizations when compared to the complexity inherent to real systems .
this paper is concerned with the study of the magnetic properties of the cu - al - mn intermetallic alloy . after a suitable heat treatment @xcite
, this material displays an l2@xmath3 structure .
the stoichiometric compound cu@xmath4almn is ferromagnetic with a curie temperature of 630 k and it is known that the entire magnetic moment of the system must be attributed to the mn atoms @xcite coupled via the rkky interaction @xcite .
ac susceptibility measurements have revealed the existence of a spin - freezing process at low temperatures .
this freezing has been shown to correspond to a spin glass phase transition for low enough mn content @xcite . beyond this problem , here we focus on the study of the low temperature magnetic hysteresis for such systems .
we have studied a family of composition related alloys which undergo , prior to the spin - freezing process , a diffusionless structural transition to a close - packed phase ( martensitic transition ) . in these materials , magnetic disorder has two main origins : the lack of stoichiometry and the existence of incomplete configurational ordering @xcite .
the important fact is that the mn configurational disorder increases as the mn content decreases .
we show that increasing the degree of disorder , the low temperature hysteresis cycle changes from sharp to smooth .
this change is suggested to be related to a disorder induced phase transition . since the experimental evidences are not definitively conclusive
, we reinforce our interpretation by developing a microscopic model for such site - diluted magnetic systems . by fitting the model parameters to cu - al - mn available data
, we reproduce approximately the observed phenomenon and associate it with the disorder induced phase transition .
the paper is organized as follows . in section
ii , we present the experimental results ; in section iii we introduce a model based on a site diluted ising model suitable for the study of hysteresis in spin glass systems and compare the results obtained numerically with the experimental ones corresponding to cu - al - mn . finally , in section iv we discuss the results and conclude .
we have studied a family of alloys , with compositions close to the cu@xmath5almn@xmath6 line .
polycrystalline ( grain size @xmath7 100 @xmath8 m ) cylindrical - shaped ingots were obtained by melting pure elements @xmath9 .
rectangular shaped samples ( @xmath10 mm@xmath11 ) were cut using a low - speed diamond saw .
the samples were mechanically polished down to a thickness of @xmath12 mm .
samples were annealed for @xmath13 s at @xmath14 k , quenched into a mixture of ice and water , and annealed at room temperature for several weeks .
this heat treatment leads to a long - lived metastable state in which the material displays an l2@xmath3 structure ( @xmath15 space group ) .
this structure can be viewed as defined on a bcc lattice divided in four distinguishible fcc sublattices ( @xmath16 , @xmath17 , @xmath18 and @xmath19 ) as shown in figure [ fig1 ] . for samples with @xmath20 ( as in the present work ) , sublattices @xmath18 and @xmath19 ( for instance ) are filled with cu - atoms , and mn - atoms sit preferentially in sublattice @xmath16 . however , for kinetic reasons , a fraction of mn atoms also occupies the @xmath17 sublattice .
experiments based on atom location by channelling enhanced microanalysis ( alchemi ) performed on similar samples have shown that almost no mn atoms occupy sublattices @xmath18 and @xmath19 .
@xcite the spin - freezing temperatures @xmath21 of the alloys studied have been obtained from ac - susceptibility measurements @xcite . as shown in table [ samples ] , in all the samples studied @xmath21 is below the martensitic transition .
we also present the values of the paramagnetic curie temperature @xmath22 and the temperature @xmath23 of the structural transition taken from the same ref . .
concerning @xmath22 , note that it is positive for the three samples , which indicates the mean ferromagnetic character of the interactions .
magnetization measurements have been carried out , at low field rates ( @xmath24 oe / s ) using a squid magnetometer enabling applied fields up to 5 t ( the relative error in the field control is @xmath25 % ) .
the hysteresis observed in the glassy phase reflects the non - equilibrium state of the system characterized by a logarithmically slow response of the magnetization to an applied field .
limiting hysteresis loops have been measured after a zero - field cooling ( zfc ) and subsequent ( isothermal ) field sweep between @xmath26 t and @xmath27 t at different temperatures ( controlled within @xmath28 k ) . in figure [ fig2 ]
we show the hysteresis cycles obtained at @xmath29 k in the three samples studied .
loops are symmetrical and display small coercivity and remanence .
it is worth mentioning that for the sample with @xmath30 ( @xmath31 mn ) , the hysteresis loop has also been measured after a field cooling ( fc ) .
the cycle obtained is identical from the one measured after a zfc .
this proves the absence of global uniaxial exchange anisotropy in the polycrystalline samples considered . from the results given in fig .
[ fig2 ] it is clear that the magnetization can be supposed to be composed of an irreversible and a reversible contributions:@xcite @xmath32 only the irreversible term contributes to hysteresis . actually , the reversible contribution must be ascribed to the competition of the external field with minoritary antiferromagnetic bonds in the system , and can be assumed to be linear with the magnetic field : @xmath33 from linear fitting , we have determined @xmath34 for the alloys investigated .
the values obtained are : @xmath35 emu/(molmn oe ) , @xmath36 emu/(molmn oe ) and @xmath37 emu/(molmn oe ) for the samples with @xmath38 ( @xmath39 mn ) , @xmath30 ( @xmath40 mn ) and @xmath41 ( @xmath42 mn ) respectively .
therefore , within errors , @xmath34 can be assumed independent of @xmath43 , at least in the range of compositions studied .
after subtraction of the reversible term in the magnetization measured , we have estimated a saturation magnetization @xmath44 . it is represented ( per mn atom ) as a function of @xmath43 in fig .
we obtain that @xmath44 increases linearly with @xmath43 . the value of @xmath44 corresponding to the stoichiometric cu@xmath4almn alloy ( @xmath45 ) has been taken from the literature @xcite . as temperature
is increased the width of the cycle ( coercivity ) decreases and , as expected , we have not observed hysteresis above @xmath46 . in figure [ fig4 ]
we show the effect of temperature on hysteresis by plotting the central part of the hysteresis loops at three different temperatures . in all cases
the cycles are smooth and no discontinuities ( or avalanches ) have been detected .
however the cycle corresponding to @xmath41 obtained at @xmath47 k , shows practically a complete reversal of the remanent magnetization in a very reduced interval of fields .
this changes the shape of the hysteresis loop which becomes sharper in its central part . in order to quantitatively
characterize the shape of the hysteresis loops , we have represented in fig .
[ fig5 ] the remanence @xmath48 , the coercivity @xmath49 and the area of the loops @xmath50 , as a function of temperature for the three different compositions studied .
the area of the loops , which gives the dissipated energy per cycle , decreases with @xmath51 and increases with the mn content @xmath43 .
notice also that for the sample with @xmath41 the coercive field does not change for @xmath52 k. this fact is associated with the change in the shape of the hysteresis loop outlined in the previous paragraph .
we argue that the observed change in the hysteresis loop can be related to the fact that , as has recently been shown @xcite , a spin - glass phase transition occurs for samples with @xmath53 , but not for samples with higher mn concentration . for these compositions
, the cusp in the magnetic susceptibility curve as a function of temperature is related to the freezing of ferromagnetic clusters , being the high - temperature phase superparamgnetic and the low - temperature phase what we will call in this article a ferromagnetic spin - glass @xcite .
therefore we suggest that the change in the hysteresis loop may be related to a phase transition which has the origin on the different ground state properties of the system as @xmath43 is increased .
actually the value of @xmath43 at the transition point , should be close to the percolation limit for the system studied .
it would be expected that remanent magnetization reverses at a single value of the field ( large avalanche ) .
the fact that this does not strictly occur is due , as we will discuss in the next section , to the polycrystalline character of the samples studied .
we have also studied first - order reversal cycles within the limiting loop obtained as follows : starting from @xmath54 t the external field is increased to a value @xmath55 ( @xmath56 t ) , before saturation . at this point
the field evolution is reversed back to @xmath54 t. for different @xmath55 one obtains a family of trajectories , some of which are displayed in fig .
[ fig51 ] for the sample with @xmath30 and @xmath47 k. it is worth noting that in all cases the decreasing reversal trajectories join tangentially the limiting cycle .
the shape of these reversal trajectories depends on @xmath55 . in the inset of fig .
[ fig51 ] we show the coercive field of the reversal trajectories @xmath57 as a function of @xmath55 . note that for the sample with @xmath41 ( circles ) , @xmath57 remains approximately constant with @xmath55 when aproaching @xmath49 from above .
this is indicative of the fact that the reversal of magnetization occurs at a fixed field for high @xmath43 .
an interesting feature observed in a number of systems exhibiting flfo phase transitions is the return point memory effect .
this property is well illustrated by analysing the characteristics of complete internal hysteresis loops .
these internal loops are obtained when , coming from @xmath54 t and being in a state corresponding to a field @xmath58 t on the limiting cycle , the field evolution is reversed back until reaching @xmath59 , and later is increased up again to @xmath60 .
the return point memory effect is verified if , after this path , the system returns exactly to the same state that was left in the limiting cycle , that is , if it has memory of the former state . in this case , the same memory effect extends to any subcycle within a given cycle , and therefore this indicates that the state of the system can remember an entire hierarchy of turning points in its past evolution .
this phenomenon has been observed in some , but by no means all , systems with flfo phase transitions .
we have performed such complete internal cycles by reversing the field at different states in the limiting cycle .
typical results are shown in fig .
it is observed that after performing an internal loop and returning to the former field , the magnetization has increased . by performing internal cycles at different sweeping field rates ( in the positive and negative applied field regions ) ,
we have checked that this increase of the magnetization is significantly larger than the increase expected from logarithmic relaxational effects .
this enables to conclude that , in our system , the return point memory property is not satisfied .
in order to gain understanding of the observed experimental phenomena we develop , in this section , a model containing two basic ingredients : randomness of the lattice position of the magnetic atoms and competing interactions .
these two ingredients are crucial for the existence of spin - glass phenomena @xcite .
the proposed model is a modified version of the diluted ising model at @xmath61.@xcite we only consider sublattices @xmath16 and @xmath17 as possible sites for mn atoms ( see fig .
[ fig1 ] ) .
therefore we define a simple cubic lattice @xcite with @xmath62 ( @xmath63 ) sites and with periodic boundary conditions and lattice spacing @xmath64 . on each site
we define a variable @xmath65 which indicates the presence ( 1 ) or absence ( 0 ) of mn on that site , and an ising variable @xmath66 which accounts for two possible magnetic states of mn atoms .
the hamiltonian of the model is written as : @xmath67 where the first term is a sum over site pairs , @xmath68 is the exchange energy , @xmath69 the external field and @xmath8 the effective magnetic moment per atom .
a number @xmath70 of mn atoms are randomly placed as follows : a fraction @xmath71 is placed in sublattice @xmath16 while the rest @xmath72 is `` misplaced '' in sublattice @xmath17 .
values of @xmath73 ranging from @xmath74 to @xmath75 have been considered .
the concentration of mn is given by @xmath76 .
the exchange constants @xmath68 are distance dependent .
for the stoichiometric alloy , @xmath68 have been fitted from inelastic neutron scattering experiments @xcite .
the fitted values are the ones corresponding to neighbors of 2nd , 4th , 6th and 7th order in table [ jij ] .
we assume that @xmath68 do not change significantly when altering the mn content
. nevertheless we still have to interpolate the values of the @xmath68 corresponding to distances which are not present in perfectly ordered stoichiometric alloys . to do that we have fitted a standard rkky dependence to the experimental @xmath68 : @xmath77 where @xmath78 is proportional to the fermi moment @xmath79 .
since it is difficult to fit such an oscillating function we have first checked different values of @xmath78 in order to get the correct signs of the @xmath68 found experimentally .
the signs are only correctly fitted for values of @xmath78 around @xmath80 .
@xcite a standard least square fitting , then , renders @xmath81 mev . using this fit
we then interpolate the values of @xmath68 for the neighbors of order @xmath82st , @xmath83rd and @xmath29th indicated in table [ jij ] with asterisks . a strong antiferromagnetic interaction between n.n . sites
( that can only occur due to misplaced mn atoms ) is obtained .
this is consistent with the antiferromagnetic character found in @xmath18-mn with fcc structure @xcite ( lattice parameter @xmath84 ) .
nevertheless we have checked that the mean character of the interactions ( which can be estimated as a sum @xmath85 over all mn pairs @xmath86 ) is always ferromagnetic for all values of @xmath43 and @xmath73 .
this is consistent with the positive paramagnetic curie temperatures obtained experimentally ( see table [ samples ] ) . for the simulations we consider the exchange energy for next - nearest neighbors @xmath87 mev as the energy unit .
we define a reduced hamiltonian as : @xmath88 where @xmath89 and @xmath90 .
we define the reduced magnetization as : @xmath91 with this definition , the maximum reduced magnetization for the stoichiometric alloy is @xmath92 .
hysteresis cycles ( @xmath93 ) are obtained by starting with all the spin variables @xmath94 and the external field @xmath95 .
the field is then slowly increased until a spin becomes locally unstable , i.e. there is a decrease of energy if it is flipped .
then the external field is stopped , the spin is flipped and an avalanche starts : more spins may become unstable .
these are simulatenously flipped , which may trigger some new unstable spins , and so on until the avalanche stops . the number of steps necessary to reach this stable situation defines the avalanche duration and the corresponding magnetization change the avalanche size .
then the external field is increased again .
this is the so - called synchronous dynamics .
@xcite we have studied lattices with @xmath96 and for many runs , averages over @xmath97 different initial random atomic configurations have been taken .
we have systematically studied systems with concentration @xmath98 , @xmath99 , @xmath100 , and @xmath75 and disorder values @xmath101 .
examples of the hysteresis cycles are shown in fig .
[ fig7 ] . for a fixed value of @xmath73 ,
the hysteresis loop becomes sharper as @xmath43 is increased , while for a fixed value of @xmath43 the loop becomes smoother when increasing @xmath73 .
note also that for high values of @xmath73 and @xmath43 the loops become antiferromagnetic - like with long tails and with a tendency to split into two subloops . in all cases , when looking in detail at the loops , they consist of a sequence of discrete jumps ( avalanches ) that join metastable states .
this is illustrated in fig .
[ fig8 ] . due to the existence of antiferromagnetic interactions , inverse avalanches , decreasing the magnetization when increasing the field , may occur . concerning this point
, one should remember that the evolution is controlled by a local ( and not global ) equilibrium condition . in order to study in more detail the evolution from smooth to sharp hysteresis cycles
we have studied the statistical distribution of avalanche sizes .
sharp loops will exhibit few large avalanches while smooth loops will exhibit small avalanches only .
the avalanche distributions have been computed by averaging over 50 different realizations of the random distribution of mn atoms .
the resulting histograms are presented in fig .
[ fig9 ] for different values of p and x. we find that , in this p - x diagram , a critical line exists at which avalanches of all sizes are observed and the distributions are power - law .
this behavior has been analyzed following the techniques developed for the study of avalanches in a previous work on the study of flfo phase transitions in the random bond ising model @xcite .
the critical line can be found by studying two quantities : ( i ) the normalized average size @xmath102 of the largest avalanche in a cycle and ( ii ) the average duration @xmath103 of the largest avalanche in a cycle . as an example
we show the behavior of these two quantities along a line with @xmath104 and changing the concentration @xmath43 ( fig .
[ fig10 ] ) and along a line with @xmath105 and changing the disorder parameter @xmath73 ( fig .
[ fig11 ] ) .
the phase transition is smoothed due to finite size effects , which can be corrected as explained in ref .. actually , estimations of the transition points can be obtained by locating , using a polinomial fitting , the maximum on @xmath106 and the inflection point on @xmath107 . a systematic study of the behavior of these two quantities as a function of @xmath43 and @xmath73 allows to obtain the phase diagram shown in fig .
[ fig12 ] .
the discrepancy between the two estimations of the critical line can be attributed to finite - size - effects .
we find critical behaviour , depending on @xmath73 , for @xmath108 .
it is difficult to locate on this @xmath109 diagram the position where this phase transition will be expected to occur in the cu - al - mn system , since the value of @xmath73 for our samples is unknown .
actually we can only have some rough estimations of the @xmath110 dependence in the samples studied .
first , it is possible to obtain @xmath73 from monte carlo simulations of the atomic ordering process in a realistic lattice model for cu - al - mn presented in ref . .
the model considered in that paper contains only interactions to first and second nearest neighbors and its parameters have been fitted from the order - disorder transition temperatures in the real experimental system . the behavior of @xmath73 as a function of @xmath43 obtained from that model is represented by the two dotted lines in fig .
[ fig12 ] .
the lower line corresponds to the values of @xmath73 at @xmath61 , and the upper represents the values of @xmath73 at temperatures just below the order - disorder transition .
according to this result , the actual values of @xmath73 would be somewhere in the middle of these two dotted lines .
the intersection of these lines with the phase transition ( open symbols ) would give the critical concentration where the change in the hysteresis cycle is expected .
however , these estimations of @xmath110 correspond to equilibrium configurations , but kinetic effects ( samples have been quenched from high @xmath51 ) could modify the amount of disorder substantially .
moreover , the model used for the mc simulations does not contain any magnetic interaction term , which could have some relevance in the mn ordering process .
a second estimation can be obtained from the linear behavior of the saturation magnetization per atom @xmath44 with @xmath43 shown in fig .
[ fig3 ] . note that the system contains two kinds of mn atoms : the minoritary misplaced ones ( in sublattice @xmath17 ) that exhibit antiferromagnetic interaction with its nearest neighbours and the majoritary ones ( in sublattice @xmath16 ) with ferromagnetic interaction with its second neighbours
therefore , when reaching the remanence point by decreasing the applied field from saturation , the misplaced mn atoms will present inverted magnetic moment .
thus , a naive approximation is @xmath111 , where @xmath8 is the magnetic moment per atom and @xmath112 is the saturation magnetization per atom corresponding to a sample with mn atoms completely disordered over the @xmath16 and @xmath17 sublattices ( @xmath113 ) .
note that @xmath114 is symmetric arround @xmath115 as can be expected . the above equation combined with the linear behaviour @xmath116 presented in fig .
[ fig3 ] , leads to a parabolic dependence @xmath117 .
@xcite although the value of @xmath78 is unknown , it should be very small .
at least smaller than the value @xmath118 found for our sample with @xmath119 .
we can , therefore , sketch this behaviour on the @xmath109 diagram shown in fig .
[ fig12 ] .
the intersection of the two parabolas ( corresponding to @xmath120 and @xmath121 ) with the transition line found from our model gives a different estimation of the transition point .
taking into account the above arguments , from fig .
[ fig12 ] it seems reasonable to assume that the transition is located at @xmath122 .
although this value is slightly above the experimental estimation @xmath123 , the values are still close enough to suggest that , given the similarity of the effects observed in the experimental and simulated hysteresis cycles , the phenomenon may be an example of the kind of phase transition found in the model . from the model developed in the present work , we can also study the internal hysteresis loops in order to compare with the behavior found experimentally .
[ fig13 ] shows , as an example , an internal loop for a system with @xmath30 and @xmath124 . the qualitative agreement with the experimental case is extremely good , and in both cases an increase of the magnetization is found after the internal loop .
this shows that the rpm property does not hold .
actually , two conditions must be satisfied for the rpm property to be fullfilled .
the first one is the adiabatic character of the system evolution and the second one is the so - called no - passing rule ( preservation of partial ordering of metastable states by the dynamics ) @xcite .
the former condition is satisfied in our simulations ( synchronous dynamics ) and approximately in the experiments ( slow driving rate ) .
nevertheless , the existence of antiferromagnetic interactions producing inverse avalanches ( see fig . [ fig8 ] ) destroys the second condition .
the model developed above describes the behavior of a single crystal with an easy axis of magnetization ( eam ) that corresponds to the direction of the external field . in a polycrystalline sample
the situation is quite different : grains present different crystallographic orientations with respect to the applied field .
the corresponding angular distribution of eam results on two main effects on the hysteresis cycles : ( i ) the cycles become smoother ( hence the avalanches are much more difficult to detect ) and ( ii ) the hysteresis cycle changes its shape .
we can estimate this change in the case of uniaxial anisotropy .
let @xmath125 be the cycle corresponding to a single crystal oriented with the eam along the external field @xmath126 , and let @xmath93 the cycle corresponding to the polycrystal .
assuming that there is a uniform angular distribution of eam in the polycrystal , we will have : @xmath127 where @xmath128 is the differential solid angle .
this integral has to be performed over all the possible directions of the eam in the sample .
this is not straightforward and requires the knowledge of the direction of the eam . for a cu - al - mn
single crystal it is not known .
nevertheless we can obtain an approximate solution to eq .
[ anis ] as : @xmath129 where the factor @xmath130 is the differential of solid angle and @xmath131 is the maximum angle between the field and the eam .
for instance we have @xmath132 for a @xmath133 and @xmath134 eam , and @xmath135 for a @xmath136 eam .
in the extreme case of a monoclinic symmetry ( a single eam ) one will have @xmath137 .
figure [ fig14 ] shows the results of this correction on a given hysteresis cycle .
as expected the jump at the phase transition becomes broadened and extends in a certain range of fields .
in section ii we have presented low - temperature measurements of the hysteresis cycles of cu - al - mn in three samples of different mn content .
we have shown that for temperature @xmath138k and mn content @xmath139 the loops exhibit a sharp reversal of magnetization in their central part .
we have proposed that this change is related to the existence of a phase transition from a spin - glass phase to a ferromagnetic spin - glass phase . in order to gain insight into this phase transition
we have proposed ( in section iii ) a model for the study of the magnetic properties of this system , including the basic physical ingredients of the problem .
the model is based on a @xmath61 diluted ising model with exchange interactions that have been fitted to the experimental values corresponding to stoichiometric cu@xmath140almn .
the model reproduces the existence of magnetic disorder arising from the non - stoichiometry and the misplaced mn atoms , as found experimentally .
it should be remarked that we have not studied the exact equilibrium ground state of the model .
rather we have focused on the hysteresis loops that are obtained by using a synchronous dynamics evolution that drives the system through a sequence of metastable states . 1 . the change from smooth to sharp cycles by changing the mn content . in the simulations this change corresponds to a phase transition , of the same kind as the one reported in the random field @xcite and random bond ising models @xcite , and it is found at @xmath141 , which is close to the experimental estimation @xmath142 2 . the existence of long tails for large fields .
the experimental cycles are not saturated even at fields of @xmath29 t. from the model , we have deduced that these tails arise ( i ) from the existence of minoritary antiferromagnetic interactions and ( ii ) as a consequence of the polycrystalline character of the samples used in the experiments .
the magnetic moments subjected to an effective antiferromagnetic interaction are very diluted and do not exhibit any cooperative behavior .
consequently they show a reversible response to the external field . from the model ,
it is reasonable to suppose that these magnetic moments are associated to misplaced mn atoms having other mn atoms as nearest neighbors .
concerning the polycrystallinity of the samples , its main effect is the broadening of the hysteresis due to the averaging over all possible crystallographic orientations of the grains with respect to the external field .
the fact that the rpm effect is not satisfied has been found both experimentally and in the simulations .
the reason is the existence of inverse avalanches associated with the presence of antiferromagnetic interactions . in these avalanches
@xmath143 is negative , contrary to the stability conditions deduced from equilibrium thermodynamics .
the possibility of such a result is due to the fact that the process considered is not an equilibrium one ( even if it is very close to a quasiestatic process ) .
in fact the system evolves through a series of metastable states separated by large energy barriers . in the simulations metastability
arises from the fact that relaxation is performed according to the local effective field acting on each magnetic moment .
an important question to be answered is why avalanches are not observed in the experiments .
it is clear that polycrystallinity limits the maximum size of the avalanches to that of a grain . given that the typical size of the grains is of the order of 100 @xmath8 m , and the volume of the sample is @xmath7 3 mm@xmath144 , one will need a relative resolution much better than @xmath145 in the magnetization measurement .
this means a resolution better than 0.1 emu/(mol mn ) , while the actual resolution of our experimental system is of the order of 10 emu/(mol mn ) .
the inset of fig .
[ fig6 ] is illustrative of this point .
it would be very enlightening to perform experiments in single crystals , where we expect that avalanches could be observed .
firstly , as mentioned before , the phase transition found in our diluted ising model ( and in other similar diluted isind models @xcite ) is of the same kind as those found for the random field @xcite and random bond ising models @xcite .
for these last two models it is known that the phase transition in the hysteresis loop is associated with an equilibrium ground state phase transition @xcite .
it is reasonable to think that for our version of the site diluted ising model , there will also exist such a ground state transition from a completely disordered ground state ( small @xmath43 region ) to a ground state with a non - vanishing magnetization ( large @xmath43 region ) . experimentally , there are evidences that in cu - al - mn these two phases correspond to a true spin - glass phase and to a ferromagnetic spin - glass @xcite . in the model ,
whether the two phases can be catalogued as being spin - glass - like would depend on its thermal behavior .
this study is out of the scope of the present work .
secondly , the transition line in the @xmath109 diagram has been located from the maximum of the duration and the inflection of the size of the largest avalanche .
an additional estimation can be obtained by locating the points where the distribution of avalanche sizes becomes power law . except for finite - size effects
, the three estimations should be coincident .
obtaining the transition points accurately using the third method is quite difficult because it is necessary to perform non - linear fits of a power - law with an exponential correction , @xmath146 , in order to locate the point ( @xmath147 ) where the distribution changes from super- ( @xmath148 ) to subcritical ( @xmath149 ) behaviour .
it is much easier to evaluate the exponent @xmath150 characterizing the critical distribution of avalanches , which is not affected by the small exponential corrections close to the critical line .
the obtained value is @xmath151 .
this value can be compared with the values @xmath152 and @xmath153 found for the 3d - random field ising model and the 3d - random bond ising model @xcite .
this agreement of the numerical values reinforces the hypothesis that there exists universality in this kind of models exhibiting flfo phase transitions .
finally , a much deeper question concerns the nature of this critical line . some authors have stated that critical distributions of avalanches do not occur at a certain line but on a broad region due to the existence of the so - called self - organized criticality .
this theory has been proposed for externally driven complex dissipative systems with spatial and temporal degrees of freedom . according to bak
@xcite these systems naturally evolve to a critical state characterized by avalanches with no intrinsic time and length scales .
this results in power - law distributions for avalanches . in this sense
the results we have obtained from the model point in the opposite direction : we are dealing with a true critical line . indeed , for selected values of the model parameters controlling the disorder @xmath73 and @xmath43 , the system exhibits the critical behavior , as can be seen in fig .
[ fig9 ] .
we acknowledge benjamn martnez for a helpful colaboration and critical reading of the manuscript .
we also thank ll .
maosa , j. ortn and c. frontera for fruitful discussions .
this work has received finantial support from cicyt ( spain ) project number mat98 - 0315 , and cirit ( catalonia ) project number sgr119 . by `` ferromagnetic spin - glass ''
we understand a phase which , in absence of an applied field , exhibits a large degree of frozen disorder but with clusters with non - zero magnetization .
such structures have also been called cluster glasses or mictomagnetic phases .
see , for instance , c.m.hurd , contemp .
* 23 * , 469 ( 1982 ) .
actually the structure of the system in the martensitic phase is monoclinic but , due to the diffusionless character of the martensitic transition , the atomic environment is very similar in both the bcc and martensitic phase .
since there are antiferromagnetic interactions , inverse avalanches may occur and , in few occasions , a degenerate situation may appear with two energetically equivalent configurations which successively alternate .
in such cases we decide randomly which is the stable one and proceed with the field increment .
|
we study isothermal magnetization processes in the cu - al - mn intermetallic alloy .
hysteresis is observed at temperatures below the spin - freezing of the system .
the characteristics of the hysteresis cycles as a function of temperature and mn content ( magnetic element ) are obtained . at low temperature
( @xmath0 k ) a change from smooth to sharp cycles is observed with increasing mn content , which is related to the decrease of configurational disorder .
we also study a zero - temperature site - diluted ising model , suitable for the description of this cu - al - mn system .
the model reproduces the main features of the hysteresis loops observed experimentally .
it exhibits a disorder - induced critical line separating a disordered phase from an incipient ferromagnetic ground - state .
the comparison between the model and the experiments allows to conclude that the observed change in the experimental hysteresis loops can be understood within the framework of the theory of disorder - induced criticality in fluctuationless first - order phase transitions .
|
a rich diversity in network latency and packet loss rates have become essential for experimental evaluation of bittorrent and other communication protocols used in the internet .
the need for such a diversity in network latency and packet loss rates is because of the heterogeneous nature of the internet @xcite .
the heterogeneity of the internet is the primary motivation for the creation of testbeds such as planetlab @xcite .
however , due to the shared nature of the planetlab platform , the results of the experiments performed on planetlab are not reproducible @xcite . in contrast , dedicated clusters , such as grid5000 @xcite , not only offer a reproducible environment but also enable the scaling of bittorrent experiments by supporting a large number of bittorrent instances on a single machine .
the primary shortcoming of experiments performed on clusters is the absence of the diverse network latency and packet loss rates that can be found in the internet . as _ the impact of the network latency and packet loss on bittorrent performance is not known _ , there exists _ a dilemma while selecting a testbed for the bittorrent experiments_. the bittorrent protocol uses tcp to efficiently distribute the _ pieces _ of a file to a large number of peers using peer - to - peer ( p2p ) connections @xcite .
the peers contribute to the file distribution by uploading the pieces that have been downloaded . during the file download , the peers exchange control messages to select the pieces of the file to upload and the peers to whom these pieces are to be uploaded .
bittorrent also allows the users to limit the rate at which data is uploaded and downloaded .
these rate limits allow the users to restrict the network bandwidth that bittorrent can compete for during a bittorrent session . as the users can control the upload process by limiting the upload rates , and bittorrent uses control messages to decide the connections to upload to , bittorrent is inherently different from other tcp based file transfer protocols such as http and ftp . over the years bittorrent performance
has received considerable attention @xcite .
as these studies do not evaluate the interaction between tcp and bittorrent , the impact of network latency and packet loss rates on bittorrent performance is not known .
network latency and packet loss introduce a _ ramp - up _ period which is required by tcp to attain the maximum upload rate that can be achieved @xcite .
bittorrent users that limit the upload rate can therefore limit the impact of tcp ramp - up .
apart from the tcp throughput , the bittorrent performance is also dependent on the control messages exchanged by the peers .
the control messages generated by a peer can generate a delayed response at a remote peer because of the network latency and the packet losses .
the delayed response can affect the various algorithms used by bittorrent , such as the peer selection algorithm which is used to decide the peer to upload to .
due to the above reasons , the analytical models for tcp can not be directly used to provide the impact of network latency and packet loss on bittorrent performance . in this paper
, we use the download completion time , i.e , the time required to download a file using bittorrent , as a metric to study the impact of network latency and packet loss . in our experiments , we observe that network latency and packet loss have a marginal impact ( less than 15% ) on the download completion time of a file .
we first study the impact of network latency without packet loss .
network latency causes not only delays in receiving control messages but also tcp ramp - up .
we therefore study the impact of network latency by studying the impact of the delays in receiving control messages and tcp ramp - up on the download completion time .
we observe that the download completion time , when the round - trip time ( rtt ) between _ any two peers _ in the torrent is 1000 ms , is not more than 15% of the download completion time when the rtt is 0 ms .
we then study the impact of network latency with packet loss .
we observe that an rtt of 400 ms _ between any two peers _ and a packet loss rate of 5% does not increase the download completion time by more than 15% of the download completion time observed when the rtt is 0 ms and the packet loss rate is 0% .
we also observe that an rtt of 1000 ms _ between any two peers _ with a packet loss rate of 5% can increase the download completion time by more than 15% of the download completion time observed when the rtt is 0 ms and the packet loss rate is 0% . as an rtt greater than 400 ms between _ any two peers _
is unrealistic , our results show that for upload rates typically seen in the internet , the download completion time is not sensitive to the network conditions that can be found in the internet ; _ dedicated clusters , such as grid5000 , can be used to perform bittorrent experiments_. the remainder of this paper is structured as follows .
the network topologies and the technique used to emulate the latency and packet loss are presented in section [ sec : methodology ] .
we first present the impact of latency without packet loss on download completion time . in section [ sec : homogeneouslatency ] , we emulate the same latency between any pair of peers in the torrent .
we use this topology to study the impact of tcp ramp - up and the impact of the delays in receiving the bittorrent control messages on the download completion time . in section
[ sec : heterogeneouslatency ] , we emulate torrents to study the impact of network latency when the condition of same network latency between any two peers is relaxed
. we then study the impact of network latency and packet loss in section [ sec : impactlosses ] .
we finally conclude in section [ sec : conclusion ] .
in this paper , we use the terminology used by the bittorrent community .
a _ torrent _ , also known as a bittorrent session or a swarm , consists of a set of peers that are interested in having a copy of the given file .
a peer in a torrent can be in two states : the _ leecher _ state when it is downloading the file , and the _ seed _ state when it has a copy of the file being distributed .
the peers distribute the file in chunks called _ pieces _ ; a piece is further split into _ blocks _ to facilitate the piece upload and download .
a peer is said to _ unchoke _ a remote peer if it is uploading the blocks of a piece .
a _ tracker _ is a server that keeps track of the peers present in the torrent . for our experiments ,
we use a private torrent with one tracker , one initial seed ( henceforth called the seed ) , and 300 leechers .
we use the _ download completion time _ , the time required by the leechers to download the file distributed using bittorrent , as the metric to study the impact of network latency and packet loss .
all the experiments were performed using an instrumented version of the _ bittorrent mainline client _
@xcite on machines running linux as the host operating system .
the bittorrent mainline client internally uses tcp .
we used tcp cubic @xcite , the default tcp implementation for the current series of the linux kernel , for our experiments . the latest version of utorrent , a bittorrent client , is based on utp which uses udp as the transport layer protocol @xcite . as in the case of tcp
, utp has a window based congestion control mechanism .
the design of utp also ensures that utp ramp - up is not faster than tcp @xcite .
the control messages generated by utorrent using utp are similar to those that are present in bittorrent clients that use tcp .
hence , we believe that the results presented in this paper are valid for bittorrent clients using utp . in this section , we first present the various scenarios used in the experiments .
we then present the procedure to emulate the network latency and packet loss .
this is followed by an overview of grid5000 , the experimental platform we used . finally , we present the parameters used while performing the experiments .
we now enumerate the scenarios used to study the impact of network latency and packet loss on the download completion time of a file
. 1 . _ scenario of homogeneous latency .
_ in this scenario , we emulate _ a fixed network latency between any two peers _ in the torrent .
the fixed network latency , though unrealistic , provides a controlled environment to study the impact of network latency on the download completion time of a file .
we use this scenario to get an insight on the threshold of network latency beyond which the network latency affects the download completion time .
the scenario also gives the download completion time when the maximum latency between any two peers in a torrent in known .
the results of this study can also be used to give the impact of network latency when the hosts are geographically distributed and are connected with links that have negligible packet loss .
scenario of heterogeneous latency .
_ in this scenario we relax the condition of fixed network latency to emulate a more realistic network topology .
we use this scenario to confirm that the results obtained in the scenario of homogeneous latency are valid even when the condition of fixed network latency between any two peers is relaxed . for this scenario ,
we group peers that have the same network latency among themselves in an emulated autonomous system ( as ) .
we assume these ases to be fully meshed and that the inter - as latency is greater than the intra - as latency ; we assume the network latency between the peers in a given as is the same .
bittorrent allows its users to limit the upload and download rate .
these rate limits restrict the network bandwidth that bittorrent can compete for with other applications such as web browsers . in this paper
, we do not place any restrictions on the download rate .
we set the upload rate limit of the peers from a wide range of values , from 10 kb / s to 100 kb / s .
we use this range of upload rates as choffnes _ et al . _
@xcite show that 90% of the hosts present in public torrents upload at rates that are smaller than 100 kb / s . in our experiments
, we assume the same upload rate limit at all the leechers while studying the impact of network latency and packet loss .
we use the following torrent configurations to set the limit on the upload rate of the peers . 1 . _ slow seed and slow leechers .
_ in this scenario , we assume all the peers in the torrent have a low upload rate .
we also assume the same upload rate at the seed and leechers .
we performed two experiments for this scenario ; we limit the upload to 10 kb / s for the first experiment and 20 kb / s for the second experiment .
fast seed and slow leechers . _
some torrents have seeds that upload faster than the leechers .
public torrents are also known to have leechers that are capable of high upload and download rates . in torrents where the seed favors fast leechers @xcite , these leechers are able to download the file faster than their slower peers .
as these leechers are capable of downloading pieces faster than their slower peers , they act like a fast seed to the slow peers in their peer - set .
we emulate torrents that have a fast seed by limiting the upload rate of the seed to 50 kb / s and the upload rate of the leechers to 20 kb / s .
fast seed and fast leechers .
_ we perform these experiments to emulate torrents where all the peers are capable of high upload rates .
we performed two experiments for this scenario ; we limit the upload to 50 kb / s for the first experiment and 100 kb / s for the second experiment .
we use the network emulation ( netem ) module for the linux kernel @xcite to emulate the network latency and packet loss between the peers .
the netem module operates between the tcp / ip implementation and the device driver for the network device . in our experiments ,
the peers used the ethernet and the loopback device to communicate with each other .
the netem module emulates network latency by en - queuing the packets at the ingress interface and the egress interface of a network device ; packets losses are emulated by dropping the packets at the ingress and egress interfaces . when the loss is on the egress interface , due to cross - layer optimizations performed by the tcp implementation in linux , tcp retransmits the packet without considering it as a loss in the network . to avoid such retransmissions , we introduce packet losses on the ingress interface of the network device .
a shortcoming of netem is that the network latency and packet losses are _ inherently _ bound to the network device .
netem can be used to emulate network latency and losses for each connection , however such topologies are hard to verify and manage . in this paper
we set the network latency and packet loss rates for a particular network device .
the network latency between a pair of peers is the sum of the network latency at the devices used ; the loss rate seen by a packet is the loss rate at the ingress interface of the destination .
we use the publicly available iplane measurements @xcite to obtain the values of the network latency and loss rate to emulate .
figure [ fig : iplanemeasurements ] shows the distribution of rtt and loss rate observed in 5 iplane samples ; each sample contains the measurement of about @xmath0 links . in figure
[ fig : latencyinternet ] , we observe that 98% of the links have an rtt less than 400 ms , and 99.8% of the links have an rtt less than 1000 ms .
we use these results to emulate a wide range of rtt values , from 0 ms to 1000 ms , between a pair of peers in the torrent . in figure
[ fig : lossrateinternet ] , we observe that 95% of the links probed have a loss rate less than 0.05 , i.e , 5% packet loss .
we emulate a 5% packet loss on the ingress of each interface of the machines while studying the impact of packet losses .
we performed our experiments on the grid5000 experimental testbed @xcite .
grid5000 consists of a grid of clusters that are geographically distributed across france .
each cluster consists of several machines connected by a dedicated and high speed lan ; these clusters are inter - connected by high speed links .
we performed our experiments on _ a single cluster of grid5000_. in this cluster , we observe an rtt of less than 1 ms between a pair of machines when network latency is not emulated . due to the very small network latency and the absence of packet loss between the machines in the lan
, the cluster does not reflect the network conditions present in the internet . unlike other testbeds such as planetlab ,
the machines in the cluster are not shared .
grid5000 thus provides a reliable and robust platform for performing reproducible experiments .
we scale our experiments by running 100 leechers on a single machine of the cluster .
we performed the following test to ensure that the machines can support up to 100 leechers while emulating the desired network latency .
we distribute a 50 mb file in a private torrent with a single seed and a single tracker .
the seed and tracker ran on one of the machines of the cluster .
we used another machine in the cluster to run the leechers .
we limit the upload rate of each of the peers to 100 kb / s and we vary the number of leechers running on a machine from 2 to 100 . to study the impact of the number of leechers , we monitor the rtt estimate of tcp in the following manner .
all the peers in the torrent use the socket interface of tcp to communicate with each other .
the ` send ` method of this interface is used by the peers to send data to other peers in the torrent . on each call of ` send ` , we sample the rtt estimate of tcp by using the ` tcp_info ` option of the ` getsockopt ` method .
figure [ fig : impnumnodes ] shows the average rtt estimate of tcp when we emulate an rtt of 0 ms , 400 ms , and 1000 ms , between the peers ; the error bars represent the minimum and maximum rtt estimate observed in five iterations .
we observe that the number of leechers running on a given machine has a less than 15% impact on the average rtt estimate of tcp .
the netem module buffers the tcp frames which are in flight for a time period equal to the rtt being emulated .
an rtt of 1 second ( 1000 ms ) for an upload rate of 100 kb / s would require 100 frames of 1 kb to be in flight . for our experiments , we use a buffer size of @xmath1 frames to support 1000 frames of each of the 100 peers to be in flight .
we perform our experiments in a private torrent with one tracker , one initial seed ( henceforth called the seed ) , and a flash crowd of 300 leechers .
we distribute a 50 mb file in this torrent . for our experiments , we assume that the peers remain in the torrent until all the leechers have finished downloading the file .
as shown in figure [ fig : grid5ktopology ] , we use one machine for the tracker and the seed .
we use three machines for the 300 leechers ; each machine runs 100 instances of the leechers .
a pair of peers in the torrent use either the loopback interface or the ethernet interface to communicate with each other .
the maximum segment size ( mss ) of tcp specifies the maximum payload length that can be exchanged over a connection @xcite .
one factor that contributes to the mss is the maximum transmission unit which is typically 1500 bytes for devices used in lans .
the mtu is set to 16436 bytes for the loopback interface on many linux distributions . to ensure that the payload length exchanged by the peers does not depend the interface used by the connection
, we set the mtu on the loopback interface to 1500 bytes . despite this limit
, we observed that a significant number of tcp segments have a size larger than the mss negotiated during connection establishment .
we observe these large segments because of tcp segmentation offloading ( tso ) which is enabled by default in the 2.6 series ( the current series ) of the linux kernel .
tcp segmentation offloading ( tso ) enables the host machine to offload some of the tcp / ip implementation , such as segmentation , and calculation of ip checksum , to the network device .
tso also supports the exchange of data in frames of sizes that can be greater that the underlying mtu size @xcite .
the increase in the frame size can result in significant improvement in throughput ; the improvement depends on various factors such as cpu processing power and the amount of data being transferred @xcite .
clusters , such as those present in grid5000 , include hosts that support tso .
we now study the impact of rtt on the tcp payload length when tso is enabled .
we use four machines to create a private torrent with 300 leechers , one tracker , and one initial seed , to distribute a 50 mb file .
we assume the same rtt between any two peers for the results presented in figure [ fig : pktlentsoenabled ] and figure [ fig : pktlentsodisabled ] .
we also limit the upload rate to 20 kb / s .
figure [ fig : pktlentsoenabled ] shows the distribution of payload length for different rtt values when tso is enabled ; the results present the distribution of the payload lengths observed in 5 iterations .
we observe that an increase in the network latency between the peers results in an increase in the number of tcp segments with a large payload length .
this was observed when the mtu was set to 1500 bytes indicating that tso can result in payload lengths greater than the mtu value specified .
figure [ fig : pktleninternet ] shows the payload lengths from the publicly available traces of the internet traffic observed in the wide backbone @xcite ; the values presented are from the sample taken on the wide backbone on november 29 , 2009 . in figure
[ fig : pktlentsoenabled ] and figure [ fig : pktleninternet ] , we observe that the maximum payload length of the packets sent over the internet is smaller than the maximum payload length observed in grid5000 when tso is enabled .
we observe small payload lengths in figure [ fig : pktleninternet ] because hardware support on all the intermediate devices is essential for exchange of large segments .
when tso is disabled and the mtu is set to 1500 bytes , figure [ fig : pktlentsodisabled ] shows the distribution of the payload length for different rtt values ; the results present the distribution of the payload lengths observed in 5 iterations .
we observe that when tso is disabled and the mtu is set to 1500 bytes , the maximum payload length is similar to that observed in the internet .
we set the mtu to 1500 bytes on the loopback interfaces and disabled tso on the ethernet and loopback interface for the experiments presented in the section [ sec : homogeneouslatency ] and section [ sec : heterogeneouslatency ] .
the outcome of the experiments with tso enabled are discussed with the impact of packet loss in section [ sec : impactlosses ] .
] in this section , we assume the same network latency between any two peers in the torrent because _ this scenario gives the worst case impact of a given network latency for a given upload rate limit_. the same network latency results in the same rtt between any two peers in the torrent because the sum of the other delays , such as queuing delays at the intermediate routers , is less than 1 ms in a grid5000 cluster . in each experiment , we choose an rtt to emulate from a wide range of values , from 0 ms to 1000 ms . in section [ sec : emullatency ] , we observed that 99.8% of the internet links have an rtt less than 1000 ms .
_ an rtt of 1000 ms between any two peers thus gives a worst case impact of the network latency that can be observed in the internet_. we use a wide range of values , from 10 kb / s to 100 kb / s , to set the upload rate limit of the peers in the torrent ; we assume the same upload rate limit for all the leechers in the torrent .
for a given upload rate limit , figure [ fig : symmetricdownloadcompletion ] shows the impact of the rtt between any two peers on the download completion time . each point on the plot
is the average download completion time in 10 iterations ; the error bars indicate the minimum and maximum download completion time observed in 10 iterations .
the two factors that can be affected by network latency and cause an increase in the download completion time are tcp ramp - up and the delays in receiving the control messages .
we study the impact of network latency on the download completion time by studying the impact of network latency on these two factors in section [ sec : impactrampup ] and section [ sec : controlmessagesimpact ] respectively . in figure
[ fig : symmetricdownloadcompletion ] , for a given upload rate limit from 10 kb / s to 100 kb / s , we observe that an rtt of up to 1000 ms between any two peers does not increase the average download completion time by more than 15% of the average download completion time when the rtt is 0 ms .
when the maximum upload rate of the seed and leechers is limited to 10 kb / s , figure [ fig : symmetricdownloadcompletion ] shows that the download completion time increases when the rtt is greater than 200 ms . in section [ sec : impactrampup ] , we show how tcp ramp - up is responsible for this increase in the download completion time .
when the upload rate of all the peers is limited to 20 kb / s , in figure [ fig : symmetricdownloadcompletion ] , we observe that the download completion time is not a monotonous function of the rtt . peers having an rtt of 1000 ms have a lower download completion time compared to peers having an rtt of 400 ms .
we show that this is the impact of network latency on the delays in receiving bittorrent control messages in section [ sec : controlmessagesimpact ] .
an rtt of 1000 ms between any two peers gives the worst case impact of network latency .
figure [ fig : symmetricdownloadcompletion ] thus shows that _ network latency has a marginal impact , less than 15% , on the average download completion time for the upload rates observed in public torrents_. the bittorrent application uses the ` send ` system call of the socket interface of tcp to upload the pieces of a file .
bittorrent limits the upload rate for a given tcp connection by periodically calling the ` send ` system call .
the time between successive calls of ` send ` is high when the upload rate is low .
each call of ` send ` typically results in creation of a tcp segment which is transmitted .
if the acknowledgment for the transmitted segment arrives before the subsequent call of ` send ` , then the upload rate of the application does not require a congestion window ramp - up . _
a ramp - up is required on the tcp connections where the time between successive calls of _ ` send ` _ is smaller than the rtt_. bittorrent also enables a peer to simultaneously upload pieces of the file to many peers of a torrent in parallel . an increase in the number of parallel unchokes results in an increase in the time between successive calls of ` send ` .
this is because the upload rate limit is divided over the connections used for the unchoke .
we present the distribution for the time between successive calls of the ` send ` at all the leechers ( over 5 iterations ) in figure [ fig : iatleecher ] ; the rtt between any two peers is 0 ms in figure [ fig : leecher0 ] and 400 ms in figure [ fig : leecher400 ] . for an upload rate of 10 kb / s , in figure [ fig : leecher0 ] and figure [ fig : leecher400 ] , we observe that the time between successive ` send ` calls is greater than 200 ms for a significant number of calls . for these calls , when the rtt between the peers is less than or equal to 200 ms a ramp - up shall not be required . due to the absence of ramp - up , in figure [ fig : symmetricdownloadcompletion ] for an upload rate limit of 10 kb / s , we observe that an rtt less than or equal to 200 ms does not increase the average download completion time by more than 5% of the average download completion time observed when the rtt is 0 ms .
for the upload rate limit of 10 kb / s , an rtt larger than 200 ms may require a tcp ramp - up .
similarly , for the upload rate limits of 20 kb / s , 50 kb / s , and 100 kb / s , in figure [ fig : iatleecher ] , we observe that ramp - up may be required for an rtt greater than 120 ms , 20 ms , and 20 ms respectively . hence , for an upload rate limit in the range of 10 kb / s to 100 kb / s , tcp ramp - up will be required for an rtt greater than 200 ms . in figure
[ fig : symmetricdownloadcompletion ] for an rtt of up to 1000 ms , we observe that tcp ramp - up does not increase the average download completion time by more than 15% of the download completion time observed the rtt is 0 ms .
tcp ramp - up has a marginal impact , less than 15% , on the average download completion time for the upload rates observed in public torrents_. in figure [ fig : symmetricdownloadcompletion ] for an upload rate of 20 kb / s , we observe that the download completion time is not a monotonous function of the rtt .
the download completion time when the rtt is 1000 ms is smaller than the download completion time when the rtt is 400 ms .
we now present the reasons for this behavior .
we would like to comment that this discussion is specific to the implementation of the bittorrent client we used in the experiments and that _ these observations may not be true for other bittorrent clients that are available_. once every 10 seconds , a seed selects from its peer - set a set of leechers to unchoke .
the leechers are selected based on their download rate and time since the start of their last unchoke . during an unchoke , the leecher downloading the piece requests for multiple blocks of a piece to pipeline the blocks for the unchoke .
the number of blocks requested is a function of the estimated download rate at the leecher .
this estimate of the download rate is a moving average and it grows slowly to attain the rate at which the seed uploads to the given leecher .
this growth is even slower if the connection requires a tcp ramp - up .
the slow growth in the estimated download rate thus results in a slow increase in the number of blocks that are in the pipeline at the seed .
we observe that this slow increase , along with the latency in receiving the block requests , results in time periods during which the seed unchoking the leecher is idle , i.e. , awaiting block requests . if the leecher selection algorithm of the seed is invoked during these _ idle periods _ , then the leecher _ will not be selected _ for an unchoke resulting in an abrupt termination of the unchoke .
we observe that this abrupt termination of the unchoke occurs frequently when the upload rate limit is 20 kb / s and when the rtt between the peers is greater than 120 ms and less than 800 ms .
we now show how the abrupt termination of an unchoke affects the availability of pieces at the leechers .
once a piece is downloaded , the leecher can then upload this piece to other leechers in its peer - set .
after the piece download , the leecher sends a _ have _ message with the piece identifier to the peers in its peer - set .
the _ have _ message indicates that the piece can be downloaded from this leecher . in figure
[ fig : rtthaveimpact ] , we use the _ have _ messages sent by the leechers to show the evolution of the average number of pieces ( over 10 iterations ) that are available at the leechers . for reasons mentioned above
, when the upload rate limit of the seed and leechers is 20 kb / s , an rtt of 400 ms between the seed and a leecher typically results in the abrupt termination of the unchoke .
the abrupt termination results in an incomplete download of a piece and affects the availability of pieces as shown in figure [ fig : have20 ] .
we do not observe abrupt termination of the unchoke when the rtt is 0 ms and when the rtt is 1000 ms .
hence , the average download completion time when the rtt is 400 ms is greater than the average download completion observed when the rtt is 1000 ms .
we do not observe abrupt termination of unchokes when we have a fast seed ( upload rate limited to 50 kb / s ) in a torrent with slow leechers ( upload rate limited to 20 kb / s ) ; figure [ fig : have5020 ] shows the evolution of pieces for this torrent configuration .
we observe a similar evolution for the _ have _ messages when the upload rate at the peers is limited to 50 kb / s and 100 kb / s . in this section , we emulated a large range of rtt values , from 0 ms to 1000 ms , to study the impact of network latency on the download completion time .
we use an rtt of 1000 ms to give the worst case impact of network latency for a given upload rate .
we observe that an rtt of up to 1000 ms between any two peers has a marginal impact , less than 15% , on the average download completion time of a file .
for an upload rate of 20 kb / s , we observe that the download completion time is not a monotonous function of the rtt .
this behavior emphasizes _ that the models for tcp throughput can not be directly used to study the impact of network latency on the time required to download a file using bittorrent_. in the next section , we relax the condition of same latency between any two peers in the torrent to confirm that the scenario of same latency gives us the worst case impact of network latency .
we now present results to confirm that the observations made in section [ sec : homogeneouslatency ] are valid even when the condition of fixed latency is relaxed .
we relax the condition of same latency between any two peers by emulating ases in the following manner .
public torrents have peers that are spread out geographically .
a pair of peers in the same as typically have a smaller network latency compared to a pair of peers that are present in different ases . for our experiments
, we use a private torrent with peers distributed in emulated ases . for our experiments ,
we emulate an as using one machine .
we assume the same intra - as latency and we also assume that the intra - as latency is less than the inter - as latency .
we assume that all the ases are fully meshed .
.latency values on the loopback and ethernet device while emulating an as on a machine .
[ cols="^,^,^ " , ] in section [ sec : homogeneouslatency ] , we observed the impact of network latency when tso is disabled and packet loss is not emulated .
we observe that for an rtt of up to 1000 ms , tcp ramp - up and the delays in receiving the control messages have a marginal impact , less than 15% , on the average download completion time of a file . when tso in enabled and packet losses are not emulated , we observe that the impact of network latency on the download completion time is similar to that observed in the case of homogeneous latency , which has been presented in section [ sec : homogeneouslatency ] .
these results show that _ bittorrent experiments performed on clusters that support tso shall produce results that are similar to those performed on clusters that do not support tso_. we now discuss the scenario of homogeneous latency with packet loss when tso is disabled .
while emulating a packet loss , each packet en - queued by netem is dropped with a probability controlled by the loss rate . a packet loss result in retransmission of the packet by the source .
the source reduces the congestion window in response to the loss and then ramps - up its congestion window to attain the desired upload rate .
we observe that for an upload rate limit of 20 kb / s and 50 kb / s , an rtt of 1000 ms and a loss rate of @xmath2 between any two peers does not increase the average download completion time by more than @xmath3 of the average download completion time observed when _ neither latency nor packet loss was emulated_. however , when the upload rate limit is 10 kb / s or 100 kb / s , the rtt between any two peers is 1000 ms , and when the loss rate is 5% , we observe that the average download completion time is more than 15% of the download completion time when the rtt is 0 ms and the loss rate is 0% .
an rtt of 1000 ms between any two peers is unrealistic as 99.8% links probed by iplane have an rtt less than 1000 ms .
however , _ an rtt of up to 400 ms is realistic _ as 98% of the links probed by iplane have an rtt less than 400 ms . for the upload rate limit from 10 kb / s to 100
kb / s , when the rtt between any two peers is 400 ms and the loss rate is 5% , we observe that the download completion time is not more than 15% of the download completion time observed when the rtt is 0 ms and loss rate is 0% . in section [ sec : heterogeneouslatency ] we observe that the scenario of homogeneous latency gives an upper bound on the impact of a given network latency .
therefore , the results presented in table [ tab : impactlosstso ] show that , for an upload rate limit from 10 kb / s to 100 kb / s , _ a loss rate of up to 5% and an rtt of up to 400 ms between any two peers in the torrent has a marginal impact , less than 15% , on the average download completion time of a file . _
in this paper we present the impact of network latency and packet loss on the download completion time of a file distributed using bittorrent .
we use the download completion time as the metric for evaluation because the bittorrent users are primarily interested in the download completion time of a file .
we first studied the impact of network latency on the download completion time . for a given upload rate limit
, we emulated the same latency among the peers to give a worst case impact of network latency on the download completion time .
the download completion time can be affected by tcp ramp - up and the delays in receiving the bittorrent control messages .
we therefore studied the impact of network latency on the download completion time by studying the impact of network latency on tcp ramp - up and the delays in receiving the control messages . for our experiments
, we varied the upload rate limit from 10 kb / s to 100 kb / s and the rtt from 0 ms to 1000 ms .
we observe that the tcp ramp - up and the delays in receiving the bittorrent control messages only have a marginal impact , less than 15% , on the average download completion time .
the high rtt values used in our experiments also emulate torrents with peers that are not only geographically apart but also connected with high capacity links that support the bittorrent upload rate without causing congestion .
our results show that _ experiments performed on grid5000 give results similar to those performed on testbeds such as planetlab that have geographically distributed hosts that are connected by high capacity links_. we also study the impact of network latency on the delays in receiving the control messages ; this impact can not be captured using the traditional models for tcp throughput .
we then studied the impact of packet loss by emulating a loss rate of 5% . for the upload rates seen in public torrents , from 10 kb / s to 100 kb / s
, we observe that realistic rtt values of up to 400 ms and a packet loss rate up to 5% , have a marginal impact , less than 15% , on the average download completion time .
we performed our experiments over a wide range of rtt values , from 0 ms to 1000 ms , and a wide range of packet loss rates , from 0% to 5% , to study the impact of network latency and packet loss .
our results show that _ experiments can be performed on dedicated clusters , such as those present in grid5000 , without explicitly emulating latency and packet loss between the peers in a torrent_. we also studied the impact of using devices that support tso on the outcome of bittorrent experiments performed on clusters .
our results show that bittorrent experiments performed on clusters that support tso produce results that are similar to those performed on dedicated clusters that do not support tso .
our main conclusion is that , for upload rates seen in public torrents , network latency and packet loss have a marginal impact on the download completion time of a file , hence , dedicated clusters such as grid5000 can be safely used to perform realistic and reproducible bittorrent experiments .
experiments presented in this paper were carried out using the grid5000 experimental testbed , being developed under the inria aladdin development action with support from cnrs , renater and several universities as well as other funding bodies ( see https://www.grid5000.fr ) .
m. dischinger , a. haeberlen , i. beschastnikh , p. k. gummadi , and s. saroiu , `` satellitelab : adding heterogeneity to planetary - scale network testbeds , '' in _
sigcomm 08 : proceedings of the acm sigcomm 2008 conference on data communication_.1em plus 0.5em minus 0.4emnew york , usa : acm , 2008 , pp .
315326 .
d. qiu and r. srikant , `` modeling and performance analysis of bittorrent - like peer - to - peer networks , '' in _
sigcomm 04 : proceedings of the 2004 conference on applications , technologies , architectures , and protocols for computer communications_.1em plus 0.5em minus 0.4emnew york , ny , usa : acm , 2004 , pp .
367378 .
a. r. bharambe , c. herley , and v. n. padmanabhan , `` analyzing and improving a bittorrent networks performance mechanisms , '' in _
infocom 2006 .
25th ieee international conference on computer communications .
_ , april 2006 , pp . 1 12 .
a. legout , n. liogkas , e. kohler , and l. zhang , `` clustering and sharing incentives in bittorrent systems , '' in _ proceedings of the 2007 acm sigmetrics international conference on measurement and modeling of computer systems_.1em plus 0.5em minus 0.4emacm , 2007 , pp .
301312 .
h. madhyastha , t. isdal , m. piatek , c. dixon , t. anderson , a. krishnamurthy , and a. venkataramani , `` iplane : an information plane for distributed services , '' in _ proceedings of osdi _ , 2006 , pp .
367380 .
d. freimuth , e. hu , j. lavoie , r. mraz , e. nahum , p. pradhan , and j. tracey , `` server network scalability and tcp offload , '' in _
atec 05 : proceedings of the usenix annual technical conference_.1em plus 0.5em minus 0.4emberkeley , ca , usa : usenix association , 2005 , pp . 1515 .
|
network latency and packet loss are considered to be an important requirement for realistic evaluation of peer - to - peer protocols .
dedicated clusters , such as grid5000 , do not provide the variety of network latency and packet loss rates that can be found in the internet .
however , compared to the experiments performed on testbeds such as planetlab , the experiments performed on dedicated clusters are reproducible , as the computational resources are not shared . in this paper , we perform experiments to study the impact of network latency and packet loss on the time required to download a file using bittorrent . in our experiments ,
we observe a less than 15% increase on the time required to download a file when we increase the round - trip time between any two peers , from 0 ms to 400 ms , and the packet loss rate , from 0% to 5% .
our main conclusion is that the underlying network latency and packet loss have a marginal impact on the time required to download a file using bittorrent .
hence , dedicated clusters such as grid5000 can be safely used to perform realistic and reproducible bittorrent experiments .
bittorrent , experiment , performance , clusters , latency , loss rate .
|
the dominant thermal flux during type i bursts and superbursts offer an ideal opportunity for studying neutron star ( ns ) properties through spectral measurements of their photospheres ( e.g. london , taam & howard 1984 , 1986 ; foster , ross & fabian 1987 ) . though never confirmed , the reports by waki et al .
( 1984 ) , turner & breedon ( 1984 ) , nakamura , inoue & tanaka ( 1988 ) and magnier et al .
( 1989 ) of 4 kev absorption lines during type i bursts motivated theoretical work on the formation of the fe ly@xmath0 line and edge ( e.g. foster et al . 1987 ; day , fabian & ross 1992 ) .
cottam , paerels & mendez ( 2002 ; hereafter cpm ) observed the accreting ns using the reflection grating spectrometer ( rgs ) aboard the xmm - newton satellite and claimed detection of hydrogen - like and helium - like fe absorption features from the @xmath6 h@xmath0 transitions during type i bursts .
letter _ is an initial study on the implications of the hydrogen - like feature and predictions for heavy element abundances near the photosphere .
the helium - like feature will be left for future work .
xmm / newton observed for 335 ks during which 28 type i x - ray bursts were seen .
cpm combined the burst spectra and split them into a high luminosity ( @xmath7 ) part ( from the initial rise and fall ) and a low @xmath7 part , which covers the fading to the baseline @xmath7 . in the high @xmath7 part
, the detected absorption line was identified as the h@xmath0 transition of hydrogenic fe . in the low @xmath7 spectrum , a line identified as the he - like h@xmath0 transition was seen . for these identifications ,
cpm measured the ns s gravitational redshift at @xmath8 , giving @xmath9 from the line centroid ( neglecting the rotational corrections of fujimoto 1985 ; zel & psaltis 2003 ) .
these spectra have photospheric temperatures @xmath10 and @xmath11 , using the measured color temperatures ( @xmath12 and @xmath13 ; cpm ) and correcting for the redshift and spectral hardening from comptonization ( london et al . 1986 ; madej 1997 ) . in 2 , we calculate the fe abundance when accretion is active , showing that proton spallation of fe ( bildsten , salpeter & wasserman 1992 ; hereafter bsw ) naturally yields @xmath14 for accretion of solar metallicity material .
the spallation also produces large columns of elements with @xmath3 , which could be observable via their photoionization edges during the cooling phase of the burst or a radius expansion burst . in 3 ,
we show that stark broadening ( paerels 1997 ) dominates the h@xmath0 line equivalent width and that the measured equivalent width ( properly redshifted to the ns surface ) of @xmath15 ev implies an fe column of @xmath16 depending on line depth . however , these abundance calculations are highly uncertain due to non - lte ( nlte ) effects and resonant line transport which we believe play an important role in the line formation .
we summarize in 4 and note that the radiative flux during the burst can levitate hydrogenic fe .
the continuum photosphere for the 1.276 kev photons of the fe h@xmath0 transition is at an electron column density of @xmath17 , where @xmath18 is the thomson cross section .
if @xmath19is uniformly abundant at the solar value ( @xmath20 by number ; grevesse & sauval 1999 ) above the continuum photosphere at 1 kev then @xmath21 . joss ( 1977 ) showed that convection during the burst can not bring burning fluid elements to the photosphere as the entropy there exceeds that in the burning layers by large factors .
hence , we look to fe in the accreted material as the likely source of the photospheric elements .
bsw studied the deposition and separation of accreting elements , and we use their results here .
imagine first that accretion halts during the burst . in a pure h atmosphere ,
the upward pointing atmospheric electric field is @xmath22 , so that the force on fe is @xmath23 .
the counter - balancing drag force on the drifting fe nucleus is @xmath24 , where @xmath25 is the electron - proton coupling from bsw .
the downward drift speed is @xmath26 so that it takes @xmath27 for fe to fall to a column depth @xmath28 .
hence , if accretion halts , fe would sink beneath the photosphere in much less than the burst duration . given the sub - eddington peak luminosity , accretion likely remains active during the bursts .
kuulkers et al .
( 2002 ) have shown this to be the case for gx 17 + 2 . the persistent flux from when type i bursts occur implies @xmath29 ( gottwald et al . 1986 ; cpm ) .
in order to obtain a lower limit on the accretion rate per unit area we assume that the accretion covers the whole ns , so that @xmath30 .
the measured redshift implies @xmath9 ( cpm ) , so that the ns s surface is inside the last stable orbit regardless of its rotation rate ( kluzniak & wagoner 1985 ; kluzniak , michelson & wagoner 1990 ) .
the resulting accretion gap allows for the accreting beam to strike the atmosphere with a velocity comparable to free - fall ( kluzniak & wilson 1991 ) , realizing the coulomb stopping scenario of zeldovich and shakura ( 1969 ) .
the integration of a free trajectory from the last stable orbit to the surface yields an impact angle @xmath31 and kinetic energy of @xmath32 mev / nucleon . since some angular momentum and energy can be lost in the accretion gap , we presume that @xmath33 and fix the incident energy at @xmath34 mev / nucleon , resulting in stopping a particle of charge @xmath35 and mass @xmath36 after traversing a vertical column @xmath37 .
bsw showed the repercussions of the @xmath38 scaling : ( i ) the incident electrons stop first , ( ii ) hydrogen and helium penetrate furthest to @xmath39 , and ( iii ) all heavier elements stop prior to the h / he beam .
for example , fe stops at @xmath40 and is spalled by the energetic protons ( of current @xmath41 ) until it has sunk to safety ( i.e. @xmath42 ) .
this is the mechanism shown by bsw to destroy accreted heavy elements .
grazing incidence reduces the distance that must be fallen to reach safety and increases the surviving fraction of the incident cno elements relative to that in bsw .
the infalling fe could have bound electrons as it spirals into the ns surface .
however , the combination of varying doppler shifts and the short flight time compared to the residence time in the atmosphere ( derived below ) means that the inflow is an unlikely environment for narrow line formation .
we thus focus our discussion on the thermalized fe in the atmosphere . at 200 mev per nucleon ,
@xmath43% of the incident fe suffer a charge changing nuclear reaction before traversing @xmath44 and thermalizing .
accretion is thus depositing thermal fe into the ns atmosphere at @xmath45 at a rate @xmath46 .
for the spallation cross section ( i.e. that cross section for @xmath47 to be chipped to lower z ) @xmath48 ( silberberg et al . 1998 ) the fe lifetime to spallation from the proton beam is @xmath49 .
the fe must sink to a depth of @xmath50 for safety , so that when @xmath51 ( or @xmath52 ) , destruction dominates settling ( i.e. @xmath53 ) .
hence , as long as @xmath54 during the burst exceeds 1% of that occurring prior to the burst , the destruction by the proton beam completely determines the fe residence time in the upper atmosphere .
this seems a condition easily satisfied for many bursters and greatly simplifies the calculation of @xmath55 as well as the stark broadening discussed in 3 , since , in this limit
, the fe is present only near the stopping point . because of the complicated interplay of the currents and forces in the upper atmosphere ( see bsw ) , the local abundance near the stopping point ranges from @xmath56 times the incident abundance .
the proton spallation reduces the fe abundance as it sinks , imprinting an abundance profile @xmath57 , which is an exponential decay with depth .
the fe rich region above the fe poor layer puts a small inverted gradient in the mean molecular weight ( @xmath58 ) that would trigger a rapid overturn in an isothermal layer if @xmath59 .
gradients this large will not be reached as long as the accreting material has an abundance less than ten times solar .
doubly diffusive instabilites might still occur , but those timescales are likely no shorter than @xmath60 .
hence , for finding @xmath55 , we only need to balance the fe destruction rate ( neglecting the @xmath0 beam is less than a 20% effect , as those cross sections are rarely larger than twice those induced by protons ; ferrando et al .
1988 ) @xmath61 , with the fe deposition rate @xmath62 , giving @xmath63 due to neutron knockout reactions , about 10% of those fe nuclei are @xmath64 and 5% are @xmath65 .
the fe column is comparable to the uniform solar abundance calculation because @xmath66 .
since many spallation reactions occur as the nuclei sink , we also found the abundances of all lower @xmath67 and @xmath68 nuclei .
we calculated this nuclear cascade by following the evolution of each isotope under the constraint that production equals destruction .
the cross - sections from silberberg et al .
( 1998 ) were used .
we found that the sum of all nuclei with @xmath69 outnumber fe by a factor of 2.5 - 3.0 ( see table 1 ) . in the absence of the bsw spallation scenario
, these nuclei would not be present .
the single most abundant heavy nucleus is @xmath70 .
the immediate question that comes to mind is whether these spallation products can be detected .
they may be detected via their h@xmath0 lines , but the calculation of their predicted ew is left to future work . in table 1 , we find the @xmath71 at which @xmath72 of these nuclei would be in the hydrogenic state ( for @xmath73 in table 1 ) , allowing for a possible photoionization edge detection once @xmath74 ( presuming lte ) . to fit all @xmath67
, we used the explicit cross - section from rybicki & lightman ( 1979 ) and set the bound - free gaunt factor to @xmath75 , giving @xmath76 where @xmath77 is the edge energy ( shown in table 1 , value in parentheses is the redshifted edge energy ) from bethe & salpeter ( 1957 ; hereafter bs ) .
this @xmath78 agreed to within ( 3 - 5)% to verner et al.s ( 1996 ) analytic fits for ar , ca and fe .
for now , we estimate the impact of such an edge on the spectrum by considering the integral over the optical depth above the photoionization edge , @xmath79 , assuming an effectively cold atmosphere ( i.e. no reemission above the edge ) @xmath80 . for @xmath81 , the @xmath82 integral
is simply expanded to obtain @xmath83 integrated from @xmath84 to @xmath85 which is independent of @xmath67 .
table 1 shows @xmath82 assuming the element is present at the level implied by the spallation scenario and finding the amount of it in the hydrogenic ground state from saha equilibrium for @xmath86 .
as the burst cools or during a radius expansion burst , these edges could become prominent , and the entry of @xmath87 in table 1 presumes all of the spallation created elements have at least one bound electron .
it is intriguing to note that these edge energies and @xmath82 are close to those previously reported by waki et .
al . ( 1984 ) , turner & breedon ( 1984 ) , nakamura et al .
( 1988 ) , magnier et al .
( 1989 ) and kuulkers et al .
we now estimate the fe column needed to give the h@xmath0 line observed by cpm . motivated by our accretion scenario , we initially presume that this line forms in a thin isothermal layer of fe in lte at temperature @xmath88 above the hotter continuum photosphere .
the observed redshift implies a surface gravity of @xmath89 .
though rapid rotation ( @xmath90 ) might explain the observed line width ( @xmath91 ) , the line s equivalent width depends most strongly on the intrinsic broadening mechanisms .
the scale for thermal doppler broadening is @xmath92 .
the natural line width , @xmath93 ( bs ) , is given by the spontaneous decay rate for the @xmath94 transitions , which in units of the thermal doppler shift is @xmath95 .
we now show that stark broadening is larger by an order of magnitude than either of these effects for the h@xmath0 transition .
stark broadening results from the shift of line energies due to the electric field from neighboring ions , @xmath96 , where @xmath97 is the mean ion spacing ( mihalas 1978 ) .
the energy perturbation from the linear stark effect is @xmath98 , where @xmath99 is the size of the orbital . using the linear stark effect
is justified since its energy shift at the densities and temperatures considered exceeds that of the degeneracy breaking lamb shift for the relativistic h - like ion ( bs ) . for the @xmath100 state of fe at a thomson depth unity , we find @xmath101 , an order of magnitude larger than thermal doppler . since the majority of the intrinsic line width is attributed to stark broadening , @xmath55 must be found from the measured ew given by the curve of growth ( mihalas 1978 ) , @xmath102 where @xmath103 is the ew in energy units , @xmath104 is the normalized line depth at line center , @xmath105 is the intrinsic energy width ( in our case the stark shift ) , @xmath106 is the ratio of the line opacity to the continuum opacity and @xmath107 .
we use the electric field distribution for a uniform ion sea ( e.g. the holtsmark distribution , @xmath108 ; mihalas 1978 ) , giving @xmath109 where @xmath110 is the ratio of line optical depth to continuum optical depth at line center . depending on density either thermal doppler or stark broadening dominates ,
so we convolve the holtsmark distribution with the voigt profile to cover the complete range of densities .
we also ignore debye screening effects on the electric field distribution since the deviation from the holtsmark distribution is small for the typical @xmath88 and @xmath111 .
because rotational broadening acts to lower the apparent line depth and is likely to be important , the observed value of @xmath112 is a lower limit .
indeed , the true line depth could be significantly larger than 0.3 .
however , such a large value of @xmath104 is difficult to understand for realistic temperature gradients and presuming lte . for a gray atmosphere ( a reasonable approximation for the temperature profile of a type i burst atmosphere )
@xmath104 attains a maximal value of 0.2 when the fe resides at very low column depths .
the resolution of this dilemma may be the role of nlte effects and resonant line transport in the h@xmath0 line .
resonant line transport could generate such an absorption feature with a large line depth without invoking large temperature contrasts as in lte ( jefferies 1968 ) .
nlte effects clearly play a role in determining abundances from the ew measurements .
for instance , an immediate measure of the role of nlte effects is that the radiative ionization rate for the fe ground state exceeds the collisional ionization rate ( london et al .
1986 ) , complicating the ionization balance for fe . to resolve the role of nlte effects ,
the radiative transfer equation must be solved self - consistently for both the resonant transport of the h@xmath0 and ly@xmath0 lines with detailed balanced of the fe ionization and excitation in the strong radiation field of a type i burst . for
now , we simply present results for @xmath113 , which is set by observations and @xmath114 , which is characteristic of a gray atmosphere .
we calculate the @xmath55 required to reach @xmath115 presuming all fe resides at the column depth , @xmath116 .
we assume lte and that the dominant states for fe are either fully ionized , or in the hydrogenic ground state or helium - like ground state to approximate the partition function . for radial infall , with the fe layer between @xmath117 and @xmath118 , @xmath119 .
this temperature range is consistent with the absence of helium - like fe in the high @xmath7 phase of the burst ( cpm ) .
since the occupation of the @xmath120 state exceeds that in @xmath121 by @xmath122 , we also calculate @xmath123 for the implied @xmath55 values .
for this line , thermal doppler and natural line broadening dominate stark broadening ( madej 1989 ) . when @xmath124 and the fe resides at the pressure implied by radial accretion , @xmath125 .
foster et al .
( 1987 ) also calculated @xmath126 for a uniformly distributed fe abundance at the solar value and found that it would not exceed @xmath127 .
it would require an @xmath128 to give an fe column consistent with that calculated from spallation .
however , such large temperature contrasts ( and hence such large values of @xmath104 ) are not possible in these atmospheres and we need to understand the nlte line transport problem before we can address this apparent inconsistency . day , fabian & ross ( 1992 ) pointed out that a uniform presence of fe at the solar value would create an observable photo - ionization edge .
the fe column require to produce the observed h@xmath0 line would yield enormous edges . however , nlte effects would weaken the strength of these photoionization features significantly ( london et al .
the reported broad spectral features in the 6 - 9 kev range ( van paradijs et al .
1990 , strohmayer & brown 2002 ) during type i bursts and superbursts need to be re - analysed as possible fe photo - ionization edges .
active accretion in the context of a beam hitting the atmosphere naturally gives @xmath129 which only depends on the physics of proton spallation of fe . in this scenario ,
heavy elements with @xmath3 are produced in quantities comparable to fe and may be observed via their photoionization edges or atomic lines ( see table 1 ) during the cooling phase of type i bursts or superbursts .
our initial lte calculations have found that an fe column of @xmath130 is needed to produce the measured @xmath131 of cpm .
however , until a more complete calculation of the resonant line transport for the h@xmath0 and ly@xmath0 line has been undertaken in this strong radiation environment , we can not say whether this factor of twenty discrepency is certain . throughout this paper
, we have presumed that fe is present because of active accretion during the burst . however , the radiative force on hydrogenic fe in the ground state is @xmath132 where @xmath133 is the flux per photon energy at the ly@xmath0 transition ( @xmath134 ) .
since @xmath133 is nearly constant across the line , @xmath135 in the optically thin limit ( michaud 1970 ) , where @xmath136 is the oscillator strength from bs . if @xmath137 is given by a black - body , then at @xmath138 , the radiative acceleration on fe is @xmath139 , a factor of three larger than @xmath140
. the radiative force will be even larger for realistic type i burst spectra , which are harder than a blackbody at the observed @xmath141 ( london et al .
1986 ; pavlov , et al .
the fe diffusion time is less than the burst duration , so there is adequate time for the accelerated fe to reach a new diffusive equilibrium .
the calculation of the resulting fe column , which could well saturate the line and lower the radiative force , is beyond the scope of this _ letter _ , as it requires a simultaneous solution of the radiative transfer and the fe diffusion equations , like that done for hot white dwarfs ( e.g. chayer , fontaine & wesemael 1995 ) .
we thank a. piro for calculating the infall energy and incident angles and n. barghouty for providing the spallation cross - section code of silberberg et al .
we thank the referee , marten van kerkwijk , for critical comments and ira wasserman for many insightful conversation and for pointing out the critical role nlte effects can play in line formation .
this work was supported by the national science foundation under grants phy99 - 07949 and ast02 - 05956 and nasa through grant nag 5 - 8658 .
l. b. is a cottrell scholar of the research corporation .
fe & 9.279(6.87 ) & 0.32 & 1.00 & 277 & 1.12 + mn & 8.573(6.35 ) & 0.18 & 0.36 & 66 & 1.05 + cr & 7.897(5.85 ) & 0.26 & 0.48 & 56 & 0.99 + v & 7.248(5.37 ) & 0.26 & 0.43 & 32 & 0.93 + ti & 6.627(4.91 ) & 0.31 & 0.48 & 22 & 0.86 + sc & 6.035(4.47 ) & 0.41 & 0.58 & 17 & 0.81 + ca & 5.471(4.05 ) & 0.47 & 0.59 & 11 & 0.75 + k & 4.934(3.65 ) & 0.48 & 0.55 & 7 & 0.69 + ar & 4.427(3.28 ) & 0.53 & 0.54 & 4 & 0.65 +
|
the gravitational redshift measured by cottam , paerels and mendez for the neutron star ( ns ) in the low - mass x - ray binary exo 0748 - 676 depends on the identification of an absorption line during a type i burst as the h@xmath0 line from hydrogenic fe .
we show that fe is present above the photosphere as long as @xmath1 during the burst . in this limit , the total fe column is @xmath2 for incident material of solar abundances and only depends on the nuclear physics of the proton spallation .
the fe destruction creates many heavy elements with @xmath3 which may imprint photo - ionization edges on the ns spectra during a radius expansion event or in a burst cooling tail .
detecting these features in concert with those from fe would confirm a redshift measurement .
we also begin to address the radiative transfer problem , and find that a concentrated fe layer with @xmath4 and column @xmath5 ( depending on the line depth ) above the hotter continuum photosphere is required to create the h@xmath0 line of the observed strength .
this estimate must be refined by considerations of non - lte effects as well as resonant line transport . until these are carried out
, we can not say whether the fe column from accretion and spallation is in conflict with the observations .
we also show that hydrogenic fe might remain in the photosphere due to radiative levitation from the high burst flux .
|
the study of instabilities in low dimensional strongly correlated electron systems has received much attention in the last few years .
one of the main reasons is that a charge instability phenomenon ( phase separation ( ps ) ) often shows up in the vicinity of the superconducting transition in cuprates . in the case of double exchange models for manganese oxides that exhibit the colossal " magnetoresistance effect @xcite
, this charge instability arises close to the transition to ferromagnetism and , interestingly , finite size studies of both the two - dimensional realistic model and its one - dimensional version display similar features in this respect .
the case of magnetic instabilities ( metamagnetism ( mm ) ) has also received recent attention in connection to the one dimensional antiferromagnetic ( af ) @xmath2 spin chain with next - nearest - neighbor interactions ( nnn ) , where it was found that mm arises in a finite region of the phase space @xcite .
generically , at the point where charge or magnetic instabilities occur , a divergent compressibility or magnetic susceptibility arises .
this divergence is in turn associated respectively with the coexistence of two phases with different hole concentrations or magnetizations .
the aim of this paper is to discuss a general way to determine whether an instability could show up in a given one - dimensional model , using abelian bosonization .
our discussion applies to any one - dimensional model describable as an integrable model plus perturbations whose effect is to renormalize the luttinger liquid ( ll ) parameters , @xmath1 and @xmath0 .
the effect of irrelevant operators is also discussed .
our discussion could be also relevant in the study of certain 2d systems which can realize the so called sliding luttinger phase @xcite , since a divergence in the 1d susceptibility leads to a singularity in the 2d one @xcite .
generically , our approach provides a quick tool to study the tendency of different perturbations to produce instabilities in the above mentioned systems . as a sample case for magnetic systems
, we analyze the @xmath2 chain with nnn exchange , treating first this last interaction perturbatively within a bosonization approach and find qualitative agreement with the results obtained in @xcite in finite systems ( this problem was also studied in @xcite ) .
we also study the opposite coupling limit , which we call hereafter zig - zag " limit , where the system can be reinterpreted as a two - leg zig - zag ladder . in this case , we determine a region in the parameter space where mm occurs . to analyze the case of charge instabilities we consider the charge sector of the hubbard model at incommensurate fillings , perturbed by a nearest - neighbor density - density interaction @xmath3 . in this case
we find that for small coulomb repulsion @xmath4 there is a region where the system phase separates which corresponds to negative values of @xmath3 , in agreement with previous studies @xcite .
we consider a generic situation in which the low energy degrees of freedom ( which could correspond to either charge or spin variables ) are described by a tomonaga - luttinger liquid in the unperturbed case .
the corresponding hamiltonian once interactions are taken into account will generically be of the form ( modulo irrelevant terms ) h_0 + ( _ x ) ^2 + ( _ x ) ^2 + ( _ x ) , [ maineq ] where @xmath5 corresponds to the tomonaga - luttinger hamiltonian h_0 = 1 2 dx ( v k ( _ x ) ^2 + v k ( _ x ) ^2 ) . [ ll ] the first two terms can be readily absorbed into a redefinition of the ll parameters @xmath6 and @xmath7 , while the third one changes the chemical potential in the case of charge variables and the magnetic field in the spin case . under the above mentioned conditions
the compressibility for the charge modes , described by an effective ll with parameters @xmath8 and @xmath9 can be shown to be given by @xcite , [ comp ] and similarly , the magnetic susceptibility for the spin modes described by an effective ll with parameters @xmath10 and @xmath11 is given by .
[ susc ] one readily observes that a divergence in these quantities arises either when @xmath12 or @xmath13 vanish .
it should be pointed out that these two things could happen simultaneously , but this is not always the case and hence ps or mm instabilities are to be identified with the vanishing of the effective velocity .
we argue that this is true , provided that the quantities @xmath14 and @xmath15 remain positive definite and irrelevant perturbations do not change the large scale behavior . in ref .
@xcite an attempt to characterize these instabilities in one dimensional systems exhibiting a universal character described by a tomonaga - luttinger model was made . in this paper
the authors identified the divergence of the above mentioned thermodynamic quantities with a divergence in the so called luttinger parameter @xmath1 . although this interpretation led to a consistent analysis for the cases studied in @xcite
, we argue that a more general criterion consists in identifying the instability regions with those where the velocity vanishes .
this last statement can be understood as follows : by analogy with the unperturbed @xmath2 chain in a magnetic field , we see that when the fermi velocity goes to zero we approach a ground state of fm nature ( and @xmath1 does not necessarily diverge , though this happens for the particular case of the @xmath2 chain for @xmath16 and zero field @xcite ) .
then the magnetic susceptibility diverges as @xmath17 and , if this happens before reaching saturation , the magnetization curve as a function of the applied magnetic field presents a jump .
once we have shown that there is a jump in the magnetization curve , for some critical value of the applied magnetic field , @xmath18 , we can conclude that two different values of the magnetization will coexist at this point @xcite . in the next two sections we study two sample cases where this general discussion applies , the @xmath2 chain with nnn interactions where mm has been shown to occur @xcite , @xcite and the extended hubbard model with nearest - neighbour interactions @xmath3 , where ps appears for negative @xmath3 @xcite .
the hamiltonian of the @xmath2 af with nnn interactions in a magnetic field is given by h_xxz = j _ i=1^n ( s^x_is^x_i+1 + s^y_is^y_i+1 + s^z_is^z_i+1 ) + + j _ i=1^n ( s^x_is^x_i+2 + s^y_is^y_i+2 + s^z_is^z_i+2 ) - h _ i=1^n _
[ xxz ] the large scale behaviour of the @xmath2 chain can be described by a @xmath19 free boson theory with hamiltonian ( [ ll ] ) .
the field @xmath20 and its dual @xmath21 are given by the sum and difference of the light - cone components , respectively .
the constant @xmath1 governs the conformal dimensions of the bosonic vertex operators and can be obtained exactly from the bethe ansatz solution of the @xmath2 chain ( see e.g. @xcite for a detailed summary ) .
we have @xmath22 for the @xmath23 symmetric case ( @xmath24 ) and is related to the radius @xmath25 of @xcite by @xmath26 . in ( [ ll ] )
@xmath0 corresponds to the fermi velocity of the fundamental excitations of the system . in terms of these fields ,
the spin operators read s_x^z & ~ & 1 _ x + a : ( 2 k_f x + ): + , [ sz ] + s_x^ & ~ & ( -1)^x : e^i ( b ( 2 k_f x + ) + c ) : , [ s+ ] where the colons denote normal ordering with respect to the groundstate with magnetization @xmath27 .
the fermi momentum @xmath28 is related to the magnetization of the chain as @xmath29 .
the effect of an @xmath2 anisotropy and/or the external magnetic field is then to modify the scaling dimensions of the physical fields through @xmath1 and the commensurability properties of the spin operators , as can be seen from ( [ sz ] ) , ( [ s+ ] ) .
the constants @xmath30 , @xmath31 and @xmath32 were numerically computed in the case of zero magnetic field @xcite ( see also @xcite ) . in what follows we study both the weak coupling ( @xmath33 ) and the zig - zag ( @xmath34 ) limits . *
i ) @xmath33 limit * using ( [ sz ] ) and ( [ s+ ] ) the nnn interaction term in the bosonic language reads h_nnn= dx ( g_1 ( _ x ) ^2 + g_2 ( _ x ) ^2 + _ x + ( ) ) , [ nnn ] where @xmath35 and @xmath36 , @xmath37 and @xmath38 depend on @xmath39 and the non - universal constants @xmath30 , @xmath31 and
@xmath32 as g_1 = ( - 2a^2 ( 4k_f ) ) + 2b^2 ( 4k_f ) , [ g1 ] g_2 = -4(c^2 - ( 4k_f ) ) , [ g2 ] and = m , = b c . [ lambda ] the first two terms in ( [ nnn ] ) have the effect of renormalizing both the compactification radius and the fermi velocity in the following way v_eff^2= ( v / k + 2 g_1)(vk + 2 g_2 ) , k_eff^2= , [ rentot ] now we can make contact with the analysis of ref .
@xcite : the mm region is identified within our approach as the set of phase space points in which the susceptibility diverges for @xmath40 .
one immediately observes from ( [ rentot ] ) that , for zero magnetic field and without taking into account the effect of the @xmath38 perturbation , both @xmath41 and @xmath42 vanish simultaneously and that this happens when @xmath43 ( see fig . 1 ) .
however , this situation can change for two reasons : first , an extra term arises from the nnn perturbation , which renormalizes the external magnetic field @xmath44 and hence changes both @xmath41 and @xmath45 through the bethe ansatz equations , generically in a way which removes the above mentioned simultaneity .
second , if one goes beyond the zero loop order , both @xmath45 and @xmath41 will renormalize in a different way due to the @xmath38 term @xcite .
we will discuss this issue again in the context of charge instabilities in the next section .
in particular , the boundary between the mm and the fm phases is obtained as the set of points in which @xmath41 and/or @xmath42 vanish / es for @xmath46 , and that between the mm and the af phases as the set of points in which @xmath47 and @xmath48 for all values of the magnetization @xmath49 .
let us focus on the mm - fm boundary , which is easily obtained using ( [ susc ] ) and ( [ rentot ] ) .
in this case , we can use the numerical values for the non - universal constants @xmath30 , @xmath31 and @xmath32 appearing in ( [ sz ] ) and ( [ s+ ] ) obtained in @xcite . the boundary obtained in this way agrees qualitatively with that obtained in @xcite ( see fig . 1 ) .
the lack of quantitative agreement is presumably due to the perturbative treatment of the nnn interactions , which could be improved by considering higher loop contributions in a renormalization group analysis from the bosonization side and due to finite size effects from the numerical one .
however , our main aim is to discuss the appearance of instabilities in a generic and simple way .
one should be careful also about the regime of validity of this approach , since due to the renormalization of the luttinger parameter @xmath45 , the scaling dimensions of the many discarded irrelevant perturbations change , and nothing prevents one of these to become relevant .
in fact , by analyzing the scaling dimension of the leading irrelevant @xmath38 perturbation , we observe that it reaches the limiting value 2 at @xmath50 , and hence our approach ceases to be valid for @xmath51 .
this critical line separates the massless regime from a massive one , and our results compare qualitatively well with those obtained in @xcite ( dashed line in fig . 1 ) .
the study of the af - mm boundary is in this case more involved due to the appearance of the non - universal constants @xmath30 , @xmath31 and @xmath32 in the bosonized operators which are not available for non - zero magnetic field .
( the estimation of the field dependence of these constants for @xmath52 has been done in @xcite but its evaluation is more delicate on the ferromagnetic side @xmath53 @xcite . ) * ii ) zig - zag limit ( @xmath34 ) * in this limit
we reinterpret the hamiltonian ( [ xxz ] ) as a two leg zig - zag ladder in the weak interchain coupling limit @xcite . in this description one
represents each of the chains by one free compactified @xmath19 boson .
thus , the whole ladder is represented by two bosons @xmath54 , @xmath55 , each governed by an action given by ( [ ll ] ) , plus the perturbative terms arising from the interchain zig - zag coupling .
this system , in the presence of an external magnetic field , has been studied in @xcite , where it was shown that for non - zero magnetization a relevant interaction gives a mass to the relative " field @xmath56 , while the diagonal " field , @xmath57 , remains massless .
the effective hamiltonian governing the large scale behavior of @xmath58 is then given by h_eff = 1 2 dx ( v_eff k_eff ( _ x _ diag)^2 + v_effk_eff ( _ x _ diag)^2 + _ x _ diag ) , [ hamdiag ] where , calling @xmath59 , @xmath60 and @xmath41 and @xmath45 are given by v_eff^2= v^2(1 + ) , k_eff^2=k^2(1 + ) ^-1 .
[ rendiag ] there is however an extra term that mixes the fields @xmath61 and @xmath58 h_mix = - dx _ x _ diag ( _ rel ) .
[ mix ] the effect of this term was studied in ref .
@xcite where it was shown that it gives rise to a spin nematic phase close to @xmath62 .
the main point for our analysis is that the diagonal field is still described by a ll , and we expect the same picture to apply for @xmath63 .
for this reason , we do not take into account this last term in what follows .
one then observes from ( [ rendiag ] ) that the non - universal constants do not appear in the perturbing terms , and hence we can obtain in this case both the fm - mm and mm - af boundaries .
the results are presented in fig . 2 .
we consider now the hubbard model , defined as h = -_j,(^_j , _
j+1 , + h.c . ) + u_j n_j , n_j , + _ j , n_j , , [ hubham ] where n_j , = ^_j,_j , . in the absence of an external magnetic field
it has been shown that this model presents charge - spin separation .
it has also been shown ( c.f .
@xcite ) that the large scale behavior of the spin and charge degrees of freedom can be described by two decoupled boson field theories with dynamics governed by the tomonaga - luttinger hamiltonian .
the parameters @xmath1 and @xmath0 can in each case be exactly obtained for all values of @xmath64 and @xmath4 via numerically solving the bethe ansatz equations in @xcite .
approximate expressions for the velocity of the charge sector in the small and large @xmath4 regimes are given in @xcite .
the addition of a density - density interaction between nearest neighbors leads to the so called extended hubbard model , whose hamiltonian is given by ( [ hubham ] ) plus the term h = v_jn_j n_j+1 , [ perthubham ] where n_j = n_j,+n_j , .
it is a simple matter to show that , modulo irrelevant operators , the effect of the perturbation ( [ perthubham ] ) is to renormalize the parameters @xmath64 and @xmath4 as follows : + v , u u + 2v .
therefore , the low energy behavior of the charge sector of the extended hubbard model is that of a luttinger liquid with parameters k_eff = k(u + 2v,+ v ) , v_eff = v(u + 2v,+ v ) , where @xmath65 y @xmath66 , are given by the exact bethe ansatz solution of the non - extended hubbard model ( [ hubham ] ) .
this allows for the determination of the divergences in the compressibility @xmath67 , as given by ( [ comp ] ) , which we identify with the zeros of the effective velocity .
we have only considered small values of @xmath3 , since our approach is a perturbative one . within this restriction , we have found no singularities , and therefore no ps , in the large @xmath4 regime . on the other hand , regarding the small @xmath4,@xmath3 region , in which the effective velocity is given by v_eff = v(u + 2v,+ v ) ~(1 + ) ( + v ) + , [ veffhub ] we have found that the roots of the above expression present the same qualitative behavior as that described in @xcite .
it should be stressed that @xmath45 remains finite in the region where ( [ veffhub ] ) vanishes . following this approach , one could also study the appearance of instabilities in the magnetic sector .
we have discussed the appearance of mm and ps in systems that can be described as perturbed luttinger liquids . specifically , we have studied the @xmath2 spin chain with nnn interactions and the extended hubbard model with nearest - neighbor density - density interactions .
we have found fm , af and mm regions in the zig - zag limit of the @xmath2-nnn spin chain . in the weak coupling limit ,
qualitative agreement with previous results was found for the mm - fm transition .
we were not able to complete our search for mm due to lack of specific numerical data , an issue to be discussed elsewhere .
the instabilities were in all cases identified as the roots of the inverse susceptibility ( [ susc ] ) .
concerning the extended hubbard model , according to our analysis , it does not present ps in the large @xmath4 and small @xmath3 limit .
on the other hand , in the small @xmath4 , @xmath3 limit instabilities do show up , through the roots of the effective fermi velocity , while the effective luttinger parameter remains finite .
we argue that all the instabilities studied are associated in general with a vanishing effective velocity @xmath41 . as a matter of fact
, it is the roots of @xmath41 that give a divergent compressibility of the extended hubbard model , though in the other cases studied , this coincides with a divergence of the effective luttinger parameter @xmath45 .
this should not be regarded as the actual hallmark of the instabilities : although our treatment of the @xmath2-nnn spin chain gives at the same time @xmath68 , this is only true in the absence of an external magnetic field and moreover ceases to be valid beyond the zero loop order since both parameter renormalize in a different manner @xcite .
it should be stressed that our approach gives a simple way to qualitatively analyze instabilities in generic charge and magnetic systems provided they can be described as perturbed luttinger liquids .
henceforth , it provides a quick tool to study whether different perturbations could produce such instabilities . _ acknowledgments : _ it is a pleasure to acknowledge fruitful discussions with m. arlego , e. dagotto , a. furusaki , m.d .
grynberg , t. hikihara , a. honecker and p. pujol .
the authors acknowledge partial financial support of conicet and fundacin antorchas .
r. chitra , s.k .
pati , h.r .
krishnamurthy , d. sen , s. ramasesha , phys .
* b52 * , 6581 ( 1995 ) . s.r .
white , i. affleck , phys .
rev . * b54 * , 9862 ( 1996 ) .
d. allen , d. snchal , phys .
* b55 * , 299 ( 1997 ) .
nersesyan , a.o .
gogolin , f.h.l .
eler , phys .
* 81 * , 910 ( 1998 ) .
e. srensen , i. affleck , d. augier , d. poilblanc , phys .
* b58 * , r14701 ( 1998 ) .
cabra , a. honecker , p. pujol , eur .
* b13 * , 55 ( 2000 ) .
|
we discuss the appearance of magnetic and charge instabilities , named respectively metamagnetism ( mm ) and phase separation ( ps ) , in systems which can be described by a perturbed luttinger liquid .
we argue that such instabilities can be associated with the vanishing of the effective fermi velocity @xmath0 , which in some cases coincides with a divergence of the effective luttinger parameter @xmath1 .
we analyze in particular an @xmath2 chain with next - nearest - neighbor interactions in different limits where mm shows up and an extended hubbard model where in turn , ps occurs .
qualitative agreement with previous studies is found .
pacs numbers : 75.10.jm , 75.60.ej # 1 # 1#1
|
in their recent studies dorey , dunning and tateo ( ddt ) [ 1 ] have considered the manifestly non - hermitian schrdinger equation , in @xmath4 units , @xmath5 \,{\large \psi }
_ { k,\ell } ( r)=e_{k,\ell } { \large \psi } % _ { k,\ell } ( x).\ ] ] they have rigorously proved that the spectrum @xmath6 is real and discrete in the domain of the sufficiently large angular momenta , @xmath7 \equiv \ell _ { ddt}(\alpha ) .\ ] ] this inspired our subsequent study of this model [ 2 ] where we have shown that in the strong coupling regime with @xmath8 , the low lying ddt bound states may be very well approximated by the harmonic oscillators . at the same time
we have noticed that the quality of such an asymptotic approximation may deteriorate quite significantly with both the increase of excitation @xmath9 and/or with the decrease of @xmath10 .
such a situation is , obviously , challenging .
firstly , our study [ 2 ] revealed that the manifest non - hermiticity of the models of the type ( 1 ) leads to the reliable leading order approximation _ only after _ we select our harmonic oscillator approximant as lying _ very far _ from the real axis ( i.e. , from the hermitian regime ) .
such a recipe is , apparently , deeply incompatible with a smooth modification of the traditional zero - order approximants occurring in current hermitian @xmath11 recipes ( cf . a small sample of some references in [ 3 ] ) . at the same time
, the smallness of @xmath12 still supports the feeling that the similar perturbation techniques _ should _ prove efficient after their appropriate modification .
this observation offered a sufficiently strong motivation for our continued interest in the complex , non - hermitian model ( 1 ) which may be understood as a characteristic representative of a very broad class of the so called pseudo - hermitian models with real spectra , the analyses of which became very popular in the recent literature [ 4 ] - [ 18 ] . within this class , the strong - coupling version of ddt oscillators ( 1 ) with @xmath13 forms a particularly suitable testing ground as it combines the necessary reality of its spectrum with the smallness of the inverse quantity @xmath14 .
moreover , the phenomenologically appealing non - hermitian models like ( 1 ) are rarely solvable in closed form so that the presence of a universal " small parameter @xmath15 offers one of not too many ways towards their systematic approximate solution . in our present paper
ii we intend to discuss such a possibility in more detail .
the first stages of interest in the non - hermitian oscillators ( 1 ) date back to an old paper by caliceti et al [ 5 ] .
it studied the imaginary cubic problem in the context of perturbation theory and , more than twenty years ago , it offered the first rigorous explanation why the spectrum in such a model may be real and discrete . in the literature , this result has been quoted as a mathematical curiosity [ 6 ] and only many years later , its possible relevance in physics re - emerged and has been emphasized [ 7 ] .
this initiated an extensive discussion which resulted in the proposal of the so called @xmath16 symmetric quantum mechanics by bender et al [ 8 ] .
the key idea of the new formalism lies in the empirical observation that the ( phenomenologically desirable ) existence of the real spectrum need not necessarily be attributed to the hermiticity of the hamiltonian .
the current hermiticity assumption @xmath17 is replaced by the mere @xmath16 symmetry @xmath18 . here
, @xmath19 denotes the parity ( @xmath20 ) while the anti - linear operator @xmath21 mimics the time reflection ( @xmath22 ) .
it is easy to verify that example ( 1 ) exhibits such a type of symmetry [ 2 ] and may serve as an elementary illustration of the latter extension of quantum mechanics . the bender s and boettcher s conjecture that @xmath23 may imply @xmath24 is fragile .
the extent as well as limitations of its validity are most easily analyzed in the language of linear algebra using the biorthogonal bases [ 4,9 ] and/or exactly solvable hamiltonians [ 10 ] .
nevertheless , the relevance of many unsolvable oscillators originates from their applicability in physics [ 11 ] and field theory [ 12 ] .
in such a setting it is necessary to develop and test also some efficient approximation methods .
new and intensive studies employed the ideas of the strong - coupling expansions [ 13 ] as well as the complex version of wkb [ 14 ] , hill determinants and fourier transformation [ 15 ] , functional analysis [ 16 ] , variational and truncation techniques [ 17 ] , and linear programming [ 18 ] . in what follows we intend to use the method based on the smallness of the inverse angular momentum parameter @xmath14 .
various versions [ 19 ] of such an approach are available for hermitian models where the combination of the central repulsive core @xmath25 with a confining ( i.e. , asymptotically growing ) interaction @xmath26 forms a practical effective potential @xmath27 which possesses a pronounced minimum .
near such a minimum the shape of the potential is naturally fitted by the elementary and solvable harmonic oscillator well .
corrections can be then evaluated in an unambiguous and systematic manner [ 20 ] . as long as we intend to move to the complex plane
, the leading - order approximation may become non - unique .
one finds several different complex and/or real minima of @xmath27 even in our oversimplified examples ( 1 ) [ 2 ] .
for all the similar non - hermitian hamiltonians , even the most sophisticated forms of the perturbation expansions in the powers of our small parameter @xmath11 lose their intuitive background and deserve careful new tests , therefore .
some of the related reopened questions will be clarified by our forthcoming considerations inspired , basically , by the pseudo - perturbative shifted@xmath28 expansion technique ( pslet ) in its form designed for the standard hermitian hamiltonians and described , say , by mustafa and odeh in ref .
[ 20 ] . in this recipe ,
the shifts of @xmath10 were admitted as suitable optional auxiliary parameters while the use of the prefix
pseudo- " just indicated that @xmath10 itself is an artificial , kinematical parameter rather than a genuine dynamical coupling .
as we already mentioned , one of the first @xmath29symmetric models with an immediate impact on physics has been the buslaev s and grecchi s quartic anharmonic oscillator [ 7 ] described by the radial schrdinger equation in @xmath30dimensional space , @xmath31 \chi _ { k,\ell } ( r)=e_{k,\ell } ~\chi _ { k,\ell } ( r).\ ] ] in this model they shifted the coordinate axis to the complex plane , @xmath32 with a constant @xmath33 and variable @xmath34 @xmath35 .
they also required that @xmath36 at all the partial waves @xmath37 and dimensions @xmath38 .
this example may find the various sophisticated generalizations some of which will be also mentioned in due course in what follows .
for example , a @xmath39dependent shift @xmath40 may be needed both for the exactly solvable coulombic model of ref.[21 ] and for all the more general and purely numerically tractable potentials @xmath41of ref.[8 ] with the positive exponents @xmath42 .
fortunately , the transition to @xmath40 remains particularly elementary , being mediated by the mere change of the variable in eq.(3 ) within this class ( c.f . ,
e.g. , ref.[21 ] for an explicit illustration ) .
another remark might concern the assumption that the wave functions are square integrable .
for the most elementary @xmath29 symmetric hamiltonians this assumption seems very natural but in certain more sophisticated models its use may require a more careful analysis as presented , e.g. in refs .
[ 4 ] and [ 16 ] .
the practical experience with the hermitian version of the pseudo - perturbation shifted - @xmath0 expansion technique of mustafa and odeh [ 20 ] may serve as a key inspiration of an appropriate complexified new pslet recipe .
firstly we notice a formal equivalence between the assumed smallness of our parameter @xmath43 and of its ( arbitrarily ) shifted form .
thus we introduce a new symbol @xmath44 and , simultaneously , move and rescale our coordinates @xmath45 to @xmath46 here @xmath47 is an arbitrary point , with its particular value to be determined later .
equation ( 3 ) thus becomes @xmath48 ^{2}}+% \frac{\bar{l}^{2}}{q}~v(x(r))\right\ } \psi _ { k,\ell } ( x(r))=e_{k,\ell } \psi _ { k,\ell } ( x(r)),\ ] ] where @xmath49 is a constant that scales the potential @xmath50 at large-@xmath51 limit and is set , for any specific choice of @xmath51 and @xmath52 equal to @xmath53 at the end of calculation .
expansions about this point , @xmath54 ( i.e. @xmath55 ) , yield @xmath56 ^{2}}% = \sum_{n=0}^{\infty } ~(-1)^{n}~\frac{(n+1)}{r_{0}^{n+2}}~x^{n}~\bar{l}% ^{-n/2},\ ] ] @xmath57 it is also convenient to expand @xmath6 as @xmath58 of course one may consider also energy coefficient of half - entire power of @xmath59 in ( 8) but all these coefficients vanish ( c.f . , e.g. , ref.[19,20 ] ) . equation ( 5 ) , therefore , reads @xmath60 where @xmath61 equation ( 9 ) is to be compared with the non - hermitian @xmath16 - symmetrized perturbed harmonic oscillator in the one dimensional schrdinger equation @xmath62 \phi _ { k}(y)=\lambda _ { k}~\phi _ { k}(y),\ ] ] where @xmath63 is a complexified perturbation - like term and @xmath64 is obviously a constant .
such a comparison implies @xmath65 @xmath66+}\frac{1}{\,\bar{l}}\left [ \beta ( \beta + 1)t_{0}+\lambda _ { k}^{(0)}\right ] + \sum_{n=2}^{\infty } \lambda _ { k}^{(n-1)}~% \bar{l}^{-n } \nonumber \\
\quad \quad \quad \ \ \ \ \ & = & e_{k,\ell } ^{(-2)}~~\bar{l}+e_{k,\ell } ^{(-1)}+\sum_{n=1}^{\infty } e_{k,\ell } ^{(n-1)}~\bar{l}^{-n}\end{aligned}\ ] ] the first two dominant terms are obvious@xmath67 @xmath68 and with appropriate rearrangements we obtain@xmath69 @xmath70 here @xmath47 is chosen to minimize @xmath71 , i. e. @xmath72 equation ( 13 ) in turn gives , with @xmath73 , @xmath74 consequently , @xmath75 @xmath76 adds a constant to the energy eigenvalues and @xmath77 the next leading correction to the energy series , @xmath78 , consists of a constant term and the exact eigenvalues of the unperturbed one - dimensional harmonic oscillator potential @xmath79 where @xmath80 evidently , equation ( 19 ) implies that @xmath47 is either pure real , @xmath81 or pure imaginary , @xmath82 .
next , the shifting parameter @xmath83 is determined by choosing @xmath84 .
that is @xmath85 , \ ] ] where @xmath86 for @xmath47 pure real and @xmath87 for @xmath88 pure imaginary .
then equation ( 9 ) reduces to @xmath89 ~\psi _ { k,\ell } ( x)=\left [ \sum_{n=-1}^{\infty } e_{k,\ell } ^{(n)}~% \bar{l}^{-(n+1)}\right ] ~\psi _ { k,\ell } ( x).\ ] ] with @xmath90 and for @xmath91 @xmath92 equation ( 21 ) upon setting the wave functions @xmath93 readily transforms into the riccati - type equation:@xmath94 -2f_{k,\ell } ^{^{\prime } } ( x)u_{k,\ell } ^{^{\prime } } ( x)-f_{k,\ell } ^{^{\prime \prime } } ( x)=0,\end{aligned}\ ] ] where the primes denote derivatives with respect to @xmath95 .
it is evident that this equation admits solutions ( c.f . , e.g. , ref.[20 ] ) of the form @xmath96
@xmath97 with@xmath98 @xmath99 obviously , equating the coefficients of the same powers of @xmath59 and @xmath100 ( for each @xmath9 ) , respectively , one can calculate the energy eigenvalues and eigenfunctions ( following the uniqueness of power series representation ) from the knowledge of @xmath101 , @xmath102 , and @xmath103 in a hierarchical manner . in order to test the performance of our strategy ,
let us first apply it to the two trivial , exactly solvable @xmath16- symmetric examples .
using the potential @xmath104 ( where @xmath105 is a real coupling constant ) in the above @xmath29symmetric pslet setting , one reveals the leading - order energy approximation @xmath106 the unique minimum at @xmath107 occurs in the upper - half of the complex plane . in this case
@xmath108 the leading energy term reads @xmath109 and higher - order corrections vanish identically .
therefore , the total energy is @xmath110 where @xmath111 is the principle quantum number .
evidently , the degeneracy associated with ordinary ( hermitian ) coulomb energies @xmath112 is now lifted upon the complexification of , say , the dielectric constant embedded in @xmath105 .
moreover , the phenomenon of _ flown away states _ at @xmath113 emerge , of course if they do exist at all ( i. e. the probability of finding such states is presumably zero , the proof of which is already beyond our current methodical proposal ) .
therefore , for each @xmath114 state there is an @xmath115 state to _ fly away_. next , let us replace the central - like repulsive / attractive core through the transformation @xmath116 , i.e. @xmath117 with @xmath118 denoting _ quasi - parity _ , and recast ( 25 ) as @xmath119 which is indeed the exact result obtained by znojil and lvai [ 21 ] . equation ( 27 ) implies that _ even - quasi - parity _ , @xmath120 states with @xmath121 _
fly away _ and disappear from the spectrum . nevertheless , _
quasi - parity - oscillations _ are now manifested by energy levels crossings .
that is , a state-@xmath9 with _ even - quasi - parity _ crosses with a state-@xmath122{0xb4}}% % beginexpansion { \acute{}}% % endexpansion $ ] with _ odd - quasi - parity _ when @xmath123{0xb4}}% % beginexpansion { \acute{}}% % endexpansion \right ) /2.$ ] however , when @xmath124 the central - like core becomes attractive and the corresponding states cease to perform _ quasi - parity - oscilations .
_ for more details on the result ( 27 ) the reader may refer to znojil and lvai [ 21 ]
. for this potential the leading energy term reads @xmath126 and supports four eligible minima ( all satisfy our conditions in ( 17 ) ) obtained through @xmath127 as @xmath128 and @xmath129 . in this case
@xmath130@xmath131 for @xmath129 , and @xmath132 for @xmath133 .
whilst the former ( 29 ) yields @xmath134 the latter ( 30 ) yields @xmath135 in both cases @xmath136 the higher - order corrections vanish identically .
yet , one would combine ( 31 ) and ( 32 ) by the superscript @xmath137 and cast @xmath138 therefore , the @xmath29 symmetric oscillator is exactly solvable , by our recipe , and its spectrum , non - equidistant in general , exhibits some unusual features ( cf . [ 22 ] for more details ) .
however , it should be noted that for the one - dimensional oscillator ( where @xmath139 and@xmath140 even and odd parity , respectively ) equation ( 33 ) implies ( i ) @xmath141 @xmath142 , for @xmath143 , and ( ii ) @xmath144 , @xmath145 , for @xmath146 . which can be combined together to form the exact well known result @xmath147 with a new , redefined quantum number @xmath148 .
in our @xmath29symmetric schrdinger equation ( 1 ) with the practical effective potential @xmath150 the general solutions themselves are analytic functions of @xmath151 ( c.f . , e.g. , [ 6 ] ) .
we may construct them in the complex plane which is cut , say , from the origin upwards .
this means that @xmath152 with the length @xmath153 and with the span of the angle @xmath154 .
compact accounts of the related mathematics may be found in bender et al [ 8 ] .
let us proceed with our @xmath16-pslet and search for the minimum / minima of our leading energy term for the _ ddt_-oscillators ( 1 ) @xmath155 evidently , condition ( 17 ) yields @xmath156 = 0.\ ] ] obviously , a closed form solution for this equation is hard to find ( if it exists at all ) and one has to appeal to numerical techniques to solve for @xmath157 _ a priori _ , it is convenient to do some elementary analyzes , in the vicinity of the extremes of @xmath158 ( mandated by condition ( 2 ) ) , and distinguish between the two different domains of @xmath158 . for this purpose
let us denote @xmath159 , re - scale @xmath160 and abbreviate @xmath161 with @xmath162 @xmath163 .
this gives the following new algebraic transparent form of our implicit definition of the minimum / minima in ( 37 ) , @xmath164 in the weak - coupling domain , vanishing @xmath165 equation ( 38 ) becomes trivial ( @xmath166 . it is easy to verify that ( 36 ) , with @xmath167 has a _
unique absolute minimum _ at _ _ @xmath168 in the strong - coupling regime , @xmath169 equation ( 38 ) yields @xmath170 at this point , one may choose to work with @xmath171 ( i.e. @xmath172 ) and obtain the leading ( zeroth)-order approximation @xmath173 , \\ ] ] and , with @xmath174,\qquad r_{0}=-ig^{2/5}z^{1/5},\ ] ] the first - order correction @xmath175 \nonumber \\ & = & \left ( \frac{g}{z^{2}}\right ) ^{1/5}\left [ \sqrt{\frac{15~(1+z)}{4}}% ( 2k+1)-\sqrt{\frac{3}{2}}\right ] .\
\end{aligned}\ ] ] consequently , the energy series ( 8) reads , up to the first - order correction @xmath176 .\ ] ] nevertheless , one may choose to work with @xmath177 ( i.e. @xmath178 ) and obtain @xmath179 .\ ] ] thus the zeroth - order approximation yields @xmath180 ~;\ ~g_{s}=\sqrt{\frac{2}{3}}\left [ \ell _ { d}+% \frac{1}{2}-(2k+1)\sqrt{\frac{5~(1+z)}{8}}\right ] .\ ] ] in figure i we plot the energies of ( 44 ) vs @xmath181 at different values of @xmath182 @xmath183 .
obviously , our results show that even with @xmath184 the spectrum remains real and discrete .
moreover , once we replace @xmath185 @xmath186 i.e. @xmath187 with @xmath118 denoting quasi - parity , _ quasi - parity - oscillations _ are manifested by the unavoidable energy levels crossings , see figure ii .
table i shows that our results from eqs.(42 ) and ( 44 ) compare satisfactorily with those obtained by znojil , gemperle and mustafa [ 2 ] , via direct variable representation ( dvr ) .
we may mention that even in the domain of not too large @xmath188 the difference between the exact and approximate energies remain small , of order of @xmath189 from eq.(42 ) , with @xmath190 and @xmath191 from eq.(44 ) , with @xmath192 for the ground state .
such a prediction should not mislead us in connection with the related convergence / divergence of our energy series ( 8) , which is in fact the genuine test of our present @xmath29symmetric pslet formulae .
the energy series ( 8) with @xmath193 converge more rapidly than it does with @xmath194 nevertheless , our leading energy term remains the benchmark for testing the reality and discreteness of the energy spectrum . in table
ii we compare our results ( using @xmath195 hereinafter , numerically solve for @xmath47 and following the procedure of section iii ) for ( 1 ) with @xmath196 using the first ten - terms of ( 8) and the corresponding pad approximant , again with those from the dvr approach .
they are in almost exact accord . hereby , we may emphasize that the digital - precision enhances for larger @xmath59 ( smaller @xmath197 ) values , where the energy series ( 8) and the related pad approximants stabilize more rapidly .
extending the recipe of our test beyond the weak - coupling regime @xmath198 ( i.e. @xmath199 ) we show , in table iii , the energy dependence on the non - vanishing @xmath200 evidently , the digital - precision of our @xmath16 - pslet recipe reappears to be @xmath201-dependent and almost @xmath158 - independent . in table
iv we witness that the leading energy approximation inherits a substantial amount of the total energy documenting , on the computational and practical methodical side , the usefulness of our pseudo - perturbation recipe beyond its promise of being quite handy . yet , a broad range of @xmath158 is considered including the domain of negative values , safely protected against any possible spontaneous @xmath16 -symmetry breaking .
our purpose was to find a suitable _ perturbation series like _ expansion of the bound states .
we have started from the observation that the physical consistency of the model ( 1 ) ( i.e. , the reality of its spectrum ) is characterized by the presence of a strongly repulsive / attractive core in the potential @xmath27 .
this is slightly counterintuitive since the phenomenologically useful values of @xmath0 are usually small and only the first few lowest angular momenta are relevant in the hermitian schrdinger equations with central symmetry .
only exceptionally , very high partial waves are really needed for phenomenological purposes ( say , in nuclear physics [ 23 ] ) .
the strong repulsion is required there , first of all , due to its significant _ phenomenological _ relevance and _ in spite _ of the formal difficulties .
fortunately , efficient @xmath8 approximation techniques already exist for the latter particular realistic hermitian models .
they have been developed by many authors ( cf . ,
e.g. , their concise review in [ 24 ] ) .
their thorough and critical tests are amply available but similar studies were still missing in the non - hermitian context .
our present purpose was to fill the gap at least partially .
we have paid thorough attention to the first few open problems related , e.g. , to the possible complex deformation of the axis of coordinates .
our thorough study of a few particular @xmath16 symmetric examples revealed that the transition to the non - hermitian models is unexpectedly smooth .
we did not encounter any serious difficulties , in spite of many apparent obstacles as mentioned in section ii ( e.g. , an enormous ambiguity of the choice of the most suitable zero - order approximation ) . in this way
, our present study confirmed that the angular momentum ( or dimension ) parameter @xmath0 in the `` strongly spiked '' domain where latexmath:[$|\ell artificial perturbation - like parameter @xmath197 which may serve as a guide to the interpretation of many effective potentials as a suitably chosen solvable ( harmonic oscillator ) zero - order approximation followed by the systematically constructed corrections which prove obtainable quite easily .
the participation of mz in this research was partially supported by ga as ( czech republic ) , grant nr .
a 104 8004 .
* figure 1*. cubic oscillator ( 35 ) eigenenergies @xmath203 in ( 44 ) vs @xmath0 for @xmath204 and @xmath205 at different values of @xmath206 @xmath207 znojil m 2001 what is pt symmetry ? `` ( arxiv : quant - ph/0103054 , unpublished ) ; conservation of pseudo - norm in pt symmetric quantum mechanics '' ( arxiv : math - ph/0104012 , unpublished ) and a generalization of the concept of pt symmetry " ( arxiv : math - ph/0106021 , submitted ) ;
|
the pseudo - perturbative shifted - @xmath0 expansion technique pslet is shown applicable in the non - hermitian @xmath1- symmetric context .
the construction of bound states for several @xmath2- symmetric potentials is presented , with special attention paid to @xmath3 oscillators .
[ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary
|
the tested device consists in a system of electrodes designed to hold an electron in a well - defined position over the liquid helium surface .
a single electron transistor ( set ) located below the electron is used to monitor its quantum state .
the whole structure is represented in fig.[sample ] .
it comprises an electron reservoir and the electron trap , which is supplied in electrons from the reservoir , and a guard electrode made out of a thick ( @xmath30.5 @xmath4 m ) layer of nb .
the structure is covered by a saturated helium film @xmath5200 - 400 in thickness .
two electrodes , made out of thin niobium and lying at the bottom of the reservoir , fix its electrostatic potential with respect to the guard .
a gorge through the guard electrode connects the reservoir to the trap . when this electrode is negatively biased relative to the set , a potential barrier forms at this gorge , isolating the trap from the reservoir .
the voltage across the set is modulated by the charge on its island , @xmath6 .
the first sum is over all the conductors in the systems having capacitance @xmath7 and potential @xmath8 with respect to the island .
the second sum is over the image charges induced on the island by stray charges in the rest of the system , such as charges or dipoles in the substrate and the electrons over helium .
the voltage @xmath9 is _ a priori _ unknown .
it depends on the bias current and changes from run to run .
however , it must be periodic with period @xmath10 , the charge of the electron .
to reduce the noise due to fluctuations of the _ dc _ voltage across the set , we apply a low - frequency ( 80 - 150 hz ) modulation of the order of 100 @xmath11v to the guard electrode , which is strongly coupled to the set island .
the voltage variation across the set produced by this modulation is detected by a lock - in amplifier .
the amplitude of this signal @xmath12 is proportional to the derivative of @xmath9 function and , thus , is periodic itself .
we determine @xmath12 experimentally by sweeping the potential , @xmath13 , of one of the electrodes , usually the one which will be swept in subsequent measurements .
we then choose a part of the response to the sweep where background charges did not change , so that several periods of the function @xmath14 can be superimposed by transforming @xmath13 modulo @xmath15 , where the period @xmath15 is adjusted self - consistently . after the modulo transformation the chunk of data is averaged and interpolated by a smoothing spline function @xmath16 .
the rest of the data is fitted piecewise with the function @xmath17 , where the amplitude @xmath18 , and the phase @xmath19 are the fitting parameters .
parameter @xmath19 is the charge , expressed as fraction of @xmath10 , induced on the set island by stray charges in the system .
an electron seeding and monitoring experiment proceeds as follows . 1 .
we seed the electrons on the helium surface by igniting a corona discharge in a small chamber separated from the rest of the cell by a transmission electron microscope sample grid .
the cell is heated to @xmath31.1 k before a discharge .
the presence of the electron normally is easily detected by applying a voltage @xmath20100 @xmath4v at 100 khz to the right reservoir electrode and measuring with a lock - in amplifier the voltage induced on the left electrode .
when electrons appear on the surface , the signal changes by 10 to 200 nv rms . 2 .
after electrons are seeded , we let the system cool down from 1 k and proceed to trap the desired number of them in the trap . typically , at this point , the guard electrode is biased to a negative potential , between @xmath21 and @xmath22 v , and the set is biased to a positive potential between 0 and @xmath23 v. first , we lower the voltage applied to the reservoir electrodes to charge the trap .
care should be taken not to lower the voltage too much , since this can cause irreversible loss of all the electrons .
normally , we find the suitable potentials by trial and error . in typical cases , at least one of the reservoir electrodes stays more positive than the guard , although the electrons can be kept even with both electrodes negative with respect to the guard .
reduced charge on the set island in terms of the potential of the right reservoir electrode .
set potential is + 0.6 v , guard electrode potential 0 .
the bottom insert schematically shows calculated profile of the potential along the centerline of the sample , the dashed line schematically represents local energy of the electrons .
the top insert shows small portion of a raw set signal containing one of the jumps ( indicated by the arrow ) .
( color online ) , height=5 ] 3 .
after the trap is charged , we start ramping up the potential of the right reservoir electrode .
typical results are shown in fig . [ charging ] , where we show the dependence of the phase of the set signal oscillations on the voltage applied to that electrode .
when we sweep the voltage from large positive values down , the charge induced on the set changes little , until we reach the charging voltage .
small changes seen in fig .
[ charging ] are probably due to motion of charges and dipoles in the substrate and occur even before the electrons are seeded for the first time after cooldown from room temperature . at the charging voltage ( @xmath30.26 v in the figure )
the set detects a sudden change in the charge .
since a set measures charge modulo @xmath10 , we can not determine the real change and adjust it by integer number of @xmath10 to `` close the loop '' .
the scheme of distribution of potentials at the moment of charging is shown in the insert of fig . [ charging ] .
the solid line represents the potential created by the electrodes along the center of symmetry of the sample in the plane of the electrons .
the dashed line indicates the energy of the electrons , which , due to coulomb repulsion , is higher than this potential by @xmath24 , where @xmath25 is the electron number surface density and @xmath26 is their distance from the electrode . when ramping down the potential , the energy of the electrons over the reservoir drops , but electrons in the potential well stay trapped . as the voltage on the right reservoir electrode
is raised further , the barrier height decreases and electrons start leaking out one by one .
each electron leaving the trap monitored on the set signal as a sudden change in the phase of the oscillations , shown in the top insert . finally , at @xmath30.63 v the last electron leaves the trap . between @xmath30.57 and @xmath30.63 v only one electron
is left in the trap .
the amplitude of the jump depends on the number of the electrons residing in the trap and goes to @xmath27 for a few electrons .
this value is to be compared with @xmath28 obtained from the numerical simulation .
the length of the plateaus between the jumps depend on the coulomb repulsion between the electrons .
these results are rather similar to the results of the royal holloway group@xcite . however , in the present experiments , the steps are more stable and better defined , being both `` higher '' and `` longer '' , because of the enhanced coupling to the pyramidal set , compared with the flat - island set and the smaller trap used in london .
having developed the technique of trapping the electrons , we proceeded to the measurement of their escape from the trap .
having prepared the trap with one electron , we momentarily open it partially by applying short ( @xmath29s ) pulses to the right reservoir electrode .
we then examine the set signal to determine the presence of the electron .
if it has escaped , we re - populate the trap and repeat the measurement . after accumulating @xmath30 attempts , we plot the resulting electron escape probability _ vs. _ the amplitude of the pulse opening the trap , as shown in fig .
[ escape ] .
the error bars on the figure come from the statistical analysis of the data .
the probability of the escape of a single electron from a trap , in terms of the amplitude of the pulse applied to the right reservoir .
error bars and fits are discussed in the text .
v , @xmath32 v , @xmath33 v. the calculated potential barrier , used in the fit is @xmath34 .
( color online ) , height=5 ] the dependence of the escape probability from a potential well on some parameter can be calculated as in ref .
[ ] : @xmath35 , where @xmath36 is the attempt frequency of the escape , @xmath37 is the pulse length and @xmath38 is the height of the energy barrier , which depends on parameter @xmath39 .
this formula applies also to the case of quantum tunneling through a barrier if it can be approximated by cubic+parabolic potential . in this case
one should use the effective temperature @xmath40 , where @xmath41 is the frequency of the oscillations of the particle in the well .
we have calculated the barrier height in terms of the voltage on the right reservoir numerically , using the known sample geometry . to account for possible contact potentials between electrodes
, we have assumed that at the lowest temperature the escape of the electron is due to quantum tunneling .
then the calculated probability of escape has been fitted to the data with the set potential as the fitting parameter . here
, we did not make assumption about the shape of the potential ; calculated profiles of the potential barrier have been used . using this adjusted value of the set potential , the @xmath42 function
is computed , which is then used in the subsequent fits .
the fits are quite insensitive to the values of the attempt frequency and the pulse length , so we have used the fixed numbers of 40 ghz and @xmath43s , correspondingly .
the escape curve changes as a function of the current through the set . to make this measurement
, we apply the desired current to the set , apply the measuring pulse to the right reservoir and then change the set current to a value suitable for measurements ( @xmath33 na ) . when the current through the set is increased , the curve shifts to the left and becomes less steep .
both effects can be explained by an increase of the electron temperature .
such an increase is not surprising since the current through set generates an electric field in its vicinity as electrons tunnel into and out of the island , changing the potential of the island by @xmath44 , where @xmath45 is the total capacitance of the island .
knowing the barrier , we can calculate the temperature of the electron from the fitting parameters .
we find that the temperature of the electron at zero current through the set is 270 mk , rising to 0.4 k at @xmath46 na while the cell temperature is @xmath4713 mk .
finally , when raising the cell temperature above 270 mk , we observe that the temperature of the electron increases more slowly than that of the cell .
also , we tried to estimate the thermalisation time of the electron .
after heating the electron by @xmath48 we turn off the current and , after a delay , apply the measuring pulse .
we determine the electron temperature by repeating the measurement a number of times and using the procedure outlined above .
we find that the temperature always returns to a stationary value even for the shortest time delay that we could apply of @xmath49s .
both the anomalous behavior of the electron temperature and its short thermalisation time indicate a rather strong coupling with parts of the environment .
we suspect that these features are due to the strong coupling with the pyramidal set and by the large pressing field acting on the electrons .
2 p.m.platzman and m.i.dykman , _ science _ * 284 * , 1967 ( 1999 ) .
lea , p.g .
frayne , and yu .
mukharsky , _ fortschr .
* 48 * , 1109 ( 2000 ) .
p. glasson , g. papageorgiou , k. harrabi , et al . , _ j. physics and chemistry of solids _ * 66 * , 1539 ( 2005 ) .
e. varoquaux and o. avenel , _ phys .rev .
b _ * 68 * , 054515 ( 2003 ) .
|
we have observed trapping of electrons in an electrostatic trap formed over the surface of liquid @xmath0 .
these electrons are detected by a single electron transistor located at the center of the trap .
we can trap any desired number of electrons between 1 and @xmath1 . by repeatedly ( @xmath2 times ) putting a single electron into the trap and lowering the electrostatic barrier of the trap , we can measure the effective temperature of the electron and the time of its thermalisation after heating up by incoherent radiation . in 1999 ,
platzman and dykman@xcite proposed that single electrons electrostatically trapped on the surface of a liquid helium film could be used as qubits and hence form the basis of a quantum computer .
this proposal quickly aroused experimentalistsinterest.@xcite here we present experimental results on the trapping of a single ( or other desired number ) electron and measurements of the upper limit of its relaxation time in view of defining its usefulness as potential qubit .
|
let @xmath1 be the @xmath0-dimensional euclidean space endowed with standard norm @xmath4 and inner product @xmath5 .
we denote the set of full rank lattices by @xmath6 , where a ( full rank ) _ lattice _ @xmath2 of @xmath1 is a set of the form @xmath7 with @xmath8 .
the columns of @xmath9 are called a _ basis _ of @xmath2 .
the determinant @xmath10 of the lattice is independent of the chosen basis . for a lattice @xmath2 ,
let @xmath11 be the family of all centrally symmetric convex bodies , that is , compact convex sets @xmath12 with @xmath13 , which do not contain a non - zero lattice point in their interior .
we denote the volume ( @xmath0-dimensional lebesgue measure ) of a convex body @xmath14 by @xmath15 and its diameter by @xmath16 .
notice that for centrally symmetric sets the diameter is twice the circumradius . _ minkowski s first fundamental theorem _
( see e.g. ( * ? ? ?
* , th . 2 ) and
@xcite ) gives an upper bound for the volume of a convex body @xmath14 : [ mink - theorem ] if @xmath17 and @xmath18 then @xmath19 here we consider the following problem : [ iso - problem ] given a lattice @xmath20 and a real number @xmath21 $ ] , determine the minimum diameter @xmath22 and the bodies @xmath23 for which this minimum is attained .
since the problem is trivial for @xmath24 , we consider it only for @xmath25 . in theorem [ satz1 ]
we solve it partially by giving a description of convex bodies attaining the minimum .
we conjecture ( conjecture [ conj1 ] ) that these convex bodies are the only ones attaining the minimum and we show that the conjecture is valid in many cases , e.g. , for a wide class of lattices ( corollary [ coro_small - radii - theorem ] and proposition [ prop : ridge ] ) and for many values of @xmath26 ( corollary [ small - radii - theorem ] ) .
in particular we give a complete answer for all lattices of dimension @xmath27 , for the integral lattice @xmath28 , the leech lattice @xmath29 , all root lattices @xmath30 with @xmath31 , @xmath32 with @xmath33 , @xmath34 with @xmath35 and all their reciprocals ( theorem [ t : sol_lattices ] ) .
the @xmath36-dimensional case was already completely solved in @xcite by different methods .
before stating our main results we need some more notation . for a lattice @xmath17
we consider its _ dirichlet - voronoi cell _ @xmath37 its volume is equal to @xmath38 , its circumradius is equal to the _ inhomogeneous minimum _ ( also called covering radius ) of @xmath2 , @xmath39 and its inradius ( also called packing radius ) is half of the _ homogeneous minimum _ of @xmath2 , @xmath40 here @xmath41 denotes the @xmath0-dimensional ball centered at the origin with radius @xmath42 . for sets @xmath43 we write @xmath44 to denote the minkowski addition ( vector sum ) .
we write @xmath45 instead of @xmath46 . for a set @xmath47 and a real number @xmath48
we define @xmath49 .
then , if @xmath12 is a @xmath0-dimensional convex body we have @xmath50 and @xmath51 . clearly , @xmath52 is a _ parallelohedron _ ,
i.e. , a convex polytope which tiles @xmath1 by ( lattice ) translations .
furthermore , every @xmath53-dimensional face ( facet ) of @xmath54 contains exactly one lattice point in its relative interior , and it is centrally symmetric with respect to this point ( see e.g. @xcite ) . for @xmath55 $ ]
we define the convex body @xmath56 by choosing the ( unique ) positive real number @xmath57 such that @xmath58 has volume @xmath26 ( see figure [ extremal_sets ] for an example of @xmath56 in @xmath59 ) .
[ satz1 ] let @xmath17 be a lattice and let @xmath55 $ ] .
then for all @xmath14 with @xmath60 the inequality @xmath61 holds .
we present a proof in section [ sec : proofs ] .
we say that a convex body @xmath23 is an _ extremal body _ if it is a solution of problem [ iso - problem ] , hence if @xmath62 .
[ conj1 ] for @xmath17 and @xmath55 $ ] , @xmath56 is the unique extremal body .
notice that conjecture [ conj1 ] is trivially true if @xmath63 because of the classical isodiametric inequality without lattice - point constraints ( see e.g. @xcite ) , which says that for a fixed volume the ball is the only set with minimum diameter . in section [ sec : equality - cases ] we verify conjecture [ conj1 ] for many particular cases : [ t : sol_lattices ] @xmath56 is the only extremal body for any @xmath64 $ ] and for the following lattices ( for explicit descriptions we refer to @xcite ) : all lattices in dimension @xmath36 and @xmath65 , the integral lattice @xmath28 , the leech lattice @xmath29 , all root lattices @xmath30 with @xmath31 , @xmath32 with @xmath33 , @xmath34 with @xmath66 and all their reciprocals .
our proof of theorem [ satz1 ] relies on an equivalent point of view . instead of minimizing the diameter among all convex bodies in @xmath11 with fixed volume ,
we maximize the volume among all convex bodies in @xmath11 with fixed diameter .
it turns out that extremal bodies also maximize volume among all convex bodies in @xmath11 with fixed diameter . for any @xmath67 $ ] the minimum diameter of a body @xmath23 with volume @xmath26
is at most the diameter of @xmath54 , which is equal to @xmath68 .
theorem [ satz1 ] is equivalent to the following statement .
[ satz2 ] let @xmath20 be a lattice and let @xmath69 be at most @xmath68 .
define @xmath26 by @xmath70 .
then for every @xmath14 with @xmath71 the inequality @xmath72 holds . here
we show that theorem [ satz2 ] implies theorem [ satz1 ] . using an analogous argument
the other direction can be proved . for @xmath73 $ ]
let @xmath14 be a convex body with @xmath60 .
suppose that @xmath74 .
then theorem [ satz2 ] yields a contradiction : @xmath75 hence , @xmath76 . for the proof of theorem [ satz2 ] we recall some standard notions from the geometry of numbers .
a lattice @xmath20 is called _ admissible _ for a subset @xmath77 if @xmath12 has no lattice point except the origin in its interior . in particular the lattice
@xmath2 is admissible for all convex bodies in @xmath11 . on the other hand , a lattice @xmath20
is called _ packing lattice _ for @xmath12 if @xmath78 for all distinct @xmath79 . then for a centrally symmetric convex body @xmath12 ,
a lattice @xmath2 is a packing lattice for @xmath12 if and only if it is admissible for @xmath80 ( see e.g. ( * ? ? ?
* , th . 1 ) ) . as mentioned above , the case @xmath81 is covered by the classical isodiametric inequality without lattice - point constraints .
thus we suppose @xmath82 .
let @xmath14 be a convex body with @xmath83 . by definition of @xmath11
the lattice @xmath2 is admissible for @xmath12 .
therefore @xmath3 is a packing lattice for @xmath12 . since @xmath54 is a parallelohedron , @xmath84 ( roughly speaking , the dirichlet - voronoi cell @xmath54 contains the full set @xmath12 in pieces " ) . on the other hand , since @xmath83 we have @xmath85 for all @xmath86 .
then , @xmath87 \cap \operatorname{dv}(2l).\ ] ] the right hand side of is equal to @xmath88 , because if @xmath89\cap \operatorname{dv}(2l)$ ] for @xmath90 , then by the definition of the dirichlet - voronoi cell we have @xmath91 .
therefore , @xmath92 , as required .
in this section we further investigate equality cases . we give geometric conditions on the vertices ( corollary [ coro_small - radii - theorem ] ) and on the facets ( proposition [ prop : ridge ] ) of the dirichlet - voronoi cell @xmath54 which assure the validity of conjecture [ conj1 ] for the corresponding lattice @xmath2 . as a consequence of these results
we obtain a proof of theorem [ t : sol_lattices ] .
let @xmath20 be a lattice and let @xmath55 $ ] . then @xmath93 .
we define @xmath94 let @xmath23 be an extremal body
. then 1 .
@xmath95 , 2 .
the intersection @xmath96 is empty , 3 .
@xmath97 . for every @xmath98 there exists @xmath86 such that @xmath99 . from the definition of @xmath54
we know that @xmath100 is a nearest point of the lattice @xmath3 to @xmath101 .
if @xmath102 , then @xmath103 . consequently , since @xmath100 is a nearest point to @xmath101 , we have @xmath104 .
hence , @xmath105 .
this shows i ) .
\ii ) is obvious since @xmath83 . for iii )
suppose that @xmath106 is strictly contained in @xmath1 .
then there exists a set @xmath107 with positive volume such that @xmath108 does not intersect @xmath106 .
the latter is invariant with respect to translations of @xmath3 which contradicts the extremality of @xmath12 .
let @xmath14 be an extremal body with respect to the lattice @xmath2 and a value @xmath55 $ ] .
since @xmath3 is admissible for @xmath12 , there exists a closed halfspace @xmath109 for every @xmath110 with bounding hyperplane @xmath111 through @xmath112 so that @xmath113 .
since @xmath12 is contained in @xmath114 , we have the representation @xmath115 \|{\boldsymbol{y}}\| < { r_{\!_l}}(v ) } } g_{{\boldsymbol{y}}}^- \cap b_d\bigl({r_{\!_l}}(v)\bigr).\ ] ] notice that we do not have to care about separation of lattice points outside or on the boundary of @xmath114 .
we now consider those @xmath116 with @xmath117 which are facet - centers of the dirichlet - voronoi cell @xmath54 .
notice that all the facets of @xmath54 have a center since they are centrally symmetric .
in some cases we can prove that facet defining hyperplanes of @xmath54 coincide with those of @xmath12 .
in the following @xmath118 denotes the set of lattice points being centers of facets of @xmath54 .
let the facets of @xmath54 be defined by hyperplanes @xmath119 passing through @xmath120 and let @xmath121 denote the corresponding closed halfspaces bounded by @xmath119 and containing @xmath54 , so that @xmath122 is a non - redundant description of @xmath54 .
the following proposition allows to prove conjecture [ conj1 ] for many values of @xmath26 and many lattices .
[ prop : equalfacets ] let @xmath20 be a lattice and let @xmath69 be at most @xmath68 . define @xmath26 by @xmath70 .
let @xmath23 , given as in , be extremal with respect to @xmath2 and @xmath26 , and let @xmath120 with @xmath123 . if the facet @xmath124 of @xmath54 intersects @xmath125 then @xmath126 .
suppose @xmath127 for some @xmath120 satisfying the assumptions of the proposition .
we construct a ball @xmath107 having the following properties : 1 .
@xmath128 , 2 .
@xmath129 , 3 .
@xmath130=\emptyset$ ] .
then the lattice @xmath3 is a packing lattice for @xmath131 . using the argument which was applied in the proof of theorem [ satz2 ]
we obtain the strict inequality @xmath132 , which contradicts the assumption that @xmath12 is extremal .
it remains to construct @xmath107 .
since the facet @xmath133 of @xmath54 intersects @xmath134 by our hypothesis , there exists a vertex @xmath101 of @xmath133 with @xmath135 ( see figure [ fig1 ] )
. lies inside the shaded region.,width=245 ] since @xmath133 is centrally symmetric with respect to @xmath112 , either @xmath101 or @xmath136 lies in the open halfspace @xmath137 . without loss of generality
we assume @xmath138 . on the line segment connecting @xmath101 and @xmath112
there is a point @xmath139 with @xmath140 ( see figure [ fig1 ] ) .
we have @xmath141 and , since @xmath139 lies in the relative interior of a facet of a dirichlet - voronoi cell , for any other @xmath142 it holds @xmath143 .
let @xmath144 where @xmath145 is chosen so that @xmath146 lies in the interior of @xmath147 .
thus there exists @xmath148 sufficiently small such that the ball @xmath107 centered in @xmath146 with radius @xmath149 satisfies properties i)iii ) .
the following result is an immediate consequence of proposition [ prop : equalfacets ] .
[ small - radii - theorem ] let @xmath20 be a lattice and let @xmath69 be at most @xmath68 .
define @xmath26 by @xmath70 .
if every facet of @xmath54 contains a vertex @xmath101 with @xmath135 , then @xmath56 is the unique extremal body with respect to @xmath2 and @xmath26 .
so corollary [ small - radii - theorem ] proves conjecture [ conj1 ] for many values of @xmath26
. we can apply corollary [ small - radii - theorem ] also to prove conjecture [ conj1 ] for certain classes of lattices and any value of @xmath55 $ ] : [ coro_small - radii - theorem ] let @xmath17 be a lattice such that every facet of @xmath54 contains a vertex @xmath101 with @xmath150 .
then @xmath56 is the unique extremal body with respect to @xmath2 and any @xmath55 $ ] .
the case @xmath151 is covered by corollary [ small - radii - theorem ] .
so we assume that @xmath152 . if @xmath12 is an extremal body for the diameter @xmath152 then @xmath153 .
hence for arbitrarily small @xmath145 the convex body @xmath154 is extremal for diameter @xmath155 , since @xmath156 . thus applying proposition [ prop : equalfacets ] to every facet of @xmath157
we get that @xmath158 .
hence @xmath159 .
notice that even in dimension @xmath65 there are lattices to which corollary [ coro_small - radii - theorem ] can not be applied .
we give an explicit example using the notation of _ selling parameters _ from @xcite : let @xmath160 be the lattice defined by the selling parameters @xmath161 and @xmath162 for @xmath163 with @xmath164 and the pair @xmath165 .
a gram matrix of a basis @xmath9 of @xmath2 is for example @xmath166 the dirichlet - voronoi cell of @xmath2 is a truncated octahedron ( or permutohedron , see figure [ f : permutohedron ] ) .
all the vertices of its @xmath167-gonal facet given by @xmath168 , @xmath169 , @xmath170 , @xmath171 have norm @xmath172 whereas the vertex denoted by @xmath173 has norm @xmath174 . in order to overcome this problem in @xmath59 and to solve more equality cases , we prove the following proposition .
it uses the fact that the dirichlet - voronoi cell is a parallelohedron and hence every projection along a @xmath175-face ( _ ridge _ ) is a centrally symmetric hexagon or , as a limiting case , a parallelogram .
this was independently proved by mcmullen @xcite and venkov @xcite .
the facets of the parallelohedron adjacent to the @xmath167 or @xmath176 translates of a ridge are said to form a _
@xmath167-belt _ or _
@xmath176-belt _ respectively .
[ prop : ridge ] let @xmath20 be a lattice and let @xmath69 be at most @xmath68 . define @xmath26 by @xmath70 .
let @xmath23 , given as in , be extremal with respect to @xmath2 and @xmath26 .
let @xmath54 be given as in and let @xmath120 be the center of a facet belonging to a @xmath176-belt of @xmath54 .
if @xmath177 for all @xmath178 , then also @xmath126 and hence @xmath179 . we suppose that @xmath127 , respectively @xmath180 .
consider the sets @xmath181 and @xmath182 , see figure [ f : ridge ] . since @xmath56 is centrally symmetric ,
the isometry @xmath183 maps the closure @xmath184 to @xmath185 and hence @xmath186 . clearly , @xmath187 . by assumption ,
the facet @xmath188 of @xmath54 with center @xmath112 contains two ridges @xmath189 and @xmath190 which are also ridges of other facets in a @xmath176-belt .
since @xmath133 is centrally symmetric with respect to @xmath112 we find a relative interior point @xmath101 of either @xmath189 or @xmath190 which is contained in @xmath191 .
otherwise @xmath111 would be equal to the affine hull of @xmath192 and hence equal to @xmath119 . without loss of generality
we assume @xmath193 and @xmath194 .
we can suppose that the points of @xmath189 all lie in @xmath195 and @xmath196 , because otherwise we could apply corollary [ small - radii - theorem ] to show @xmath126 .
therefore there exists @xmath197 such that @xmath198 lies in the interior of @xmath199 and @xmath200 .
so there exists a ball centered in @xmath198 and contained in @xmath201 with positive volume .
it shows that @xmath202 .
hence @xmath56 has larger volume than @xmath12 which contradicts the extremality of @xmath12 .
corollary [ coro_small - radii - theorem ] and proposition [ prop : ridge ] can be applied to many lattices so that the isodiametric problem is solved for them completely , that is , for every volume respectively every diameter . in the proof of theorem [ t : sol_lattices ] below
we work out this argument for several prominent lattices , where we did not try to be exhaustive .
if the automorphism group of a lattice @xmath2 acts transitively on the facet centers of @xmath52 , then every facet of @xmath54 contains a vertex @xmath101 with @xmath150 .
lattices with this transitivity property are for example the integral lattice @xmath28 and the root lattices @xmath30 , @xmath32 and @xmath34 ( see ( * ? ? ? * chap . 4 , chap . 22 , cor .
to th . 5 ) ) . with a similar argument
it can be shown that corollary [ coro_small - radii - theorem ] applies to the leech lattice @xmath29 and the lattices @xmath203 .
the facets of @xmath204 are given by the lattice vectors of squared length @xmath167 and @xmath176 .
the automorphism group of @xmath29 acts transitively on each of these two sets .
a vertex @xmath101 of `` type @xmath205 '' ( see ( * ? ? ?
23 ) ) satisfies @xmath206 .
this vertex is incident to @xmath207 facets which correspond to vectors of length @xmath167 and to @xmath208 facets which correspond to vectors of length @xmath176 .
hence in every facet a vertex of type @xmath205 can be found .
the case @xmath203 is analogous .
other lattices where corollary [ coro_small - radii - theorem ] can be applied are all the @xmath36-dimensional lattices and the reciprocals @xmath209 , @xmath210 , @xmath211 of the root lattices , since they are examples of lattices @xmath2 for which every vertex @xmath101 of @xmath54 satisfies
@xmath150 ( see ( * ? ? ?
* chap . 4 , chap . 22 , th . 7 ) ) . finally in order to get the solution for @xmath212
, we can use the knowledge of all combinatorial types of dirichlet - voronoi cells in dimension @xmath65 .
it is well known ( fedorov , @xcite ) that there are five combinatorial types of dirichlet - voronoi cells : _ cube _ , _ hexarhombic dodecahedron _ , _
rhombic dodecahedron _ , _ hexagonal prism _ and the _ truncated octahedron_. the first four are degenerations of the last one . the vertices of @xmath54 can be partitioned into equivalence classes by identifying those ones which are lattice translates or point reflections of each other .
vertices of the same class have all the same distance to the origin . for a vertex @xmath101 of a facet with center @xmath112 , the opposite vertex @xmath136 of that facet belongs to the same class . using this symmetry
it is easily checked for all but the truncated octahedron , that every facet of @xmath54 contains a vertex @xmath101 with @xmath150 .
then corollary [ coro_small - radii - theorem ] can be applied to derive that @xmath56 is the unique extremal set . in the remaining type ,
i.e. , the one corresponding to the truncated octahedron ( see figure [ f : permutohedron ] ) , we can apply proposition [ prop : ridge ] .
we find that @xmath54 has three equivalence classes of eight vertices each .
each hexagonal facet of the truncated octahedron contains two opposite vertices of every class and each vertex class has vertices in four quadrilateral facets ( see points on vertices in figure [ f : permutohedron ] ) . assuming that only one class attains the radius @xmath213 , we know by corollary [ small - radii - theorem ] that the condition of proposition [ prop : ridge ] holds , and therefore we obtain @xmath56 is the unique extremal set
for a lattice @xmath20 , the explicit isodiametric inequality for convex bodies @xmath18 can be stated by computing the volume @xmath214 in terms of the diameter @xmath215 . notice that the function defined as @xmath216 is clearly an increasing function of @xmath42 .
then it follows that for any convex body @xmath23 @xmath217 and equality holds when ( and , in many cases , only when ) @xmath179 . for instance , in the case of the 3-dimensional euclidean space and the integral lattice @xmath218 ,
the isodiametric inequality is expressed in the following way ( we write @xmath219 for the sake of brevity ) : @xmath220 bonnesen , t. , fenchel , w. ( 1974 ) _ theorie der konvexen krper_. berlin : springer .
english translation : _ theory of convex bodies_. edited by l. boron , c. christenson and b. smith .
moscow , i d , 1987 : bcs associates
|
in this paper , the isodiametric problem for centrally symmetric convex bodies in the euclidean @xmath0-space @xmath1 containing no interior non - zero point of a lattice @xmath2 is studied .
it is shown that the intersection of a suitable ball with the dirichlet - voronoi cell of @xmath3 is extremal , i.e. , it has minimum diameter among all bodies with the same volume .
it is conjectured that these sets are the only extremal bodies , which is proved for all three dimensional and several prominent lattices .
2000 mathematical subject classification : primary 52a20 , 52c07 ; secondary 52a40 keywords : isodiametric problem , lattices , dirichlet - voronoi cells , parallelohedra
|
the mathematical technique of fourier transform has ubiquitous usage in physics . to demonstrate fourier transform , there are experiments available which use electronic circuits , computer programs@xcite,@xcite and bass guitar strings@xcite . in this paper
, we present a _
mechanical _ means of physically realizing a fourier transform .
a mechanical demonstration of a fourier transform can be achieved using a kundt s tube .
kundt s tube is a simple and an easily available apparatus in undergraduate teaching labs where it is used to measure the speed of sound . in this experiment
, we use the resonance property of the kundt s tube to find the frequencies and amplitudes of component sinusoidal waves which make up an electric signal .
the representation of a function as a sum of sine and cosine terms is called a fourier series@xcite . that is to say , that the right hand side of @xmath0\ ] ] is the fourier series of the function @xmath1 .
@xmath2 are called the fourier coefficients of ` component ' frequencies @xmath3 and can be mathematically calculated using the orthogonality properties of the above trigonometric functions .
historically , kundt ( 1866 ) had used the concept of stationary waves to measure the speed of sound with better accuracy@xcite . even today , we find this tube in undergraduate teaching labs to introduce undergraduate students to one of the methods of measurement of speed of sound .
the kundt s tube resonates when sound waves of certain frequencies pass through it . for a tube closed at one end and the source at the other end ,
the resonating frequencies are , @xmath4 where @xmath5 is length of tube , @xmath6 is speed of sound in the medium encapsulated by tube and @xmath7 is an integer .
the setup consists of the kundt s tube , a cork which is movable along the length of it and a speaker at other end .
cork dust or thermocol pieces are used as mediums for observing striations in the tube .
open source softwares , _ audacity _ and _ praat _ are used , for producing input signals and recording the signal from a microphone in the kundt s tube respectively .
+ * legend - * r : cork , k : kundt s tube , m : microphone , s : speaker , c : computer , i : input from microphone , o : output to speaker this kundt s tube is closed only at one end as the air near the speaker - end forms an anti - node forced by the vibrations of speaker s diaphragm .
_ audacity _ is used to add two sinusoidal waves to make an electric signal which is then fed to the speaker . this electric signal is treated as a signal known to be composed of two sinusoidal waves but whose frequencies and amplitudes are unknown . the aim of the experiment is a primitive one : decompose the electric signal to it s component frequencies and amplitudes . the procedure for achieving the former and latter parts of this aim
is presented in sections ( [ a ] ) and ( [ b ] ) respectively .
the procedure to find the two unknown frequencies of the input signal may be named the qualitative fourier transform . before finding the component frequencies of the electric signal ,
a length - frequency calibration is necessary .
the kundt s tube is filled with an optimum amount of cork dust and the length of the tube is fixed by the cork .
the resonant frequency for this length is estimated by feeding the speaker a range of frequencies using a function generator and judging which frequency sets up the tallest striations in the cork dust .
the resonant frequencies for a certain number of lengths is similarly found and plotted ( fig.([c ] ) ) .
note that for the whole calibration , one should decide upon a single mode @xmath8 .
we chose @xmath9 and selected frequencies which set up the corresponding pattern in the cork dust .
now , the speaker is fed with the unknown signal and the length of the cork is continually changed till the tube resonates at mode @xmath9 .
this happens at two different lengths ) . ] which are noted .
the resonant frequencies corresponding to these lengths is found using the calibration plot ( [ c ] ) .
this concludes the qualitative fourier transform as the two unknown frequencies that comprises the signal is found . in the experiment that we performed ,
the electric signal fed to the speaker comprised of the frequencies : 358 hz and 448 hz and the qualitative ft found resonance at first mode at the lengths of 69 cm and 54 cm corresponding to the frequencies 358 hz and 448 hz .
quantitative fourier transform is the procedure to find the amplitudes corresponding to the two unknown frequencies found using the qualitative ft . to quantify the fourier transform ,
we need two more calibration curves .
the cork is placed at one of the resonant lengths of the unknown signal and the cork - dust which has served its purpose is removed .
a microphone is placed at the displacement anti - node .
the speaker is fed with the corresponding resonant frequency at different input voltages using _ audacity _ and the average sound intensity measured using _
praat _ is recorded for each input voltage .
an average recorded intensity - input voltage calibration plot is thus obtained . at this point , when we attempted to verify the shape of the waveform recorded by _ praat _ , we found that the waveform was being truncated at the crests and troughs for high input voltages . to remove the associated errors in the average intensity recorded
, we scaled down the volume in the computer appropriately . once the calibration for a particular resonant length is done , the same is repeated for the other resonant length .
the calibration curves got are shown in fig.([d ] ) . to find the amplitudes of the component frequencies of the electric signal ,
the kundt s tube is kept at one of the resonant length and the microphone at the displacement anti - node .
the unknown signal is fed to the speaker ( at the same scaled down computer volume ) and the average intensity is recorded .
the corresponding input voltage is found using the calibration plots ( [ d ] ) .
similar procedure is repeated for the other resonant length .
note that due to the scaling down of computer volume , the amplitudes here correspond to the scaled down electric signal and not the original electric signal which was used in sec.([a ] ) .
this concludes the quantitative fourier transform as the amplitudes of the two component frequencies are estimated . in the experiment we performed , we fed these two frequencies at different pairs of amplitudes and the quantitative ft s estimated amplitudes are tabulated in table ( [ e ] ) .
.quantitative fourier transform [ cols= " < , < , < , < , < , < , < , < , < " , ] it can be observed from table ( [ e ] ) that other than readings 8 , 9 and 10 , the amplitude measurements fall under a resolution of 0.15 v hence giving a rough estimate of the input amplitudes in the unknown electric signal .
the complete process of finding out the component frequencies of input signal by scanning the length of kundt s tube and finding out the amplitudes of the corresponding frequencies using calibration curves may be called as performing the fourier transform of an electric signal using kundt s tube .
mechanical systems having discrete resonant frequencies can be considered as candidates for performing a fourier transform of an electric signal .
preceding the mechanical system should be a device which can convert electric signals to mechanical signals which can be fed to a resonace capable mechanical system .
example - springs of different resonant frequencies put on a hanger .
a mechanical vibrator can be used to convert electric signals to mechanically vibrate the hanger .
only those springs whose resonant frequencies match with the frequency of the sinusoidal waves in electric signal will show resonance .
though unnecessary for the present discussion , it is instructive to look at the fourier transform that _
audacity _ can perform on the sound signal recorded by the microphone when the speaker is fed with a sinusoidal resonant frequency of the tube .
fig.([f ] ) shows such a plot where the speaker was fed with a 367hz signal which was a resonant frequency for a particular length of the tube .
it can be seen that though the speaker was fed with a single frequency , the constructive interference of certain frequencies contained in the noise are seen as peaks .
these are different modes of resonance of the tube at that particular length .
as undergraduate students have more intuition of sound waves than of electronics or computer codes , we expect them to relate more easily to the meaning of fourier transforms .
we would like to thank dr .
k. senapati for getting us into this problem .
we thank n. anand for suggesting the kundt s tube for performing a mechanical ft .
we thank dr .
r. das for his encouragement and school of physical sciences - national institute of science education and research , bhubaneswar - india for funding this experiment .
m. courtney , n. althausen _ arxiv : physics/0605154 _ [ physics.ed-ph ] ( 2006 ) .
mathematical methods for physics and engineering - riley , hobson , bence , 3rd ed . _ cambridge university press _ ( 2006 ) .
_ annalen der physik ( leipzig : j. c. poggendorff ) _ 127 ( 4 ) : p.497523 ( 1866 ) .
|
an experiment to demonstrate the fourier transform of an electric signal using the kundt s tube is described .
the results of finding the component frequencies and an approximation to the amplitudes of two sinusoidal signals which compose an input electric signal is presented .
undergraduate students are expected to more easily relate to the meaning of a fourier transform through such mechanical demonstrations .
|
the famous hubble expansion drives more distant objects to recede faster from us .
recent observations of supernovae argue that this expansion is in fact accelerating , implying an existence of dark energy with negative pressure @xcite .
this dark energy , which contributes two third of energy density in the present universe , is mysterious in its physical origin .
precise measurements of its time evolution will be crucial to uncover the identity of this energy component .
one of most promising probes to measure the dark energy is a standard ruler called baryon acoustic oscillations in large - scale clustering @xcite baryon acoustic oscillations ( hereafter bao ) arise from sound waves that propagated in the hot plasma of tightly coupled photons and baryons in the early universe . as the universe expands and cools
, photons finally decouple from baryons 400,000 years after the big bang , with the sound waves revealed as the acoustic oscillations in the anisotropies of the cosmic microwave background ( hereafter cmb ) @xcite .
the equivalent but attenuated feature exists in the clustering of matter , as baryons fall into dark matter potential well after the recombination @xcite .
these were recently detected in galaxy redshift surveys @xcite .
good cmb anisotropy data provides the absolute physical scale for these baryonic peaks ; comparing this to the observed location of the baryonic peaks in the two - point correlation function or power spectrum provides measurements of cosmological distance scale . clustering in the transverse direction probes the angular diameter distance , and clustering in the line - of - sight direction probes the hubble parameter .
the cosmological distance scale as a function of redshift is then the record of the expansion history of the universe , which in turn measures the evolution of dark energy density .
many studies have been devoted to estimations of precision on cosmological distance scale achievable using the bao from future galaxy redshift surveys @xcite .
a common scheme is to assume the gaussian statistical errors on the power spectrum , which are constructed straightforward from the finite volume of the survey and the incomplete sampling of the underlying density field with galaxies @xcite , and then to propagate these errors to constrain the errors on cosmological parameters including cosmological distance scale whether by using monte - carlo simulations or using analytic method such as a fisher matrix formalism . in @xcite
, we used a fisher matrix formalism to show that the bao from future galaxy redshift surveys can constrain the cosmological distance scale to percent level precision , thereby providing robust measurements of present - day dark energy density and its time - dependence that are comparable to future type ia supernova surveys . the most important element in these methods , other than the survey specifications , is how to include the effects of nonlinear structure formation , which depends on redshift as well as bias .
the primary nonlinear effect on the standard ruler test is that the bao signature is reduced with time by the nonlinear growth of density fields into present - day structures @xcite .
this nonlinear mode - coupling also alters the overall shape of the power spectrum by increasing the small - scale power relative to the linear growth of power spectrum ( * ? ? ? * and references therein ) . at the same time , this mode - coupling increases statistical errors above our gaussian approximations @xcite .
the nonlinear effects first appear on small scales and then proceed to larger scales , tracing the hierarchical formation of structures .
nonlinear redshift distortions and nonlinear galaxy bias can further degrade the bao ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and modify the statistical variance . in @xcite , we used a sharp wavenumber cut to exclude the nonlinear wavenumbers from the fisher matrix formalism .
that is , we treated all wavenumbers up to a certain threshold as linear and excluded all larger wavenumbers from our calculations , while the threshold varied depending on redshift .
our choice of linear wavenumbers was based on the magnitude of underlying mass density fluctuations rather than direct estimations of degradation of baryonic signature with redshift , as we did not have any means to estimate the latter without n - body simulations .
in addition , the effects of bias or redshift distortions on the bao were too ambiguous to parameterize and include in the fisher matrix .
however , the nonlinear effect does not turn on sharply with wavenumber but instead appears gradually . in @xcite , we used n - body simulations to study the effect of nonlinear growth , bias , and redshift distortions on the appearance of the bao . while we have shown in that paper
that our choice of linear scales in @xcite reasonably represents the amount of standard ruler information at different redshift , we still lacked a quantitative description of the gradual erasure on the bao as a function of redshift , bias , and redshift distortions .
recently , a quantitative model has been provided by @xcite . in a two - point correlation function ,
the bao signature is an excess of pairs separated by its characteristic scale of @xmath0 .
@xcite models the nonlinear process of erasing the bao signature as differential motions or lagrangian displacements of these pairs of tracers . using the n - body simulations from @xcite
, we estimated the amount of differential motions caused by nonlinear growth , halo bias , and redshift distortions at different redshifts .
we then showed that the nonlinear erasure of the bao signature is successfully modeled by approximating these displacement field as gaussian and convolving this lagrangian displacement distribution with the bao peak in the two - point correlation function , or multiplying the linear power spectrum by the fourier transform of the gaussian function .
as noted in @xcite , estimating the lagrangian displacement fields with a reasonable precision does not require a simulation as large as estimating the degradation in the bao signature does : lagrangian displacements are calculated from all pairs separated by @xmath0 , while the bao depend on the small number of excess pairs at that separation . in this paper , we use the lagrangian displacement distribution calculated from @xcite to modify the fisher matrix formalism presented in @xcite so as to correctly reflect the gradual loss of standard ruler information from the bao in scale and in time due to not only nonlinear growth but also bias and redshift distortions .
we isolate the cosmological distance information from the bao more strictly than in @xcite , i.e. , without information from @xcite test or from the broad - band shape of power spectrum , and derive distance error estimates using the fisher matrix calculations .
we show that the full , 12-dimensional ( for one redshift bin ) fisher matrix in @xcite can be contracted to a 2-dimensional or even 1-dimensional covariance matrix of @xmath1 and @xmath2 with physically motivated approximations .
we present the resulting fitting formula , for errors from the full dimensional fisher matrix , based on the reduced 2-dimensional covariance matrix , which is a function of survey volume , galaxy number density , and parameterized nonlinearity .
note that the effect of bias and redshift distortions on the bao is now straightforward to parameterize as an increase in lagrangian displacement field and easy to mingle into the fisher matrix calculation .
we show that the application of the formula extends to the photometric redshift surveys .
this fitting formula will save considerable amount of computational efforts in forecasting distance errors for future galaxy surveys .
we compare these results with distance constraints from a @xmath3 analysis of the full n - body results from @xcite .
our fitting formula differs from @xcite in many details , such as the different treatments of nonlinear degradation of standard ruler test and ours having a structure based on physically motivated model . in
[ sec : full ] , we present the full fisher matrix calculation that is upgraded by using the lagrangian displacement field . in [ sec:1da ] , we show a 1-dimensional model as an approximation of the full fisher matrix in
[ sec : full ] . in
[ sec:2da ] , the 1-dimensional model is extended to a 2-dimensional model .
we present the resulting fitting formula .
we also discuss the extension of this formula for photometric redshift surveys . in
[ sec : result ] , we compare distance errors from the fitting formula with the estimates from the full fisher matrix . in
[ sec : nbody ] , we compare the distance errors from the full fisher matrix calculations ( or from the fitting formula ) with the distance errors from a @xmath3 analysis of the n - body data .
in this section , we construct the fisher matrix that assesses the amount of distance information available exclusively from using the bao as a standard ruler .
this enables us to use the lagrangian displacement distribution @xcite to account for the nonlinear effects on baryonic peaks and therefore nonlinear effects on the distance precision . assuming the likelihood function of the band powers of galaxy power spectrum to be gaussian , the fisher matrix is approximately @xcite : @xmath4 where @xmath5 is the observed power spectrum at @xmath6
, @xmath7 is the cosine of the angle of @xmath6 with respect to the line - of - sight direction , @xmath8 are the cosmological parameters to be constrained , and @xmath9 is the effective volume of the survey , given as @xmath10 ^ 2 d\vec{r } = \left [ \frac{{n}p(k,\mu)}{{n}p(k,\mu)+1 } \right ] ^2 { v_{\rm survey}}=\left [ \frac{{n}p(k)(1+\beta \mu^2)^2}{{n}p(k)(1+\beta \mu^2)^2 + 1 } \right ] ^2 { v_{\rm survey}}. \label{eq : veff}\ ] ] here @xmath11 is the comoving number density of galaxies at the location @xmath12 , @xmath13 is the linear redshift distortion parameter , and @xmath14 is the survey volume .
this formulation assumes that the density field is gaussian and that boundary effects are not important . the second equality in equation ( [ eq : veff ] )
holds only if the comoving number density @xmath15 is constant in position , and the third equality assumes linear redshift distortions @xcite . from @xcite ,
the lagrangian displacement fields due to nonlinear growth and nonlinear redshift distortions at separation of @xmath16 can be approximated as an elliptical gaussian function . since most of the effect comes from the bulk flow , galaxy bias does not have much effect on lagrangian displacement .
the rms radial displacements across ( @xmath17 ) and along the line of sight ( @xmath18 ) at this separation follow @xmath19 and @xmath20 , where @xmath21 for a cosmology with @xmath22 at the present - day , @xmath23 , and @xmath24 is the growth function while @xmath24 is normalized to be @xmath25 at @xmath26 such that @xmath27 at high @xmath28 .
@xmath29 should be scaled proportionally to @xmath30 .
then the surviving baryonic features in the nonlinear power spectrum can be expressed as : @xmath31}\end{aligned}\ ] ] where @xmath32 is the portion of the linear power spectrum @xmath33 with the acoustic signature , including the effects of silk damping @xcite . from equations ( [ eq : fij ] ) and ( [ eq : veff ] ) , the distance precision from a galaxy redshift survey depends on @xmath14 , @xmath34 , and the redshift of the survey .
the redshift dependence enters because of the loss of information due to nonlinear effect , such as the erasure of baryonic features as well as the nonlinear growth of power at large wavenumber . in @xcite ,
the scales degraded by nonlinear effects were removed from the fisher matrix calculation by cutting off the wavenumber integral at a @xmath35 , where @xmath35 depends on redshift by requiring @xmath36 at a corresponding @xmath37 . in reality , the nonlinear effect is progressive with @xmath38 : there is acoustic scale information even beyond @xmath35 while some information is lost within @xmath39 @xcite . now that we have a good model to describe the degradation in baryonic features ( eq . [ [ eq : surpbao ] ] ) , we can remove the redshift dependent @xmath35 .
we set @xmath40 for all redshifts and change @xmath18 and @xmath17 to account for the erasure of baryonic features . as we seek to isolate the information from the acoustic scale , we attach the exponential suppression in equation ( [ eq : surpbao ] ) to the full power spectrum when computing @xmath41 .
this means that we are removing all information , baryonic and non - baryonic , from smaller scales , where the broad - band shape of the power spectrum might give distance information .
moreover , when we compute the derivatives of @xmath42 , we take the exponential factor in equation ( [ eq : surpbao ] ) outside of the derivatives .
this is equivalent to marginalizing over a large uncertainty in @xmath17 and @xmath18 when computing distance errors ; we do not want to include any distance information from the anisotropy and scale of the nonlinear damping of the acoustic peaks . with these prescriptions ,
the fisher matrix becomes @xmath43}\frac{2\pi k^2 dk d\mu}{2(2 \pi)^3}\nonumber \\\end{aligned}\ ] ] where @xmath44 is the linear power spectrum . to this point , we have not excluded the distance information on linear scales from non - baryonic features .
there are two sources of such information : first , from the @xcite test ( hereafter ap test ) using angular anisotropy of power spectrum , and second , from the power spectrum not following a simple power law .
first , at a given @xmath38 , both the redshift distortions from large scale infall and the cosmological distortions give rise to angular anisotropies in power .
when the linear redshift distortion parameter @xmath13 becomes very small , both distortions converge to an identical angular signature , a quadratic in @xmath7 , causing the effect of redshift distortions to be degenerate from the effect of @xmath45 . in this case
, the distance information from the ap test will be suppressed .
however , when @xmath13 is non - negligible , as we assume a specific angular dependence of redshift distortions ( e.g. , linear redshift distortions ) , the two effects can be in principle distinguished as both distortions have different higher multiples , and we can therefore isolate the cosmological distance information on @xmath45 ( i.e. , the ap test ) from the effect of @xmath13 .
second , if the galaxy power spectrum follows a simple power law without any preferred scale , the cosmological distortions remain degenerate from the redshift distortions for a negligible @xmath13 .
however , the galaxy power spectrum contains another preferred scale , the horizon scale at the epoch of matter - radiation equality .
this effect produces much broader feature in power spectrum , but in principle can provide an extra standard ruler @xcite .
however , we consider both of these sources of information less robust than the bao . for the ap test
, we do not believe we understand the quasilinear and nonlinear behavior of redshift distortions very accurately . for the broadband shape of the power spectrum
, the resulting distance information will be susceptible to the systematic effect such as tilt , nonlinear bias , or nonlinear growth in power .
therefore we want to remove distance information from both non - baryonic features . for the distance information from the broad - band shape ,
we remove this by computing the fisher matrix @xmath46 with @xmath47 to produce a reasonable cmb information . ] and subtracting it from the fisher matrix @xmath48 with the fiducial @xmath49 .
in addition , in order to remove any distance information from a presumed form of redshift distortions , we assume @xmath50 in the derivatives in equation ( [ eq : fullfij ] ) in computing both @xmath51 and the fiducial fisher matrix . that is , the angular ap test is removed .
meanwhile , we hold the amplitude of power in @xmath9 and therefore the statistical error per each @xmath38 bin constant by keeping @xmath13 in the redshift distortions @xmath52 in @xmath9 ( eq . [ [ eq : veff ] ] ) unchanged . note
that our inclusion of an arbitrary growth function already implies that any constraints on @xmath13 are not yielding distance scale information from the amplitude of the power spectrum ; setting @xmath53 in derivatives will not have any further effects .
the resulting fisher matrix after subtraction by @xmath51 ( while @xmath54 is the same ) will only contain the distance information from the bao .
we ignore the minor effect that the broad - band shape of the fiducial power spectrum is slightly different from that of the @xmath55 case .
the resulting fisher matrix is combined with the cmb information ( _ planck _ satellite including polarization ) and then inverted to give a covariance matrix .
we calculate the distance errors marginalized over a total of 12 parameters including the angular diameter distance ( @xmath56 ) and the hubble parameter ( @xmath57 ) at the redshift of the galaxy survey and 10 others : @xmath58 , @xmath59 , @xmath60 , tilt ( @xmath61 ) , @xmath62 , the normalization , @xmath63 , @xmath64 , an unknown growth rate ( @xmath65 ) , and an unknown shot noise . in detail
, we use the 1st year and the 3rd year wilkinson microwave anisotropy probe ( hereafter , wmap1 and wmap3 , respectively ) results as our fiducial cosmologies @xcite .
our fiducial model for wmap1 is then @xmath66 , @xmath67 , @xmath68 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , and @xmath73 . for wmap3
, we use @xmath74 , @xmath75 , @xmath76 , @xmath69 , @xmath77 , @xmath78 , @xmath79 , and @xmath73 . by adopting @xmath80 ( i.e. , @xmath17 and @xmath18 ) , we no longer need to specify the redshift of survey , and the distance error will only depend on @xmath14 , @xmath34 , and @xmath80 .
we assume a fiducial galaxy survey of @xmath81 . while there is a simple scaling relation for varying @xmath14 , the effect of @xmath34 depends on the value of @xmath80 .
we try various @xmath34 and a large range of @xmath80 .
we characterize @xmath34 as a value at @xmath82 , i.e. , @xmath83 , where @xmath42 is real - space power . in redshift space ,
the linear theory redshift distortions enhance the power along the line - of - sight direction by @xmath84 against shot noise .
motivated by the single peak in the correlation function , we next consider whether acoustic distance scale precision can be modeled simply as the problem of centroiding a peak in the presence of the noise generated from shot noise and the cdm power spectrum .
we study this in spherical geometry in this section and generalize to the anisotropic case in the next .
nonlinear effects will broaden the peak and therefore increase the uncertainty in measuring the location of the peak .
we approximate the full fisher matrix based on how well we can centroid the location of the peak , that is , the sound horizon @xmath85 at the drag epoch when observed in the reference cosmology : @xmath86 ^2 \frac{4\pi k^2 dk}{2(2 \pi)^3}. \end{aligned}\ ] ] we first find an appropriate form of @xmath87 . if the peak in the correlation function is a delta function at @xmath88 , the fourier transformation of the peak takes the following form in the power spectrum : @xmath89 a broadened peak will have the same power spectrum multiplied with an additional damping envelope set by the fourier transformation of the broadened peak shape . for example , if the peak in the correlation function is broadened with a gaussian function due to the silk damping effect ( @xmath90 ) and lagrangian displacement distribution ( @xmath80 ) , then the convolution in configuration space with the gaussian function is fourier - transformed to a multiplicative exponential factor in fourier space . @xmath91 where @xmath92 . in reality , the decoupling between photons and baryons is not instantaneous at the recombination , and therefore the resulting silk - damping effect @xmath93 integrated over time deviates from a gaussian and can be better approximated by @xmath94 with the silk - damping scale @xmath95 @xcite .
we therefore have @xmath96 starting from physical models of the transfer function gives the same results .
let @xmath97 .
@xmath98 ^ 2\\ & = & a k^n \left [ \left ( \frac{{\omega_c}}{{\omega_m } } t_c(k ) \right ) ^2 + 2\frac{{\omega_c}}{{\omega_m}}\frac{{\omega_b}}{{\omega_m } } t_b(k)t_c(k)+\left ( \frac{{\omega_b}}{{\omega_m } } t_b(k ) \right ) ^2 \right ] \end{aligned}\ ] ] as @xmath99 is a slow function of @xmath38 relative to @xmath100 , the leading order term for baryonic features can be approximated to @xmath101 from @xcite ( and references therein ) , @xmath102 where the silk damping effect @xmath103 . including the nonlinear damping returns equation ( [ eq : ksilk ] ) .
@xmath104 as the amplitude of @xmath105 grows with redshift by @xmath106 , we describe the normalization of @xmath107 as : @xmath108 where @xmath109 is galaxy power at @xmath82 at the given redshift .
we now differentiate @xmath105 to calculate @xmath87 .
@xmath110 \exp{(-(k{\sigma_s})^{1.4})}\exp{(-k^2 { \sigma_{\rm nl}}^2/2)}.\ ] ] then the fisher matrix becomes : @xmath111 ^ 2\exp{(-2 ( k{\sigma_s})^{1.4})}\exp{(-k^2 { \sigma_{\rm nl}}^2 ) } \frac{4\pi k^2 dk}{2(2 \pi)^3}. \nonumber\\\end{aligned}\ ] ] the comoving sound horizon is @xmath112 @xmath16 and most of baryonic information comes from @xmath113 , which keeps @xmath114 large over the wavenumber of our interest .
therefore , the sinusoidal terms oscillate rapidly relative to the wavenumber dependence of the exponential factor . treating these oscillations as rapid relative to all other wavenumber dependence , we approximate the leading term @xmath115 as @xmath116 . the large @xmath114 values also leave @xmath117 small relative to @xmath115 and so we drop this term .
these approximations then yield @xmath118 the resulting error on the location of the baryonic peak is @xmath119 ^{-1/2}\label{eq : sig1d}.\end{aligned}\ ] ] the fractional error on the location of the peak from the observed galaxy redshift surveys , @xmath120 ( @xmath121 ) , is equivalent to the fractional error on the distance estimation when the physical location of the peak , that is , the true value of the sound horizon @xmath122 at the drag epoch , is well known from the cmb .
more specifically , it is equivalent to the distance information exclusively from baryonic peaks as a standard ruler , as we have used only @xmath123 in deriving equation ( [ eq : f1d ] ) . in comparison to the full fisher matrix ( eq . [ [ eq : fullfij ] ] ) , knowing that @xmath124 in equation ( [ eq : fullfij ] ) is dominated by @xmath125 , we hypothesize that equation ( [ eq : f1d ] ) is a good approximation to equation ( [ eq : fullfij ] ) in spherical symmetry .
in this section , we upgrade the spherically symmetric model in the previous section to a 2-d model . in allowing anisotropic behavior of the correlation function ,
we want to measure the location of the baryonic peak in the correlation function along ( @xmath126 ) and across the line of sight direction ( @xmath127 ) in the reference cosmology , that is , two axes of an oblate ellipsoidal ridge in the correlation function .
the fourier transform of the ellipsoid is @xmath128 where @xmath129 .
measuring the fractional errors on @xmath127 and @xmath126 is equivalent to measuring the fractional errors on @xmath130 and @xmath131 , respectively , where @xmath122 is the true physical value of the sound horizon .
when the precision on the sound horizon @xmath122 from the cmb data is much better than the precision on @xmath131 and @xmath130 from galaxy redshift surveys , again , the errors on @xmath127 and @xmath126 are virtually equivalent to the errors on @xmath1 and @xmath2 . when @xmath132 , the baryonic peak along the line of sight direction is further broadened . in fourier space ,
the modes along the line of sight are damped further than other modes at given @xmath38 , introducing an angular dependence within the integrand of @xmath133 .
we choose parameters @xmath134 and @xmath135 .
we choose @xmath136 instead of @xmath137 to be consistent with the sign of derivatives for @xmath1 in equation ( [ eq : fullfij ] ) .
using @xmath137 instead would only change the sign of the off - diagonal term of the final fisher matrix we will derive . the derivatives with respect to the anisotropic distances are @xmath138 when we evaluate the derivatives @xmath139 and @xmath140 at the fiducial cosmology , i.e. , the true cosmology where @xmath141 . using these
we can write the 2-d fisher matrix as @xmath142 ^ 2 \left [ \frac{\partial \ln x}{\partial p_i}\frac{\partial \ln x}{\partial p_j } \right ] \frac{2\pi k^2 dk\;d\mu}{2(2 \pi)^3 } \nonumber \\\end{aligned}\ ] ] where @xmath143 is the linear redshift distortions .
we can make a similar approximation to equation ( [ eq : dpbao ] ) for @xmath144 : @xmath145 we subsequently approximate that @xmath146 , as before .
@xmath147 } \nonumber \\ & & \int_{-1}^{1}\frac{d\mu}{2 } \frac{(\sqrt{8 \pi^2 } a_0 { p_{0.2}}r(\mu))^2}{(p(k)r(\mu)+{n^{-1}})^2 } f_i(\mu ) f_j(\mu ) \exp{\left [ -k^2(1-\mu^2){\sigma_\perp}^2-k^2\mu^2{\sigma_\parallel}^2 ) \right ] } \\[6pt ] & = & { v_{\rm survey}}a_0 ^ 2 \int_{0}^{1 } d\mu\ ; f_i(\mu ) f_j(\mu ) \int_{0}^{\infty } dk\,{k^2 \exp\left [ -2(k{\sigma_s})^{1.4}\right ] \over \left({p(k)\over{p_{0.2}}}+{1\over n{p_{0.2}}r(\mu)}\right)^2 } \exp\left[-k^2(1-\mu^2){\sigma_\perp}^2-k^2\mu^2{\sigma_\parallel}^2\right ] \label{eq : f2d } \nonumber \\\end{aligned}\ ] ] where @xmath90 is the silk damping scale , @xmath17 is the real - space nonlinear lagrangian displacement , and @xmath18 is the redshift - space nonlinear displacement along @xmath148 .
equation ( [ eq : f2d ] ) holds even if the redshift distortion effect @xmath52 deviates from the linear redshift distortions : e.g. , @xmath149 ( see [ subsec : redd ] ) .
the quantity @xmath150 becomes the effective shot noise @xmath151 .
any additive power due to nonlinear growth or bias will decrease @xmath152 below @xmath153 .
we propose to use equation ( [ eq : f2d ] ) as our fitting function to the full fisher matrix calculation ( eq . [ [ eq : fullfij ] ] ) . as usual , the covariance matrix is the inverse of the fisher matrix .
unless @xmath17 or @xmath18 is very large , the @xmath7 dependence of the integrand is mild and therefore easy to compute .
we find excellent convergence using a simple midpoint method with a grid of 20 points in @xmath7 and 50 points in @xmath38 with @xmath154 ( up to @xmath155 ) .
one can precompute @xmath156 at the gridpoints in @xmath38 , as these do not depend on @xmath157 , @xmath13 , @xmath17 , or @xmath18 .
the integral over @xmath38 can be done once for all three matrix elements , so that computing the full matrix takes negligibly more time than computing one element . by comparing the numerical results from equation ( [ eq : f2d ] ) with those from equation ( [ eq : fullfij ] ) , which is presented in
[ sec : result ] , we derive @xmath158 .
we calculate @xmath159 from the equation given in @xcite h^{-1 } ( { h{\rm\;mpc^{-1}}})$ ] ] . for wmap1 , @xmath160 and @xmath161 . for wmap3 , @xmath162 and @xmath163 .
note that @xmath14 should be in @xmath164 if @xmath38 is in @xmath165 units .
the derived @xmath158 values are consistent with the analytic estimates from equation ( [ eq : fullp ] ) although not exact .
approximating the silk damping effect as a gaussian function , that is , @xmath166 , is less exact in the limit of large @xmath80 and a small @xmath83 but is convenient because we can add the effect of the silk damping and the nonlinear damping quadratically ( i.e. , @xmath92 in eq .
[ [ eq : gausilk ] ] ) .
we quote @xmath167 and @xmath168 for this case as well .
for wmap1 , we calculate @xmath169 with a choice of @xmath170 , and for wmap3 , @xmath171 with a choice of @xmath172 .
the contribution of @xmath153 in equation ( [ eq : f2d ] ) implies that the effect of redshift distortion @xmath52 offsets the behavior of the shot noise @xmath173 . on large scales ,
the anisotropic contribution from @xmath174 relative to @xmath175 will increase @xmath176 along the line of sight , decreasing the effective shot noise not only for the measurement of @xmath2 but also for @xmath1 , albeit by a smaller amount .
if the distance to galaxies are uncertain in an uncorrelated way , e.g. , due to thermal peculiar velocities ( i.e. , the finger - of - god effect ) or due to photometric redshifts , it is straightforward to include this effect in equation ( [ eq : f2d ] ) by fixing @xmath174 : we simply need to include a gaussian uncertainty that corresponds to halo velocity dispersion due to the finger - of - god effect or redshift uncertainty for photometric surveys .
that is , @xmath177 note that we do not increase values of @xmath18 by the amount of uncertainty in the distance .
the reason why we only need to modify @xmath52 , whether due to the nonlinear redshift distortions or due to photometric redshift errors , is because the resulting exponential suppression in power spectrum ( @xmath178 ) not only decreases the signal but also decreases the variance from the cdm power spectrum in the line - of - sight direction .
the net effect is thus equivalent to a relative increase of shot noise . in
[ subsec : photoz ] , we present distance error estimates from a photometric redshift survey using equation ( [ eq : rsig ] ) . in the case of photometric redshift errors
, there is an additional drawback other than the exponential suppression of power in the line - of - sight direction : features in the transverse power spectrum are smeared by projections of clustering at different redshifts onto a mean redshift .
we ignore this effect in our formulation .
as we presented in @xcite , this effect will increase the error on @xmath1 by @xmath179 for photometric uncertainty ( @xmath180 ) of @xmath181 in @xmath182 at @xmath183 .
meanwhile , using @xmath184 may be a redundant correction in the case of thermal peculiar velocities , as we may already be including these peculiar velocity term in the computation of @xmath18 ( of course , in redshift space ) from simulations .
in fact , it is more conservative to put this term into @xmath18 than it is to put into @xmath185 , because while both cases decrease the signal by the same amount , the latter case also decreases the noise by the same amount while the former does not .
therefore , we do not take any steps to isolate thermal velocities in the lagrangian displacements and move that contribution to @xmath185 .
we simply reserve @xmath90 for the inclusion of observational uncertainties such as spectroscopic or photometric redshift errors . when @xmath186 and @xmath175 , the nonlinear exponential damping becomes isotropic , and the 2-d fisher matrix ( eq . [
[ eq : f2d ] ] ) reduces to 1-d fisher matrix ( eq .
[ [ eq : f1d ] ] ) multiplied by simple angular integrals .
@xmath187 this gives a @xmath188 matrix of @xmath189 then the covariance matrix is @xmath190 the off - diagonal term is not zero , implying that the constraints on @xmath191 and @xmath126 , and hence the constraints on @xmath1 and @xmath2 , are not independent even in the limit of superb data .
the off - diagonal term being positive means that @xmath1 and @xmath192 are anticorrelated . for all modes except those at @xmath193 and @xmath194 ,
there exist positively correlated changes in @xmath1 and @xmath2 that leave the measured quantity of the mode unchanged .
summing all the modes , with relative weight depending on @xmath7 , leaves a net anticorrelation between @xmath1 and @xmath192 .
geometrically , we are attempting to measure the axes of an ellipsoidal shell , but a quadrupole distortion of the shell is less well constrained because it leaves the intermediate angles relatively unchanged . the value of @xmath195 from equation ( [ eq : c1d ] ) .
in fact , we find that @xmath196 regardless of the choice of @xmath17 , @xmath18 , @xmath157 , and @xmath13 .
we recommend that this covariance be included in assessing the implications of acoustic scale measurements for dark energy .
we note that @xmath1 is an integral of @xmath192 , so the fact that the acoustic scale measurements of these two are anticorrelated means that the constraints are slightly stronger than the diagonal errors would imply .
if one is fitting a model in which the transverse and radial distance scales are required to be the same , for example at low redshift , then this implies a contraction of the fisher matrix with the vector @xmath197 .
this yields an error @xmath198 where @xmath199 and @xmath200 .
for the case of spherical symmetry , this reduces to @xmath201 @xmath202 as expected from equation ( [ eq : c1d ] ) .
we next compare the forecasts on @xmath1 and @xmath2 from the full fisher matrix ( eq . [ [ eq : fullfij ] ] , hereafter , ` full - d errors ' ) to those from our fitting formula ( eq . [ [ eq : f2d ] ] , hereafter , ` 2-d errors ' ) for various values of @xmath157 , @xmath17 , and @xmath18 .
we hold @xmath175 , as this is a minor term
. we will present the comparisons for the wmap3 cosmology ; the performance is similarly good for the wmap1 cosmology .
we first study the case where @xmath203 .
the upper panels of figure [ fig:2dc ] show expected errors on @xmath1 and @xmath2 using the baryonic peaks as a function of nonlinear parameter , @xmath17 and @xmath18 respectively , for various @xmath83 .
the blue lines show the full - d errors and the black lines show the 2-d errors .
the figure shows that the errors from our fitting formula are in excellent agreement with the errors from the full fisher matrix calculations .
deviations are typically only a few % . the deviation becomes only noticeable for very large shot noise @xmath157 ( @xmath204 ) , and very large @xmath18 ( @xmath205 ) where our assumptions to derive equation ( [ eq : f2d ] ) break down .
we next present the distance errors when @xmath206 . as nonlinear redshift distortions elongate the lagrangian displacement distribution of pairs at @xmath16 along @xmath148 by @xmath207 , where @xmath208 , the distance errors for @xmath206 represent the effect of nonlinear redshift distortions on baryonic features .
we consider cases where @xmath209 and @xmath210 , while @xmath211 corresponds to more realistic redshift distortions , and @xmath212 depicts an extreme case of anisotropy in lagrangian displacement fields . in the lower panels in figure [ fig:2dc ] ,
we show the full - d errors and the 2-d errors from equation ( [ eq : f2d ] ) for @xmath211 . again .
the 2-d model reproduces the full - d errors to a good extent .
we find a similarly good agreement between the full - d errors and the 2-d errors for @xmath212 .
note that errors from the fitting formula predict @xmath213 regardless of @xmath214 and @xmath174 .
values of @xmath215 from the full - d calculations are fairly close to the expected values .
we use our fitting formula to derive distance errors for photometric redshift surveys .
we do this by modifying @xmath216 as described in [ subsec : redd ] .
we use wmap3 cosmology and assume a photometric redshift error ( @xmath180 ) of @xmath217 , which corresponds to @xmath218 rms in @xmath219 at @xmath183 .
we use equation ( [ eq : rsig ] ) to properly include the photometric redshift errors into the fisher matrix .
figure [ fig : photoz ] shows good agreement between errors on @xmath1 from our fitting formula and errors from the full - d fisher matrix . although the discrepancy is larger than spectroscopic cases ( i.e. , figure [ fig:2dc ] ) , the offset is at most 8% .
fortunately the deviation happens to be small at the common values of @xmath83 for photometric redshift surveys , e.g. , @xmath220 to @xmath221 .
in this section , we compare our revised full fisher matrix formalism ( eq . [ [ eq : fullfij ] ] ) to the distance estimates from a @xmath3 analysis of n - body data .
we find that our revised fisher matrix formalism provides an excellent forecast for the distance errors from the n - body data .
we use the n - body simulations from @xcite and perform a @xmath3 analysis to extract the acoustic scale from the simulated power spectrum .
@xcite used wmap1 cosmology and the hydra code @xcite to generate 51 sets of cosmological n - body simulations with a box size of @xmath222 that were evolved from @xmath223 to @xmath224 ( 30 sets at @xmath224 ) , @xmath183 , and @xmath225 . in the @xmath3 analysis ,
the spherically averaged real - space power spectra @xmath226 from the n - body data are fitted to model power spectra including a scale dilation parameter @xmath227 , @xmath228 . in detail
, we use an additive polynomial function to represent the effect of shot noise and nonlinearity : @xmath229 the additive polynomial function should also suppress the distance information from the broad - band shape of the power spectrum . recall that , in the fisher matrix calculations , we explicitly subtracted from the fisher matrix the distance information from the broadband shape ( [ sec : full ] ) .
@xcite used the linear power spectrum for @xmath228 ; we improve this by modifying the linear power spectrum to include the nonlinear erasure of the acoustic peaks based on @xcite . in detail
, we use @xmath230}+p_{\rm smooth}(k),\ ] ] where @xmath231 is the `` no wiggle '' form from @xcite .
@xmath80 here does not need to be precisely the same as @xmath80 measured in @xcite and used in the fisher matrix calculations .
@xmath80 is used here merely to improve the template power spectrum .
the resulting errors on @xmath227 presented below are a smooth function of our choice of @xmath80 for @xmath232 . in equation ( [ eq : chi ] ) ,
the fit parameters are @xmath227 , a multiplicative bias @xmath233 , a scale - dependent bias @xmath234 , and additive terms @xmath235 , @xmath236 , and @xmath237 from nonlinear growth , bias , or shot noise .
the dilation parameter @xmath227 is the ratio of the true distance to our estimate from a @xmath3 analysis . for simplicity
, we set the true cosmology as the reference cosmology , and therefore the mean value of @xmath227 is expected to be unity . the error on @xmath238 represents the combined errors on @xmath239 and @xmath240 , more specifically @xmath241 ( [ subsec : symm ] ) , as we use the power spectra in real space .
since we do not know the true covariance matrix of power spectrum , we do not trust differences in @xmath3 to give an accurate error on @xmath227 .
instead , we calculate the mean value and error of @xmath238 using jackknife subsampling of simulations while assuming an error on the band power to be given by a gaussian random field assumption , i.e. , based on the number of modes contributing to the band power .
we derive 51 ( 30 ) subsamples by removing one simulation each time from our 51 ( 30 at @xmath224 ) simulations .
each subsample is then fitted assuming a gaussian random error on the band power .
the error and mean value of @xmath238 is computed from the variations among the jackknife subsamples . as we emphasized in @xcite , assuming a gaussian error ignores mode - coupled errors between wavenumbers and therefore underestimates the statistical noise on small scales
nevertheless , the variations among jackknife sampling should reflect the true non - gaussian error to a reasonable extent , as these subsamples are drawn from actual nonlinear n - body data .
that is , @xmath3 statistic slightly misweights the data on small scales in each subsample but the variation among subsamples should not produce an overly optimistic @xmath242 compared to the true error .
the resulting @xmath242 from the revised @xmath3 analysis is , in general , similar to or slightly smaller than the quoted values in @xcite . at all
redshifts , the value of @xmath243 is stable with respect to different fitting ranges of wavenumber @xmath244 ( @xmath245 , @xmath246 , and @xmath247 ) or the inclusion of @xmath234 . note that in @xcite , we had to use a narrower fitting range ( @xmath248 ) at @xmath225 .
we will quote errors for @xmath249 without @xmath234 .
when fitting to the matter power spectrum , we will use @xmath250 . meanwhile , in all cases but one , the mean values of @xmath227 are within @xmath251 of unity , including 17 out of 24 cases within @xmath252 , indicating that there is no detection of a bias on the mean value of @xmath227 .
the worst out of 24 cases , without @xmath234 with @xmath253 at @xmath225 , gives a mean value of @xmath227 that is @xmath254 off from unity . in detail , in @xcite , we reasoned that using @xmath234 will slightly bias @xmath227 above unity , beyond @xmath218 in worse cases , because the fitting process will favor a negative @xmath234 to match the erased portion of the bao , and the resulting phase shift in the baryonic peaks will be compensated by @xmath227 above 1 .
now that we have accounted for the erased features in generating @xmath255 , this bias on @xmath227 has decreased compared to @xcite . at @xmath224
, we find @xmath256 for @xmath257 from the n - body data .
this corresponds to @xmath258 for @xmath81 .
the full - d error ( @xmath259 , @xmath260 and @xmath250 ) predicts @xmath261 , which is about @xmath262 smaller than the n - body data .
note that these fisher matrix errors are different from the values quoted in @xcite , mainly because of the different fiducial cosmology . at @xmath183 , we find @xmath263 for @xmath264 . for @xmath81 ,
this rescales to @xmath265 .
the corresponding full - d error for @xmath266 and @xmath250 predicts @xmath267 for @xmath268 and @xmath250 , in excellent agreement . at @xmath225
, we calculate @xmath269 for @xmath264 of simulations . for @xmath81 ,
the error rescales to @xmath270 . in comparison ,
the full - d error for @xmath271 and @xmath250 predicts @xmath272 .
we also compute the results when the baryonic signature in these nonlinear density fields at @xmath225 is reconstructed using the simple scheme presented in @xcite .
we use a @xmath273 gaussian filter to smooth gravity , displace the real particles and a set of smoothly distributed reference particles by the linear theory motion predicted from the nonlinear density fields , find new density fields from the difference of the density fields of the real particles and the reference particles , and compute the power spectra of these new fields . fitting these spectra ,
we find @xmath243 of @xmath274 for @xmath275 of simulation volume .
hence , we use the full 51 only when the non - linearities have reduced the role of the @xmath276 . ] and thus @xmath277 for @xmath278 .
this is a considerable improvement over the 1.57% measured without reconstruction .
this matches the fisher matrix prediction if @xmath279 , while we measured @xmath280 in @xcite . whether this difference is due to sample variance or due to our technical difficulties in estimating @xmath80 for the reconstructed density fields remains to be studied . while this result is an example for a negligible shot noise without galaxy bias , our @xmath3 result implies an impressive prospect for the reconstruction in the existing and future galaxy surveys .
we will investigate the effects of shot noise and galaxy bias on reconstruction in future papers .
for the biased case at @xmath225 , we use the mass case with @xmath281 from @xcite , which has galaxy bias similar to the lrg samples in sdss @xcite . from the @xmath3 analysis , we find @xmath282 for @xmath264 .
the error is equivalent to @xmath283 for @xmath81 . to evaluate errors using the fisher matrix
, we need to know the value of @xmath83 that corresponds to this case .
this is tricky because the nonlinear growth and bias effect in the mass case with @xmath281 increases the small - scale power above a nominal shot noise that would be strictly from the inverse number density .
in addition , this anomalous power is not likely constant in scale .
thus , we roughly estimate @xmath284 using the best fit additive polynomial in equation ( [ eq : chi ] ) evaluated at @xmath82 : we derive @xmath285 , which corresponds to @xmath286 for @xmath81 . in general
, we find excellent agreement between the errors from the @xmath3 analysis of the n - body data and the analytic full - d/1-d errors .
the discrepancy is at most @xmath179 for the cases we have shown .
some of the discrepancy is probably from sample variance . with only 51 values of @xmath227
, we can not estimate @xmath243 perfectly .
in addition , while the small - scale power added due to various nonlinearities can be accounted for by properly modifying @xmath152 below @xmath83 , variance due to nonlinearity are not accounted for in our fisher matrix formalism : this will induce slightly larger distance errors from the n - body data than the full - d/1-d errors . while our fitting formula in the @xmath3 analysis produces reasonable estimates of distance errors , it is not necessarily a most optimized one ( other examples can be found in * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
more sophisticated fitting schemes may also reduce the discrepancy between the @xmath3 analysis and the fisher matrix estimates .
we have computed the cosmological distance errors available from the bao in future galaxy redshift surveys using a fisher matrix formalism that incorporates the lagrangian displacement field to account for the erasure of the bao due to nonlinear growth , bias , or redshift distortions .
the resulting formalism is only a function of survey volume , shot noise , and a nonlinear parameter that can be measured quantitatively .
we have derived physically motivated lower dimensional approximations to the full fisher matrix and showed excellent agreement between distance error estimates from the approximations and the full fisher matrix .
we present the resulting fitting formula to calculate a 2-dimensional covariance matrix for @xmath239 and @xmath240 .
the fitting formula straightforwardly applies to photometric redshift surveys with a simple modification to the effective shot noise , although its agreement to the estimates from the full fisher matrix becomes slightly degraded .
the merit of the fitting formula is its simplicity in terms of input variables and computation .
finally , we compared the error estimates from the revised fisher matrix with the error estimates from a @xmath3 analysis of n - body simulations . for the @xmath3 analysis , we also used the lagrangian displacement field to account for the nonlinear effect on baryonic peaks in the template power spectrum .
this improved various aspects of the error estimates .
we showed that both error estimates are in excellent agreement .
the discrepancies ( at most @xmath179 ) could be due to non - gaussianity contributions to the variance not included in the fisher matrix calculation , but may simply be sample variance due to the limited number of n - body simulations . to use the formula , one must construct @xmath17 , @xmath18 , and @xmath34 at the redshift of the survey in question .
@xmath17 and @xmath18 are dominated by the bulk flows in the universe ; their simple scalings were given in
[ sec : full ] , although one should probably impose a lower @xmath287 and rescale @xmath29 for wmap3 cosmology .
@xcite found that highly biased tracers could increase the lagrangian variances by small amounts , although this has not been calibrated in detail .
the value of @xmath83 is easy to calculate by using @xmath288 for wmap3 , and @xmath289 for wmap1 .
reconstructing the density field reduces the lagrangian displacements and restores the baryonic peaks from nonlinear degradation @xcite . in @xcite
, we found that our simple scheme was able to decrease @xmath17 and @xmath18 about by half at @xmath225 ( from @xmath290 and @xmath291 ) . in the middle panel of figure [ fig:2dc ] , this decrease due to the reconstruction happens where the performance is improving rapidly with @xmath292 , making future galaxy redshift surveys more promising .
the @xmath3 results in [ sec : nbody ] implies that the improvements may be better than 50% .
we have not yet quantified these improvements in @xmath293 as a detailed function of @xmath34 and redshift , but we recommend , to be conservative , that one consider a 50% drop in @xmath293 as an estimation of what reconstruction can do ; figure 1 of @xcite estimates that correcting for bulk flow on scales larger than @xmath294 will decrease the displacement by 50% .
although equation ( [ eq : f2d ] ) appears complicated in that one must compute three two - dimensional integrals , it is far simpler than the full fisher matrix .
the 12-dimensional problem requires that one compute many integrals of oscillatory integrands to high precision to avoid degeneracies .
in contrast , the few integrals in the approximation are all smooth and can be computed rapidly and robustly . meanwhile
, the approximation includes the anisotropic effects of redshift distortions in the shot noise and the nonlinear degradations of the acoustic scale .
the 1-d model presented in [ sec:1da ] offers an estimate of how the performance scales with the non - linear degradation .
let us rework equation ( [ eq : sig1d ] ) , replacing the silk damping form with a gaussian and approximating the denominator as a power - law @xmath295 .
the performance then becomes proportional to @xmath296 where @xmath297 .
hence , we find that the distance - scale performance scales as @xmath298 . for the case of white noise , we have @xmath299 and the result that the precision scales as @xmath300 .
this is familiar with white noise : a peak has a higher signal - to - noise ratio proportional to the inverse square root of its width , while the measurement of its centroid scales as the width divided by the signal - to - noise ratio .
however , the cold dark matter power spectrum is better approximated by @xmath301 near @xmath302 .
therefore , if our acoustic scale measurements are sample - variance limited , we expect to use @xmath303 , which implies a precision scaling as @xmath304 .
alternatively stated , the survey volume to reach a given precision scales as @xmath305 . in the language of centroiding a peak
, what is happening is that the noise is not white and instead has large correlations between neighboring separations .
shifts in the centroid of a narrower peak require larger changes on smaller scales that the noise model disfavors .
this rapid scaling with @xmath292 implies that one gains rapidly with improvements in reconstruction so long as @xmath80 is not much smaller than the silk damping scale of @xmath306 .
the difference between @xmath307 and @xmath308 is a factor of 1.8 in distance and 3.2 in survey area .
it should be noted that the redshift - space boosts of the displacements along the line of sight cause @xmath18 to be interestingly large even at @xmath309 .
finally , we consider the cosmic variance limits of the acoustic oscillation method .
figure [ fig:2dcv ] presents the fractional errors on @xmath130 and @xmath131 from a survey of @xmath310 steradians , i.e. , the reasonably accessible extragalactic sky .
redshift bins of width 0.1 are used .
the bottom lines show the precision available for a survey with perfect reconstruction and no shot noise .
the top lines show the precision for a survey with the unreconstructed non - linear degradations and shot noise of @xmath220 .
the middle lines show a survey with reconstruction halving the values of @xmath311 and @xmath312 , again with @xmath220 .
this last case is what we would suggest as a reasonable estimate for a densely sampled survey .
one sees that the acoustic oscillation distance scale can reach precisions of about 0.4% and 0.7% on @xmath130 and @xmath131 , respectively , for each @xmath313 bin at @xmath314 , improving slightly toward @xmath224 . of course
, most dark energy models predict smooth trends in @xmath1 and @xmath2 on scales of @xmath313 , so in practice one would combine several bins in constructing tests . hence , effective precisions better than 0.2% in distance are available with the acoustic oscillation method
. we thank martin white and nikhil padmanabhan for useful discussions .
hs and dje were supported by grant ast-0407200 from the national science foundation . hs was supported in part by the national optical astronomy observatory and gemini observatory through gemini work scope no . 0525280-gem03021 .
the gemini observatory is operated by the association of universities for research in astronomy , inc .
, under a cooperative agreement with the nsf on behalf of the gemini partnership : the national science foundation ( united states ) , the particle physics and astronomy research council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , cnpq ( brazil ) and conicet ( argentina ) .
dje was further supported by a alfred p.sloan research fellowship .
|
we present the cosmological distance errors achievable using the baryon acoustic oscillations as a standard ruler .
we begin from a fisher matrix formalism that is upgraded from @xcite .
we isolate the information from the baryonic peaks by excluding distance information from other less robust sources .
meanwhile we accommodate the lagrangian displacement distribution into the fisher matrix calculation to reflect the gradual loss of information in scale and in time due to nonlinear growth , nonlinear bias , and nonlinear redshift distortions .
we then show that we can contract the multi - dimensional fisher matrix calculations into a 2-dimensional or even 1-dimensional formalism with physically motivated approximations .
we present the resulting fitting formula for the cosmological distance errors from galaxy redshift surveys as a function of survey parameters and nonlinearity , which saves us going through the 12-dimensional fisher matrix calculations .
finally , we show excellent agreement between the distance error estimates from the revised fisher matrix and the precision on the distance scale recovered from n - body simulations .
|
one of the strongest lines of evidence for an accelerating universe is the observation that @xmath6 type ia supernovae ( sne ia ) are @xmath7 mag fainter than predicted by an @xmath8 cosmology ( riess et al .
1998 ; perlmutter et al . 1999
; riess 2000 ) .
however , there are several effects such as gray dust ( aguirre 1999 ) or luminosity evolution ( drell , loredo , & wasserman 2000 ) that could mimic the effect of a cosmological constant .
riess et al .
( 2001 , hereafter r2001 ) have recently concluded that sn 1997ff , one of the two sne discovered by gilliland , nugent , & phillips ( 1999 ) in the hubble deep field north ( hdfn ; williams et al .
1996 ) , has a redshift of @xmath9 making it the highest redshift probable sn ia identified so far .
( the sn ia classification is based on the nature of the host galaxy , an evolved , red elliptical ; also the observed colors and temporal evolution of sn 1997ff are most consistent with those of sne ia . )
this object is particularly interesting because at @xmath10 the preceding epoch of deceleration causes objects to appear brighter than a persistence of astrophysical effects invoked as alternative hypotheses to explain the sn ia faintness at @xmath11 .
r2001 measured a distance modulus for sn 1997ff which is more than a magnitude brighter ( @xmath12 mag brighter assuming the spectroscopic indication of @xmath13 ) than the predictions from uniform gray dust or toy evolution models , disfavoring these alternatives to a cosmological constant .
lewis & ibata ( 2001 ) stated that sn 1997ff could be magnified by a factor of @xmath14 by two
elliptical systems " close to its line of sight .
r2001 anticipated a magnification of @xmath15 mag from these same sources , in good agreement with lewis & ibata .
as we see below , these are in fact two spiral galaxies ( # 709 and # 720 in table 1 ) which magnify sn 1997ff by only @xmath16 mag , about half of the value estimated by lewis & ibata , and close to the upper limit inferred by r2001 for these two galaxies from the induced shear on the host - galaxy shape and orientation .
m " ortsell , gunnarsson , & goobar ( 2001 ) also tackled this problem using rest - frame @xmath17-band and ultraviolet luminosities to infer the masses of the lenses . after performing numerous ray - tracing simulations
, they concluded that the range of possible magnifications was too large to obtain a firm value for the distance modulus of sn 1997ff based on our current knowledge of the lensing galaxies .
here we use the superb photometric and positional information provided by the hdfn to estimate velocity dispersions for the galaxies lensing sn 1997ff using the tully - fisher ( tf ) and faber - jackson relations , and correcting for evolution effects . with this information ,
we calculate , using the multiple - plane lensing formalism ( schneider , ehlers , & falco 1992 , hereafter sef ) , the magnification of sn 1997ff and the shear induced on the host galaxy , which is fully compatible with its expected intrinsic shape .
the structure of the paper is as follows .
section 2 describes the environment of sn 1997ff , and identifies the main lensing foreground galaxies .
the multiple - plane lens formalism is discussed in 3 , while
4 presents our estimates of the magnification and shear .
section 5 summarizes our main results and conclusions .
figure 1 shows the positions of the galaxies around sn 1997ff with @xmath18 mag and @xmath19 .
the coordinates , i d number , and @xmath20 magnitudes are taken from the fernndez - soto , lanzetta , & yahil ( 1999 ) catalog and photometric redshifts have been obtained using the bpz package ( bentez 2000 , http://acs.pha.jhu.edu/txitxo/bpz.html ) .
table 1 lists the characteristics of the galaxies contributing significantly to the magnification of sn 1997ff .
the @xmath17 and @xmath21 absolute magnitudes @xmath22 and @xmath23 have been estimated from the @xmath24 and _
hst_/nicmos f110w and f160w magnitudes , respectively .
this was done by generating k - corrections based on their spectral type , drawn from the 4 main coleman , wu , & weedman ( 1980 ) galaxy types and two kinney et al .
( 1996 ) starbursts ; the k - corrections are @xmath25 mag in the case of the four strongest lenses .
the circular velocities @xmath26 for the late - type galaxies in table 1 have been estimated using the recent results of ziegler et al .
( 2002 ) , which measure the @xmath17-band tf slope and normalization for @xmath27 mag , @xmath28 fors deep field galaxies quite similar , in statistical terms , to our lens sample .
we assume a relative error of @xmath29 estimated from the scatter around the best fit in figure 3 of ziegler et al .
these errors are propagated in the calculation of the einstein radius , etc . , and
account for most of the uncertainty in the final value of the magnification . as a cross - check ,
we have also estimated the expected values of @xmath26 usingthe @xmath21-band tf relation , as determined by giovanelli et al .
( 1997 ) : the results are very similar to those of the ziegler et al .
( 2002 ) @xmath17-band tf . to determine their circular velocities with another method , we observed galaxies # 709 and # 720 on 2001 may 29 and 30 ut with the echellette spectrograph and imager ( esi ; sheinis et al .
2002 ) on the keck - ii 10-m telescope .
a slit width of @xmath30 was chosen , which gave an instrumental resolution of @xmath31 70 km s@xmath32 ( based on fits to the night - sky emission lines ) .
the spectrum of galaxy # 709 shows strong , single - component emission lines with fwhm = 78 km s@xmath32 ( corrected for the instrumental resolution ) .
the spectrum of galaxy # 720 exhibits double - peaked emission - line profiles , with a peak - to - peak splitting of 200 km s@xmath32 . after correcting for the assumed inclination angles ( @xmath33 for galaxy # 709 and @xmath34 for galaxy # 720 ) , we derive circular velocities of 70 and 110 km s@xmath32 , respectively .
in contrast , the respective circular velocities expected according to the tf relation are substantially higher , @xmath35 and @xmath36 km s@xmath32 ( table 1 ) .
although both values are marginally consistent at the @xmath37 level , these results , especially in the case of # 709 , seem to support the conclusion of gillespie & van zee ( 2002 ) that emission lines from the central regions of star - forming galaxies can underestimate the total circular velocity by a factor of @xmath38 .
hence , we consider the circular velocities derived from the keck data to be lower limits on the true quantities , and we use the tf estimates and their uncertainties in the calculations below . taking the spectroscopic results at their face value would yield @xmath39 mag from the # 709#720 pair alone , which can be considered to be the lower limit on the magnification of sn 1997ff .
note also that this would make the total magnification smaller ( by @xmath40 mag ) , and would further strengthen our cosmological conclusions on gray dust or evolution models .
for the two early - type galaxies , # 694 and # 653 , we use the faber - jackson relation between the halo velocity dispersion and the @xmath41-band luminosity as defined by kochanek ( 1994 , 1996 ) , @xmath42 , where @xmath43 corresponds to @xmath44 mag ( efstathiou , ellis , & peterson 1988 ) and @xmath45 km s@xmath32 mpc@xmath32 ) . owing to the redshift of these galaxies , @xmath46 , we expect evolution effects to be considerable ( e.g. , treu et al . 2002 ; although see also crampton et al .
keeton , kochanek , & falco ( 1998 ) studied the properties of a sample of 17 lensing galaxies with @xmath47 , and measured an evolution rate for the mass - to - light ratio of @xmath48 for the cosmological model considered here .
after accounting for this effect , the velocity dispersions of galaxies # 694 and # 653 become about half of their value in the absence of evolution .
because of the configuration of the lenses , placed at different redshifts , it is necessary to use the multiple lens - plane formalism ( sef ) to estimate the magnification and shear on the sn 1997ff position
. the magnification matrix for @xmath49 lenses at , in general , different positions and redshifts , can be represented as @xmath50 where @xmath51 ( the identity matrix ) , the coefficients @xmath52 are distance ratios ( @xmath53 if all the galaxies are on the sample plane ) , and @xmath54 can be represented as @xmath55 in this equation @xmath56 and @xmath57 correspond to the convergence and shear induced by each @xmath58-galaxy on the source in the absence of other lenses . assuming that the mass distribution of the galaxies can be well represented by a singular isothermal sphere , we have @xmath59 the einstein radius @xmath60 can be represented as @xmath61 ( bartelmann & schneider 2001 ) , where @xmath62 is the circular velocity , @xmath63 the distance between the @xmath58-galaxy and the source , and @xmath64 is the distance between the source and the observer . finally ,
if the matrix @xmath65 is symmetric , it can be represented in the usual way as @xmath66 where the total shear is @xmath67 and the total magnification is @xmath68^{-1 } $ ] . to describe the effects of lensing on the shape of a galaxy
, we can represent it with the complex ellipticity @xmath69e^{i2\phi } $ ] , where @xmath70 is the axis ratio , @xmath71 . given
a reduced shear @xmath72 , the intrinsic ellipticity of the source can be recovered as ( schneider & seitz 1995 ) @xmath73 the above equations have been implemented as a python module which can be obtained by contacting n. bentez .
using the above expressions , we find that the lensing correction to the brightness of sn 1997ff is a magnification of @xmath74 mag , for an @xmath75 , @xmath76 cosmology and using the filled - beam approximation to the angular distance .
table 1 shows the einstein radius and magnification @xmath77 induced on the position of sn 1997ff by each individual galaxy .
if we naively multiply all these magnifications , we arrive at a value of @xmath78 , which represents only a @xmath79 difference from the multiple lens - plane formalism result .
r2001 estimated that , from the ellipticity and orientation of the host galaxy , the probability of its being magnified by more than 0.34 mag by the galaxy pair # 709 and # 720 ( the two strongest lenses ) is only 18% .
our new result for the total shear is @xmath80 , half of the value expected for an equivalent single isothermal sphere , and along the direction perpendicular to the # 709 + # 720 pair .
given this reduced shear , the host ellipticity is significantly less constraining in providing an upper limit to the magnification ; the 90% confidence limits for the magnification are ( 0.086 , 1.37 ) .
the cause of this reduced leverage of the distortion constraint is that the shear vectors from the galaxies whose contribution was not considered in r2001 partially cancel out .
the result is that , unlike the magnification ( @xmath81 mag ) , the distortion they induce on the sn 1997ff host is very small , and slightly counteracts that caused by the # 709 + # 720 pair , further reducing the total shear .
r2001 measured a distance modulus of @xmath82 mag , assuming @xmath83 .
after correction for lensing , the new value is @xmath84 mag , in excellent agreement with the expected value of @xmath85 mag for the @xmath86 ( @xmath87 ) universe expected from the @xmath88 population of sne ia ( riess et al . 1998 ) . using the preceding calculation to correct the sn 1997ff distance modulus for lensing , we can compare the result to toy models that invoke astrophysical sources of dimming instead of a cosmological constant to match the observations of sne ia at @xmath88 .
here we will assume the spectroscopic redshift measurement of @xmath13 for sn 1997ff to simplify the evaluation of the astrophysical dimming expected at a specific redshift .
the astrophysical dimming described in r2001 would take the functional form of @xmath89 mag relative to an empty universe . at @xmath13
this results in a predicted distance modulus of @xmath3146.4 mag ( @xmath90 ) .
the lensing - corrected distance modulus of sn 1997ff is 45.49 mag , smaller ( i.e. , the sn is brighter ) than the astrophysical dimming model by 0.91 mag ( reduced from 1.25 mag without the lensing correction ) . as discussed by r2001 , the fit to the sn data yields non - gaussian and asymmetric uncertainties beyond the @xmath91 interval . returning to the fit from r2001 to the sn data ( corrected by the slight increase in the distance modulus error due to the lensing uncertainty ) yields confidence intervals in the direction of larger distances of @xmath92 , @xmath93 , and @xmath94 equal to 0.34 , 0.67 , and 0.90 mag , respectively .
we therefore conclude that the simplest alternatives to an accelerating universe appear to be inconsistent with sn 1997ff at the @xmath95 confidence level .
however , we caution that the astrophysical model of dimming is naive and before tests of astrophysical sources can be considered robust , independent confirmation of this result will be required from additional sne ia at @xmath96 .
mannheim ( 2001 ) has shown that the observations of @xmath97 sne can be accommodated within a conformal gravity cosmological model ( mannheim 1992 ) with @xmath98 , which predicts a distance modulus of @xmath99 at @xmath100 .
this agrees within @xmath101 with our new result , implying that more data are necessary to discriminate between this hypothesis and a cosmological constant .
the magnification of sn 1997ff , @xmath102 , is high but compatible with the expectation from simulations .
for instance , barber et al .
( 2000 ) find @xmath103 for @xmath104 of the lines of sight to @xmath105 .
however , we would not attribute such magnification to a selection effect ; the dispersion of observed magnitudes due to lensing is only @xmath310.2 mag , no larger than the intrinsic dispersion of sne ia , and it has been shown to cause negligible selection bias ( riess et al . 1998 ) . it is unlikely that lensing by large - scale structure could significantly affect our results .
dalal et al .
( 2002 ) show that most of the scatter in the magnification comes from scales of less than @xmath106 , corresponding to the range considered here .
the expected residual root - mean - square on larger scales is @xmath107 mag , smaller than our uncertainty .
it is noteworthy that by combining an `` individual lenses '' approach like the one followed here , with a shear analysis which estimates the large - scale magnification as proposed by dalal et al . , it could be possible to significantly reduce the uncertainty in supernova distances due to lensing .
dark haloes ( without optically visible galaxies ) may be a possible problem with this approach , but so far there is very little , if any , evidence for their existence .
the major source of error in our calculations is the dispersion in the tf and faber - jackson relations at high redshift , but in the future improved calibrations using near - infrared data should reduce this scatter considerably .
the amount of magnification we expect for sn 1997ff is similar to the values determined by lewis & ibata ( 2001 ) and r2001 , and roughly agrees with the results of m " ortsell et al .
( 2001 ) for realistic values of their input parameters .
however , our use of the multiple lens - plane formalism , individual galaxy k - corrections , and tf and faber - jackson relationships , corrected for evolution , provides an estimate that is more accurate and robust than previous work . in summary ,
sn 1997ff , with @xmath108 , is the highest redshift sn ia discovered so far . it is shown here that gravitational lensing by nearby foreground galaxies is likely to have magnified sn 1997ff by @xmath109 mag . due to the spatial configuration of the foreground galaxies , the shear from the individual lenses partially cancels out , and the total distortion induced on the host galaxy is less than half of that produced by a single lens with the same magnification . after correcting for lensing , and assuming @xmath110 , the distance modulus to sn 1997ff is @xmath111 mag , in better agreement with the @xmath112 cosmology expected from lower - redshift sne ia , but inconsistent with uniform gray dust or simple evolution models as an explanation for the dimming of @xmath97 sne ia at the @xmath95 confidence level .
it is noteworthy that our new result also agrees within @xmath101 with the predictions of conformal gravity cosmological models ( mannheim 2001 ) . since the study of high - z sne
ia can not avoid the effects of gravitational lensing , future use of sne to constrain cosmological parameters will need to consider and ultimately contend with the effects of lensing .
the authors thank john blakeslee for useful comments and suggestions .
n.b . acknowledges financial support from the nasa acs grant .
p.e.n . acknowledges support from a nasa ltsa grant and by the director , office of science under u.s .
department of energy contract no .
de - ac03 - 76sf00098 .
a.v.f . is grateful for nsf grant ast-9987438 and a guggenheim foundation fellowship .
the keck observatory was made possible by the generous financial support of the w. m. keck foundation .
author(1982 ) aguirre , a. 1999 , , 525 , 583 barber , a. j. , thomas , p. a. , couchman , h. m. p. , & fluke , c. j. 2000 , , 319 , 267 bartelmann , m. , & schneider , p. 2001
, physics reports , 340 , 291 ben ' itez , n. 2000 , , 536 , 571 cohen , j. , et al . 2000 , , 538 , 29 coleman , g. d. , wu , c .- c . , & weedman , d. w. 1980 , , 43 , 393 crampton , d. , schade , d. , hammer , f. , matzkin , a. , lilly , s. j. , & le f ` evre , o. 2002 , , 570 , 86 dalal , n. , holz , d. e. , chen , x. , & frieman , j. 2002 , astro - ph/0206339 drell , p. s. , loredo , t. j. , & wasserman , i. 2000 , , 530 , 593 efstathiou , g. , ellis , r. s. , & peterson , b. a. 1988 , , 232 , 431 fern ' andez - soto , a. , lanzetta , k. m. , & yahil , a. 1999 , , 513 , 34 gillespie , e. b. , & van zee , l. 2002 , in press ( astro - ph/0201274 ) gilliland , r. l. , nugent , p. e. , & phillips , m. m. 1999 , , 521 , 30 giovanelli , r. , haynes , m. p. , herter , t. , vogt , n. p. , da costa , l. n. , freudling , w. , salzer , j. j. , & wegner , g. 1997 , , 113 , 53 keeton , c. r. , kochanek , c. s. , & falco , e. e. 1998 , , 509 , 561 kinney , a. l. , calzetti , d. , bohlin , r. c. , mcquade , k. , storchi - bergmann , t. , & schmitt , h. r. 1996 , , 467 , 38 kochanek , c. s. 1994 , , 436 , 56 kochanek , c. s. 1996 , , 466 , 638 lewis , g. f. , & ibata , r. a. 2001 , mnras , 324 , l25 mannheim , p. d. 1992 , , 391 , 429 mannheim , p. d. 2001 , astro - ph/0104022 m " ortsell , e. , gunnarsson , c. , & goobar , a. 2001 , , 561 , 106 perlmutter , s. , et al .
1999 , , 517 , 565 riess , a. g. 2000 , , 112 , 1284 riess , a. g. , et al .
1998 , , 116 , 1009 riess , a. g. , et al .
2001 , apj , 560 , 49 ( r2001 ) schneider , p. , ehlers , j. , & falco , e. e. 1992 , gravitational lenses ( berlin : springer verlag ) ( sef ) schneider , p. , & seitz , c. 1995 , , 294 , 411 sheinis , a. i. , et al .
2002 , pasp , 114 , 851 treu , t. , stiavelli , m. , casertano , s. , mller , p. , & bertin , g. 2002 , , 564 , l13 williams , r. e. , et al .
1996 , , 112 , 1335 ziegler , b. l. , et al .
2002 , , 564 , l69 rlrrrrrrrcrrrr 709 & 4 - 402.32&@xmath113&@xmath114 & @xmath115&0.555&21.68 & 21.34 & @xmath116 & sb2 & @xmath117 & 1.106 & 0.30650 & 4 - 430.0 & @xmath118&@xmath119&@xmath120&0.875&23.42 & 23.07 & @xmath121 & sb2 & @xmath122 & 1.031 & 0.14 720 & 4 - 402.31&@xmath123 & @xmath124 & @xmath125 & 0.557&21.01 & 20.00 & @xmath126 & sbc & @xmath127 & 1.097 & 0.50694 & 4 - 493.0 & @xmath128 & @xmath129&@xmath130&0.849&21.66 & 20.19 & @xmath131 & el & @xmath132 & 1.035 & 0.25625 & 4 - 378.0 & @xmath133 & @xmath134&@xmath135&1.225&24.17 & 23.22 & @xmath136 & i m & @xmath137 & 1.014 & 0.11653 & 4 - 254.0 & @xmath138&@xmath139&@xmath140&0.900&22.34 & 21.14 & @xmath141 & el & @xmath142 & 1.013 & 0.18
|
with a redshift of @xmath0 , sn 1997ff is the most distant type ia supernova discovered so far .
this sn is close to several bright , @xmath1 galaxies , and we consider the effects of lensing by those objects on the magnitude of sn 1997ff .
we estimate their velocity dispersions using the tully - fisher and faber - jackson relations corrected for evolution effects , and calculate , applying the multiple - plane lensing formalism , that sn 1997ff is magnified by @xmath2 mag . due to the spatial configuration of the foreground galaxies ,
the shear from individual lenses partially cancels out , and the total distortion induced on the host galaxy is considerably smaller than that produced by a single lens having the same magnification .
after correction for lensing , the revised distance to sn 1997ff is @xmath3 mag , which improves the agreement with the @xmath4 cosmology expected from lower - redshift sne ia , and is inconsistent at the @xmath5 confidence level with a uniform gray dust model or a simple evolution model .
|
ghost imaging with thermal light @xcite has been studied extensively in recent years .
bennink and coworkers @xcite first pointed out that ghost imaging can also be realized using a classical source with the appropriate correlations .
a thermal or quasi - thermal source can exhibit such a classical correlation .
a very close formal analogy was demonstrated between ghost imaging with thermal and quantum - entangled beams in refs .
@xcite , which implied that classically correlated beams were able to emulate the relevant features of quantum ghost imaging .
number of experiments on thermal light ghost imaging have been performed in refs .
a unified treatment of classical and quantum ghost imaging was established in terms of gaussian - state analysis in ref .
@xcite .
recently , some attention has been paid to the physics of thermal - light ghost imaging @xcite and higher - order - coherence or correlation effects of thermal light @xcite .
the higher - order coherence or correlation has shown attractive properties in practical applications .
multiphoton imaging with thermal light is one of these exciting areas . in a previous work ,
our group theoretically initiated a study of thermal - light ghost imaging in terms of higher - order correlated thermal light @xcite .
we proposed a thermal - light ghost imaging scheme with third - order correlated thermal light . in this scheme , a third - order correlated thermal - light source , a test optical arm and two reference optical arms are used to produce two ghost images .
two ghost images are created at two different places in a nonlocal fashion as a consequence of the third order correlation of the involved optical fields .
it was shown that the third - order correlated imaging includes richer correlated imaging effects than the second - order correlated one . in this paper , we want to propose a thermal - light ghost imaging scheme with @xmath0th - order intensity correlation and investigate the visibility of the ghost images . we show that the usual second- and third - order ghost imaging are only two particular examples of our present scheme . the paper is organized as follows . in sec .
ii , begin with the thermal - light source with @xmath0th - order correlation , we will propose the thermal - light ghost imaging scheme with @xmath0th - order intensity correlation . in sec .
iii , the visibility of the ghost images in the proposed scheme will be investigated .
finally , we shall conclude our paper with discussions and remarks in the last section .
the basic setup for ghost imaging with @xmath0-order correlated thermal light is indicated in figure 1 which includes one test optical arm and @xmath1 reference optical arms .
first arm is the test arm .
an unknown object with transmission function @xmath2 and a collective lens with focal length @xmath3 are placed on the test arm while there is one imaging lens with focal lengths @xmath4 on each reference arm .
the object is placed at the focal plane of the collective lens at distance @xmath5 from the thermal - light source .
the distance between @xmath6th imaging lens and the thermal - light source is @xmath7 with @xmath8 .
the first detector @xmath9 is a bucket detector placed at the focal plane of the collective lens on the right - hand side .
the other detectors are scanning detectors @xmath10 with @xmath8 which are placed at distances @xmath11 from @xmath6th imaging lens .
the signals from the @xmath0 photon counting detectors are sent to an electronic coincidence circuit to measure the rate of coincidence counts .
the thermal light source with @xmath0th - order intensity correlation , usually obtained by illuminating a laser beam into a slowly rotating ground glass , is divided into @xmath0 beams , which can be implemented by an appropriate combination of @xmath1 beam splitters .
consider a monochromatic plane wave described by the field @xmath12 $ ] illuminating a material containing disordered scattering centers .
after scattering , the field can be written as @xmath13d\mathbf{q}$ ] where @xmath14 is the transverse wave vector introduced by the random scattering and satisfies the relation @xmath15 .
hence , @xmath16 is a stochastic variable obeying gaussian statistics .
however , the scattered waves with different transverse wave vectors are statistically independent .
if @xmath17 , the scattered field can be approximately written as @xmath18 $ ] where @xmath19d\mathbf{q}$ ] is the slowly varying envelope . as a result
, we have defined a monochromatic thermal light random in both strength and propagation direction . according to the wiener - khintchine theorem
, the first - order spectral correlation satisfies the following expression @xmath20 where @xmath21 is the power spectrum of the spatial frequency . for any field with thermal statistics ,
all high - order correlations can be expressed in terms of the first - order ones due to @xmath22 @xcite .
then , @xmath0th - order spectral correlation of thermal light can be written as @xmath23 where the prime in the summation means that all of repeating terms are subtracted from the summation over @xmath24 and @xmath25 . after passing through a combination of @xmath1 beam splitters ,
the thermal light with @xmath0th - order spectral correlation is divided into @xmath0 correlated thermal - light beams which are input light beams of the test and reference arms . for simplicity ,
we consider the one - dimensional case , and let @xmath26 and @xmath27 with @xmath28 be the transverse coordinates of the source plane and detection planes , respectively . here
we have assumed that the starting planes of the @xmath0 thermal - light beams are overlapped .
let @xmath29 be the impulse response function of the test arm , and @xmath30 with @xmath31 be the impulse response function of @xmath32th reference arm , respectively .
assume that the light field on the @xmath32th detection planes is denoted as @xmath33 , which is connected with the optical field of the thermal - light source through the following relation @xmath34 where @xmath35 is the fourier transformation of the impulse response function @xmath30 .
then , the @xmath0th - order correlation function of the joint intensity at the @xmath0 detection planes can be expressed as @xmath36 the rate of coincidence counts is governed by the @xmath0th - order correlation function given by equation ( [ 4 ] ) which can be calculated in terms of the impulse response functions of the relevant optical systems .
the impulse response functions @xcite in the reference arms can be written as @xmath37 where @xmath38 with @xmath39 , and the impulse response function in the test arm is given by @xmath40\nonumber \\ & & \times\int t\left ( x\right ) \exp \left [ -i\left ( \frac{kx_{1}}{f_{c}}+q\right ) x\right ] dx.\end{aligned}\ ] ] th - order correlated thermal light .
it consists of one test arm and @xmath1 reference arms .
the test arm includes the object @xmath2 , the collection lens @xmath41 , and the bucket detector @xmath9 . the object and the detector
are placed in its two focal planes of the collection lens , respectively .
each reference arm includes one imaging lens and a scanning detector denoted by @xmath42 and @xmath10 with @xmath43 , respectively . ] in order to obtain the @xmath0th - order correlation function given by equation ( 4 ) , we first calculate the following integrations : @xmath44 where both @xmath45 and @xmath46 may take @xmath47 , but @xmath48 . in the broadband limit , s(q )
can be regarded as a constant s(0 ) .
hence we have the total intensity of the thermal light @xmath49 , where @xmath50 is the spectral bandwidth of the source . substituting the impulse response functions of the test and reference arms given by equations ( [ 5 ] ) and ( [ 7 ] ) into equations ( [ 8 ] ) and ( [ 9 ] ) , we obtain @xmath51 where both @xmath45 and @xmath46 may take @xmath52 but @xmath48 , and we have introduced two phase functions @xmath53 where the scaled positions @xmath54 . from equations ( 11 ) and ( 13 ) ,
it is straightforward to see that when the positions of the object , ghost images , and the lenses obey the following gaussian thin lens equations for the @xmath0th - order correlated imaging @xmath55 the cross - correlation functions of intensity fluctuation can be simplified as @xmath56 when gaussian thin lens equations are satisfied , and @xmath57 the @xmath0th - order correlation function of the joint intensity at the @xmath0 detection planes given by equation ( [ 4 ] ) can be reduced to the following form @xmath58 where the prime on the right - hand side of above equation means that repeating terms in the summation over @xmath24 and @xmath25 should be subtracted .
the last term on the right - hand equation ( 18 ) is the background term which is the multiplication of the intensity distribution at the @xmath0 detectors .
it does not contribute to the correlated imaging , but it may affect the visibility of produced ghost images . each term in the summation over @xmath24 and @xmath25
is the multiplication of the intensity distribution at detector @xmath10 and the intensity fluctuation correlation between @xmath9 and @xmath10 with @xmath59 , which gives the information of the object imaged at @xmath10 .
in particular , when the @xmath1 reference arms are identic , i.e. , @xmath60 , @xmath61 and @xmath62 , we have @xmath63 , @xmath0th - order correlation function becomes @xmath64 where we have introduced the parameter @xmath65 $ ] .
equation ( 19 ) indicates that for the object placed at the test arm , @xmath1 ghost images can be produced in the @xmath1 reference arms through coincidence count measurements upon @xmath0 detectors .
each ghost image is an amplified image of the object with the amplifying rate @xmath66 with @xmath39 since @xmath54 . from the gaussian thin lens equations of @xmath0th - order correlated imaging given by equation ( 15 )
we can see that the higher - order correlated imaging exhibits richer imaging effects .
in fact , the imaging equation ( 15 ) indicates that @xmath67 is the object distance while @xmath68 is the image distance for @xmath45th joint path with @xmath69 .
just as the ordinary imaging law , when the object distance of the @xmath45th joint path @xmath67 is greater ( less ) than the focal length @xmath70 , the correlated image is real ( virtual ) . as a specific example
, we consider the case of @xmath71 . in this case , from the imaging equation ( 15 ) we can obtain the following eight ghost - image configurations : ( 1 ) when @xmath72 with @xmath73 and @xmath74 , the three ghost images are real ; ( 2 ) when @xmath75 with @xmath73 and @xmath74 , the three ghost images are virtual ; ( 3 ) when @xmath76 , @xmath77 and @xmath78 , one ghost image is real at the position @xmath79 , the other two ghost images are virtual at the positions @xmath80 and @xmath81 ; ( 4 ) when @xmath82 , @xmath83 and @xmath78 , the ghost image at the position @xmath80 , two ghost images at the positions @xmath79 and @xmath81 are virtual ; ( 5 ) when @xmath84 , @xmath83 and @xmath77 , the ghost image at the position @xmath81 is real , two ghost images at the positions @xmath79 and @xmath80 are virtual ; ( 6 ) when @xmath85 , @xmath77 and @xmath78 , two ghost images at the positions @xmath79 and @xmath80 are real , the ghost image at the position @xmath81 is virtual ; ( 7 ) when @xmath76 , @xmath86 and @xmath77 , two ghost images at the positions @xmath79 and @xmath81 are real , the ghost image at the position @xmath80 is virtual ; ( 8) when @xmath82 , @xmath78 and @xmath83 , two ghost images at the positions @xmath80 and @xmath81 are real , the ghost image at the position @xmath79 is virtual .
it is obvious that when @xmath87 , the usual results of the second- and third - order thermal - light ghost imaging are recovered .
the visibility is an important parameter to estimate the quality of a ghost image .
we first generalize the visibility of the second - order correlated imaging @xcite to the case of @xmath0th - order correlated imaging as follows @xmath88 _ { \max } } { \left [ g^{\left(n\right ) } \right ] _ { \max } } .\ ] ] for our present scheme of the @xmath0th order thermal - light imaging , substituting equations ( 19 ) into equation ( 20 ) , we obtain the visibility @xmath89_{\max}}{\left[q_{0}\int \left\vert t\left ( x\right ) \right\vert ^{2}dx+\sum^n_{i=2}|t(x_i)|^2\right]_{\max}},\ ] ] which indicates that the visibility depends on not only the spectral bandwidth of the thermal - light source @xmath90 and the transmission function @xmath2 but also the order number of the correlation of the thermal - light source .
the visibility can be enhanced with the decrease of the spectral bandwidth .
an increase of the transmission leads to a decrease of the visibility since when the transmission area increases , more points contribute to the background that directly makes the visibility decrease . in order to see the relationship between the visibility and the order number of the correlation , making use of equations ( 10 ) and ( 11 )
, we have @xmath91 where @xmath92 . according to the cauchy - schwartz inequality @xmath93 , which means that @xmath94 .
hence , we find that @xmath95_{\max}}{\left[1+\sum^n_{i=2}|t(x_i)|^2\right]_{\max}}\leq \frac{n-1}{n},\ ] ] which implies that the visibility of the ghost images can be dramatically enhanced when the order of correlation becomes larger . for @xmath0th order correlated thermal - light imaging ,
the upper bounds of the visibility is given by @xmath96 . as expected , when @xmath97 the upper bound of the visibility of the second correlated imaging @xcite is @xmath98 .
in conclusion , we have proposed a theoretical scheme of ghost imaging in terms of @xmath0th - order correlated thermal light .
our scheme includes one test arm and @xmath1 references arms .
one object is placed on the test arm .
@xmath1 ghost images are produced in the reference arms in a nonlocal fashion by means of a higher - order correlated imaging process with an @xmath0th - order correlated thermal source and correlation measurements .
we have derived the gaussian thin lens equations which the positions of the ghost images obey in the ghost imaging protocol .
we have also investigated the visibility of the ghost images in the scheme , and obtained the upper bounds of the visibility for the @xmath0th - order correlated thermal - light ghost imaging .
it has been shown that the visibility depends on not only the spectral bandwidth of the thermal - light source and the transmission area of the object but also the order number of the correlation of the thermal - light source .
it is found that the visibility of the ghost imagines can be dramatically enhanced when the order of correlation becomes larger .
our present scheme is a many - ghost imaging protocol .
on one hand , it gives rise to a theoretical origin for developing many - ghost imaging technology .
this gives rise to the possibility of experimentally producing correlated many ghost images .
in fact , the higher - order correlated imaging opens up new avenues for realizing multi - port information processing . on the other hand ,
physically these ghost images stem from higher - order coherence or correlation of optical fields . in this sense , the appearance of the many ghost images reveals an observable physical effect of higher - order coherence or correlation of optical fields .
hence , it is of very significance to study higher - order correlated imaging not only for well understanding the essential physics behind the higher coherence or correlation of optical fields but also for developing multi - port information processing technology .
the experimental realization for the many ghost imaging protocol proposed here and practical applications of the many ghost imaging phenomenon deserves further investigation .
this work was supported by the national fundamental research program grant no .
2007cb925204 , the national natural science foundation under grant nos .
10775048 and 10325523 , and the education committee of hunan province under grant no .
08w012 .
99 r. s. bennink , s. j. bentley , and r. w. boyd , phys . rev . lett . * 89 * , 113601 ( 2002 ) .
a. gatti , e. brambilla , and l. a. lugiato , phys .
. lett . * 90*,133603 ( 2003 ) .
a. gatti , e. brambilla , m. bache , and l. a. lugiato , phys .
lett . * 93 * , 093602 ( 2004 ) .
a. gatti , e. brambilla , m. bache , and l. a. lugiato , phys .
rev . a * 70 * , 013802 ( 2004 ) .
j. cheng and s. s. han , phys . rev . lett . * 92 * , 093903 ( 2004 ) . y. cai and s. y. zhu , opt
* 29 * , 2716 ( 2004 ) .
k. w. c. chan , m. n. osullivan , and r. w. boyd , phys .
a * 79 * , 033808 ( 2009 ) .
e. brambilla , a. gatti , m. bache , and l. a. lugiato , fortschr . phys .
* 52 * , 1080 ( 2004 ) .
a. valencia , g. scarcelli , m. dangelo , and y. h. shih , phys .
lett . * 94 * , 063601 ( 2005 ) .
f. ferri , d. magatti , a. gatti , m. bache , e. brambilla , and l. a. lugiato , phys .
lett . * 94 * , 183602 ( 2005 ) .
d. zhang , y. h. zhai , l. a. wu , and x. h. chen , opt .
lett . * 30 * , 2354 ( 2005 ) .
r. meyers , k. s. deacon , and y. h. shih , phys , rev . a * 77 * , 041801 ( 2008 ) .
b. i. erkmen and j. h. shapiro , phys .
a * 77 * , 043809 ( 2008 ) ; _ ibid_. * 79 * , 023833 .
a. gatti , m.bache , d. magatti , e. brambilla , f. ferri , and l. a. lugiato , j. mod . opt . * 53 * , 739 ( 2006 ) .
m. bache , d. magatti , f. ferri , a. gatti , e. brambilla , and l. a. lugiato , phys .
a * 73 * , 053802 ( 2006 ) .
y. cai and s. y. zhu , phys .
e * 71 * , 056607 ( 2005 ) .
y. h. zhai , x. h. chen , d. zhang , and l. a. wu , phys .
a * 72 * , 043805 ( 2005 ) .
g. scarcelli , v. berardi , and y. h. shih , phys .
lett . * 96 * , 063602 ( 2006 ) .
m. dangelo , y. h. kim , s. p. kulik , and y. h. shih , phys .
92 * , 233601 ( 2004 ) .
h. liu , x. shen , d. m. zhu , and s. han , phys . rev . a * 76 * , 053808 ( 2007 ) .
m. zhang , q. wei , x. shen , y. liu , h. liu , j. cheng , and s. han , phys .
a * 75 * , 021803 ( 2007 ) .
a. gatti , m. bondani , l. a. lugiato , m. g. a. paris , and c. fabre , phys .
lett . * 98 * , 039301 ( 2007 ) .
l. g. wang , s. qamar , s. y. zhu , and m. s. zubairy , phys .
a * 79 * , 033835 ( 2009 ) .
m. e. brezinski and b. liu , phys .
a * 78 * 063824 ( 2008 ) .
y. h. shih , front .
china , * 2 * , 125 ( 2007 ) .
j. h. shapiro , phys .
a * 78 * , 061802(r ) ( 2008 ) .
l. h. ou and l. m. kuang , j. phys .
b * 40 * , 1833 ( 2007 ) .
y. f. bai and s. s. han , phys .
a * 76 * , 043828 ( 2007 ) .
j. b. liu and y. h. shih , phys .
a * 79 * , 023819 ( 2009 ) .
d. z. cao , j. xiong , s. h. zhang , l. f. lin , l. gao , and k. g. wang , appl .
. lett . * 92 * , 201102 ( 2008 ) .
i. n. agafonov , m. v. chekhova , t. sh .
iskhakov , and a. n. penin , phys .
a * 77 * 053801 ( 2008 ) .
richter , phys .
a * 42 * , 1817 ( 1990 ) .
l. mandel and l. wolf , _ optical coherence and quantum optics _
( cambridge university press , cambridge , 1995 ) p428 .
d. z. cao , j. xiong , and k. g. wang , phys .
a * 71 * , 013801 ( 2005 ) ; d. z. cao and k. g. wang , _ phys .
lett . _ a * 333 * , 23 ( 2004 ) .
m. bache , d. magatti , f. ferri , a. gatti , e. brambilla , and l. a. lugiato , phys .
a * 73 * , 053802 ( 2006 ) .
|
in this paper , we propose a theoretical scheme of ghost imaging in terms of @xmath0th - order correlated thermal light .
we obtain the gaussian thin lens equations in the ghost imaging protocol .
we show that it is possible to produce @xmath1 ghost images of an object at different places in a nonlocal fashion by means of a higher - order correlated imaging process with an @xmath0th - order correlated thermal source and correlation measurement .
we investigate the visibility of the ghost images in the scheme , and obtain the upper bounds of the visibility for the @xmath0th - order correlated thermal - light ghost imaging .
it is found that the visibility of the ghost images can be dramatically enhanced when the order of correlation becomes larger .
|
gravitational lensing has the ability to probe the internal mass distributions of galaxies at cosmological distances .
the ability to form complete samples , based solely on mass , has allowed studies of the formation and evolution of early - type galaxies at intermediate redshifts ( e.g. @xcite ; @xcite ) .
anomalous flux - density ratios observed in merging gravitational lens images have recently been used to argue for cold dark matter substructure within the haloes of lensing galaxies @xcite .
the magnification of the lensed source has also allowed studies of star - formation at high redshift ( e.g. @xcite ) and the dust content of quasar host galaxies to be estimated @xcite .
furthermore , gravitational lenses have proved to be powerful tools for determining the cosmological parameters .
lenses with well - constrained mass models and accurate time - delay measurements have been used to calculate the hubble parameter ( e.g. @xcite ) .
the gravitational lensing statistics from large systematic surveys have provided complementary and independent constraints on the cosmological constant and density parameters ( @xcite ; @xcite ; @xcite ) to those obtained from sn1a @xcite , large - scale structure @xcite and cosmic microwave background @xcite observations .
in addition , the nature of the curvature parameter can also be determined from the image - separations and source redshifts via the @xmath8-@xmath9 relation @xcite . however , each of these applications of gravitational lensing is critically dependent on the redshifts of the lensing galaxy and lensed source being known . with the objectives outlined above in mind
, a spectroscopic survey of galaxy - scale gravitational lenses using the w. m. keck and palomar observatories has been underway over the last few years @xcite . in this paper
we present our latest results .
the observing sample consisted of four gravitational lenses discovered during the course of the cosmic lens all - sky survey ( class ; @xcite ) . in section [ spec - targets ] a short review of each of these gravitational lenses is given .
the observations with the w. m. keck - i telescope are presented in section [ spec - obs ] and the optical spectra are presented in section [ spec - results ] .
the resulting implications for each gravitational lens are discussed in section [ spec - disc ] .
finally , in section [ spec - conc ] a summary of the observations and analysis presented in this paper is given .
throughout we adopted an @xmath10 , @xmath11 flat - universe , with a hubble parameter @xmath12 km s@xmath13 mpc@xmath13 .
the targets of our observations were class b0128 + 437 , class b0445 + 123 , class b0631 + 519 and class b0850 + 054 .
a short review of each gravitational lens is given below .
the quadruple - imaged gravitational lens system class b0128 + 437 is described in @xcite and @xcite .
merlin ( multi element radio linked interferometer network ) 5 ghz imaging detected four compact components , with a maximum image separation of @xmath14 , arranged in a classic quad - lens formation .
the 325 mhz to 8.46 ghz radio - spectrum was found to be strongly peaked around @xmath15 ghz , which suggested the lensed source may be a compact symmetric object ( cso ; e.g. @xcite ) .
follow - up vlba ( very long baseline array ) imaging at 5 ghz found sub - structure consistent with the lensing of a cso hypothesis .
however , it should be noted that the recent vlbi observations of @xcite have placed the cso interpretation in some doubt since a flat - spectrum compact core appears to be present in 3 of the 4 images , suggesting a core - jet structure .
optical hst ( hubble space telescope ) and infra - red ukirt ( united kingdom infra - red telescope ) photometry found class b0128 + 437 to be highly reddened ( @xmath16 ) , with the dominant infra - red emission being from the lensed images .
therefore , the background source in class b0128 + 437 is thought to be highly redshifted and/or the lensing galaxy is extremely dusty .
recent modelling attempts favour lensing by a spiral galaxy , although conclusive observational evidence of this has still to be obtained @xcite . class b0445 + 123 ( @xcite ) is a double - imaged gravitational lens system .
8.46 ghz vla ( very large array ) and 5 ghz merlin and vlba observations found two compact components separated by @xmath17 with identical radio - spectra .
more sensitive 8.46 ghz vla imaging detected extended , arc - like structure in component a. imaging with the wht ( william herschel telescope ) detected faint extended optical emission with an @xmath18-band magnitude of @xmath19-mag . however , the resolution was not sufficient to separate the lensed images from the lensing galaxy . class b0631 + 519 ( york et al . , in preparation ) is a new double - imaged gravitational lens system discovered by class .
this lens was found to have two compact components separated by @xmath20 when observed at 8.46 ghz with the vla .
observations with merlin and the vlba at 5 ghz detected structure in the radio components which was consistent with gravitational lensing .
optical imaging with the wht detected emission ( @xmath18-band @xmath21-mag ) mainly from the lensing galaxy .
the wealth of observational constraints available in class b0631 + 519 make this gravitational lens an exciting prospect for detailed modelling of the lensing potential in the future .
class b0850 + 054 @xcite is a double - imaged gravitational lens system .
the two components , separated by @xmath22 , are compact when observed with the vla and merlin at 8.46 ghz and 5 ghz respectively .
however , high resolution imaging with the vlba at 5 ghz shows sub - structure in component a consistent with gravitational lensing .
ukirt @xmath23-band imaging detected infra - red emission ( 18.4-mag ) believed to be mainly from the lensing galaxy .
unfortunately , the resolution was insufficient to determine the position of the lensing galaxy relative to the lensed images and mass - modelling has not been possible . a flux - density monitoring program at 8.46 ghz with the vla has not detected any variability in class b0850 + 054 .
like class b0128 + 437 , the radio - spectrum of class b0850 + 054 turns - over at @xmath24 ghz which , coupled with the lack of variability , suggests the background source may be a cso .
( 15,5.5 ) ( -1.1,-0.05 )
the spectroscopic observations were carried out using the w. m. keck - i 10-m telescope on 2001 november 16 and 17 . on both nights the conditions were photometric and the fwhm seeing was typically @xmath25 .
the optical spectroscopy was taken through the low resolution imaging spectrograph ( lris ; @xcite ) with a @xmath26-wide longslit .
lris has a red ( lris - r ) and a blue ( lris - b ) camera ; each with a @xmath27 pixel array .
together they can provide spectral coverage between @xmath28 .
the plate - scale of both arrays is @xmath29 pixel@xmath13 and the field of view is @xmath30 .
the light beam was split between lris - b and -r using a 6800 dichroic .
a @xmath31 grooves mm@xmath13 grism blazed at 5000 produced a dispersion of @xmath32 pixel@xmath13 and a wavelength coverage of @xmath33 through lris - b .
a @xmath34 grooves mm@xmath13 grating blazed at @xmath35 and centred at @xmath36 was used through lris - r .
this had a dispersion of @xmath37 pixel@xmath13 and a wavelength coverage of @xmath38 . throughout the observing run lris - b and -r
were set to single and dual amp readout modes , respectively .
observations of the spectrophotometric standard star bd@xmath39 @xcite were taken to flatten the spectra and identify sky absorption features during the reduction stage . at the end of each night ,
spectroscopic dome flat - field exposures , bias frames and arcs were taken . during the first night
the lris autoguider and rotator were not operational .
as a consequence imaging and accurate guiding were not possible .
spectroscopy of class b0631 + 519 was therefore attempted as it was optically bright enough for accurate manual guiding .
the rotator was functioning on the second night which allowed accurate offsetting to the fainter gravitational lenses .
however , slit position angles were set such that each gravitational lens and a suitably bright nearby object were in the slit . this allowed confirmation that each gravitational lens had been detected .
a summary of the observing run is given in table [ log ] .
the exposure times were typically @xmath40 seconds .
however , for class b0850 + 054 the exposure times for lris - b and -r differed due to a software problem with the lris - b control system during the second observation . .observing logs for spectroscopic observations of class gravitational lenses class b0128 + 437 , class b0445 + 123 , class b0631 + 519 and class b0850 + 054 .
slit position angles ( pa ) are given east of north .
for the class b0631 + 519 observation there was no defined slit pa because the instrument rotator was not operational and therefore the slit pa was not fixed .
[ cols="^,^,^,^ , > " , ] an intriguing feature of the class b0631 + 519 spectrum , shown in fig .
[ lens-0631 ] , is the presence of a low redshift galaxy at @xmath42 . in the analysis presented above ,
this galaxy has been assumed to be one of two lensing galaxies . here , we investigate the single lensing galaxy scenario by determining whether a @xmath43 deflector is capable of producing @xmath44 image - splitting of a @xmath45 background source .
single deflectors at low redshifts are uncommon , but nevertheless do exist .
for example , mg1549 + 3047 ( @xmath46 ; @xcite ) and q2237 + 030 ( @xmath47 ; @xcite ) have low redshift lensing galaxies .
however , hst imaging of mg1549 + 3047 @xcite and q2237 + 030 @xcite has also shown both to have highly luminous ( @xmath48-mag ) and therefore massive lensing galaxies . unfortunately , the resolution of the york et al .
wht @xmath18-band image was insufficient to resolve the individual lensing galaxies .
therefore , only an upper - limit can be placed on the luminosity of the nearby class b0631 + 519 lensing galaxy by i ) assuming it is responsible for all of the @xmath49-mag @xmath18-band emission , ii ) converting this emission to the rest - frame @xmath50-band magnitude , @xmath51 , using the galaxy colour differences and non - evolutionary @xmath52-correction , @xmath53 , of @xcite , and iii ) comparing with the luminosity of an @xmath54 galaxy @xcite .
the absolute @xmath50-band magnitude , @xmath55 , was calculated using , @xmath56 where @xmath57 is the luminosity distance .
this was found to be @xmath58 which corresponds to a luminosity of @xmath59 or @xmath60 . assuming that this galaxy is producing the @xmath61 image splitting of the distant @xmath45 galaxy results in a mass enclosed within the einstein radius of @xmath62 and a mass - to - light ratio of @xmath63 .
however , it is clear from the spectrum of class b0631 + 519 that the @xmath45 galaxy is the dominant source of emission in @xmath18-band . therefore , this calculation provides a firm lower limit of the mass - to - light ratio of the low redshift galaxy .
as the typical mass - to - light ratios of galaxy - scale deflectors are at most only a few tens of @xmath64 and there is currently no evidence that the lensing is cluster assisted , this arrangement appears to be unlikely .
however , the possibility that the low redshift galaxy is a dark or dust enshrouded lens can not be ruled out at this stage .
there is currently only one ` dark lens ' candidate , class b0827 + 525 @xcite , in the entire class gravitational lens sample , which along with class b0631 + 519 would result in a class ` dark lens ' fraction of @xmath65 per cent .
alternatively , the narrow emission line spectrum of the low redshift object is consistent with a nuclear starburst galaxy which has undergone or is currently undergoing a period of intense star - formation .
therefore , it is possible that the under - luminous nature of the low redshift galaxy could be due to extinction caused by the presence of dust .
further observations would need to be conducted to investigate this possibility . whether class b0631 + 519 is a gravitational lens with two lensing galaxies or
a single dark / dusty lensing galaxy is unclear .
however , it is most likely that the @xmath66 galaxy is the main deflector and the @xmath42 object is a dwarf star - forming galaxy which happens to be coincident along the line - of - sight .
we have presented new optical spectra of four recently discovered gravitational lenses from class .
the redshift of the background source in class b0128 + 437 was found to be @xmath0 and the lensing galaxy redshifts in class b0445 + 123 and class b0850 + 054 were found to be @xmath67 and @xmath68 respectively .
interestingly , we discovered that class b0631 + 519 may have two lensing galaxies .
one is a low luminosity galaxy at @xmath3 which has little effect on the lensing potential , and the other is the primary lensing galaxy at @xmath4 .
the other possibility is that the low redshift galaxy is a dark / dust enshrouded gravitational lens .
however , the scenario in which the main lens is at @xmath66 is currently favoured .
we find the lensing galaxies in class b0445 + 123 and class b0631 + 519 ( @xmath5 ) to be early - type galaxies , with einstein radii of @xmath69 and @xmath70 kpc respectively .
the lensing galaxy in class b0850 + 054 was found to be consistent with a late - type galaxy with an einstein radius of 1.6 @xmath71 kpc .
for the redshift data presented in this paper to be useful for studying galaxy formation at high redshift or investigating the cosmological parameters , they must be coupled with the redshifts of the missing components .
we have identified isolated emission lines in the class b0128 + 437 and class b0850 + 054 spectra which we hypothesised identifications for . however , these tentative identifications will only be confirmed by further observations .
of the 22 gravitational lenses discovered by class , 17 have measured lensing galaxy redshifts and 12 have background source redshifts recorded .
obtaining spectra of the remaining gravitational lenses with unknown source or lens redshifts is an essential step in the follow - up of the class gravitational lensing survey .
to achieve this goal , further deep spectroscopic observations are required in the optical and near - infrared .
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 recognise 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 .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
jpm acknowledges the receipt of a pparc studentship .
lvek acknowledges the support from an stsci fellowship grant .
rdb is supported by an nsf grant ast0206286 .
|
we present the optical spectra of four newly discovered gravitational lenses from the cosmic lens all - sky survey ( class ) .
these observations were carried out using the low resolution imaging spectrograph on the w. m. keck - i telescope as part of a program to study galaxy - scale gravitational lenses . from our spectra
we found the redshift of the background source in class b0128 + 437 ( @xmath0 ) and the lensing galaxy redshifts in class b0445 + 123 ( @xmath1 ) and class b0850 + 054 ( @xmath2 ) .
intriguingly , we also discovered that class b0631 + 519 may have two lensing galaxies ( @xmath3 , @xmath4 ) .
we also found a single unidentified emission line from the lensing galaxy in class b0128 + 437 and the lensed source in class b0850 + 054 .
we find the lensing galaxies in class b0445 + 123 and class b0631 + 519 ( @xmath5 ) to be early - type galaxies with einstein radii of @xmath6 kpc .
the deflector in class b0850 + 054 is a late - type galaxy with an einstein radius of @xmath7 kpc .
[ firstpage ] gravitational lensing - quasars : individual : class b0128 + 437 , class b0445 + 123 , class b0631 + 519 , class b0850 + 054
|
human beings are adept and drawing context - sensitive associations and inferences across a broad range of situations ranging from the mundane to the creative inferences that lead to scientific discovery .
such reasoning has a strong pragmatic character and is transacted with comparatively scarce cognitive assets . however , despite our apparent proficiency at drawing inferences , and our ability to express words in such a manner that other people can ( usually ) understand the meaning that we are trying to convey , our theoretical understanding of how this process occurs has been slow to develop .
the field of cognitive science has recently produced an ensemble of semantic models which have an encouraging , and at times impressive track record of replicating human information processing , such as human word associations norms @xcite .
the term `` semantic '' derives from the intuition that words seen in the context of a given word contribute to its meaning , or , more colloquially expressed , the meaning of a word is derived from the `` company it keeps '' @xcite . in order to progress in our understanding of how meaning is generated from sets of words in a language we must understand the way in which the mental lexicon of that language is generated during language acquisition , and how it works once created in the mind of a specific individual .
the mental lexicon of a language refers to the words of a language , but its structure is represented by the associative links that bind this vocabulary together .
such links are acquired through experience and the vast and semi - random nature of this experience ensures that words within this vocabulary are highly interconnected , both directly and indirectly through other words .
for example , during childhood development and the associated acquisition of english , the word _ planet _ becomes associated with _ earth _ , _ space _ , _ moon _ , and so on .
even within this set , _ moon _ can itself become linked to _ earth _ and _ star _ etc .
words are so associatively interconnected with each other that they meet the qualifications of a ` small world ' network wherein it takes only a few associative steps to move from any one word to any other in the lexicon @xcite . because of such connectivity individual words are not represented in long - term memory as isolated entities but as part of a network of related words .
however , depending upon the context in which they are used , words can take on a variety of different meanings and this is very difficult to model @xcite .
much evidence shows that for any individual , seeing or hearing a word activates words related to it through prior learning . understanding how such activation affects memory requires a map of links among known words , and free association provides one reliable means for constructing such a map @xcite .
in free association experiments , words are presented to large samples of participants who produce the first associated word to come to mind . the probability or strength of a pre - existing link between words
is computed by dividing the production frequency of a response word by its sample size .
for example , the probabilities that _ planet _ produces _ earth _ and _ mars _ are 0.61 and 0.10 , respectively , and we say that _ earth _ is a more likely or a stronger associate of _ planet _ than _
just like the nonlocality experiments of quantum theory , human memory experiments require very careful preparation of the state to be tested .
for example , in extralist cuing , participants typically study a list of to - be - recalled target words shown on a monitor for 3 seconds each ( e.g. , _ planet _ ) .
the study instructions ask them to read each word aloud when shown and to remember as many as possible , but participants are not told how they will be tested until the last word is shown .
the test instructions indicate that new words , the test cues , will be shown and that each test cue is related to one of the target words just studied ( e.g. , _ universe _ ) .
these cues are not present during study ( hence , the name extralist cuing ) . as each cue
is shown , participants attempt to recall its associatively related word from the study list .
in contrast , during intralist cuing the word serving as the test cue is presented with its target during study ( e.g. , _ universe planet _ ) .
participants are asked to learn the pairing , but otherwise the two tasks are the same .
it appears that more associate - to - associate links benefit recall , and two competing explanations for this phenomenon have been proposed : spreading activation , and spooky activation at a distance .
this paper will demonstrate an intriguing connection between these two explanations , obtained by making the assumption that words can become entangled in the human mental lexicon .
entanglement is a phenomenon unique to quantum behaviour .
if a system consisting of two components becomes entangled then it can not be thought of as separate anymore ; a description of one component without reference to the other will , in some cases , fail .
indeed , an entangled quantum system will generally exhibit an intercomponent agreement with reference to any combination of measurement settings .
this is of particular importance for a system that becomes spatially extended , as in the case where the two components are taken a long way from each other . here
we find that quantum systems display correlation instantaneously in response to what might even be a delayed choice of measurement setting @xcite , and yet can not be used to transmit information between two observers , and thus does not actually violate special relativity @xcite .
this is in contrast with classical scenarios of correlation . in a classical situation
a system is in a pre - existing state , and this is discovered through the process of measurement .
not so with a quantum system , where the process of measurement can actively influence the outcome itself .
this fundamental difference between the two types of system was first alluded to in the by now famous epr debate , but was only inescapably highlighted with the more subtle ( and recent ) results surrounding the contextuality of quantum systems ( see @xcite or @xcite for a good introduction to these ideas ) .
an entangled quantum system is very different from a correlated classical system ; no pre - existing elements of reality @xcite have been found that can explain the agreement that is obtained between distant measuring devices that are set to determine the state of a quantum system . to make these ideas more concrete ,
let us consider a specific example of classical correlation .
if the same number is written on two pieces of paper , enclosed in two envelopes , and sent to alice and bob at two distant ends of the universe , the information obtained upon opening of one of the envelopes will instantly correlate with the state of the other envelope at the other end of the universe .
however , these correlated pieces of paper are not entangled .
the number on the two pieces of paper can be regarded as a hidden variable , or element of reality ; even before we open the envelope it exists , in both envelopes . upon opening the envelope at one end of the universe we find out what that number is , and
hence know what number is already inscribed upon the other piece of paper .
the quantum analogue of this scenario would be far stranger .
the situation most similar to the nonlocal effects exhibited by entangled quantum systems would involve alice , at one end of the universe choosing to write a number upon her blank piece of paper when she opens her envelope , and then finding that bob , upon opening his envelope found exactly the same number upon his piece of paper at the other end of the universe .
obviously this does not happen .
however , we might ask if similar cases of intercomponent dependency , or spooky - activation - at - a - distance , exist for systems beyond the field of physics .
we shall now look at the problem of modelling associate - to - associate links in the human mental lexicon , before showing how the assumption that associates might be entangled in a subject s cognitive state can lead to a new model of word associations .
figure [ target - t ] shows a hypothetical target word having two target - to - associate links in a subject s cognitive state . there is also an associate - to - associate link between associates 1 and 2 , and an associate - to - target link from associate 2 to the target @xmath0 .
the values on the links indicate relative strengths estimated via free association .
nelson et al . have investigated reasons for the more likely recall of words having more associate - to - associate links @xcite .
two competing explanations for why associate - to - associate links benefit recall have been proposed .
the first is the spreading activation equation , which is based on the classic idea that activation spreads through a fixed associative network , weakening with conceptual distance ( e.g. , @xcite ) : @xmath1 where @xmath2 is the number of associates and @xmath3 .
@xmath4 denotes the strength of implicit activation of target @xmath0 due to study , @xmath5 target - to - associate activation strength , @xmath6 associate - to - target activation strength ( resonance ) , and @xmath7 associate - to - associate activation strength ( connectivity ) .
multiplying link strengths produces the weakening effect .
activation ostensibly travels from the target to and among its associates and back to the target in a continuous chain , and the target is strengthened by activation that returns to it from pre - existing connections involving two and three - step loops .
more associate - to - associate links create more three - step loops and theoretically benefit target recall by increasing its activation strength in long - term memory .
importantly , note that the effects of associate - to - associate links are contingent on the number and strength of associate - to - target links because they allow activation to return to the target .
if associate - to - target links were absent , even the maximum number of associate - to - associate links would have no effect on recall because activation could not return to the target .
in contrast , in the ` spooky activation at a distance ' equation , the target activates its associative structure in synchrony : @xmath8 where @xmath3 ; @xmath5 , target - to - associate @xmath9 strength ; @xmath6 , associate @xmath9-to - target strength ( resonance ) ; @xmath7 , associate @xmath9-to - associate @xmath10 strength ( connectivity ) this equation assumes that each link in the associative set contributes additively to the target s net strength .
the beneficial effects of associate - to - associate links are not contingent on associate - to - target links .
stronger target activation is predicted when there are many associate - to - associate links even when associate - to - target links are absent .
in fact , associate - to - target links are not special in any way .
target activation strength is solely determined by the sum of the link strengths within the target s associative set , regardless of origin or direction .
how should we represent the combination of words in the human mental lexicon ? qt uses the tensor product , @xmath11 , to denote composite systems .
consider the case of @xmath12 study words : @xmath13 and @xmath14 presented to a group of subjects .
let us assume that , when cued , the subjects recall neither target word .
in this case we could write : @xmath15
where the notation @xmath16 is just shorthand for the tensor product state @xmath17 describing the composite system of two negative outcomes .
if word @xmath13 alone was recalled then we would write @xmath18 , whereas in the converse case we would write @xmath19 and finally , if both words were recalled then the tensor product would yield the state @xmath20 . however , this straightforward scenario is not the only form of situation possible in the quantum formalism .
superposition states can also occur , and these are important as they can represent the situation where the words @xmath13 and @xmath14 may be more likely to be recalled in one context than another .
assume that we can represent one subject s cognitive state with reference to the combined targets @xmath13 and @xmath14 as a 2 @xmath21-bit register that refers to their states of ` recalled ' and ` not recalled ' in combination .
thus , if we represent the target words using the standard superpositions @xmath22 and @xmath23 , ( where @xmath24 and @xmath25 ) , then it is possible to denote the state of the combined system by writing the tensor product @xmath26 where @xmath27 .
this is the most general state possible .
it represents a quantum combination of the above four possibilities , obtained using a tensor multiplication between the states @xmath28 and @xmath29 .
in contrast to the simple cases discussed above , here no state of recall is ` the ' state , rather , we must cue the subject and elicit a response from them before we can talk about a word being ` recalled ' or ` not recalled ' .
indeed , a different cue might elicit a very different response , and the quantum formalism could deal with this via a change of basis .
it is important to realise however , that is _ not _ the only form of state that can be obtained from combination of @xmath28 and @xmath29 in the quantum formalism .
the other form of state , an _ entangled state _ is one that it is impossible to write as a product . as an example of an entangled state
, we might consider the the state @xmath30 where the words @xmath13 and @xmath14 are either _ both _ recalled , or both _ not _ recalled in relation to a cue @xmath21 .
one representation of this scenario is given by the following state : @xmath31 this seemingly innocuous state is one of the so - called _ bell states _ in qt .
it is impossible to write as a product state , thus it differs markedly from .
the fact that entangled systems can not be expressed as a product of the component states makes them non - separable . more specifically , there are no coefficients which can decompose equation into a product state exemplified by equation which represents the two components of the system , @xmath13 and @xmath14 , as independent of one another . for this reason
@xmath30 is not written as @xmath32 as it ca nt be represented in terms of the component states @xmath28 and @xmath29 .
.matrix corresponding to hypothetical target shown in figure [ target - t ] [ cols="^,^,^,^",options="header " , ] nelson and mcevoy have recently begun to consider the spooky - activation - at - a - distance formula in terms of quantum entanglement , claiming that the activation - at - a - distance rule assumes that the target is , in quantum terms , entangled with its associates because of learning and practicing language in the world .
associative entanglement causes the studied target word to simultaneously activate its associate structure " ( * ? ? ?
the goal of this section is to formalise this intuition . at the outset , it is important that the quantum formalism be able to cater for the set size and connectivity effects described elsewhere @xcite . recall both set size and associative connectivity have demonstrated time and again robust effects on the probability of recall . because the spooky - activation - at - a - distance formula sums link strengths irrespective of direction , it encapsulates the idea that a target with a large number of highly interconnected associates will generate a high activation level during study .
table [ tab : target - t ] is a matrix representation of the associative network of the hypothetical target @xmath0 shown in figure [ target - t ] .
the last line of the matrix represents the summation of free association probabilities for a given word .
for example , @xmath33 these free association probabilities may be added as it is assumed that each free association experiment is independent , that is associate @xmath34 being recalled in relation to cue @xmath0 is assumed independent of it being recalled in relation to the cue @xmath35 . since free association experiments require a subject who has not been primed in any way beyond that provided by the cue , this appears to be a reasonable simplifying assumption .
both spreading activation and spooky - activation - at - a - distance approaches assume that free association probabilities determine the strength of activation of a word during study ; they only differ in the way this activation strength is computed .
viewing free association probabilities in this way allows the matrix to be considered as a many bodied quantum system modelled by three qubits .
figure [ fig : qubit - register - t ] depicts the system , here , each word is in a superposed state of being activated @xmath36 , or not @xmath37 .
note how each summed column with a non - zero probability leads to a qubit . for ease of exposition in the following analysis
, we shall change variables .
the probabilities depicted in table [ tab : target - t ] are related to the probability densities of figure [ fig : qubit - register - t ] by taking their square root : e.g. @xmath38 .
using such a change of variables , the state of the target word @xmath0 would be written as : @xmath39 where the probability of recall due to free association is @xmath40 , and @xmath41 represents the probability of a word not being recalled .
thus , the states of the individual words are represented as follows in order to avoid cluttering the analysis with square root signs : @xmath42 where @xmath43 and @xmath44 . as detailed in the previous section ,
tensor products are used to model many bodied quantum systems .
the state @xmath45 of the most general combined quantum system is given by the tensor product of the individual states : @xmath46 the intuition behind this expression is an enumeration of the possibilities of the states of the qubits .
so , @xmath47 represents the state in which all respective qubits collapse onto their state @xmath36 .
in other words , @xmath47 denotes the state of the system in which words @xmath0 , @xmath35 and @xmath34 have all been _ activated _ due to study of target @xmath0 .
the probability of observing this is given by the taking the square of the product @xmath48 .
conversely , the state @xmath49 corresponds to the situation in which none of the words have been activated .
the state @xmath45 of the three - bodied system does not capture nelson & mcevoy s intuition that the studied target word @xmath0 simultaneously activates its associative structure .
recall their suggestion that target @xmath0 activates its associates in synchrony : when target @xmath0 is studied it activates _ all _ of its associates , or none at all .
interestingly , this idea is quite easily captured by the assumption that the state @xmath45 evolves into a bell entangled state @xmath50 of the form : @xmath51 this formula represents a superposed state in which the entire associative structure of a subject is in a state of potential activation ( @xmath47 ) or potential non - activation ( @xmath52 , with the probabilities of these states occurring given by @xmath53 and @xmath54 respectively .
how should we ascribe values to the probabilities @xmath53 and @xmath54 ?
in qt these values would be determined by the unitary dynamics evolving @xmath45 into @xmath50 .
as such dynamics are yet to be worked out for cognitive states , we are forced to speculate .
one approach is to assume the lack of activation of the target is determined solely in terms of lack of recall of any of the associates .
that is , @xmath55 consequently , the remaining probability mass contributes to the activation of the associative structure as a whole .
thus , starting from the assumption of a bell entangled state , we find that the probability @xmath54 expresses nelson & mcevoy s intuition : @xmath56 term @xmath57 corresponds to the summation of the free association probabilities in the above matrix .
in other words , term @xmath57 corresponds exactly to the spooky - activation - at - a - distance formula ( see equation [ eq : spooky ] ) . at such
, the assumption of a bell entangled state provides partial support for the summation of free association probabilities which is embodied by the spooky - activation - at - a - distance equation .
ironically perhaps , term @xmath58 corresponds to free association probabilities multiplied according to the directional links in the associative structure .
this is expressed in the second term of the spreading activation formula ( see equation [ eq : spreadact ] ) .
in other words , departing from an assumption of entanglement leads to an expression of activation strength which combines aspects of both spooky - activation - at - a - distance and spreading activation .
the third term @xmath59 is more challenging to interpret . in a more complete model capable of generating some form of evolutionary dynamics we might expect that it would arise from the underlying structure of the hilbert space used . here
, it has arisen from apparently sensible assumptions about how probabilities should be amassed in the the bell entangled state .
what significance can be drawn from this term ?
the beginnings of an answer can be formulated by returning to figure [ target - t ] . when seen in the context of actual values ,
the term @xmath59 has a significant compensating effect , subtracting 0.77 from the summation of the spooky term with the spreading activation term : @xmath60 thus , according to this analysis , the strength of activation @xmath54 lies somewhere between spreading activation and spooky - activation - at - a - distance .
based on a substantial body of empirical evidence , nelson and mcevoy have argued persuasively spreading activation _ underestimates _ strength of activation @xcite .
here we have seen that when departing from an assumption that the associative structure is bell entangled , a cautious preliminary conclusion is that spooky - activation - at - a - distance overestimates the strength of activation and it is term @xmath59 which compensates for this .
we conclude this section with some remarks how the entanglement model above assists in the explanation of some experimental results that have not been well accounted for in current models of the mental lexicon .
attempts to map the associative lexicon soon made it clear that some words produce more associates than others .
this feature is called ` set size ' and it indexes a word s associative dimensionality @xcite .
it was also revealed that the associates of some words are more interconnected than others .
some words have many such connections ( e.g. , _ moon - space _ , _ earth - planet _ ) , whereas some have none , and this feature is called connectivity " @xcite .
thus , experiments have shown that in addition to link strengths between words , the set size and connectivity of individual words have powerful effects on word recall , which existing theories can not generally explain .
this is an interesting result for the model suggested here , as according to the above analysis we might surmise that the more associates a target has , the more qubits are needed to model it .
when these are tensored the resulting space will have a higher dimensionality .
therefore a large set size is catered for by a tensor space of higher dimensionality .
conversely , interconnectivity is catered for by larger probabilities in the initial superposed states of the respective qubits .
consider once again the matrix in table [ tab : target - t ] . in the general case ,
when the associative structure is highly interconnected the sums of probabilities in the last row will tend to be higher .
these will contribute to higher activation strength as they are summed ( as defined by probability @xmath54 ) in the same way as in spooky - activation - at - a - distance .
so it is possible to have a high activation strength even though the tensor space has high dimensionality . when the associative structure in not interconnected these probabilities will be low and hence lead to lower strength of activation
obviously the model presented here is overly simple .
the human mental lexicon is a vast and highly interconnected network consisting of thousands of words , and its associative structure is far more complex than the simple toy model in figure [ target - t ] could ever hope to represent .
however , some promising initial results have been obtained from this very simplified analysis .
we have seen some evidence that a quantum approach can model set size and connectivity effects , and a prediction has been made that spooky - activation - at - a - distance overestimates associative behaviour by some factor . while equation makes a very concrete suggestion about how large this overestimation might be , it is unlikely that the situation will prove this simple .
firstly , this is a very basic toy model , addressing the behaviour of only one target and two associates .
a fully developed theory would have to be stated in terms of all the targets and all of their associates .
it is very hard to extrapolate from this model to a more realistic one .
one possibility would involve applying the apparatus of statistical mechanics in a density matrix approach to model sets of targets and their associates , but no results have been obtained here to date .
a related issue surrounds the meaning of the term @xmath59 in equation as more associative structure is added .
here , we would see @xmath59 gaining more and more terms ( since @xmath57 and @xmath58 essentially correspond to the two extreme values in the expansion of the bracketed term in ) .
this does not seem plausible , and the way in which this term would act as it became larger is yet to be established .
in addition to these immediate issues , much work remains to be done in this area .
given the large amount of data collected about word association norms @xcite we might expect that experiments can be performed that might distinguish between the varying predictions of the different models , and work is underway to generate results here .
a physically and cognitively motivated time evolution equation capable of generating the bell - type state is essential before we can consider this model to be truly quantum(like ) and initial ruminations about how this might be achieved are presented in @xcite .
ending on a slightly more positive note , we consider the most important result of this article to be the indication that a straight tensorial combination of associate words in the mental lexicon is not particularly representative of the intuition that words and their associates are activated in synchrony .
given that a bell - type entangled state provides a far more likely candidate for the behaviour of word association networks this result thus provides some first steps towards establishing evidence that human cognitive structures have some quantum(like ) behaviour .
this project was supported in part by the australian research council discovery grant dp0773341 to p. bruza and k. kitto , and by grants from the national institute of mental health to d. nelson and the national institute on ageing to c. mcevoy .
j.p . levy and j.a . bullinaria . in r.f .
french and j.p .
sounge , editors , _ connectionist models of learning , development and evolution : proceedings of the sixth neural computation and psychology workshop _ , pages 273282 .
springer , 1999 .
d. nelson and c.l .
entangled associative structures and context . in p.d .
bruza , w. lawless , c.j .
van rijsbergen , and d. sofge , editors , _ proceedings of the aaai spring symposium on quantum interaction_. aaai press , 2007 .
m. patel , j.a .
bullinaria , and j.p . levy . .
french and j.p .
sounge , editors , _ connectionist models of learning , development and evolution : proceedings of the fourth neural computation and psychology workshop _ , pages 199212 .
springer , 1997 .
|
following an early claim by nelson & mcevoy @xcite suggesting that word associations can display ` spooky action at a distance behaviour ' , a serious investigation of the potentially quantum nature of such associations is currently underway .
this paper presents a simple quantum model of a word association system .
it is shown that a quantum model of word entanglement can recover aspects of both the spreading activation model and the spooky model of word association experiments .
|
multiplayer non - cooperative game theory is a mathematical formulation of competition in which players compete against one another to obtain some resource or reward [ 1 ] . at its most abstract level
a game is simply a mapping , via some function , of the elements of one set to another .
although the mathematics of game theory can be phrased in terms of such abstract symbols , those symbols are given an interpretation in terms of the actions , or strategies , of the players and the eventual rewards they receive .
implicit in this interpretation is the notion that , should it be so desired , the players could actually _ play _ such a game .
in other words , there exists a real physical implementation of a game in terms of physically realizable actions and tangible rewards .
we call any game that can be implemented in a physical reality a _ playable game_. in a world described by classical physics the correspondence between the mathematical abstractions and the real physical objects that might be used to play a game is straightforward
. thus we might implement a game using counters and the strategies would correspond to moves made with those counters .
the resultant state of the counters after the players strategies have been implemented determines the outcome of the game and we could imagine the players receiving some tangible reward such as cash or cupcakes .
informally , in such a scenario it is assumed that rational players will choose a strategy that will maximise this quantity , given that the other players are choosing their strategies to achieve the same maximisation of their reward .
the world as we know it , however , is governed by the laws of quantum mechanics and classical properties emerge as a macroscopic limit to this more fundamental description . in a quantum description
it is not so straightforward to assign elements of physical reality to the various components we might choose for the implementation of a game [ 2 ] .
thus in a game played with quantum objects obeying the laws of quantum mechanics it is not immediately obvious how to draw the correspondence between the mathematical abstractions of game theory and the various physical elements needed to implement a game . by focusing carefully on the notion of _ playability _ for games
we believe that these difficulties can be resolved .
the perspective we develop shows how the standard notions from game theory can be applied to games played with quantum mechanical objects .
it is fair to say that there has been a mixed reaction to the merger of game theory and quantum mechanics , first formally introduced over ten years ago [ 3,4 ] .
these original seminal papers of meyer , and eisert , wilkens and lewenstein , however , have created an exciting and intriguing new area of research .
possibly motivated by the recognition that the model of computation based on turing machines depends fundamentally on the nature of physical reality [ 5 ] , and the subsequent development of quantum algorithms [ 6 ] , there has perhaps been the hope that quantum mechanics will , somehow , lead to a similar revolution in game theory . such a revolution has not yet occurred , but there have been indications of some tantalizing results .
it is not uncommon to hear the opinion that quantum mechanics has nothing to offer game theory and vice versa .
this pessimism may be due , in some part , to the various approaches to terminology in which terms such as quantum games , quantized games , quantum strategies and sometimes even quantized strategies , can be used in slightly different ways .
a major step forward in the clarification of the terminology was taken by bleiler [ 7 ] who introduced the term proper quantization to refer to quantum games that are correct extensions of some underlying classical game .
much work has focused on specific examples of games played with entangled quantum systems ( see , for example , [ 3,4 , 8 - 18 ] ) in which at first sight it appears that in some circumstances quantum mechanics allows the players to reach a more advantageous equilibrium , when compared to the classical games upon which these quantum versions are based .
such comparisons are , however , not straightforward and it is not always clear to what extent the quantum extension can be said to be the correct quantum version of the classical game that inspired it . by focusing here on _
playable _ games we adopt the approach that games are physical processes that are played with objects and actions that have a physical existence . in what follows
we shall concentrate on 2-player games for convenience .
it is not difficult to extend the description to multiplayer non - cooperative games with a greater number of players than 2 . in typical expositions of multiplayer
non - cooperative game theory it is assumed that the players each choose a strategy from some set of strategies available to them . in a 2-player game with players _ a _ and _ b _ we might denote the set of strategies available to player _ a _ by @xmath0 and similarly for player _
b_. the strategies are thus no more than elements of a set .
the sets @xmath1 and @xmath2 do not need to be equal , nor do they need to intersect . the choices the players make become the input to some function that calculates their reward based on the individual choices they have made .
the rewards , or outcomes , for each player are expressed as a tuple @xmath3 drawn from a set of outcome tuples @xmath4 , the individual tuple that is calculated being determined by the input strategy tuple . thus
, if the players choice of strategy is described by @xmath5 and the outcome tuple that is generated by this input is @xmath6 we say that player _ a _ receives the outcome @xmath7 and player _ b _ the outcome @xmath8 .
if we assume that both players select their respective strategies from the same set of possibilities then the function that takes the input strategies to the outcomes can be described by the mapping @xmath9 .
we still do not quite have a structure that can be described as a _
game_. the purpose of the game is to compete , or in crude terms , to win .
there must be some outcomes from the set @xmath4 that _ a _ prefers above the others , and similarly for player _
b_. thus we need to describe a
_ preference relation _ over the outcomes for each player that describes their desired outcomes in some order of preference .
player _ a _ , knowing the values of the function @xmath10 will try to ensure that his most desired outcome will occur , given that player _ b _ is doing exactly the same thing .
it is clear that , depending on the preference relations , both players may have to compromise and accept an outcome that is not their most desired in order not to obtain an outcome that is less preferable .
if both players can select a strategy such that they would not change it , irrespective of the choice of the other player , then this is an equilibrium position known as a nash equilibrium [ 19 ] .
a nash equilibrium is often not the most optimal for the players in the sense that there can be a pair of strategy choices for _ a _ and _ b _ which will increase their respective payoffs or rewards .
a pareto optimal outcome is one in which there is no other outcome that gives at least the same payoff for every player _ and _ gives at least one player a better outcome .
pareto optimization can be thought of as the outcome such that it can not be improved upon without reducing the payoff of at least one player .
any 2-player game that can be described in the fashion we have just outlined is a _ playable game _ , provided that the function @xmath10 is _
computable_. by computable we mean that there is some algorithmic procedure for determining the output for a given input .
an excellent discussion of this can be found in [ 20 ] .
the strategy choices can be thought of as no more than symbols from some alphabet .
these symbols can be coded in binary and transmitted to some turing machine which takes the inputs and computes the output which is the outcome tuple for the players .
thus we can think of a 2-player game as a turing machine in which there are two tape inputs , one for player _ a _ and one for player _ b , _ in which each player writes their selection of symbol , in binary , on the tape .
the turing machine reads the inputs and computes the appropriate output in the form of an outcome for each player .
there are , however , two very important subtleties that are easy to overlook in this classical description .
implicit in the description of a playable classical game is the understanding that the choices of the players can be determined , that is , _
measured_. these become the inputs to the computable function that determines the outcome , but in order to perform that computation , we require that the choices can be distinguished . if we implement the game using some turing machine then that machine is implicitly assumed to be capable of reading the _ state _ of the tape .
in other words , the state of the tape is _ measured_. this becomes more transparent if we consider an implementation of our playable game in terms of coins .
each player will prepare his coins in some state , specified by a binary string that represents the chosen strategies .
this state must be read , or ` decoded ' in order to compute the outcomes .
so , for example , if the players are each given three coins then player _ a _ may choose to transmit the coins in the state _
hth _ which represents one of the 8 possible choices available to him . in order for the outcome to be correctly computed , the state of these coins , and the state of the coins of player _
b _ , must be correctly measured . also implicit in the description of a playable classical game
is the understanding that the players perform some action , or set of actions , on an _ initial state _ in order to convey their choice .
so if player _ a _ is given three coins in the state _
hhh _ in the example above then in order to achieve his desired output state he must perform the actions @xmath11 where @xmath12 is ` flip ' and @xmath13 is ` do nt flip ' .
it is tempting to describe the 8 possible output choices of each player in the 3-coin example as their possible choices of strategy .
equally , we could describe the strategy choice as a combination of _ initial state _
+ _ sequence of operations _ , because in this case there is a one - to - one correspondence .
the operations , in this case to flip or not flip , are the available actions of the players , that is , the things that the players do to achieve their desired output state .
it is more natural to think of what actions the players take , _ given an initial state _
, as a strategy because this corresponds more closely with the physical situation .
in other words a strategy is a means of answering the question given a start point , what must i do to achieve my desired end point ? .
in order to illustrate the potential difficulty with the identification of an output state as a strategy we shall consider the following _ playable _ classical game . each player is given 3 coins prepared in the state _
are placed in some device which performs the flip operation , if so desired .
furthermore , each player is given the capability of performing a flip on _ one _ of the other player s coins .
it is now no longer possible to identify the output state of player _ a _
s coins as his choice of strategy .
it is certainly possible to construct a physical device that would implement such a game , and thus the game is certainly playable . the strategy set of player _ a _ in this case is given by the list of the options open to him , which includes his possible actions on the coins of player _
b_. in this game each player has 32 possible choices of action , that is , their strategy sets have 32 elements , giving 32@xmath14 possible combinations of both players strategies .
the number of distinct output states of the 6 coins is 64 .
the example above shows that the detailed physical description of how we actually implement a game , that is how we actually play the game , is critical , because it determines the elements of our ` strategy ' set .
the output state is determined by the choice of what the players _ do _ , given an initial state , and it is this ouput state that is measured .
the measurement result becomes the input to the computable function that determines the outcomes for the players .
it is this detailed description of the physical implementation of a game that is essential when we consider what it means to play a game using quantum mechanical objects that obey the laws of quantum mechanics .
it is this detailed description that allows us to consider what it means to ` quantize ' a game .
the careful examination of the elements , implicit and explicit , that are required to implement a game in a physical reality leads us to a general model of a playable game .
this is shown in figure 1 in which these essential elements are abstracted .
the initial state is simply the starting configuration of whatever physical element is acted upon by the players in order to produce the output . in the case of an implementation of the game using a turing machine , for example , the initial state might be just be a number of blank squares upon which the players can act to produce their desired output .
the players have a set of actions they can perform upon this initial state in order to change its configuration . in the case of the coins
this is whether to flip or not .
thus this step in the implementation of a playable game can be thought of as state preparation .
the combination of the initial state and the actions performed by the players results in some output state .
this is the state the players have configured by their actions .
the configuration of the output state must be measured to produce a measurement result that captures the configuration of the output state in some way .
these results become the input to some function , which can be thought of as a look - up table , that produces the outcomes or payoffs for the players . the essential elements of a playable classical game ,
as described above , suggest how we may approach the notion of quantizing a game . strictly speaking we quantize the physical system that is used to implement a classical game .
the question of what it means to ` quantize ' a game is somewhat problematical , as we shall see . with this caveat in mind
we now allow the physical elements of our game to be quantum mechanical in nature with the operations of the players being taken from the set of those permitted by quantum mechanics .
the measurement of the state becomes a quantum measurement .
the results of the measurement become the inputs to the computable function that determines the outcomes for the players .
it is important to note that the central ` philosophy ' of the playable game has not altered by this extension to the quantum domain ; just as in the classical case , each player must consider what operations he needs to perform on the initial state in order to produce an output state the measurement of which will yield an optimal result for him , given that the other player is making the same deliberation .
of course the only real difference between games in the classical and quantum domains is that in the latter the physical elements that implement the game are quantum mechanical .
the mathematical formalism of game theory applies in both domains .
the description of a playable game implemented using classical objects implies a certain underlying physical reality . according to the copenhagen interpretation
, a quantum mechanical description of nature does not assign elements of physical reality until a measurement is made . in classical game theory
it is tacitly assumed that that the choice of strategy corresponds to some element of physical reality . in a turing machine implementation of a classical playable game , once a symbol to describe
the strategy choice has been written on the tape it is communicated to the turing machine , and those bits exist as elements of physical reality on the tape . in quantum mechanics , however , these symbols , now have to be considered as qubits on a quantum mechanical tape , and they do not have any corresponding element of physical reality until they are measured . furthermore , if we imagine two separate inputs to our quantum turing machine then not only does quantum mechanics mean that we have to inscribe qubits on our quantum tape , but also that the two quantum tape inputs can be _
entangled_. whilst the turing machine model of a playable game is instructive in highlighting the potential differences between playing games in the classical and quantum domains , it is not the best model for allowing us to draw correspondences between games played in the different domains .
the general model of a playable game given in figure 1 is more suitable .
the physical elements of a playable game described in this figure remain unchanged when we consider games in the quantum domain , the only difference being that the input and output states are now quantum states and the actions the players perform to transform the intial state to a desired output state are quantum operations . in classical playable games , the players have some preference relation over the measurement outcomes and it is these preference relations that determine the choice of operation for the players . in the quantum domain
the measurement results are the eigenstates of the measurement operator and the players have some preference relation over these eigenstates .
( we only consider von neumann measurements in this paper .
extending to more generalized quantum measurements adds unnecessary technical detail at present , and essentially introduces no significant new conceptual elements to the game - theoretic description ) . in order to be a little more specific
we shall consider 2 player games in the quantum domain .
the extension to games with more than 2 players is relatively straightforward .
we imagine a playable game with players _ a _ and _ b _ in which some physical system , prepared in an initial quantum state @xmath15 , is fed into a device .
the players each perform some unitary transformation on the state to produce an output state @xmath16 .
this output state is measured , or rather the physical property @xmath17 is measured , the result being one of the eigenstates , @xmath18 , of the operator @xmath17 .
each player has some preference relation over these eigenstates which we label @xmath19 . before enacting their operation on the initial state
, each player performs some computation which determines which operation is in their best interests , given that the other player is going to choose an operation which is in _ his _ best interests .
the preference relations , of course , determine what each player considers to be in their best interests given the constraints .
the choice of action is thus some computable function for each player which takes as input the initial state , the possible measurement outcomes , the preference relations and the available operations .
let us make this more explicit .
we let the set of possible states of our quantum system be denoted by @xmath20 .
the set of all possible unitary operations on this state will be denoted by @xmath21 where we have used the caret to remind ourselves that this is a set of _ operators_. the ouput state from our device is thus described by some mapping @xmath22 .
the choice of the element from the set @xmath21 for each player is determined by their knowledge of the initial state , the operations that can be performed on that state , the results of the measurement on the output state and the preference relations over those results .
it is important to note that if both players have full knowledge of the parameters of the physical system , including knowledge of each other s preference relations and available operations , then each player can determine the element of @xmath21 that the other player will pick , assuming rational players , and assuming that the game admits the calculation of such a preferred choice .
we are now in a position to examine some general properties of playable quantum games .
these general considerations will be made more concrete when we consider a simple , but highly non - trivial , example of a quantum game in section 3 .
it is clear that there is no fundamental difference between playing a game with quantum mechanical objects and state preparation . indeed , playing a quantum game _ is _ state preparation , with the prepared state being determined by the preference relations over the measurement outcomes ( along with knowledge of the other important parameters , as discussed above ) .
one important question , from the perspective of game theory , is whether the game parameters lead to the preparation of an _ equilibrium _ output state by rational players . before we look at the general structure of a quantum game we have proposed it is worth noting here that we are considering here only those quantum games in which the players are allowed to perform some _ unitary _ operation on the input state .
measurements are , of course , perfectly allowable quantum operations .
a more general treatment of quantum games would allow the players to perform some measurement on the quantum state , possibly followed ( or preceded ) by some unitary operation .
we shall touch upon this briefly when we consider uncertainty in quantum games shortly , but the main details of this more general treatment will be published elsewhere .
the various parameters that characterise a game played using quantum mechanical objects and operations are described below .
@xmath20 : the set of all possible states of the physical system used to implement the game @xmath23 : the initial state of the physical system @xmath24 @xmath25 : the set of possible output states @xmath26 @xmath27 : the output state , that is , the state produced by the operations of the players on the initial state .
@xmath28 @xmath21 : the set of all possible unitary transformations on an element of @xmath29 @xmath30 : the set of all unitary operations available to player _ a_(_b _ ) .
@xmath31 and @xmath32 is not necessarily equal to @xmath33 .
note that these sets may contain the indentity operator which simply means that one option that the players can choose is to do nothing to the initial state.@xmath34 @xmath35 : an element of the set @xmath32 .
@xmath36 @xmath37 : an element of the set @xmath33 .
@xmath38 @xmath17 : the hermitian operator describing the measurement on the output state @xmath27 .
the measurement produces the eigenstate @xmath39 with probability @xmath40 @xmath39 : the eigenstates of @xmath41.@xmath42 @xmath19 : an ordered list of the eigenstates @xmath43 such that the 1@xmath44 element is _ a_(_b _ ) s most preferred measurement result , the 2@xmath45 element is the 2@xmath45 most preferred result and so on .
this is the preference relation for _
a_(_b _ )
the basic sets are depicted in figure 2 .
if the players do not have access to the full set of operations permitted by quantum mechanics on the initial state then the output state will be a subset of the set of possible states of the quantum system . despite the apparent increase in formalism introduced by the quantum description
the fundamental questions of game theory remain unaltered .
what actions must a player take to ensure that his most advantageous outcome occurs , given that his opponent is making precisely the same deliberation ?
the determination of what a particular player considers to be advantageous is , of course , encapsulated in the preference relations .
another important game - theoretic question that remains unaltered for a quantum game is whether some equilibrium state is reached as a result of the players deliberations .
some immediate differences do , however , present themselves .
the most obvious of these is the difference between classical and quantum measurements .
it is not possible , in general , in quantum mechanics to determine the complete configuration of a state .
indeed , it is nt even correct to ascribe elements of reality to the elements of that configuration . in a classical game
the configuration of the output state can , in principle , be measured .
in the quantum case we can select a property of the state to measure , but this will not give us complete information about the resultant output state .
this apparent limitation is partly overcome by relating the preference relations of the players to the measurement outcomes .
however , the consequence of the quantum description is that some element of stochasticity is introduced by measurement .
in other words there will , in general , be a _ probability distribution _ over the outcomes in any game played using quantum objects .
as we have discussed , a game played with quantum mechanical objects is nothing more than state preparation followed by measurement in which the state that is prepared depends upon the players consideration of the possible measurement outcomes , the input state and their available operations on that input state .
this leads to a certain fluidity in the description of a quantum game similar to that when using the heisenberg or schrdinger pictures .
let us consider a quantum system that is used to play a game in which the input state is @xmath23 , the sets of operations available to the players are given by @xmath46 where we assume , for the moment that @xmath47 = 0~~\forall j , k$ ] .
the output state is @xmath48 , the measurement is characterized by the hermitian operator @xmath17 which has eigenstates given by @xmath49 .
now let us suppose that we change the input state , keeping all other parameters fixed .
if the new input state is described by some transformation of the initial state so that @xmath50 then , in general , a new output state @xmath51 will be produced .
if we denote the original game as game 1 and the game with the transformed input as game 2 then the output states in the games are given by @xmath52 where @xmath53 and@xmath54 describe the operator selections of the players .
we can see that game 2 can be thought of as being played with the input state @xmath23 where player _ a _ has a different set of operators to select from given by @xmath55 we shall call this equivalent game , game 3 .
alternatively we could view this as a game played with an input @xmath23 and the operator sets @xmath56 in which the state output by the players is then transformed by application of @xmath57 .
if we call this new game , game 4 , then we can see that games 2 , 3 and 4 are all equivalent and they all result in a measurement of @xmath17 on the state @xmath51 . we shall call a game for which the output state is written in the form
@xmath58 an mw - type game .
a game for which the output state can be written in the form @xmath59 we shall call an ewl - type game .
these are the forms of game considered in the previous important work on quantum games [ 4,8 ] where @xmath57 is an entanglement operator producing bell basis states from the computational basis . because the calculations the players do are based on the expected outcomes ,
the mw - type game can be thought of , somewhat informally , as an ewl - type game in which the measurement operator is transformed ( and vice versa ) . in order to make a direct comparison between the two forms of games
, however , we should consider the same input state and the same measurement , with the same preference relations over the eigenstates of the measurement operator .
an mw - type game produces an output state of the form @xmath60 and an ewl - type game produces an output state of the form @xmath61 .
if we write @xmath62 then the mw - type game is nothing more than the general form of quantum game we have introduced .
the ewl - type is , in a sense , more ` artificial ' because a subsequent rotation is imposed upon the output state .
if we use a fairly obvious notation for the operator sets so that , for example , @xmath63 then the relation between the forms of the mw- and ewl - type games can be seen in the following table for the operator sets of the players where the games in each column are equivalent .
we have written the ewl - type game in this table from two perspectives that each ` eliminate ' the rotation of the final state .
[ c]ccc & mw - type & ewl - type + input state : @xmath64 & @xmath65 & @xmath66 + input state : @xmath23 & @xmath67 & @xmath68 whether we view the operator sets as fixed or the input states as fixed it is clear that the mw- and ewl - type games will in general lead to different output states .
the two ewl - type games in the table are equivalent to an ewl - type game with input state @xmath69 , and operator sets @xmath32 and @xmath33 in which a rotation is performed on the resultant state .
the rotation clearly alters the distribution of the eigenstates of the measurement operator in the superposition .
of course both players select their operation from their available set in the knowledge that this final rotation will be performed .
alternatively we may view the ewl - type game , as in the table , as one in which no rotation is performed but the players have a transformed set of operators to select from . as we shall see
, it is an interesting question whether rational players would prefer to play a game of the form @xmath70 @xmath71 or @xmath72 for a given @xmath57 .
the next important element that a quantum mechanical description introduces is the notion that operations on the input state do not necessarily commute .
thus , in general , it matters whether player _ a _ or player _ b _ performs their operation first . if , as above , we let the set of operations available to player _
_ ) be denoted by @xmath73 where @xmath31 and @xmath32 is not necessarily equal to @xmath33 then a quantum game is _ non - commuting _ if there is at least one element @xmath74 such that @xmath75 \neq0 $ ] for at least one @xmath76 where @xmath77 . let us suppose that the players have access to a discrete set of operations such that the cardinality of @xmath32 is _ p _ and the cardinality of @xmath33 is _
q. _ if player _ a _ makes the first move then the possible output states are described by a @xmath78 matrix with elements given by @xmath79 .
if player _
b _ makes the first move then the possible output states are described by a @xmath80 matrix with elements given by @xmath81 . thus if @xmath82 we have that , in general , @xmath83 .
thus , in a non - commuting game a player s choice of operation may be determined by whether he operates on the input state before or after the other player . of course , depending on the elements of the sets @xmath30 , it is possible that a different equilibrium state is reached in the two situations .
indeed , the choice of operation for each rational player will in general be different depending on which player makes the first move .
this immediately raises the intriguing question of what happens if the players do nt know which of their operations is performed by the physical device first .
as we have seen , there are different equivalent perspectives for an mw - type game .
we shall label these as mw@xmath84 so that we have * mw@xmath85 : input state @xmath86 , operator sets @xmath87 , no final state rotation * mw@xmath88 : input state @xmath23 , operator sets @xmath89 , no final state rotation * mw@xmath90 : input state @xmath23 , operator sets @xmath91 , final state rotation by @xmath57 each of these perspectives leads to the production of the same output state and therefore the expected outcomes for the players are unchanged in each of these perspectives .
we shall suppose that mw@xmath85 is a commuting game so that @xmath92 = 0~~\forall j , k$ ] .
clearly mw@xmath88 is a non - commuting game , in general , and it is only equivalent if player _ a _ plays first . the mw - type game with input @xmath23 , operator sets @xmath93 in which player _ b _ plays first , is also equivalent to mw@xmath85 when mw@xmath85 is a commuting game .
accordingly we denote these equivalent games as mw@xmath94 and mw@xmath95 where the additional subscript denotes which player plays first .
if we denote the elements of the operator sets in mw@xmath90 as @xmath96 and @xmath97 then the commutation relation between them is given by @xmath98 = t^{-1}\left [ \hat{\alpha}_{k},\hat{\beta}_{j}\right ] t\ ] ] so that if mw@xmath85 is a commuting game mw@xmath90 is also a commuting game .
the different equivalent perspectives in ewl - type games are given by * ewl@xmath99 : input state @xmath86 , operator sets @xmath100 , no final state rotation , player _ a _ plays first * ewl@xmath101 : input state @xmath86 , operator sets @xmath102 , no final state rotation , player _ b _ plays first * ewl@xmath88 : input state @xmath23 , operator sets @xmath91 , no final state rotation , commuting game if @xmath103 = 0~~\forall j , k$ ] . *
ewl@xmath104 : input state @xmath23 , operator sets @xmath105 , no final state rotation , player _ a _ plays first * ewl@xmath106 : input state @xmath23 , operator sets @xmath107 , no final state rotation , player _ b _ plays first * ewl@xmath108 : input state @xmath23 , operator sets @xmath89 , final state rotation by @xmath109 , player _ a _ plays first * ewl@xmath110 : input state @xmath23 , operator sets @xmath111 , final state rotation by @xmath109 , player _ b _ plays first all these are different but equivalent perspectives leading to the production of the same state upon which a measurement of @xmath17 is performed , provided that we have @xmath92 = 0~~\forall j , k$ ] . as an example of the drastic difference that the order of play can make for a non - commuting game consider the perspectives ewl@xmath104 and ewl@xmath106 .
if we now swap the order of play in each of these perspectives we obtain the commuting game with input @xmath23 and operator sets @xmath32 and @xmath33 . in game theory
an equilibrium state is reached when the two players select a strategy they would not change irrespective of the other player s choice .
such a state is a nash equilibrium .
we can see from the general model of a playable game that if such an equilibrium exists then it leads to a particular output state .
in quantum games , therefore , the existence of an equilibrium implies that a particular output state is also produced ( or possibly one of a family of physically equivalent output states that yield the same distribution over the measurement outcomes ) .
if we label the eigenstates of the measurement operator as @xmath112 then the output state can be expanded in terms of these eigenstates @xmath113 and the result of the measurement is the eigenstate @xmath114 with probability @xmath115
. in general the measurement of the output state will lead to a distribution of outcomes and consequently the players must seek to optimise their _ expected _ outcomes . let us suppose that player _ a _ performs his operation first .
a _ will choose an @xmath35 so that his expected outcome is optimised whatever _ b _ subsequently does .
likewise , _
b _ will choose a @xmath37 so that his expected outcome is optimised whatever _ a _ has chosen .
let us consider a measurement @xmath17 with _ n _ eigenstates given by the list @xmath116 which for convenience we can write as the list of numbers @xmath117 .
the preference relation of player _ a _ is simply a permutation of this list such that the first state in the list is his most preferred , the second state his second most preferred , and so on , and similarly for player _
b_. if we let @xmath118 be functions that take the integers from 1 to _
n _ as input and output the number from @xmath119 that represents the _ _
k__@xmath120 most preferred state of player _ a_(_b _ )
then we can write the preference relations for the players as an ordered _ n_-tuple of the integers from 1 to _ n_. @xmath121 in order to calculate an expected _ outcome _ we must assign some numerical value , or weight , to each of the eigenstates that encapsulates the notion of preference . in crude terms this numerical value can be thought of as the payoff for each state . for simplicity
we attach the numerical weight _
n _ to the most preferred state , and the weight 1 to the least preferred state , for each player .
thus if the measurement result yields _ a _ s most preferred state then player _ a _ would receive _ n _ cupcakes , for example . the expected payoffs for the players , denoted by @xmath122 , if the operations @xmath35 and @xmath37 are chosen , are then given by @xmath123 these quantities define a matrix of expected outcomes for each player , the elements denoting each choice of operation @xmath124 . we have assumed that player _ a _ plays first and the knowledge of the order of the moves can make a difference to the strategy selection . in a non - commuting game
the order of play must be specified . for a commuting game
in which there is no specification of the order of play then if _ a _ and _ b _ can find operators @xmath125 such that @xmath126 then an equilibrium output state is obtained given by @xmath127 .
the players thus each maximise their expected outcomes over the choices of the other player . for a non - commuting game in which player _
a _ plays first , the strategy selection is influenced by the order of play .
a _ must select his strategy knowing that _ b _ has the advantage of playing second .
a _ examines the matrix of expected outcomes for player _ b _ and looks at the choices of _ b _ that maximise _ b _ s expected outcome for every choice of player _ a _ s operator .
this gives _ a _ a list of operator pairs . from these
_ a _ now selects the one that maximises his expected outcome .
when playing a game with quantum objects the players have some preference over the measurement outcomes which are the eigenstates of the measurement operator .
we have written these above simply as an ordered list of the labels for the eigenstates .
more explicitly we can write these preferences as @xmath128 any output state from the game , before measurement , can be described by @xmath129 let us consider the subsets @xmath130 of @xmath20 given by the states @xmath131 where the probability amplitudes respect the preferences so that @xmath132 the subsets @xmath130 define regions of the full hilbert space of the outputs .
these subsets are not subspaces and the elements of @xmath133 are not , in general , orthogonal to those of @xmath134 .
each player would act in such a way as to try to direct the output state into their respective regions of hilbert space .
the final output state will , therefore , be some state that would be ` as equidistant ' from their respective regions that the players can achieve with their given operator sets .
the resultant state , if this can be achieved , will be the nash equilibrium position for the quantum game .
some ramifactions of this approach are discussed in [ 21,22 ] .
let us suppose , as above , that the players have access to a discrete set of operations such that the cardinality of @xmath32 is _ p _ and the cardinality of @xmath33 is _
q. _ if player _ a _ makes the first move then the possible output states are described by an @xmath78 matrix with elements given by @xmath135 .
the players each have a preference relation over the eigenstates @xmath39 of the measurement operator @xmath17 . if _
n _ is the cardinality of the set of eigenstates then , in general , @xmath136 .
there will often be a greater number of possible output states that can be produced by the players than the number of possible measurement outcomes .
indeed , as we shall see , the possible ouput states can form a continuum whereas the measurement outcomes consist of a small discrete set of possibilities .
as we have discussed in the previous section , each player can determine a matrix of expected outcomes where the elements @xmath122 give the expected outcome for player @xmath137 for the possible output state @xmath138 .
clearly , this leads to an ordered list of expected outcomes in which each player ranks each possible output state according to its expected outcome for him .
thus the preference relation over the eigenstates of the measurement operator for player @xmath137 _ induces _ a preference relation over the possible output states .
let us suppose that the measurement operator for some 2-player quantum game is rotated by the action of an operator @xmath57 such that the new meaurement operator is @xmath139 .
the eigenstates of @xmath140 are @xmath141 .
if the players maintain the same preference relation over the new eigenstates so that @xmath142 and @xmath143 then , in order for the players to be playing the same game with respect to the new measurement basis the various components have to be transformed according to @xmath144 with these transformations the expected outcomes for player _ a _ with respect to the new measurement basis are given by @xmath145 and similarly for player _
b_. thus the expected outcomes remain unaltered , provided that both the initial state and the operator choices of the players are transformed .
we define an invertible game as one in which one , or both , players can invert the operations performed by the other player . for example , let us suppose that the set of operations available to _ a _ is given by @xmath146 then the game is invertible if _ b _ has a set of available operations that includes the inverses of these elements , that is , the set of operations available to _ b _ is given by @xmath147 .
there is nothing intrinsically quantum mechanical about such games for we could construct a playable classical game in which the operations of one , or both , players could be inverted by the other .
it is clear that an equilibrium state , in the sense that an equilibrium state is the optimal choice for both players , does not exist for an invertible game . in order to see this we shall consider a game
in which _ a _ moves first . if we assume rational players then
_ a _ s preferred choice of move given the constraints , if it exists , is computable by both _ a _ and _ b_. therefore _ b _ simply needs to follow _ a _
s move with @xmath148 where @xmath149 operating on @xmath23 maximises _ b _ s expected outcome .
of course , this implies that there is no such rational choice for _ a_. in fact , _ a _
s best strategy in this situation is to select his move with some distribution .
one immediate consequence of this is that if we allow players to access every possible operation on the input state allowed by quantum mechanics then there can not be an equilibrium output state . in order to reach some kind of equilibrium we must restrict the available strategies in some way .
this , of course , restricts the possible output states .
this consideration becomes crucial when we consider our simple , but important , example of a quantum game . if we suppose that @xmath150 and that the elements of these sets form a group then the resulting game will be invertible . for games which satisfy this condition
there can be no equilibrium state .
we define a factorable game as one in which there is no element of entanglement . in other words ,
the input states are tensor product states and the available operations only lead to output states that are also tensor product states .
factorable games can still be non - commuting and invertible .
a sequential game is one in which the players make a sequence of moves .
so , for example , player _ a _ goes first , followed by player _
b_. player _ a _ then makes his second move , followed by player _
b _ , and so on . if we let @xmath151 be the _ _ i__@xmath120 operation of player _ b _ and @xmath152 be the _ _ i__@xmath120 operation of player _ b _ then the output state after _ m
_ moves is given by @xmath153 after each move we can think of the resultant state as being equivalent to some new input state , thus a sequential game is a sequence of single - move games in which the output state from the previous game becomes the input state to the next . if we let @xmath154 be the state after the _
_ k__@xmath120 move of the players then any sequential game of _ m _ moves can be thought of as a single - move game in which the input state is given by @xmath155 with @xmath156 alternatively we can factor the sequence of operations into two new operators @xmath157 and @xmath158 where , for example , @xmath159 thus the sequential game of _ m _ moves is equivalent to a single - move game in which the operators @xmath157 and @xmath158 are included in _ a _ and _ b _ s set of available operators , respectively . note that there is no unique way to perform this factorization .
these considerations also show us that any single - move game can be represented by some equivalent sequential game of _ m _ moves .
of course there is no unique _ m_-move representation .
in fact there are an infinite number of _
m_-move representations of any single - move game . by equivalent we mean that the same output states are produced and the players receive the same expected outcomes . in game theory terms
the players are in effect playing a different game . constructing the equivalent representations of a single - move game or an _
m_-move game is not , in general , a trivial exercise .
if a single - move game has an equilibrium then all equivalent sequential
_ m_-move games also reach an equilibrium . alternatively if we construct an equivalent sequential _ m_-move game that reaches equilibrium
then the single - move game also reaches equilibrium . indeed , if it can be shown that one member of the class of equivalent _
m_-move games reaches equilibrium then all games in the class also reach equilibrium , as does the equivalent single - move game .
another way to think of the production of an equilibrium state is as the limit of a sequential game .
let us suppose that two players play a sequential game with a no limit on the number of moves they can make .
if they both reach a point such that they do not wish to make another move after a finite number of moves then the resultant output state is an equilibrium state .
conversely , if no such limit exists then there can not be an equilibrium .
if we consider an invertible sequential game with no fixed number of moves then we can see that the players would never reach a point where they would wish to stop .
if we extend a single - move game to a sequential game with no fixed number of moves then the resultant game may not have an equilibrium state even if the single - move game reaches equilibrium .
this is because the subsequent moves potentially give the players access to a larger set of operations than those for the single - move game . in order to see this
let us consider a sequential game with just 2 moves allowed .
the output state is given by @xmath160 thus , in effect , _ a _ chooses the operator @xmath161 which may not be included in his set of available operations for his first move .
we could also view this as a single - move game in which _ a _ plays @xmath162 and _ b _ plays @xmath163 thus if the original single - move game has initial state @xmath164 and sets @xmath165 and @xmath166 then the sequential game with 2 moves is equivalent to the single - move game with initial state @xmath164 and sets @xmath167 and @xmath168 .
it is also equivalent to the single - move game with initial state @xmath15 and sets @xmath169 and @xmath170 .
the 2-move sequential game also gives us a way of characterizing equilibrium for a single - move game .
if the players in a 2-move game would choose the identity operation from their set of available operators for their second move , this implies that their first move was the best they could make and they can not improve upon it by playing another .
thus equilibrium is reached for the single - move game if the second move of a 2-move sequential game is the identity operation , for both players .
equilibrium for an _
m_-move game can also be defined in a similar fashion ; if the @xmath171 move is the identity element for both players , then the game has reached equilibrium .
we have seen one way in which considerations of non - commutativity impact upon quantum games .
let us now look at another way in which this can arise within the context of a quantum game .
consider the situation shown in figure 3 where we have 2 quantum games . in both games the same input state @xmath15
is used and the players have access to the same operations @xmath172 and @xmath173 .
the games differ , therefore , only in the measurement of the output state . in game 1
the measurement performed is @xmath17 and in game 2 the measurement is @xmath174 . adopting a slightly different notation to that used previously ,
the eigenvalues and eigenstates of these operators are given by @xmath175 and we assume that @xmath176 \neq0 $ ] .
the set of possible output states in both games is the same .
the preference relations in game 1 are given by @xmath177 let us consider the unitary transformation @xmath57 which takes the state @xmath178 into the state @xmath179 .
this can be written in the form @xmath180 and we suppose that the preference relations for game 2 are given by this transformation applied to the preferences of game 1 so that @xmath181 it is clear that the players need to choose different strategies for each game , in general .
this can be most clearly seen if we assume that @xmath41 and @xmath174 are maximally conjugate .
let us further assume that the operator sets for the players are such that they produce eigenstates of @xmath17 in game 1 . in game 2 ,
these same operations lead to states that have equal probability amplitudes , up to a phase factor , in the eigenstates of @xmath174 .
thus in game 2 with these sets of operators there is no preferred strategy since all strategies lead to the same expected outcome ( this outcome is @xmath182 if we adopt the weighting convention for the outcomes of the previous sections ) .
it is an interesting question as to whether the uncertainty principle itself can be cast in the form of a game . in this viewpoint
we would treat the different measurements as competing in some sense . of course we would have to consider the more general form of a quantum game in which the players were allowed to perform a measurement . as we have seen in the above example
, localizing the output state to a particular eigenstate in game 1 produces a corresponding uncertainty in the measurement of game 2 for that output state .
this will lead to a corresponding spread in the outcomes for game 2 when the output state is localized on an eigenstate of game 1 .
intuitively we would expect that as we reduced the variance in the outcomes for game 1 we would increase those of game 2 and vice versa .
let us suppose that the output state for game 1 is the state @xmath16 the expected outcomes for player _ a _ in games 1 and 2 with this output state are @xmath183 and we have labelled the numerical weights attached to the outcomes as @xmath184 with @xmath185 . if we define the hermitian operators @xmath186 and @xmath187 by @xmath188 then the product of the variances of the outcomes in the two games satisfies the uncertainty relation @xmath189 \left\vert \psi \right\rangle \right\vert ^{2}%\ ] ] of course the function @xmath190 is defined by a look - up table being a function that has eigenvalues that are the weights arranged according to the preference relations .
the solution to the time - independent schrdinger equation can be expressed by @xmath191 where we have set @xmath192 .
the time - evolution operator @xmath193 is unitary .
it is clear that this operator can be factored into the product of two unitary operators @xmath194 and @xmath195 so that @xmath196 .
there is obviously no unique way to perform this factorisation .
the evolution of the initial state @xmath197 is fixed by the hamiltonian and yields the output state @xmath198 .
this output state can be thought of as the equilibrium state of _ some _ quantum game in which @xmath194 and @xmath195 represent the optimum strategy choices for the two players .
thus we can think of the time evolution of a quantum system as being the result of some game between two players .
constructing such a game and the associated preference relations and sets @xmath32 and @xmath33 is not , however , a trivial task .
there are , in general , an infinite number of such games for any quantum evolution that takes @xmath199 to @xmath198 .
it is clear from our previous discussion that if either of the players has access to the full set of quantum mechanically allowed operations on a given physical system used in a playable game then an equilibrium state can not be reached .
if we are to associate a playable game with the evolution of some state , then the operations available to the players are restricted by the form of the hamiltonian .
in other words the hamiltonian restricts the set of output states .
note that whilst we can always associate some game with the time evolution of a physical system , it may not always be possible to construct a physically meaningful hamiltonian that will yield the equilibrium state of a given two - player game .
however , if we can do this , then the solution of the time - independent schrdinger equation yields the equilibrium state for that game .
in what follows we shall consider a simple quantum system which can be used to play a variety of playable quantum games . in this system
we imagine that two spin-1/2 particles are prepared in some state and input into some device which allows the players to each perform a unitary operation on these particles ( or a sequence of such operations ) .
the spin-1/2 particles are then transmitted to some measurement device .
the result of this measurement determines the outcomes for the players .
we shall label the particles with _ a _ and _ b _ although this is not to be understood that particle _ a _ is only operated upon by player _ a _ necessarily , and similarly for particle _ b_. in the spin-_z _ basis where the spin - up and spin - down states are labelled with 1 and 0 , respectively , the general state of the two particles can be written as @xmath200 where we have written , for example , @xmath201 .
we shall drop the labels _ z _ , _ a _ and _ b _ for convenience , except where ambiguity might arise .
the particular quantum game that is played with this physical system is determined by the input state of the two spin-1/2 particles , @xmath23 , the set of operations @xmath30 available to player _ a_(_b _ ) , the hermitian operator characterizing the measurement , @xmath17 , and the preference relations of player _ a_(_b _ ) , @xmath19 , over the eigenstates of the measurement operator @xmath17 .
let us suppose that the initial state of our particles is given by @xmath202 and that the measurement is simply the independent determination of the spin in the _
z_-direction of both particles so that @xmath203 .
the possible results of the measurement are thus given by the eigenstates of @xmath17 which are given by @xmath204 each player will have some ( different ) preference relation over these states .
we now suppose that the device that implements the game will only allow the players to flip the spin of one particle and we further suppose that player _ a_(_b _ ) is restricted to operations only on particle _ a_(_b _ ) . one possible representation of the sets of operations available to the players is therefore given by @xmath205 where @xmath206 is the identity operator and @xmath207 is the spin operator in the _
x_-direction for particle _ a_(_b _ ) .
the preference relations for the players can be written as an ordered list of the numbers 1,2,3 , and 4 in an obvious shorthand notation .
for example , the preference relations @xmath208 are those for the iconic game of prisoner s dilemma .
the preference relations alone do not completely specify a game , however , and the strategy set for the players must also be taken into account . in classical games where there are two choices in the strategy
sets @xmath209 so that there are 4 possible outcomes , there are 432 possible pairs of preference relations in which the first members of the lists differs .
the quantum system we have described , with the strategy set limited to a spin flip or identity operation , can implement all 432 of these classical games as there is a direct one - to - one correspondence between the various elements of the classical and quantum games , that is @xmath210 the classical games can be implemented by giving each player a coin prepared in a known state .
the quantum version , with its restricted set of operators for the players , uses ` quantum coins ' and can be thought of as simply an ` expensive ' implementation of the corresponding game played with classical coins . even though there is a direct correspondence , and in fact the players would just be playing the same game albeit with more complicated objects , it is still correct to describe the quantum implementation as a quantum game because it utilises quantum mechanical objects , operations and measurements . in other words
the physical implementation is the primary determinant of whether a game is described as classical or quantum .
of course in this case where we have restricted the available operations to flip or do nt flip , the results obtained by playing the quantum game are identical with those obtained by playing the classical implementation
. it should be obvious that the games played with just two classical coins form a subset of the possible games that can be played with the quantum implementation when we ease the restriction on the set of available operations .
it is tempting to describe this procedure as ` quantizing a game ' but this terminology is fraught with difficulties .
it would certainly be correct to say we have quantized the physical system that is used to implement a game in going from classical to quantum coins .
however , a game is also characterized by a set of available strategies and by enlarging the strategy set in a quantum mechanical description , for example , we are , in effect , playing a different game in classical terms .
that being said , it would be legitimate to describe a game played with quantum coins with the preference relations of prisoner s dilemma over the 4 outcomes described by the chosen measurement operator as a single quantum game with many different playable versions according to how the available operations and input states are restricted .
we shall examine in what sense it is possible to talk of ` quantizing a game ' after we have looked at some more games that can be played with our quantum implementation . in the previous quantum game we only allowed the players operations that resulted in the production of eigenstates of the measurement operator .
in general this will not be the case and the resultant output state from the device that implements the players choices will not be an eigenstate of the measurement operator
. the action of measurement will therefore , in general , lead to some distribution over the eigenstates resulting in a distribution over the outcomes for the players .
of course , distributions over the parameters of a classical game are an integral part of game theory . in terms of our _ playable _ classical game
there are at least 3 ways that a distribution over the outcomes can emerge .
we could arrange our classical system such that there is some distribution over the initial state of the game .
if the game were to be implemented using coins , for example , then there would be some distribution over the initial state of the coins and the players would select their subsequent strategy accordingly .
alternatively , the initial state of the system could be determined , but the players could select their strategies according to some distribution .
this kind of game is usually termed a _ mixed _ game , or one in which the players adopt a mixed strategy .
a game in which the players select their strategies deterministically is usually termed a _ pure _ game , or one in which the players adopt pure strategies .
the third possibility is that the outcomes for the players are determined according to some distribution .
each of these possibilities will lead to a distribution over the outcomes for the players and they must consider their _ expected _ outcome from the game when determining their choice of strategy or distribution over those strategies
. in a playable quantum game each of these elements could also be present . however , quantum mechanics , through measurement , forces a distribution over the outcomes , in general , even when there is no other element of stochasticity introduced in the game .
a distribution over the outcomes will occur in a quantum game whenever the output state from the players is not an eigenstate of the measurement operator .
in general the output state of the players will be of the form @xmath211 , as above , where the probability of obtaining a particular result upon measurement is given by the square modulus of the amplitudes .
thus , in general , each possible output state from the players will lead to a different distribution of outcomes in a quantum game .
there is an expected outcome for each possible output state .
this is quite different from the situation in a classical mixed game in which the players select their strategies with some distribution .
we could model the quantum game with a classical game if , in the classical game , the outcomes were computed by taking the measurement as input to some function that generates a different distribution of outcomes for each possible output of the players . in the quantum case the distribution over the outcomes for each possible state arises from the physical properties of quantum measurement . in the classical case these distributions arise as the result of a _
computation_. so , provided that the function that generates the different distributions in the classical case is computable , we can always model a quantum game with some appropriate classical game .
we shall consider a simple example of this shortly .
there are at least 3 ways in which we can extend the simple quantum game of the previous section so that the resultant output state is not an eigenstate of the measurement operator .
let us briefly examine each of these possibilities . as in the previous section we consider a quantum game
in which the measurement is simply the independent determination of the spin in the _ z_-direction of both particles
so that @xmath203 .
the device that implements the game will only allow the players to flip the spin of one particle in the _ z_-direction and we further suppose that player _ a_(_b _ ) is restricted to operations only on particle _ a_(_b _ ) .
now , however , we consider the input states to have undergone some rotation before they are acted upon by the players .
the input state is now of the form @xmath212 where the rotated state of particle a is given by @xmath213 with a similar expression for particle b. the input state is an eigenstate of the spin operator @xmath214 where we have written the spin raising and lowering operators in the _ z_-direction as @xmath215 .
if we write @xmath216 and @xmath217 , then , in terms of the eigenstates @xmath43 of @xmath17 , the four possible output states are given by @xmath218 and we have taken the indices for the output state such that 1 is the identity and 2 is the spin - flip operation so that @xmath219 is the output state when player _ a _ chooses to flip and player _ b _ does not , for example . note that because of the special structure of this game in which the operations available to the players merely transform between eigenstates of the measurement operator , there are only 4 distinct amplitudes that occur in these output states .
let us label the probabilities @xmath220 as follows : @xmath221 , and @xmath222 .
the expected outcomes for player _ a _ with the preference relations of prisoner s dilemma , @xmath223 , are given by @xmath224 player _ b _ has the preference relation @xmath225 for prisoner s dilemma .
the expected outcomes for player _ b _ for each of the possible ouput states are therefore given by @xmath226 this game can be implemented entirely classically as follows .
we give each of the players a coin prepared in the state t. the players are allowed to not flip , _
f _ , or flip , _ f
_ , their respective coins .
when the measurement device receives the two coins in the state ( h , t ) , for example , it will _ compute _ an outcome for player _ a _ of 4 with probability @xmath227 , an outcome of 3 with probability @xmath228 , an outcome of 2 with probability @xmath229 and an outcome of 1 with probability @xmath230 .
the outcomes for player _ b _ are computed to be 4 with probability @xmath230 , an outcome of 3 with probability @xmath228 , an outcome of 2 with probability @xmath229 and an outcome of 1 with probability @xmath227 . in the classical version , therefore , the probabilities induced by measurement in the quantum game are replaced by a computation .
if the probabilities induced by the measurement in the quantum game are computable then any quantum game in which the players have a finite set of strategies can be replaced by an equivalent appropriate classical game played with coins in which the outcomes are determined by assignment of the computed probabilities .
we now consider that the input state remains unchanged , being the product of the spin - down states of the two particles in the _ z_-direction as in section 3.1 .
the measurement operator remains unchanged and is @xmath231 as before .
however , we now rotate the available operations of the players so that their spin flip is performed along a different axis
. this spin flip operation will have the effect of rotating the input state of the particles so that the output state will no longer be an eigenstate of the measurement .
the flip operation for player _ a _ is along the direction defined by the spin operator @xmath232 , and that of player _ b _ along the direction defined by @xmath233 .
it is easy to see that this generates different probabilities than when we rotate the input state .
consider the choice where both players do not flip . the resultant output state @xmath234 is simply the original input state and this is an eigenstate of the measurement operator leading to a particular outcome with probability 1
. the other possible output states are superpositions dependent upon the angle of rotation . in general , the probabilities occuring in the determination of the outcomes are functions of the angles @xmath235 . the overall quantum game with this set of operators for the players can , again , be modelled by an entirely classical game played with just two coins for the players .
in this case the input state remains as before being @xmath236 , the spin - flip allowed to the players is in the _ z_-direction , but the measurement is rotated so that it becomes @xmath237 .
it is clear that this is entirely physically equivalent to the game played with an input state rotated through the angles @xmath235 considered in section 3.2.1 .
the same probabilities for the outcomes will be obtained .
a potentially more interesting case occurs when the input state is entangled .
such states can not be represented by any collection of classical objects , such as coins .
let us look at the question of whether , nevertheless , the resulting quantum game can still be simulated with classical coins .
the most ` non - classical ' states of 2 spin-1/2 particles are the maximally entangled states .
these states give maximal violations of bell s inequality .
let us consider an input state of the form @xmath238 with the players allowed to spin - flip in the _ z_-direction and the measurement given by @xmath239 .
the 4 maximally - entangled bell states that form a basis for the particles are , of course , eigenstates of @xmath17 .
the 4 possible output states the players can produce are given by @xmath240 we see that there are only 2 distinct output states that can be produced . in other words , the measurement @xmath241 is not sufficient to properly distinguish between the moves of the players .
this is because the measurement can not distinguish between the 4 bell states , which give a degeneracy .
if we only allow the players to act on their ` own ' particle with a unitary transformation the output states will always be maximally entangled .
this is because independent unitary operations on the particles can not reduce the degree of entanglement . in order to properly distinguish between the output states we need to perform a bell - type measurement in which joint properties of the output states are measured .
let us , once more , consider the preference relations of prisoner s dilemma so that @xmath223 and @xmath225 .
we let the actual outcomes , that is , the numerical value of the rewards be the numbers _ a _ , _ b _ , _ c _ and _ d _ where @xmath242 . in terms of these numbers
the rewards for the players are given by @xmath243 the expected outcomes for the players are therefore given by @xmath244 in some sense the game of prisoner s dilemma is a pathological example because of the symmetries involved . if the rewards are such that @xmath245 ( as is the case with the assignment of numerical rewards we have previously used ) then the expected outcomes are all equal and thus there is no rational reason for the players to select one move above another . if , on the other hand , the rewards are such that @xmath246 then rational players would clearly prefer either of the output states @xmath234 or @xmath247 .
there is no way of ensuring this in the absence of communication between the players .
similar remarks apply if we have the condition @xmath248 but now the players would prefer either of the output states @xmath249 or @xmath250 .
we see exactly the same inability to select an advantageous strategy if we play the game of quantum chicken with the same entangled input state and the same strategy sets for the players
. the closest classical input to the entangled quantum state would be to arrange the game so that player _ a _ would receive a coin in the state h or t uniformly at random but with the coin for player _ b _ being prepared in the opposite state .
this does not reproduce the same expected outcomes , but does lead to a situation in which the selection of a strategy can not improve the results for either player , just as in the quantum case .
the expected outcomes can not be reproduced in a mixed classical game either .
it is only by adapting the function that determines the outcomes according to the probabilities from the quantum version that we can simulate this game with classical coins .
the function for simulating qpd@xmath251 with the above strategy sets and entangled input can be represented by the following rules @xmath252 where the input state to the game is ( t , t ) and @xmath253 is the probability that player _ a _ receives the reward @xmath254 , and so on .
as we have seen , a 2-player game is described by a set of objects , commonly called strategies , from which a tuple is chosen .
these tuples become the input to some function that determines the ` rewards ' for the players .
armed with knowledge of this function the players select their element of the strategy tuple in such a way as to attempt to optimise their reward , taking into account the fact that the other player is making exactly the same deliberation .
physics deals with objects that can be measured and manipulated in the ` real ' world . accordingly ,
if physics has anything at all to say about game theory , and vice versa , we must consider what it actually means to play a game in terms of actions and measurements on real physical objects .
in other words we require our games to be physically realizable , that is , _ playable _ , if we are to elucidate any potential impact that the underlying physics might have on game theory .
alternatively we might also look for new insights into a physical theory by application of the techniques of game theory .
again , however , we must look to games that are physically realizable if we are to determine whether game theory can shed any new light on a physical theory . in the quantum theory of light it is an important consideration to determine whether any predicted effects are truly quantum in nature .
that is , we try to determine whether the predicted behaviour occurs only as a result of the quantization of the electromagnetic field , or whether treating the field classically is sufficient to explain any prediction .
it would be a natural question , therefore , to ask whether there are any ` non - classical ' effects that can be obtained by playing a game with quantum objects . if the properties of a quantum game can be described by _
some _ game implemented entirely with classical objects then in what sense could the quantum version be said to be truly dependent on quantum effects ?
we would argue that the answer to this question is _ not at all_. the classical version may not look much like the quantum game , but if _ some _ equivalent classical implementation can be found then it seems clear to us that there are no genuinely quantum - mechanical effects occurring in the quantum game [ 16 ] .
let us consider a general 2-player quantum game in which the input state is @xmath23 and the players select an operation from finite sets @xmath32 and @xmath33 with @xmath255 and @xmath256 .
there are @xmath257 possible output states given by @xmath258 where we have assumed that player _ a _ moves first and that the eigenstates of the measurement operator are given by @xmath43 .
the expected outcomes for the players are given , as before , as @xmath259 the players use their calculation of the expected outcomes to determine their selection of appropriate operation .
the probabilities @xmath260 can be calculated by the players if they have full knowledge of the game parameters .
now let us consider a classical game implemented entirely by classical coins .
player _ a_(_b _ ) is given @xmath261 coins where for convenience we shall take @xmath262.the players can flip any number of coins or choose not to flip . in effect
the players are selecting a codeword that represents the choice of an element from a set of cardinality @xmath263 . recall that the outcomes for the players in a classical game are determined by the computation of some function that assigns rewards according to the input tuple that represents the selection of the players strategies . thus if this function is chosen so that the rewards are assigned probabilities according to those determined by the calculated probabilities of a quantum game we can use the classical game to simulate the quantum version .
it is clear that , provided the probabilities of a given quantum game can be _ calculated _ , we can simulate that quantum game by some classical game played entirely with classical coins .
thus , in the sense we have argued for above , there are no results obtainable by playing a game using quantum - mechanical objects that can not be simulated by playing _ some _ game with classical coins . in this sense , therefore , we do not expect to see any genuine non - classical results emerging from the process of considering games played with quantum objects , as far as the calculation of the strategies and the determination of the expected outcomes considered here .
it should be noted , however , that if the measurement is more sophisticated and the outcomes the players receive are calculated by considering the correlations in an ensemble of games then we would not expect to always be able to simulate this with classical coins when the quantum systems used to play the games violate a bell inequality [ 9 - 11 ] .
there is an assumption here that the computation performed after the measurement can be achieved classically . in order to see quantum effects , that computation would have to depend intrinsically on quantum processes . in other words , the computation that determined the output
would itself have to be quantum - mechanical in nature , and not reproducible by any classical computing device , in order for us to see any genuine non - classical effect in a quantum game in the sense we have described above . with these remarks in mind
we return to the question of what it means to ` quantize ' a game . from the perspective of physics quantization
really only makes sense when we consider objects and entities that have some physical existence and that existence is characterised by physical parameters .
the classical variables are replaced by operators and the state of the system is described by a wavefunction , or more abstractly a vector in hilbert space .
thus it really only makes physical sense to talk of quantizing a game when that game is described by physical objects .
it is legitimate to talk of quantizing the physical system that is used to implement a game , but is this strictly equivalent to quantizing a game ? in mathematical terms
a game can be abstractly described as a mapping of some tuple onto a tuple of outcomes ( one for each player ) .
the domain of the function that performs this mapping is the product of the strategy sets .
game theory is all about what the players would choose from these strategy sets given their knowledge of the potential outcomes and the potential actions of the other player .
if we extend the domain by allowing an enlarged set of moves for the players ( and , of course , extend the function to accomodate the enlarged domain ) then we are not really playing the same game in classical terms , even if we we keep the same range for the mapping .
the original , smaller , game may be in some sense contained within this larger game , or we might consider the enlarged game to be an extension of the original game , but in neither case could we consider that we are playing the same game .
so , if we ` quantize ' a game and allow an extended set of operations from those of the classical implementation we have quantized we are not really comparing like with like .
one way to approach the quantization of a game is to consider a classical game and to extend this into the quantum domain . in this way we guarantee
that , if the extension is correctly performed , the classical game sits within the quantum game and can be obtained in the appropriate classical limit .
the advantage of this is that it maintains the link to the classical game in a clear and compelling way [ 7 ] .
this aproach is particularly useful if we are interested in the question of how quantum mechanics may change the properties of a particular classical game , if at all .
however , we must be careful in our comparison . as discussed above ,
if , in performing the extension , we allow the players a greater range of ` moves ' , we are not really playing an equivalent game even if the classical game is properly contained within the extension in the appropriate limit .
an alternative viewpoint , and the one we believe to be more fruitful , is to consider the properties of playing a game with quantum objects , whether or not any classical game of interest is properly contained within it in the appropriate limit .
we could describe this viewpoint as ` gaming the quantum ' .
a quantum game can be considered to be complete , with respect to a particular measurement , if we consider the domain of the function that performs the mapping to our output states to be the set of all possible states of the physical system , together with the set of all possible unitary operations allowed by quantum mechanics on those states .
the range of this function is clearly the set of all possible states of the physical system . a particular game , with respect to a measurement ,
is then specified by the preference relations on the eigenstates of the measurement operator .
it is important to note that a quantum game is specified with respect to a measurement on a particular physical system .
if we consider a 4-state quantum system , for example , then there is a critical difference in whether these 4 states arise from a single quantum system , or whether they arise from the product space of two 2-state systems . in the latter case
we need to consider entanglement between the two physical systems , whereas in the former there is no issue of entanglement .
if we consider a 4-state quantum system which arises from the joint space of two 2-state systems , and we suppose the preference relations of prisoner s dilemma then the complete quantum game with respect to some measurement contains all possible quantum variants of prisoner s dilemma that can be played with that system . accordingly , we would describe such a game as quantum prisoner s dilemma ( qpd@xmath264 ) with respect to the measurement @xmath41 .
if we restrict the domain of the system by restricting the available operations of the players then we can play an infinite number of variants of qpd@xmath264 .
not all of these variants will contain the classical game of prisoner s dilemma in the appropriate limit .
all of these variants , however , are contained within the _ single _ quantum game of qpd@xmath264 .
the single complete quantum game qpd@xmath264 , therefore , describes a family of games characterized by the measurement and the preferences over the outcomes of the measurement .
the different members of this family can be generated by suitable restrictions of the available operations of the players . in the next section
we examine some of the features of playing the quantum game qpd@xmath264 .
the game of quantum prisoner s dilemma with respect to a measurement @xmath265 is played with two 2-state particles , conveniently represented as spin-1/2 particles . in the complete game
we allow the players any unitary operation consistent with the laws of quantum mechanics .
thus , the players are allowed to perform any operation in the joint state space @xmath266 of the two particles .
as we have noted , the complete game with pure strategies can not produce an equilibrium state if the players have perfect information .
if we assume that there exists a preferred pure strategy for player _ a _ then this is determined by some computation of player _
a_. if player _ a _ can perform this computation then so can player _ b _ and , accordingly , player _ b _ can simply apply the inverse operation which contradicts our assumption of the existence of a preferred strategy . in order for qpd@xmath264 to produce some output state that is of interest in game - theoretic terms , therefore
, the possible output states must be restricted .
this means that the operator sets available to the players are such that @xmath267 .
the set of possible output states can be partitioned into 3 disjoint sets as follows @xmath268 : the set of all output states that can be written as a tensor product state of the two particles @xmath269 @xmath270 : the set of all output states that are maximally entangled @xmath271 : the set of all output states that have a partial degree of entanglement the set of all possible output states is simply the union of these 3 disjoint sets @xmath272 . a necessary , but not sufficient , condition for the quantum game to ` contain ' the classical game in the appropriate limit is that the available operations must be capable of producing output states @xmath27 such that @xmath273 . if we suppose that our players could choose to play 4 different versions @xmath274 of our game such that @xmath275 : produces output states @xmath276 @xmath277 : produces output states @xmath278 @xmath279 : produces output states @xmath280 @xmath281 : produces output states @xmath282 then it is an interesting question as to whether rational players would choose to play game 1,2,3 , or 4 .
the games @xmath274 are produced by different restrictions on the operator sets of the players but they are all instances of the complete quantum game of qpd@xmath264 . of course , the partitioning of the output space is not unique and there are other partitions we could envisage .
the one we have adopted , however , does allow us to place the previous work on qpd@xmath264 within the framework of playable quantum games that we have developed so that we can see the connection between them and how they arise naturally from the single quantum game of qpd@xmath283 with the appropriate restrictions .
we have already touched upon a version of qpd@xmath264 when we input two maximally entangled particles in section 3.2.4 . in that case
we only allowed the players the two unitary operations of flip or do nt flip in the _ z_-direction.on one spin each . in the game envisioned by mw [ 8 ] the players are allowed to rotate the state of their respective particles in an arbitrary direction .
as we have already noted , independent unitary transformations can not change the degree of entanglement and the ouput state in this game is always maximally entangled .
thus the mw game is an instance of a game of type @xmath281 .
furthermore , the ability to generate any arbitrary rotation ensures that the set of output states from this game is just @xmath270 , that is , @xmath284 . in order to compare the various games we can play with qpd@xmath264 we shall keep the input state fixed as @xmath285 and the measurement as @xmath286 . by varying the operator sets available to the players we can restrict the possible output states
let us assume that player _ a _ moves first .
the game with a maximally entangled state as input is therefore described by the preference relations on the eigenstates of the measurement together with the operator sets @xmath287 where @xmath57 is an entanglement operator that generates the maximally entangled state @xmath288 .
if the operator sets are restricted so that @xmath289 acts only on the space @xmath290 and @xmath291 acts only on the space @xmath292 then the ouput states are always maximally entangled .
the output states will also be maximally entangled if we enlarge the sets @xmath289 and @xmath291 to include operations that can act on both spaces independently , that is , without causing the particles to interact .
this is a game of the form mw@xmath94 discussed in section 2 .
if player _
b _ makes the first move , then the game is of the form mw@xmath95 with the operator sets @xmath293 the two games mw@xmath94 and mw@xmath95 are equivalent forms of the same game and they both lead to the production of the same output state and expected outcomes for the players .
if we do not allow the players operations @xmath289 and @xmath291 that change the degree of entanglement then it is clear that the resultant quantum game , a version of qpd@xmath264 with restricted operator sets , is not a proper quantization of the classical game [ 7 ] .
as we have noted above , however , this game _ can _ be simulated by playing a game with classical coins if the outcomes are assigned probabilities in accordance with the calculated quantum mechanical distributions over the eigenstates , even though the resultant distributions can not be reproduced by any mixed strategy in the classical game with the standard assignment of outcomes .
of course , if we allow the entanglement operator to depend on some parameter so that varying this parameter changes the degree of entanglement produced then the classical game can be obtained in the appropriate limit .
entanglement often leads to a richer behaviour for a quantized system that can not be reproduced by the equivalent classical system . a playable game , however , is not just characterized by a physical system but also by some computation on the behaviours of that physical system .
it is this computation that determines the outcomes for the players for a particular output state .
the players have to perform this computation in order to select their best move .
even if we keep the computation fixed for both the classical and quantum systems , comparing the two in a game - theoretic sense is problematical because the purpose of game theory is to determine what rational players would choose from a set of available moves , or strategies .
if we allow a larger set of available moves in the quantum case , then it is not clear in what sense we are playing an equivalent game .
so is there any sense in which we may observe truly quantum effects when playing a game with quantum objects ? as we have seen , the expected outcomes in a quantum game can be reproduced by an appropriate game played with classical coins .
the choice of strategies is effectively the same in the quantum game and its classical simulation . in general , in order to see quantum effects in an entangled system we must look to an examination of the joint properties of the constituent sub - systems . in order to see a violation of the bell inequality with 2 spin-1/2 particles , for example
, we must examine the correlations between 3 different measurements .
so if we want to see non - classical effects in quantum games it is in the statistical correlations between the outcomes that we might look first [ 9 - 11 ] .
we would expect there to be a similar inequality between the correlations in the outcomes of a game that is violated when playing with quantum objects .
an examination of this question will be published elsewhere . one way to achieve these kinds of quantum games with our 2 spin-1/2 particles
is simply to input a state @xmath294 to the game of the previous section , where @xmath295 is a parameter - dependent entangling operator and we choose @xmath296 such that the input state is only partially entangled .
provided the players are not allowed operations that can increase the degree of entanglement then the resultant game will always lead to a partially entangled output ( or a completely disentangled output if we allow interactions that can only _ decrease _ the degree of entanglement ) , that is , the output state @xmath297 and this is an instance of a game of type @xmath279 . another way to achieve partially - entangled output states is to play a game of the ewl - form . in this game @xmath57
produces maximally entangled states from the ground state @xmath298 and the players are only allowed operations that act on the sub - spaces independently .
the state is then subject to a disentangling operation @xmath109 before the final measurement is made .
this version of qpd@xmath264 is that considered by ewl in their original work on quantum games [ 4 ] . if we consider the index of correlation , @xmath299 , which is a measure of the information content of the correlation between two systems [ 23,24 ] , then for our particles a and b , @xmath300 where @xmath301 is the entropy of particle a(b ) given by the usual trace over the reduced density operators so that @xmath302tr@xmath303 .
for the operator @xmath57 , which produces bell basis states from the computational basis , @xmath299 is a maximum and equal to @xmath304 .
if the players have access only to unitary operations @xmath305 that act independently on the sub - spaces @xmath290 and @xmath292 then the entropies @xmath301 are invariant under transformations of the form @xmath306 . thus , the allowed operations of the players can not change the information content of the correlation and the state remains maximally entangled after the operations of the players .
the subsequent disentangling operation can , therefore , only reduce the entanglement of the particles .
if the players have access to operations in their available sets which , when taken together , switch between eigenstates of the bell basis generated by @xmath57 then the disentanglement is perfect and will result in the output of a tensor product state .
thus the the output state from this game satisfies @xmath307 and this is an instance of a game of type @xmath277 .
as we saw in section 2 , games of the ewl type can be cast in one of several different equivalent forms .
here we are fixing the input state to be @xmath308 with the final measurement being @xmath286 and there being no ` imposed ' final state transformation so that the output state is a result entirely of the actions of the players .
there are 3 equivalent forms of qpd@xmath264 that satisfy these conditions such that the game is of the type @xmath277 : ewl@xmath88 : input state @xmath23 , operator sets @xmath91 , no final state transformation , commuting game if @xmath103 = 0~~\forall j , k$ ] .
ewl@xmath104 : input state @xmath23 , operator sets @xmath309 , no final state transformation , player _ a _ plays first ewl@xmath106 : input state @xmath23 , operator sets @xmath310 , no final state transformation , player _ b _ plays first the symmetric form , ewl@xmath88 , is the most appealing and in the version of qpd@xmath264 considered originally by ewl the available operator sets for the players are the set of rotations on a single particle so that player _ a _ operates on particle a and player _ b _ operates on particle b. this game is then commuting since , clearly , @xmath311 = 0~~\forall j , k$ ] strictly speaking , of course , the operator choices for the players in the ewl game form a continuum . playing the game with the full set of available operators amounts , therefore , to being able to select an element from an uncountably infinite set .
let us consider our players sitting down to play a game of qpd@xmath283 with input state and final measurement as above and in which the physical device that implements the game asks the players to select between the following options before actually playing the game : * option 1 * : the players can choose an arbitrary rotation on a single particle such player _
_ ) operates only on particle a(b ) .
this leads to a game of the type @xmath275 and produces output states @xmath312 .
* option 2 * : the players can choose an operator from the sets @xmath313 in which the sets @xmath32 and @xmath33 are those available in option 1 .
this leads to a game of the form ewl@xmath88 , that is , of type @xmath277 which produces output states @xmath314 .
* option 4 * : the players can choose an operator from the sets @xmath315 and @xmath33 in which the sets @xmath32 and @xmath33 are those available in option 1 .
this leads to a game of the form mw@xmath94 , that is , of type @xmath281 which produces output states @xmath316 .
this is equivalent to the game mw@xmath95 and so this is not listed as a separate option . * option * _ c _ : the players can select any operator from the complete set of unitary operations on two spin-1/2 particles .
this is the complete game of qpd@xmath264 and we label this as @xmath317 . in this game
we have @xmath318 . which of these options , or games , would rational players choose to play ? clearly ,
_ c _ _ would be discarded in the pure strategy case as there is no computation that will enable the player making the first move to choose a preferred strategy . in a sense here
we have a family of games in which the choice of which game to play is itself part of the strategy selection of the players .
the approach to quantum games we have presented here focuses on the notion of _ playability _ , emphasizing that if we actually want to play a game in some reality the constituent elements that implement the game are described by physical objects and processes . by abstracting out the essential physical elements of a playable game in this fashion we are able to develop what we believe to be the correct perspective for drawing a correspondence between playing games with classical and quantum objects . from this perspective , therefore , we are approaching the notion of a quantum game not by trying to quantize a classical game , but by considering the implications of what it means to play a game with quantum objects and processes ; in essence we might say we are ` gaming the quantum ' rather than ` quantizing a game ' .
with this viewpoint we can see that a quantum system generates a family of playable games for a given set of preference relations over the measurement outcomes .
the different members of this family are generated by the different subsets of the operations of the players .
thus , qpd@xmath264 represents a family of playable games in which the players have the preference relations of prisoner s dilemma over the eigenstates of @xmath17 .
different members of this family can be played by restricting the operations available to the players .
the quantum games considered by eisert , wilkens and lewenstein [ 4 ] , and marinatto and weber [ 8 ] are members of the family qpd@xmath264 . by shifting the perspective to that of gaming the quantum we are less interested in questions of whether any particular classical game is contained within the quantum version as a proper quantization in the appropriate limit , although it is important to be mindful of such questions if we want to draw any correct comparison between the quantum and classical versions of any particular game [ 7 ] .
it is the approach of gaming the quantum that we believe to be potentially more fruitful in addressing the kind of questions first raised by meyer ; will quantum games allow us to develop new techniques for quantum information processing ?
thus , if we are able to cast quantum control problems , or quantum algorithms , or quantum circuit design in the form of a game , can we relate the optimal outcome of the resultant game to an algorithm , or circuit , that is in some sense optimal ?
this is the direction we have taken elsewhere , where the geometric picture of considering the output state to be the result of some competition between optimal strategy choices has been found to be useful [ 21,22 ] .
in the perspective we have developed the quantization of the physical implementation of an instance of a game becomes clear . to what extent this results in something we might call a quantized game , however , is less clear .
as we have seen , if we are only interested in the expected outcomes and how they impact upon the strategy choices of the players , then any playable quantum game can be reproduced by _
some _ classical game played entirely with classical coins , provided we can compute the probabilities for the outcomes for any given output state . 1 .
k. binmore , _ playing for real : a text on game theory _ , oup usa , ( 2007 ) 2 .
wheeler and w.h .
zurek ( eds . ) , _ quantum theory and measurement _ , princeton university press , ( 1983 ) 3 .
d. meyer , _ quantum strategies , _ phys .
lett . , * 82 * , 1052 - 1055 , ( 1999 ) 4 .
j. eisert , m. wilkens and m. lewenstein , _ quantum games and quantum strategies _ , phys .
lett . , * 83 * , 3077 - 3080 , ( 1999 ) 5 .
d. deutsch , _ quantum theory , the church - turing principle and the universal quantum computer _ , proc .
soc . a , mathematical and physical sciences * 400 * ( 1818 ) , 97117 , ( 1985 ) 6 .
p. w. shor , _ polynomial - time algorithms for prime factorization and discrete logarithms on a quantum computer _ , siam j. comput . ,
* 26 * ( 5 ) , 14841509 , ( 1997 ) 7 .
s. a. bleiler , _ a formalism for quantum games i - quantizing mixtures _ , portland state university , preprint , available at http://arxiv.org/abs/0808.1389 , ( 2008 ) 8 .
l. marinatto and t. weber , _ a quantum approach to static games of complete information _ , phys .
lett . a , * 272 * , 291 - 303 , ( 2000 ) 9 .
a. iqbal and d. abbott , _ constructing quantum games from a system of bell s inequalities _ , phys .
a , * 374*/31 - 32 , 3155 - 3163 , ( 2010 ) 10 .
a. iqbal _ _ , playing games with epr - type experiments _
_ , j. phys .
a , * 38*/43 , 9551 - 9564 , ( 2005 ) 11 .
a. iqbal and s. weigert , _ quantum correlation games _ , j. phys .
a , * 37 * , 5873 - 5885 , ( 2004 ) 12 .
r. cleve , p. hoyer , b. toner and j. watrous , _ consequences and limits of nonlocal strategies _ , preprint , available at http://arxiv.org/abs/quant-ph/0404076v2 13 .
j. shimamura , .
k. zdemir , f. morikoshi and n. imoto , _ entangled states that can not reproduce original classical games in their quantum version _
lett a , * 328 * , 20 - 25 , ( 2004 ) 14
. g. brassard , r. cleve , and a. tapp . _ cost of exactly simulating quantum entanglement with classical communication _ , phys
lett . , * 83*(9):18741877 , ( 1999 ) 15 .
r. renner and s. wolf , _ towards characterizing the non - locality of entangled quantum states _ ,
preprint , available at http://arxiv.org/abs/quant-ph/0211019 , ( 2002 ) 16 .
s. j. van enk and r. pike , _ classical rules in quantum games _ , phys .
a , * 66 , * 024306 , ( 2002 ) 17 . s. e. landsburg , _ nash equilibria in quantum games _ , proc . am .
math soc . , * 139 * , 4423 - 4434 , ( 2011 ) 18 . g. dahl and s. landsburg , _ quantum strategies in noncooperative games _ , preprint , available at http://www.landsburg.com/dahlcurrent.pdf 19 . j. nash , _ equilibrium points in n - person games _ , proceedings of the national academy of sciences * 36*(1 ) , 48 - 49 , ( 1950 ) 20 .
r. penrose , _ shadows of the mind : a search for the missing science of consciousness _ , vintage , ( 2005 ) 21 .
f. s. khan and s. j. d. phoenix , _ nash equilibrium in quantum superpositions _ , proceedings of spie , * 8057 , * 80570k-1 22
. f. s. khan and s. j. d. phoenix , _ gaming the quantum _ , preprint , available at http://arxiv.org/abs/1202.1142 , ( 2012 ) 23 .
s. m. barnett and s. j. d. phoenix , _ information theory , squeezing and quantum correlations _ ,
a , * 44 * , 535 , ( 1991 ) 24 . s. m. barnett and s. j. d. phoenix , _ information - theoretic limits to quantum cryptography _ , phys .
a , * 48 , * r5 , ( 1993 )
|
we present a perspective on quantum games that focuses on the physical aspects of the quantities that are used to implement a game . if a game is to be played , it has to be played with objects and actions that have some physical existence .
we call such games playable . by focusing on the notion of playability for games we can more clearly see the distinction between classical and quantum games and tackle the thorny issue of what it means to quantize a game .
the approach we take can more properly be thought of as gaming the quantum rather than quantizing a game and we find that in this perspective we can think of a complete quantum game , for a given set of preferences , as representing a single family of quantum games with many different playable versions .
the versions of quantum prisoners dilemma presented in the literature can therefore be thought of specific instances of the single family of quantum prisoner s dilemma with respect to a particular measurement .
the conditions for equilibrium are given for playable quantum games both in terms of expected outcomes and a geometric approach .
we discuss how any quantum game can be simulated with a classical game played with classical coins as far as the strategy selections and expected outcomes are concerned .
|
the random - field ising model ( rfim ) at zero temperature is a prototype of many disordered systems which exhibit hysteretic and jerky behavior when an external parameter ( _ e.g. _ magnetic field , pressure , strain ) is slowly changed@xcite .
this behavior is related to the existence of a corrugated ( free ) energy landscape with many local minima ( or metastable states ) separated by barriers much larger than @xmath0 , therefore preventing relaxation towards equilibrium on experimental time scales . as a consequence ,
the response to an external driving field is a series of discontinuous jumps ( called avalanches ) from one metastable state to another , the number and size of these jumps varying with the amount of disorder . in the 3-dimensional rfim with gaussian random fields two regimes of avalanches
are observed@xcite : at large disorder , there are many microscopic jumps resulting in a smooth magnetization curve macroscopically ; at low disorder , there is a system - spanning avalanche resulting in a macroscopic jump in the magnetization curve .
these two regimes are separated by a critical point at which avalanches of all sizes are observed . as shown recently
, this type of nonequilibrium disorder - induced phase transition may underly the hysteretic behavior of @xmath1he adsorbed in high porosity silica aerogels@xcite . even though the @xmath2 rfim and similar models have been extensively studied
, there is currently no analytical tool to compute the saturation hysteresis loop in finite dimensions ( even approximately ) and no theory to describe the statistical properties of the metastable states , for instance their configurational entropy ( also called ` complexity ' ) as a function of magnetization .
furthermore , the behavior of the correlation functions along the hysteresis loop is unknown , although this is an issue of experimental interest@xcite . of course
, describing analytically such a nonequilibrium situation where the present state of the system depends on its past history is a difficult problem and at first sight there seems to be no other choice than to follow the dynamical evolution for given initial conditions ( _ e.g. _ one of the two saturated states corresponding to infinitely positive or negative magnetic field ) . in this way
, one can treat exactly ` mean - field ' systems , _
i.e. _ fully connected lattices@xcite or lattices with a locally tree - like structure such as the bethe lattice@xcite , and obtain analytical expressions for the saturation hysteresis loop or the avalanche size distribution . to go further and build a field - theoretical description of these phenomena
it then seems necessary to employ the martin - siggia - rose formalism@xcite . in ferromagnetic systems , however , an alternative route is possible thanks to a remarkable property of the saturation hysteresis loop in the large - disorder regime : the loop is the convex hull of the metastable states in the two - dimensional field - magnetization plane@xcite . in other words ,
the complexity is zero outside the loop and positive inside@xcite .
moreover , it exactly vanishes along the loop since there is a single ` extremal ' metastable state at a given field ( at least for a continuous distribution of the random field ) .
determining the hysteresis loop is then tantamount to counting the number of metastable states at fixed field as a function of their magnetization and finding the magnetization at which the complexity vanishes .
the main difficulty is that one must count the _ typical _ ( _ i.e. _ most likely ) number of metastable states and compute the associated _ quenched _ complexity , which of course is a nontrivial task requiring the use of replicas .
nevertheless , at least in principle , one can thereby study hysteresis and avalanche statistics without referring to the dynamical evolution .
in the present paper we test this approach by studying the soft - spin version of the rfim introduced in ref.@xcite , where the spins take on continuous values between @xmath3 and @xmath4 and are confined by a bistable potential with a linear cusp .
this model has the advantage over the standard rfim to show hysteresis for all values of the disorder in mean - field theory and is anyhow expected to behave as the hard - spin model on long length scales .
our basic strategy is to express the complexity as the legendre transform of the ` free - energy ' associated to a certain action in replica space and then to use approximations borrowed from condensed - matter theory to calculate the corresponding correlation ( or green s ) functions and thermodynamic properties .
our focus on the correlation functions is also motivated by analytical and numerical calculations on the bethe lattice which suggest that their spatial structure along the hysteresis loop is simpler than at equilibrium@xcite .
the paper is organized as follows . in section
ii , we define the model and the observables that we want to compute . in section iii
, we introduce the general formalism in the context of a replica method , focusing on the various correlation functions . in section iv , we first consider the random - phase approximation ( rpa ) which is equivalent to mean - field theory and becomes exact in infinite dimension . in section v ,
we go beyond the rpa by introducing a local self - energy approximation ( lsea ) and we compare the predictions for the magnetization , the correlation functions , and the second - moment of the avalanche - size distribution along the hysteresis loop to simulation data .
concluding remarks and directions for future work are provided in section vi .
additional details on the analytical calculations are provided in the appendix .
we consider a collection of @xmath5 soft spins placed on the sites of a @xmath6-dimensional hypercubic lattice and interacting via the hamitonian @xmath7 where @xmath8 is a ferromagnetic interaction that couples nearest - neighbor spins , @xmath9 is an external uniform field , and @xmath10 is a set of quenched random fields drawn independently from a gaussian probability distribution @xmath11 with zero mean and variance @xmath12 .
@xmath13 is a double - well potential for which we choose the same cuspy form as in ref.@xcite @xmath14 ^ 2\ ] ] so that all solutions of @xmath15 , _ i.e. _ all stationary points of the hamiltonian , are local minima .
this greatly facilitates the present study as there is no need to explicitly discard local maxima and saddle - points through the consideration of the hessian of the energy function .
these local minima are the so - called ` metastable ' states in which each spin satisfies @xmath16 where @xmath17 is a neighbor of @xmath18 on the lattice .
magnetization curves for the nonequilibrium zero - temperature random - field soft - spin model on a cubic lattice for @xmath19 , @xmath20 , @xmath21 ( black ) and @xmath22 ( red ) .
the simulation data correspond to a single disorder realization of linear size @xmath23 using an increment in the external field @xmath24.,width=302 ] at zero - temperature , one can define a local relaxation dynamics in which each spin is forced to satisfy eq .
[ eq3 ] as the external field is changed@xcite .
when adiabatically varying @xmath9 from @xmath3 to @xmath4 and backwards , the model exhibits hysteresis and the magnetization typically behaves as shown in fig . 1 .
( the figure displays the results of a single simulation on a cubic lattice with @xmath19 and @xmath20@xcite . )
these values are arbitrarily chosen and will stay fixed in the rest of this work .
the shape of the loop then changes with the coupling @xmath25 : one can see that @xmath21 and @xmath22 correspond to the large- and small - disorder regimes respectively , with a macroscopic jump in the latter case . for a given realization of the disorder , _ i.e. _ a set of random fields @xmath26 , and a given value of the external uniform field @xmath9 ,
each metastable state is characterized at a macroscopic level by its magnetization , its energy , etc . in the present study ,
we only focus on the magnetization since this is sufficient to unambiguously determine the hysteresis loop .
our goal is then to compute properties averaged over all metastable states with a given magnetization @xmath27 per site at a given external magnetic field @xmath9 ( with a flat measure ) .
a central quantity is the quenched complexity @xmath28 , which is defined as @xmath29 where @xmath30 is the number of metastable states with magnetization @xmath27 at the field @xmath9 and the overbar denotes an average over the random - field distribution .
we are also interested in the following two - point correlation ( green s ) functions : \(i ) the spin - spin correlation functions , @xmath31 with the ( disorder ) connected and disconnected components defined as @xmath32 and @xmath33 the brackets denoting an averaged over all metastable states with magnetization @xmath27 at the field @xmath9 , \(ii ) the correlation function involving the spin variable and the random field , @xmath34 \(iii ) the spin - spin correlation function for two copies of the same disordered system coupled to different external fields and with different magnetizations , @xmath35 where the subscrit @xmath36 indicates that @xmath37 are fixed and similarly for the subscript @xmath38 . along the hysteresis loop ,
the complexity is zero ( at least in the large disorder regime where the loop is continuous)@xcite and the connected spin - spin correlation is identically zero , as the fluctuations inside a metastable state vanish at zero temperature . on the other hand , the disconnected spin - spin correlation function and the spin - random - field one should be nontrivial . finally , we are also interested in the distribution of avalanche sizes along the loop ( say , along the ascending branch ) . for a given disorder realization ,
the magnetization curve @xmath39 consists of smooth parts resulting from the harmonic response to the magnetic field and a series of jumps of size @xmath40 occurring at the fields @xmath41 ( these jumps are not visible in fig . 1 due to the scale of the figure ) @xcite .
the magnetization can thus be decomposed as @xmath42 where @xmath43 is the heaviside step function . from this
, one defines the jump ( avalanche ) density @xmath44 so that @xmath45 is the number of avalanches of size between @xmath46 and @xmath47 when the field is increased from @xmath9 to @xmath48 ( note that @xmath46 is measured per site ) . in present work
we will only compute the ` unnormalized ' second moment @xmath49 ( it is ` unnormalized ' because @xmath50 as such is not a probability density and @xmath51 is the total number of avalanches between @xmath9 and @xmath48 ) .
in order to control the local magnetization , we introduce an additional source @xmath52 that is linearly coupled to the spins @xmath53 and we consider the following ( disorder - dependent ) ` partition function ' in an external magnetic field which is momentarily taken as nonuniform , @xmath54 : @xmath55&= \int { \cal d}s\ e^{\mathbf{\hat{h}}.\mathbf{s}}\prod_i\delta\big(\frac{\partial { \cal h}}{\partial s_i}\big)\nonumber\\ & = \int { \cal d}s\ e^{\mathbf{\hat{h}}.\mathbf{s}}\prod_i\delta ( v'(s_i)-j\sum_{j / i}s_j - h_i - h_i ) \end{aligned}\ ] ] where the symbol @xmath56 refers to the integration over all the spin variables , @xmath57 , and no jacobian is needed ( it would merely introduce a constant multiplicative factor ) .
this partition function can be put into a more standard form by replacing the dirac delta function by its fourier representation , @xmath58= \int { \cal d}s { \cal d}\hat{s}\ e^{-s[\mathbf{s},\hat{\mathbf{s}},\mathbf{h}]+\mathbf{\hat{h}}.\mathbf{s}+\mathbf{h}.\hat{\mathbf{s}}}\end{aligned}\ ] ] where the action @xmath46 is defined by @xmath59= \sum_i \hat{s}_i [ v'(s_i ) -j\sum_{j
/ i}s_j - h_i]\ ] ] and @xmath60 are auxiliary ( imaginary ) variables ; for conciseness , the factor @xmath61 associated to the integration of @xmath62 along the imaginary axis is adsorbed into @xmath63 .
this defines a ` free energy ' @xmath64=\ln { \cal z}[\mathbf{h},\mathbf{\hat{h}};\mathbf{h}]$ ] which is a random object whose cumulants give access to full information about the system , including the complexity and the correlation functions . the complexity is obtained from the first cumulant , @xmath65=\overline{w[\mathbf{h } , \mathbf{\hat{h}};\mathbf{h}]},\end{aligned}\ ] ] via a legendre transform , where @xmath66= \frac{\partial w_1[\mathbf{h } , \mathbf{\hat{h}}]}{\partial { \hat h}_i } , \end{aligned}\ ] ] and which for uniform sources takes the form @xmath67 ( here and below we use square brackets @xmath68 $ ] when the arguments are locally varying and parenthesis @xmath69 when they are uniform . ) as discussed in refs.@xcite , the hysteresis loop in the large - disorder regime identifies with the curve @xmath70 in the limit @xmath71 whereas the typical properties of the metastable states are obtained for @xmath72 which corresponds to the maximum of the complexity .
the information about the distribution of avalanche sizes is contained in the higher - order cumulants @xmath73,$ ] @xmath74 $ ] , ... , where @xmath75=\overline{w[\mathbf{h}^1 , \mathbf{\hat{h}}^1;\mathbf{h } ] w[\mathbf{h}^2 , \mathbf{\hat{h}}^2;\mathbf{h}]}-\overline{w[\mathbf{h}^1 , \mathbf{\hat{h}}^1;\mathbf{h } ] } \ \overline{w[\mathbf{h}^2 , \mathbf{\hat{h}}^2;\mathbf{h}]}\end{aligned}\ ] ] etc ... the correlation ( green s ) functions are obtained by derivation of the cumulants with respect to the sources . for instance , the physical two - point spin - spin correlation functions introduced above are given by @xmath76}{\partial \hat{h}_i\partial \hat{h}_j},\end{aligned}\ ] ] @xmath77}{\partial \hat{h}_{i}^1\partial \hat{h}_{j}^2}\vert_{\mathbf{h}^1=\mathbf{h}^2=\mathbf{h},\mathbf{\hat{h}}^1=\mathbf{\hat{h}}^2=\mathbf{\hat{h}}}\ , \end{aligned}\ ] ] where both right - hand sides are evaluated for uniform sources and @xmath78 is considered as a function of @xmath27 and @xmath9 through the legendre transform in eq .
( [ eq_legendre_uniform ] ) .
the spin - random - field correlation function requires a little more thought . by using the property of gaussian distributions
, @xmath79 one finds @xmath80}{\partial \hat{h}_i } h_j } = \delta \ ; \overline{\frac{\partial w[\mathbf{h},\mathbf{\hat{h } } ; \mathbf{h}]}{\partial \hat{h}_i\partial h_j } } = \delta \ ; \hat{g}_{c , ij } \ , \end{aligned}\ ] ] where @xmath81}{\partial \hat{h}_i \partial h_j}.\end{aligned}\ ] ] when all the sources are uniform , this yields @xmath82 where @xmath83 is the fourier transform of @xmath84 . (
in particular , this equation is valid in the limit @xmath71 , _
i.e. _ along the hysteresis loop : surprisingly , it seems that this extension of the ` susceptibily sum - rule ' to the nonequilibrium magnetization curve has not been noticed before . ) to compute the average over disorder , the common procedure is to replicate the system @xmath85 times and take the limit @xmath86 at the end .
( this is in contrast with the martin - siggia - rose formalism in which the partition function is directly averaged over disorder , making the use of replicas unnecessary@xcite ; here indeed , the partition function in eq .
( [ eq4a ] ) is nontrivial so that one must average @xmath87 and not simply @xmath88 . )
if one is interested in computing the cumulants of the random free - energy for generic arguments , one must introduce sources that act separately on each replica@xcite .
after performing the average over the random - field distribution , we obtain a ` replica partition function ' ,
@xmath89= \int \prod_{a=1}^{n } { \cal d}s^a{\cal d}\hat{s}^a\ e^{-s_{rep}[\{\mathbf{s}^a,\hat{\mathbf{s}}^a\}]+\sum_a[\mathbf{\hat{h}}^a.\mathbf{s}^a+\mathbf{h}^a.\hat{\mathbf{s}}^a]}\ ] ] with the replicated action given by @xmath90= \sum_i \sum_{a}\hat{s}_{i}^{a}[v'(s_{i}^{a})-j\sum_{j / i}s_{j}^{a}]-\frac{\delta}{2}\sum_i \sum_{a , b}\hat{s}_{i}^{a}\hat{s}_{i}^{b } \ .\end{aligned}\ ] ] the ` thermodynamic potential ' @xmath91=\ln { \cal z}_{rep}[\{\mathbf{h}^a,\mathbf{\hat{h}}^a\}]$ ] can then be expanded in increasing number of free replica sums@xcite , @xmath92=\sum_{p=1}^{\infty } \frac{1}{p ! } \sum_{a_1,a_2, .. ,a_p}w_p[\mathbf{h}^{a_1 } , \mathbf{\hat{h}}^{a_1};\mathbf{h}^{a_2 } , \mathbf{\hat{h}}^{a_2}; ...
;\mathbf{h}^{a_p } , \mathbf{\hat{h}}^{a_p}]\ ] ] where the @xmath93 s are continuous and symmetric functions of their arguments .
this coincides with the cumulant expansion@xcite .
the thermodynamic potential @xmath91 $ ] generates the ` magnetizations ' @xmath94 and @xmath95 @xmath96 and the correlation ( or green s ) functions , _
e.g. _ at the pair level , @xmath97 where @xmath98 denotes an average over the replicated action . as @xmath91 $ ] above ,
the magnetizations and the correlation functions can be expanded in increasing number of free replica sums : @xmath99=m_{i}^{[0]}[\mathbf{h}^a,\mathbf{\hat{h}}^a]+\sum_{e}m_{i}^{[1]}[\mathbf{h}^a , \mathbf{\hat{h}}^a \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e } ] + \frac{1}{2 } \sum_{e , f } m_i^{[2]}[\mathbf{h}^a ,
\mathbf{\hat{h}}^a \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e};\mathbf{h}^{f } , \mathbf{\hat{h}}^{f } ] + ... , \ ] ] and similarly for @xmath95 , whereas after decomposing the two - point functions as @xmath100 ( where @xmath101 does not contain any explicit kronecker delta ) , one has @xcite @xmath102=g_{c , ij}^{[0]}[\mathbf{h}^a , \mathbf{\hat{h}}^a]+\sum_{e}g_{c , ij}^{[1]}[\mathbf{h}^a , \mathbf{\hat{h}}^a \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e } ] + \frac{1}{2 } \sum_{e , f } g_{c , ij}^{[2]}[\mathbf{h}^a , \mathbf{\hat{h}}^a \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e};\mathbf{h}^{f } , \mathbf{\hat{h}}^{f } ] + ... , \ ] ] @xmath103&=g_{d , ij}^{[0]}[\mathbf{h}^a , \mathbf{\hat{h}}^a;\mathbf{h}^b , \mathbf{\hat{h}}^b ] + \sum_{e}g_{d , ij}^{[1]}[\mathbf{h}^a , \mathbf{\hat{h}}^a;\mathbf{h}^b , \mathbf{\hat{h}}^b \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e } ] \nonumber\\ & + \frac{1}{2 } \sum_{e , f } g_{d , ij}^{[2]}[\mathbf{h}^a , \mathbf{\hat{h}}^a;\mathbf{h}^b , \mathbf{\hat{h}}^b \vert \mathbf{h}^{e } , \mathbf{\hat{h}}^{e};\mathbf{h}^{f } , \mathbf{\hat{h}}^{f } ] + ... , \end{aligned}\ ] ] and similarly for @xmath104 and @xmath105 . in the present approach , however , the central object is not @xmath106 but the ` effective action ' @xmath107 which is the legendre transform of @xmath106 with respect to the two sets of sources @xmath108 and @xmath109 , @xmath110=-w_{rep}[\{\mathbf{h}^a,\mathbf{\hat{h}}^a\}]+\sum_{a}\big ( \mathbf{m}^a.\mathbf{\hat{h}}^a+\hat{\mathbf{m}}^a.\mathbf{h}^a \big ) \ , \end{aligned}\ ] ] so that @xmath111 @xmath107 is the generating functional of the ` direct ' correlation functions [ or one - particle irreducible ( 1pi ) functions , or else proper vertices , in field - theoretic language ] . at the pair level , @xmath112 as above , @xmath107 , @xmath113 , @xmath114 , etc ... can be expanded in increasing number of free replica sums . in the following , all the two - point functions will be put together as components of a matrix with both replica indices and spatial coordinates @xmath115 where @xmath116 , @xmath117 , @xmath118 have for elements @xmath119 , @xmath120 , @xmath121 ; a similar notation is used for the direct correlation functions , all collected in @xmath122 .
the matrix @xmath123 is then just the inverse of @xmath124 , _
@xmath125 the matrix inversion with respect to spatial coordinates is easily realized by fourier transformation when all the sources are taken as uniform .
the inversion with respect to the replica indices can be performed by using the expansion in free replica sums for @xmath126 and @xmath127 ( see eqs .
( [ eq_free_replica_gc],[eq_free_replica_gd ] ) ) and proceeding to a term - by - term identification .
the zeroth - order terms are then given by @xmath128}_c(\mathbf{k } ; m^a,\hat{m}^a ) = \mathbf{\underline{\underline{c}}}^{[0]}_c(\mathbf{k } ; m^a,\hat{m}^a)^{-1}\\ & \mathbf{\underline{\underline{g}}}^{[0]}_d(\mathbf{k } ; m^a,\hat{m}^a ; m^b,\hat{m}^b ) = - \mathbf{\underline{\underline{c}}}^{[0]}_c(\mathbf{k } ; m^a,\hat{m}^a)^{-1}\mathbf{\underline{\underline{c}}}^{[0]}_d(\mathbf{k } ; m^a,\hat{m}^a ; m^b,\hat{m}^b ) \mathbf{\underline{\underline{c}}}^{[0]}_c(\mathbf{k } ; m^b,\hat{m}^b)^{-1 } \ , \end{aligned}\ ] ] where @xmath129}_{c , d}$ ] are @xmath130 matrices containing the components @xmath131 , @xmath132 , @xmath133 as in eq .
( [ eq13a ] ) , and similarly for @xmath134}_{c , d}$ ] .
note that the zeroth - order functions obtained when fixing the replica sources @xmath135 and fixing the replica ` magnetizations ' @xmath136 are generically related through @xmath137}(h^a , \hat{h}^a ) , \hat{m}^{[0]}(h^a , \hat{h}^a))$ ] with @xmath138}$ ] and @xmath139}$ ] being the zeroth - order expressions as in eq .
since the zeroth - order contributions already contain the physics of the problem , we will not consider higher - order terms and will drop the superscript @xmath140 $ ] in the following .
when all replica sources or magnetizations are taken as equal , and provided that this limiting process is regular enough ( this will be discussed in more detail below ) , replica symmetry is recovered and one obtains a set of ` ornstein - zernike ' ( oz ) equations ( in the language of liquid - state theory@xcite ) which takes the following explicit form : @xmath141 and @xmath142\nonumber\\ \hat{g}_d(\mathbf{k})&=-[c_d(\mathbf{k})g_c(\mathbf{k})\hat{g}_c(\mathbf{k})+\hat{c}_d(\mathbf{k})(g_c(\mathbf{k})\hat{\hat{g}}_c(\mathbf{k})+\hat{g}_c(\mathbf{k})^2 ) + \hat{\hat{c}}_d(\mathbf{k})\hat{\hat{g}}_c(\mathbf{k})\hat{c}_c(\mathbf{k})]\nonumber\\ \hat{\hat{g}}_d(\mathbf{k})&=-[\hat{\hat{c}}_d(\mathbf{k})\hat{\hat{g}}_c(\mathbf{k})^2 + 2\hat{c}_d(\mathbf{k})\hat{\hat{g}}_c(\mathbf{k})\hat{g}_c(\mathbf{k})+c_d(\mathbf{k})\hat{g}_c(\mathbf{k})^2 ] \ , \end{aligned}\ ] ] where all functions depend on @xmath27 and @xmath143 . when the replica sources are not equal , the zeroth - order terms have the capability to capture a nonanalytic behavior in the dependence on the replica sources or replica magnetizations .
this important feature is already displayed by the noninteracting system corresponding to @xmath144 ( hereafter called the ` reference ' system ) whose properties are listed in the appendix ( eq . ( [ eqzrep0 ] ) and below with @xmath145 and @xmath146 ) . for instance , the 2-replica function @xmath147 behaves like @xmath148 when @xmath149 . we stress that this is not a spurious behavior due to the presence of a cusp in @xmath13 .
indeed , even if @xmath13 were a smooth double - well potential , it would be necessary to distinguish between the two minima in order to count the metastable states and , whatever the method , this would introduce a nonanalyticity in @xmath64 $ ] and lead to a cusp in the 2-replica function @xmath150 .
this behavior is thus an intrinsic feature of the problem under consideration and is also intimately connected to the magnetization discontinuities along the hysteresis loop , _ i.e. _ to the avalanches .
a similar connection is discussed in ref .
@xcite in the context of random elastic systems where the statistics of static avalanches ( or ` shocks ' ) is studied via the functional renormalization group .
( in this case , however , the cusp only shows up in the course of the renormalization flow whereas it is always present here , even in the large disorder regime . ) to show that the cusp is related to the ( unnormalized ) second moment of the avalanche size distribution , we consider the quantity @xmath151 , \end{aligned}\ ] ] where it is implicit that the limit @xmath152 corresponding to the hysteresis loop ( in the large - disorder regime ) has been taken .
then , using eqs .
( [ magcusp ] ) and ( [ gcusp ] ) , taking the second derivative with respect to @xmath153 and @xmath154 in the limit @xmath155 , and identifying the singular contributions proportional to @xmath156 on both sides of the equation lead to @xmath157 the nonanalyticity in @xmath64 $ ] also implies that the 2-replica correlation function @xmath158 contains a singular contribution proportional to @xmath156 .
this induces singular contributions in the three disconnected direct correlation functions via the oz equations and leads to formally diverging terms proportional to ` @xmath159 ' when the replica fields are equal .
the function @xmath160 , however , has no obvious physical meaning and this problem is in principle harmless , although it may be a source of difficulties in an approximate treatment , as will be discussed below in section v. finally , it is instructive to examine the behavior of the correlation functions in the limit @xmath152 , _
i.e. _ along the two branches of the hysteresis loop in the large - disorder regime . assuming that the general behavior is the same as in the reference system ( which seems quite reasonable , at least in the regime under consideration )
, we find @xmath161 and @xmath162 where the notation indicates that both the regular and the singular contributions of @xmath163 and @xmath164 are of order @xmath165 .
note that we have considered the dependence on the source @xmath78 .
a similar dependence is found on @xmath143 when the latter goes to plus or minus infinity ( recall that we work at the zeroth - order level of the expansion in free replica sums ) .
it is not surprising that @xmath166 vanishes as @xmath152 .
this is due to the already mentioned fact that there is only one metastable state along the loop at a given field so that the statistical fluctuations only come from the quenched disorder and are thus contained in the disconnected functions .
( as a consequence , @xmath167 also vanishes . )
the important feature is that @xmath166 and @xmath167 vanish exponentially fast with @xmath78 .
indeed , this implies that the oz equations for the physically relevant functions @xmath168 and @xmath150 take in this limit the much simpler form : @xmath169 note that @xmath170 does not contain any diverging part when @xmath152 ( or @xmath171 ) , as it must be .
equipped with the formalism of the previous section , we wish to introduce approximations based on assumptions on the structure of the direct correlation functions ( proper vertices ) .
formally , these functions may be written as @xmath172 where @xmath173 is the ` bare ' inverse propagator matrix ( not taking into account the local potential @xmath174 ) , whose components satisfy @xmath175 and @xmath176 is a ` self - energy ' matrix ( the minus sign is chosen so to match the usual definition of this quantity in field theory and many - body physics ) . in the above equations , @xmath177 is @xmath178 if @xmath18 and @xmath17 are nearest neighbors on the lattice , zero otherwise . from now on in this section , except
if explicitly stated , we only consider uniform sources that are equal for all replicas .
our first approximation , akin to the usual random phase approximation ( rpa ) , consists in assuming that the self - energies are identical to those of the noninteracting ( @xmath144 ) reference system . in fourier space , this then leads to @xmath179 where @xmath180 is the connectivity of the hypercubic lattice and @xmath181 its characteristic function .
the direct correlation functions of the reference system are obtained from the green s functions computed in the appendix [ eqs .
( [ eqgc])-([eqhhgc ] ) and ( [ eqdeltacusp])-([gdasymp ] ) with @xmath145 and @xmath146 ] .
the rpa effective action is immediately obtained by integrating the ` susceptibility sum - rule ' @xmath183 , which yields @xmath184 from this expression , one obtains @xmath185 it can be easily checked that @xmath186 is identical to the quenched complexity @xmath187 ( not to be confused with a self - energy ) obtained in the mean - field model of ref .
@xcite ( with @xmath188 replaced by @xmath25 ) .
( on the other hand , the auxiliary field @xmath78 does not coincide with the parameter @xmath189 introduced in this reference . in particular , @xmath78 satisfies the legendre relation @xmath190 , in contrast with @xmath189 . in this respect ,
the rpa is a nicer way of obtaining the mean - field limit . )
magnetization along the ascending branch of the hysteresis loop : comparison of the predictions of the random - phase approximation ( solid lines ) to simulation data ( circles ) for various values of the coupling @xmath25 ( with @xmath19 and @xmath20 ) .
the simulation data are averaged over @xmath191 disorder realizations of linear size @xmath192.,width=340 ] one expects the rpa to be valid when the coupling @xmath25 is sufficiently weak , or , equivalently , when @xmath12 and @xmath193 are sufficiently large ( all ` thermodynamic ' quantities can be expressed in terms of the reduced variables @xmath194 , @xmath195 , and @xmath196 ) .
this is indeed what is observed in fig .
[ fig2 ] where the predictions of the rpa for the magnetization along the ascending branch of the hysteresis loop are compared to numerical simulations ( the descending branch is obtained using the symmetry @xmath197 , @xmath198 ) . the theoretical value is given by @xmath199 where @xmath200 is given by eq .
( [ eqmag ] ) with @xmath145 and @xmath146 , and from the above eq .
( [ eqfield ] ) the `` displaced '' field @xmath201 is solution of the implicit equation @xmath202 .
one can see that the agreement is very good for @xmath203 and deteriorates as @xmath25 increases .
in particular , the rpa overestimates the slope @xmath204 and the magnetization curve already exhibits a reentrant behavior for @xmath205 ( typical of a mean - field theory below the critical point ) whereas the actual system is still in the large - disorder regime .
the spin - spin and spin - random - field correlation functions along the loop for @xmath206 and @xmath207 are shown in figs .
[ fig3 ] and [ fig4 ] .
they are defined by ( see section iii ) @xmath208 and @xmath209 eqs .
( [ rpa1 ] ) and ( [ eqcrpa ] ) then yield @xmath210 @xmath211\end{aligned}\ ] ] where @xmath212 and @xmath213 is the lattice green function ; in addition , @xmath214 ( @xmath215 and the correlation functions of the reference system are functions of @xmath27 or @xmath9 ) .
one can see that the rpa correctly predicts the values of the correlation functions along the hysteresis loop when @xmath25 is small ( we have checked that the agreement with the simulations remains good at the second and third nearest - neighbor distances ) but considerably overestimates these values when @xmath25 increases and the slope @xmath204 becomes large ( for instance around @xmath216 for @xmath21 ) .
this can be traced back to an overestimation of the value of @xmath215 ( note that the rpa susceptibility diverges for @xmath217 ) .
spin - random field correlation function @xmath218 at @xmath219 ( a ) and @xmath220 ( b ) along the ascending branch of the hysteresis loop for @xmath203 ( blue ) and @xmath21 ( red ) .
the predictions of the rpa ( dashed lines ) and the lsea ( solid lines ) are compared to the simulation data ( circles).,width=340 ] same as fig . 3 for the spin - spin correlation function @xmath221.,width=340 ] finally , it is also interesting to examine the rpa correlation functions for @xmath222 finite , which corresponds to metastable states _ inside _ the hysteresis loop in the @xmath223 plane .
for conciseness , we only consider the connected functions , as given by eqs .
[ eqozc ] .
we then write @xmath224 ^ 2\nonumber\\ & = [ c_c^{ref}\hat{\hat{c}}_c^{ref}-(\hat{c}_c^{ref})^2][1-z_1\lambda(\mathbf{k})][1-z_2\lambda(\mathbf{k})]\end{aligned}\ ] ] where @xmath225 which yields @xmath226\nonumber\\ \hat{g}_c^{rpa}(\mathbf{r})&=\hat{g}_c^{ref}\big[\frac{z_1-z}{z_1-z_2}p(\mathbf{r};z_1)-\frac{z_2-z}{z_1-z_2}p(\mathbf{r};z_2)\big]\nonumber\\ \hat{\hat{g}}_c^{rpa}(\mathbf{r})&=\hat{\hat{g}}_c^{ref}\big[\frac{z_1}{z_1-z_2}p(\mathbf{r};z_1)-\frac{z_2}{z_1-z_2}p(\mathbf{r};z_2)\big]\end{aligned}\ ] ] where @xmath227 as defined above .
it turns out that @xmath228 is a positive function of @xmath9 and @xmath78 and @xmath229 is negative , whereas the sign of @xmath230 may change , as shown in fig .
[ fig5 ] for @xmath72 . therefore ,
depending on whether @xmath230 is negative or positive , @xmath231 and @xmath232 are real or complex conjugates , respectively .
however , one can check that the correlation functions are always real , as it must be . on the other hand
their behavior changes with the sign of @xmath230 , as illustrated by the leading asymptotic behavior as @xmath233 .
for instance , using the asymptotic expansion of @xmath234 @xcite , one finds that @xmath235\nonumber\\\end{aligned}\ ] ] where @xmath236 $ ] .
therefore , when @xmath237 and @xmath238 are complex conjugates , the typical ornstein - zernike fall - off @xmath239 is modulated by oscillations at a wavevector @xmath240 where @xmath241 denotes the imaginary part . for
a given value of the magnetic field @xmath9 , both the correlation length and the wavevector @xmath242 depend continuously on @xmath78 ( or , alternatively , on the magnetization @xmath243 of the metastable states ) , as shown in fig .
this defines a ` disorder line ' in the @xmath223 plane where the qualitative behavior of the correlation function changes . in the case presented in fig .
[ fig6 ] ( for a small value of @xmath25 for which the rpa is expected to be valid ) , one finds that @xmath242 is non - zero for @xmath72 , that is for a typical metastable state at the field @xmath9 . on the other hand
, @xmath242 is always zero for @xmath71 , that is along the hysteresis loop .
direct correlation function @xmath244 in the reference system ( @xmath144 ) as a function of @xmath9 for @xmath72.,width=283 ] wave - vector @xmath242 as a function of the magnetization @xmath243 of the metastable states for @xmath203 and @xmath245 .
@xmath242 is non - zero for @xmath246 , the magnetization of the typical states , but zero on the hysteresis loop ( as @xmath71).,width=283 ]
the rpa is exact in the limit of infinite dimension or infinite lattice coordination but , as we have just seen , is not a good approximation for @xmath247 outside the weak - coupling regime . to go beyond this regime
, we shall assume that the fluctuations renormalize the green s functions _ without _ changing the functional form of their spatial dependence .
this amounts to assuming that the corresponding self - energies are purely local .
this assumption is the starting point of an approximate theory similar to the so - called optimized random phase approximation ( orpa ) in liquid - state theory@xcite ( see also@xcite ) , the ` locator ' approximation used in spin and coulomb glasses@xcite , and to the dynamical mean - field theory ( dmft ) for quantum problems@xcite .
this type of approximation is known to be efficient when the physics is dominated by strong short - wavelength fluctuations rather than by long - wavelength fluctuations@xcite .
accordingly , it can not properly describe the vicinity of critical points but may capture for instance hysteresis and metastable effects that are already generated by the local potential in eq .
( [ eq2 ] ) .
note that the local self - energy approximation reduces to the rpa when @xmath248 , but may be expected to provide reasonable results even in @xmath247 . from the point of view of thermodynamic consistency ,
it is useful to formulate the theory at the level of the effective action within the two - particle irreducible ( 2pi ) formalism where @xmath107 is considered as a functional of both the magnetizations @xmath249 and the green s functions@xcite ( this is known as the ` entropy functional ' in classical systems@xcite ) . up to an additive constant
, @xmath107 can be written as @xmath250 = s_{rep}[\{\mathbf{m}^{a},\mathbf{\hat{m}}^{a}\}]+\frac{1}{2 } \mbox{tr}\ln { \bf \underline{\underline{g}}}^{-1}+\frac{1}{2}\mbox{tr}\,{\bf \underline{\underline{c}}}^{(0)}{\bf \underline{\underline{g}}}+\phi_{rep}[\{\mathbf{m}^{a},\mathbf{\hat{m}}^{a}\},\mathbf{\underline{\underline{g}}}],\end{aligned}\ ] ] where @xmath251 is defined in eq .
( [ eq13a ] ) and @xmath252 is the so - called luttinger - ward functional@xcite .
( @xmath252 is also called the 2pi functional as it is in general the sum of all two - particle irreducible diagrams built with the fully dressed green s functions . in the present case ,
the singular nature of the local potential makes the definition of the vertices appearing in the diagrams tricky and eq .
( [ eqfunc ] ) must thus be considered as a nonperturbative definition of @xmath252 . ) in the above expression , the trace involves a sum over both replica and spatial indices .
the crucial point is that the explicit dependence of @xmath107 on the pair interactions is contained in the classical action @xmath253 $ ] and in the bare inverse propagator @xmath173 . on the other hand , for a given on - site potential ,
the functional @xmath252 is universal . by construction ,
the self - energies are obtained as @xmath254 etc ... , and the stationnarity of the functional @xmath107 against the variations of the green s functions provides the schwinger - dyson equations , @xmath255 where one also has from eq .
( [ eq13 ] ) that @xmath256 . at the extremum
, @xmath107 then identifies with the physical ( 1pi ) replicated effective action defined in eq .
( [ eq11 ] ) . at the level of @xmath107 ,
the local approximation consists in replacing @xmath252 by a sum of purely local contributions , @xmath257 .
this readily implies that the self - energies are purely local .
more specifically , restricting ourselves to the translationally invariant situation ( uniform configurations ) , we find in fourier space @xmath258 which expresses in a matrix form the equations for the components @xmath259 in terms of their counterparts in @xmath260 and in the self - energy . according to eq .
( [ eqsd ] ) , the only nonzero components of @xmath260 are @xmath261 and @xmath262
. the self - energies are then functions of @xmath263 and @xmath12 to be determined . as discussed for instance in ref .
@xcite , the simplest strategy for computing the above functions is to define a single - site effective action ( the so - called ` impurity ' model in strongly correlated fermi systems ) that involves the original on - site interaction and arbitrary quadratic terms .
this model is in general exactly solvable and self - consistency equations are then obtained by imposing that the green s functions of the single - site action coincide with the site - diagonal green s functions of the original lattice model .
we thus introduce the single - site action @xmath264= -\frac{1}{2}\sum_{a , b}\left[k^{ab}s^a s^b+\hat{k}^{ab}(s^a \hat{s}^b+\hat{s}^a s^b)+\hat{\hat{k}}^{ab } \hat{s}^a\hat{s}^b\right]+\sum_{a } \hat{s}^{a}v'(s^{a})\end{aligned}\ ] ] and add external sources @xmath265 that fix the same ` magnetizations ' @xmath266 and @xmath267 , respectively , as in the fully interacting system .
the so - called ` weiss fields ' @xmath268 must be chosen so that the green s functions @xmath269 of the effective single - site model coincide with the on - site green s functions @xmath270 of the original model , _
i.e. _ @xmath271 with identical self - energies @xmath272 , @xmath273 , @xmath274@xcite . by definition of the self - energies ,
the direct correlation functions of the effective model are given by @xmath275 as @xmath276 in the single - site model , with the matrix @xmath277 collecting the weiss fields @xmath268 .
subtracting from eq .
( [ eq_sd_fourier ] ) then yields @xmath278 and the consistency requirement between @xmath279 and @xmath280 provides the following equation , @xmath281 ^{-1}\ , , \end{aligned}\ ] ] where @xmath282 is a short - hand notation for @xmath283 .
( [ eq_self - consist_dmft ] ) can be considered as a self - consistent equation for @xmath277 since the direct correlation functions of the single - site effective model are obtained from the action in eq .
( [ sloc ] ) as functions of the replica ` magnetizations ' and of the weiss fields : @xmath284 $ ] .
again , we can use the expansion in number of free replica sums to derive explicit expressions .
we decompose the matrices according to @xmath285 , etc ... , and expand each component as in eqs .
( [ eq_free_replica_gc],[eq_free_replica_gd ] ) ( fixing the magnetizations instead of the sources ) . at zeroth - order ,
one finds from eq .
( [ eq_self - consist_dmft ] ) @xmath286 ^{-1}\end{aligned}\ ] ] and @xmath287 ^{-1 } \left[\mathbf{\underline{\underline{c}}}_{\:d}^{(0)}+ \mathbf{\underline{\underline k}}_{\:d}(m^a,\hat{m}^a;m^b,\hat{m}^b ) + \mathbf{\underline{\underline{c}}}_{\:0,d}(m^a,\hat{m}^a;m^b,\hat{m}^b)\right]\\ & \times \left[\mathbf{\underline{\underline{c}}}_{\:c}^{(0)}(\mathbf k)+
\mathbf{\underline{\underline k}}_{\:c}(m^b,\hat{m}^b ) + \mathbf{\underline{\underline{c}}}_{\:0,c}(m^b,\hat{m}^b)\right ] ^{-1 } \ , \end{split}\ ] ] where @xmath288 and @xmath289 are calculated from the single - site effective action in eq .
( [ sloc ] ) with the weiss fields now functions of the magnetizations through the above implicit equations ; the latter are considered at zeroth - order in the expansion in number of free replica sums , _
i.e. _ @xmath290 and @xmath291 . as the luttinger - ward functional is universal , the expression of @xmath107 is obtained by subtracting from eq .
( [ eqfunc ] ) the formal expression of @xmath292 for the single - site effective model . restricting again the calculation to the translationally invariant situation ( uniform configurations )
, we obtain @xmath293 \nonumber\\ & -\frac{1}{2}\int d{\bf k } \ \left [ \mbox{tr}\ln \big({\bf \underline{\underline{g}}}({\bf k}){\bf \underline{\underline{g}}}_{\:0}^{-1}\big)-\mbox{tr}\,{\bf \underline{\underline{c}}}^{(0)}({\bf k}){\bf \underline{\underline{g}}}({\bf k})+\mbox{tr}\,{\bf \underline{\underline{c}}}_{\:0}^{(0)}{\bf \underline{\underline{g}}}_{\:0}\right ] .\end{aligned}\ ] ] this effective action , when evaluated at the extremum corresponding to the solution of the schwinger - dyson equations , is a function of the replica magnetizations that it can also be expanded in number of free replica sums , @xmath294 the @xmath295th term then providing information on the @xmath295th cumulant ( see above ) .
since the local approximation is performed at the level of the luttinger - ward functional , the theory possesses some form of thermodynamic consistency@xcite .
indeed , one obtains the same value of @xmath107 by using the above equation or by integrating _
once _ a quantity that depends on the green s functions ( _ e.g. _ the ` internal energy ' ) with respect to the model parameters ( _ e.g. _ the coupling constants ) .
for instance , using the fact that the functional is extremal with respect to the variations of the green s functions and considering for simplicity the case where all replica magnetizations are equal , one readily derives the relations @xmath296 \nonumber\\ \frac{\partial \gamma_1/n}{\partial \delta}&=-\frac{1}{2}\hat{\hat{g}}_c({\bf r=0}),\end{aligned}\ ] ] which are exact and are thus preserved by the local approximation .
the extremal property also allows us to simply relate the fields @xmath297 and @xmath298 to the fields @xmath299 and @xmath300 in the single - site effective system ( they all correspond to zeroth - order terms in the expansion in free replica sums ) by only considering the _ explicit _ dependence of @xmath301 and @xmath302 on @xmath27 and @xmath303 .
this yields @xmath304 which leads to @xmath305m+\hat{\hat{k}}_{c}(m,\hat{m } ) \hat{m } \ .\end{aligned}\ ] ] similarly , one finds that @xmath306\hat{m}+k_{c}(m,\hat{m } ) m \ .\end{aligned}\ ] ] one may however notice that the local approximation does not really yield a fully consistent theory .
for instance , the value of @xmath307 obtained from eqs .
( [ eq_sd_fourier],[eq_schwinger - dyson_0],[eq_self - consist_dmft ] ) does not coincide with that obtained from the ` susceptibility sum - rule ' @xmath308 .
this inconsistency is a well - known flaw of this type of approximation and it can be cured by renormalizing the value of the direct correlation functions at nearest - neighbor distance as is done for instance in the so - called self - consistent ornstein - zernike approximation ( scoza)@xcite . this route , however , looks prohibitively difficult in the present case and will not be pursued .
improving the theory by introducing a @xmath309-dependence in the self - energies as is done in the cluster dynamical mean - field theory@xcite can make the theory exact up to order @xmath310 , which is an interesting property , but it does not solve the above inconsistency problem . at this stage , we must point out a serious difficulty occuring in the approximate framework developed above .
we have stressed that , just like the reference system , the exact system is most likely such that the green s function @xmath160 is singular with a diverging term proportional to @xmath159 .
the 2pi functional and the other green s functions being most likely finite , the contribution of @xmath160 must altogether vanish in the expressions of these finite quantities .
one can therefore plainly drop all dependence on @xmath160 in eq .
( [ eqfunc ] ) , which amounts to a perfect cancellation between terms in @xmath311 and in @xmath252 ( there are no @xmath160 contributions in @xmath312 ) .
this of course has consequences on the self - energies .
to avoid inconsistencies , this property must be satisfied , or at least enforced , in any sensible approximation .
as shown in the appendix , it is easily realized that @xmath313 in the single - site model may develop a singular @xmath159 term only if the weiss field @xmath314 is identically zero .
a sensible approximation scheme must therefore be compatible with ( i ) setting @xmath314 to zero , ( ii ) discarding the self - consistent equation on @xmath160 , and ( iii ) dropping all contributions involving @xmath160 in the two - particle irreducible functional of the original model and of the effective single - site model .
it turns out that these requirements are not met by the local self - energy approximation inside the hysteresis loop ( when @xmath78 and @xmath143 are finite and the quenched complexity is strictly positive ) . along the hysteresis loop ( @xmath315 )
, the situation improves and the local self - energy approximation is well behaved , at least at the level of single - replica quantities . awaiting for a proper resolution in the general case ( see the discussion in conclusion )
, we now discuss the behavior on the hysteresis loop .
we consider the solution of the lsea discussed above along the hysteresis loop , when @xmath316 or alternatively when @xmath317 .
our main assumptions , motivated by the expected exact behavior along the hysteresis loop , are that the weiss field @xmath314 and the corresponding direct correlation function @xmath318 in the effective model are identically zero ( see above ) and that the complexity is zero as well .
the former of these two assumptions implies from eq .
( [ eqozc ] ) that @xmath319 and thus @xmath320 in the original model are also zero , as anticipated .
we are then only interested in the correlation functions that are physically observable , namely @xmath321 and @xmath322 . as a result
, we only need to solve the two corresponding self - consistency equations , @xmath323 and @xmath324 , leading to @xmath325 @xmath326[\hat{c}_{0,c}(m^b)+\hat{k}_{c}(m^b)-qj\lambda({\bf k})]}\end{aligned}\ ] ] where @xmath327 , @xmath328 , @xmath329 , and @xmath330 are obtained from the effective single - site action @xmath331= \sum_{a}\left [ - \hat{k}_c^{a}s^a + v'(s^{a})\right ] \hat{s}^a - \frac{1}{2}\sum_{a , b}\hat{\hat{k}}_d^{ab } \hat{s}^a\hat{s}^b,\end{aligned}\ ] ] with the magnetizations @xmath267 fixed by the external sources @xmath332 [ @xmath333 follows from @xmath334 and the weiss fields implicitly determined as functions of the magnetizations , @xmath335 , @xmath336 .
the actual computation of the correlation functions associated to this effective single - site action is performed in the appendix . after introducing @xmath337 @xmath338 , and @xmath339
( as in section iv ) , eqs .
( [ simplifiedoz_g0c ] ) and ( [ simplifiedoz_g0d ] ) can be rewritten as @xmath340 where we have dropped the superscript @xmath36 on the magnetization , and @xmath341[\hat{c}_{0,c}(m^b)+\hat{k}_{c}(m^b)]}\ , \left [ \frac{z(m^a)p(z(m^a))- z(m^b)p(z(m^b))}{z(m^a)-z(m^b)}\right ] .\end{aligned}\ ] ] for equal magnetizations @xmath342 , eq .
( [ gd_2copies ] ) simplifies to @xmath343 ^ 2}\left [ p(z(m))+z(m)p'(z(m))\right ] .\end{aligned}\ ] ] once the coupled equations ( [ hgc ] ) and ( [ gd ] ) are solved for @xmath344 and @xmath345 , we calculate the field @xmath9 along the loop from eq .
( [ eqfields1 ] ) with @xmath346 , @xmath347 m , \end{aligned}\ ] ] and we obtain the physical correlation ( green s ) functions for equal magnetizations from eqs .
( [ rpa1 ] ) as @xmath348 @xmath349 with explicit expressions for @xmath327 and @xmath350 given in the appendix .
magnetization along the ascending branch of the hysteresis loop : comparison of the predictions of the local self - energy approximation ( solid lines ) to the simulation data ( circles ) for various values of the coupling @xmath25 .
the dashed lines for @xmath205 and @xmath351 indicate the macroscopic jump predicted by the theory.,width=340 ] the predictions of eq .
( [ eq_h_hysteresis ] ) for the ascending branch of the hysteresis loop are shown in fig .
[ fig7 ] . comparing with fig .
[ fig2 ] , we see that the improvement over the rpa is quite significant : the agreement with the simulations is now satisfactory up to @xmath205 where a small reentrant behavior is observed in the upper part of the curve .
this behavior , which erroneously indicates that the system has entered the small - disorder regime , is related to the fact that the effective action is obtained via the ` energy ' route which is not consistent with the ` susceptibility ' one , as mentioned above .
this lack to thermodynamic consistency as the coupling increases is illustrated in fig .
[ fig8 ] . of course
, the actual magnetization can not decrease as @xmath9 increases and it must jump at a ` spinodal ' field where the slope @xmath352 diverges for the first time ( as can be seen in fig .
[ fig7 ] for @xmath351 , the theoretical curve may display several spinodal fields ) .
note that due to this inconsistency the condition @xmath353 does not imply @xmath354 and thus not @xmath217 ( from eq .
( [ gsh_1copy ] ) one has @xmath355 $ ] ) .
test of thermodynamic self - consistency in the lsea : comparison of the susceptibility @xmath356 ( solid lines ) with the slope @xmath357 of the magnetization curve ( dashed lines ) along the ascending branch of the hysteresis curve . for @xmath205 the slope diverges at the spinodal field @xmath358 whereas @xmath356 remains finite.,width=283 ] the correlation functions @xmath218 and @xmath221 at @xmath219 and @xmath220 are compared to the simulation data and to the rpa in figs .
[ fig3 ] and [ fig4 ] .
the agreement with the simulations is now excellent for @xmath21 . in particular
, the lsea correctly predicts that the maximum of @xmath359 decreases as @xmath25 increases , contrary to the rpa .
the values at the second and third nearest - neighbor distances@xcite are also shown in figs .
[ fig9 ] and [ fig10 ] . the theoretical predictions for @xmath21 are still fairly good although the values of the functions are slightly overestimated .
more generally , the results displayed in these figures show that the actual dependence of @xmath360 and @xmath221 with distance is well described by rpa - like expressions in the large - disorder regime and not too close to criticality .
the two functions involve a single correlation length @xmath361 [ _ e.g. _ the second - moment correlation length defined by @xmath362 , @xmath363 , and thus related to @xmath215 by @xmath364 and the main effect of disorder fluctuations is to renormalize @xmath361 .
spin - random field correlation function at the second ( a ) and third ( b ) nearest - neighbors along the ascending branch of the hysteresis curve for @xmath203 ( blue ) and @xmath21 ( red ) .
the predictions of the lsea ( solid lines ) are compared to the simulation data ( circles).,width=340 ] same as fig .
[ fig9 ] . for the spin - spin correlation function @xmath221.,width=340 ] from the above correlation functions , we can compute the average energy per spin along the loop , which is the sum of four contributions , @xmath365 with @xmath366\nonumber\\ u_2&=-\overline{s_i}h =- mh\nonumber\\ u_3&=-\overline{s_ih_i}=-g_{sh}(r=0)\nonumber\\ u_4&=\overline{v(s_i)}=\frac{1}{2k}\overline{(h+j\sum_{j / i}s_j+h_i)^2}=\frac{1}{2k}\big[h^2+\delta+2qjmh+2qjg_{sh}(r=1)\nonumber\\ & + qj^2(g_{ss}(r=0)+m^2)+q(q-2)j^2(g_{ss}(r=\sqrt{2})+m^2)+qj^2(g_{ss}(r=2)+m^2)\big ] \ , \end{aligned}\ ] ] where , again , the dependence of @xmath9 and of the green s functions on the magnetization @xmath27 is left implicit . as can be seen in fig .
[ fig11 ] , the average energy per spin is also very well reproduced for @xmath21 and the discrepancies for @xmath205 are limited to the small range of @xmath9 where the reentrant behavior in the magnetization occurs .
energy per spin along the ascending branch of the hysteresis loop .
the predictions of the lsea ( solid lines ) are compared to the simulation data ( circles ) .
, width=340 ] additional information on the system along the hysteresis loop is encoded in the @xmath367-replica spin - spin correlation function for distinct magnetizations , @xmath368 . from the equations above ,
one derives @xmath369 , \end{aligned}\ ] ] which gives back eq .
( [ gss_1copy ] ) when the magnetizations are equal .
as discussed in section iii , this function allows one to compute the unnormalized second moment of the avalanche - size distribution via eq .
( [ s2moment ] ) , provided one extracts the nonanalytic cusp - like dependence of @xmath370 on the difference @xmath371 , and consequently on @xmath372 .
the amplitude of the linear - cusp contribution is then obtained from @xmath373 when @xmath374 .
it is clear from eq .
( [ gss_2copies ] ) that the cusp only comes from @xmath375 , _
i.e. _ from the two - replica correlation function in the single - site effective model defined by the action in eq .
( [ sloc_zeroth_order ] ) .
this function depends on the magnetizations both in an explicit way and through the weiss fields , which we write as @xmath376 .
the contribution of the weiss fields to the cusp can only come from @xmath377 [ with @xmath378 , implying that @xmath379 where the subscript @xmath380 indicates that the derivatives are evaluated for @xmath374 and the derivative with respect to @xmath377 in the right - hand side is taken with @xmath381 , @xmath382 , and @xmath383 fixed .
on the other hand , the self - consistency equation ( [ simplifiedoz_g0d ] ) imposes that the cusp in @xmath375 is directly proportional to the cusp in the weiss field @xmath377 . indeed , replacing @xmath384 by @xmath385 in the right - hand side of eq .
( [ simplifiedoz_g0d ] ) , one finds after simple manipulations @xmath386 \frac{\partial \hat{\hat k}_d(m^a;m^b)}{\partial(m^a - m^b)}\bigg|_{0^+ } \end{aligned}\ ] ] where @xmath27 is the common limit of @xmath267 and @xmath387 .
before we proceed any further , we must however deal with a difficulty that arises in eq .
( [ gd0_cusp ] ) from the behavior of the partial derivative @xmath388 when @xmath374 . at zeroth - order in the expansion in number of free replica sums , it is equivalent to use the replica magnetizations @xmath389 ) or the corresponding sources @xmath390 ) , so that we may as well consider the partial derivative of @xmath391 with respect to @xmath392 .
this quantity is most easily expressed by using the property that @xmath393 is the second derivative of the second cumulant @xmath394 with respect to @xmath395 and @xmath396 . on the other hand , from the definition of the action of the effective model ,
the derivative of @xmath397 with respect to @xmath398 is equal to @xmath399 ( using the symmetry @xmath400 ) .
after permuting the order of the derivatives , one thus has @xmath401 \ , \end{aligned}\ ] ] a quantity that is still defined in the limit @xmath402 since @xmath399 is proportional to @xmath403 [ see eqs .
( [ eqgd0reg ] ) and ( [ eqgd0sing ] ) ] .
the problem is that the green function @xmath313 contains a singular term proportional to @xmath404 , as already mentioned , so that @xmath405 diverges when @xmath406 or , equivalently , when @xmath407 .
this causes the right - hand side of eq .
( [ gd0_cusp ] ) to diverge , whereas eq .
( [ hathatkd0_cusp ] ) does not predict any divergence .
this spurious behavior is intrinsic to the local self - energy approximation but it can be circumvented by simply discarding the singular contribution of @xmath313 and only keeping in eq .
( [ derivgk ] ) the contribution of the regular term .
this is admittedly a drastic regularization procedure which however may be rationalized by recalling that the singular term , which is an exact feature of the green function @xmath163 , should not play any role in an exact treatment@xcite .
then , combining eqs .
( [ gd0_cusp ] ) and ( [ hathatkd0_cusp ] ) and considering the fields instead of the magnetizations lead to @xmath408 \frac{\partial g_{0,d}}{\partial \hat{\hat k}_d(h_0^a;h_0^b)}\bigg|_{0^+}^{reg}\right)^{-1 } \frac{\partial g_{0,d}}{\partial(h_0^a - h_0^b)}\bigg|_{\hat k_c,\hat{\hat k}_d;0^+ } \ . \end{aligned}\ ] ] where @xmath409 is the common limit of @xmath410 and @xmath411 .
it is important to note that the two derivatives of @xmath412 that appear in the right - hand side can now be computed from the effective single - site action in eq .
( [ sloc_zeroth_order ] ) by assuming that the weiss fields are replica - symmetric and independent of the sources @xmath413 and @xmath414 .
this calculation is performed in the appendix . once the amplitude of the cusp in @xmath415 is known , the amplitude of the cusp in @xmath416 , @xmath417 , which is needed for computing the unnormalized second moment of the avalanche distribution via eq .
( [ s2moment ] ) , is obtained through @xmath418 with @xmath9 and @xmath27 related via eq .
( [ eq_h_hysteresis ] ) .
one finally obtains from the fourier transform of eq .
( [ gss_1copy ] ) @xmath419 ^ 2[p(z(m))+z(m)p'(z(m))]}\nonumber\\ & = \frac{\partial g_{0,d}(h_0^a;h_0^b)}{\partial(h_0^a - h_0^b)}\bigg|_{0^+ } \frac{(\partial m/\partial h)/(\partial m/\partial h_0)}{[1-z(m)]^2[p(z(m))+z(m)p'(z(m ) ) ] } \ , \end{aligned}\ ] ] with @xmath420 given by eq .
( [ gd0_cusp_final ] ) .
this equation further simplifies if one replaces @xmath421 by @xmath422 $ ] , thereby neglecting the small thermodynamic inconsistency of the present approach .
( on the other hand , @xmath423 is exactly equal to @xmath424 in the single - site effective model . )
the unnormalized second moment of the avalanche - size distribution along the ascending branch of the hysteresis curve for @xmath203 ( a ) and @xmath21 ( b ) .
the predictions of the lsea ( solid lines ) are compared to the simulation data ( circles ) resulting from an average over @xmath425 disorder realizations of linear size @xmath426 for @xmath203 and @xmath427 for @xmath428 @xcite .
, width=340 ] the theoretical prediction for the unnormalized second moment of the avalanche - size distribution is compared to the simulation data in fig .
[ fig12 ] .
one observes a reasonable agreement , even for @xmath21 , which gives some _ a posteriori _ justification to the regularization procedure that has been used in the above calculation .
in this paper we have proposed a formalism to study the out - of - equilibrium hysteresis behavior of random field systems when submitted to an adiabatically varying external field at zero temperature .
the key ingredients consist in relating the out - of - equilibrium behavior to the statistics of the metastable states , introducing auxiliary variables to handle the latter , and building an approximation scheme based on the structure of the pair correlation functions .
we have applied this program to a soft - spin version of the random field ising model and focused on describing the system along the hysteresis loop , which is the envelope of the metastable states in the magnetization - applied field plane .
the physics of the problem involves ` avalanches ' between metastable states that in turn generates nonanalyticities in the dependence of several correlation functions on their arguments .
we have used an approximation borrowed from condensed - matter theory and most conveniently formulated in a 2pi framework , whose lowest order amounts to neglecting the spatial dependence of the self - energies .
we have derived in this way predictions for the physical , spin - spin and spin - random - field , pair correlation functions , as well as for the second moment of the avalanches , which can be compared to computer simulation data : we find a good agreement between predictions and simulation data above the ( out - of - equilibrium ) critical point , which represents a significant improvement over the mean - field ( random - phase approximation ) results also calculated here .
away from criticality , the correlation functions along the hysteresis loop keep essentially the same spatial structure as in the random - phase approximation with however a strong renormalization of the correlation length due to disorder - induced fluctuations . in spite of the encouraging accuracy of the results
, we have encountered difficulties in extending the basic local self - energy approximation either to the situation ` inside ' the hysteresis loop , for which the complexity associated with the number of metastable states is nonzero , or to the many - replica correlation functions , even along the hysteresis loop .
the presence of avalanches at @xmath2 indeed generates a strong singularity in the form of a dirac delta function in the ( unphysical ) correlation function of the auxiliary variable .
the contribution of the latter must therefore exactly vanish in physical quantities and in the effective action .
such exact cancellations are however hard to implement in an approximate theory . to cure this problem
, one must somewhat relax the self - consistency of the local self - energy approximation to allow for one more constraint enforcing the vanishing of the diverging terms without overconstraining the theory .
in addition , to improve the accuracy of the predictions and , for instance , to provide a good description of the much studied hard - spin ising model in a random field , one will have to go beyond the local approximation for the self - energies , as done in the cluster dynamical mean - field theory@xcite .
work in this direction is in progress .
the vicinity of the out - of - equilibrium critical point on the other hand can not be treated in such cluster extensions of the local approximation and requires a different treatment along the lines of the recently developed nonperturbative functional renormalization group@xcite .
in this appendix we compute the correlation ( green s ) functions of the single - site effective model whose partition function in replica space reads @xmath429+\sum_a[\hat{h}^a s^a+h^a\hat{s}^a]}\ ] ] where the action is given eq .
( [ sloc ] ) ( hereafter , for ease of notation we drop the subscript @xmath430 on all quantities . )
we consider a replica - symmetry ansatz for the weiss fields , @xmath431 with @xmath432 and @xmath433 taken as fixed .
although this ansatz neglects the fact that the weiss fields depend on the magnetizations @xmath434 and @xmath435 through the self - consistency equations ( [ eq_self - consist_cc])-([eq_self - consist_cd ] ) , the results derived in this appendix are nonetheless sufficient to compute all the quantities studied in the present work , including the amplitude of the linear cusp in the two - replica spin - spin correlation function .
the action then becomes @xmath436 -\frac{1}{2}\big[k_{d } u^2 + 2\hat{k}_{d } uv+\hat{\hat{k}}_{d } v^2\big]+\sum_{a } \hat{s}^{a}v'(s^{a})\end{aligned}\ ] ] where @xmath437 and @xmath438 .
the quadratic dependence on @xmath439 and @xmath440 can be eliminated by using a hubbard - stratonovich transformation with two auxililary fields @xmath441 and @xmath442 that play the role of correlated random fields .
this yields @xmath443}\prod_a \int d s^a d\hat{s}^a \ : e^{f_a(s^a,\hat{s}^a;h_1,h_2)}\end{aligned}\ ] ] where @xmath444 + s^a(\hat{h}_a+h_1)+\hat{s}^{a}[h_a+h_2-v'(s^{a } ) ] \end{aligned}\ ] ] and @xmath445 must be a positive quantity . to compute this quantity we first integrate @xmath449 over @xmath450 along the imaginary axis ( taking into account the factor @xmath61 that was adsorbed in @xmath451 ) .
this gives @xmath452 with @xmath453 ^ 2\nonumber\\ & = [ k_{c}\hat{\hat{k}}_{c}-(k-\hat{k}_{c})^2 ] s^2 + 2\big[(\hat{h}+h_1)\hat{\hat{k}}_{c}+(k-\hat{k}_{c})(h+h_2+k\ \mbox{sign}(s))\big]s -(h+h_2+k\ \mbox{sign}(s))^2 \ .\end{aligned}\ ] ] the second integration over @xmath454 from @xmath3 to @xmath430 and from @xmath430 to @xmath4 finally yields @xmath455 with @xmath456\end{aligned}\ ] ] and @xmath457 \ .\end{aligned}\ ] ] like @xmath458 , the quantity @xmath459 must be positive .
the green s functions can be calculated by derivation of @xmath460 with respect to the weiss fields , @xmath461 and the corresponding direct correlation functions are then obtained by inverting the ornstein - zernike equations ( [ eqozc ] ) and ( [ eqozd ] ) .
it is rather obvious that this set of equations leads in general to an analytic behavior of the correlation functions as a function of the source @xmath9 . in particular
, the function @xmath160 does not have a singular term proportional to @xmath159 although this is the expected behavior in the original lattice model , as pointed out in the main text .
one can easily see that the condition for a singular behavior to emerge from eqs .
( [ eqa9])-([eqa11 ] ) is that @xmath462 : the quantity @xmath463 then goes to @xmath464 depending on the sign of @xmath465 and this induces a heaviside step function in eq .
( [ eqa9 ] ) and a dirac delta when differentiating with respect to @xmath9 .
we now focus on the behavior along the hysteresis loop where it is sufficient to only keep the two weiss fields @xmath466 and @xmath467 ( however , the limit @xmath468 which corresponds to the ascending branch will only be taken at the end of the calculation ) .
the starting point is the simpler partition function @xmath469\ } } \end{aligned}\ ] ] which also encompasses the case of the reference system ( @xmath144 ) where @xmath145 and @xmath470 .
a single auxiliary ( random ) field @xmath471 is now required to eliminate the quadratic dependence on @xmath438 .
this leads to @xmath472 where @xmath11 is a gaussian distribution with zero mean and variance @xmath473 ( which plays the role of a ` renormalized ' disorder ) and @xmath474}=\ln \int ds \ : e^{s\hat{h } } \delta[(k-\hat{k}_c)s - h - h - k\:\mbox{sgn}s]\nonumber\\ & = -\ln(k- \hat{k}_c)+\hat{h}\frac{h+h - k}{k- \hat{k}_c}+[\frac{k\hat{h}}{k- \hat{k}_c}+\ln \big(2\cosh \frac{k\hat{h}}{k- \hat{k}_c}\big)]\theta(h+h+k)\nonumber\\ & + [ \frac{k\hat{h}}{k- \hat{k}_c}-\ln \big(2\cosh \frac{k\hat{h}}{k- \hat{k}_c}\big)]\theta(h+h - k ) \ .\end{aligned}\ ] ] as a consequence , we find @xmath475[{\cal p}(h+k)-{\cal p}(h - k)]\end{aligned}\ ] ] where @xmath476 $ ] . from this
, we readily obtain the magnetizations @xmath477[{\cal p}(h+k)-{\cal p}(h - k)]\nonumber\\ \hat{m}(h,\hat{h})&=\frac{\partial w_1}{\partial h}=\frac{\hat{h}}{k- \hat{k}_c}+\frac{2k \hat{h}}{k-\hat{k}_c}p(h+k)+[\ln \big(2\cosh \frac{k\hat{h}}{k- \hat{k}_c}\big)-\frac{k\hat{h}}{k- \hat{k}_c}][p(h+k)-p(h - k ) ] \ , \end{aligned}\ ] ] and the connected ( green s ) correlation functions @xmath478 [ { \cal p}(h+k)- { \cal p}(h - k)]\end{aligned}\ ] ] @xmath479[p(h+k)-p(h - k)]\end{aligned}\ ] ] @xmath480[\frac{h+k}{\hat{\hat{k}}_d } p(h+k)-\frac{h - k}{\hat{\hat{k}}_d } p(h - k ) ] \
.\end{aligned}\ ] ] in the limit @xmath481 , these expressions simplify to @xmath482\nonumber\\ \hat{m}(h,\hat{h } ) & \sim \frac{\hat{h}}{k- \hat{k}_c } [ 1 + 2k p(h - k ) ] \ , \end{aligned}\ ] ] and @xmath483\nonumber\\ \hat{\hat{g}}_c(h,\hat{h})&\sim \frac{2k\hat{h}}{\hat{\hat{k}}_d ( k-\hat{k}_c ) } ( h - k ) p(h - k ) \ .\end{aligned}\ ] ] at zeroth - order of the expansion in number of free replica sums , the disconnected functions are obtained from the derivatives of the second cumulant @xmath484\:w_1[h^b,\hat{h}^b]$ ] with respect to the sources .
we first consider @xmath485 and @xmath486 the presence of heaviside step functions in @xmath487 makes the result dependent on the sign of @xmath488 . for simplicity
, we set at once @xmath489 .
we then find @xmath490 \end{split}\ ] ] where @xmath491 is a symmetric function of @xmath153 and @xmath154 , @xmath492\big[(h^b - k){\cal p}(h^a+k)+(h^a - k){\cal p}(h^b+k)+\hat{\hat{k}}_d[p(h^a+k)+p(h^b+k)]\big]\nonumber\\ & + k[1-\tanh \frac{k\hat{h}}{k- \hat{k}_c}]\big[(h^b - k){\cal p}(h^a - k)+(h^a - k){\cal p}(h^b - k)+\hat{\hat{k}}_d[p(h^a - k)+p(h^b - k)]\big]\nonumber\\ & + k^2[1-\tanh^2 \frac{k\hat{h}}{k- \hat{k}_c}][{\cal p}(h^a - k)+{\cal p}(h^b - k ) ] \ , \end{aligned}\ ] ] and @xmath493 ^ 2{\cal p}(h+k)+k^2[1-\tanh \frac{k\hat{h}}{k- \hat{k}_c}]^2{\cal p}(h - k ) \
.\end{aligned}\ ] ] therefore , when @xmath149 , we obtain @xmath494 \ .
\end{aligned}\ ] ] for @xmath495 , @xmath496 is then given by @xmath497 + 4 k^2 { \cal p}(h - k)[1-{\cal p}(h - k)]}{(k- \hat{k}_c)^2}\ ] ] whereas the coefficient of @xmath498 is equal to @xmath499 the 2-replica correlation function @xmath500 consists of a regular part and a singular part proportional to @xmath156 , @xmath501 we find @xmath502 ^ 2}\big[1+k[p(h^a+k)+p(h^a - k)+p(h^b+k)+p(h^b - k)]\big]\nonumber\\ & + \frac{\hat{h}^a}{k-\hat{k}_c}\ln
\big(2\cosh \frac{k\hat{h}^b}{k- \hat{k}_c}\big)[p(h^b+k)-p(h^b - k)]\nonumber\\ & + \frac{\hat{h}^b}{k-\hat{k}_c}\ln \big(2\cosh \frac{k\hat{h}^a}{k- \hat{k}_c}\big)[p(h^a+k)-p(h^a - k)]\end{aligned}\ ] ] and @xmath503[\frac{k\hat{h}^b}{k- \hat{k}_c}+\ln \big(2\cosh \frac{k\hat{h}^b}{k- \hat{k}_c}\big)]p(h^a+k)\nonumber\\ & [ \frac{k\hat{h}^a}{k- \hat{k}_c}-\ln \big(2\cosh
\frac{k\hat{h}^a}{k- \hat{k}_c}\big)][\frac{k\hat{h}^b}{k- \hat{k}_c}-\ln \big(2\cosh \frac{k\hat{h}^b}{k- \hat{k}_c}\big)]p(h^a - k)\ .\end{aligned}\ ] ] ( there are also singular contributions proportional to @xmath504 which can be discarded as we are only interested in the vicinity of @xmath505 . ) for @xmath149 and @xmath506 , we then find @xmath507 \end{aligned}\ ] ] and @xmath508 finally , the 2-replica correlation function @xmath509 is given by @xmath510{\cal p}(h^b+k)+[1-\tanh \frac{k\hat{h}^b}{k- \hat{k}_c}]{\cal p}(h^b - k)\big]\nonumber\\ & + \frac{k\:p(h^a+k)}{k- \hat{k}_c}\big[\frac{k\hat{h}^a}{k-\hat{k}_c}+\ln \big(2\cosh \frac{k\hat{h}^a}{k-\hat{k}_c}\big)\big]\big[\frac{h^b - h^a-2k}{k}+[1+\tanh \frac{k\hat{h}^b}{k- \hat{k}_c}]\theta(h^b - h^a)\nonumber\\ & + [ 1-\tanh \frac{k\hat{h}^b}{k- \hat{k}_c}]\theta(h^b - h^a-2k)\big]+\frac{k\:p(h^a - k)}{k- \hat{k}_c}\big[\frac{k\hat{h}^a}{k- \hat{k}_c}-\ln \big(2\cosh \frac{k\hat{h}^a}{k-\hat{k}_c}\big)\big]\big[\frac{h^b - h^a}{k } \nonumber\\ & + [ 1-\tanh \frac{k\hat{h}^b}{k- \hat{k}_c}]\theta(h^b - h^a)+[1+\tanh \frac{k\hat{h}^b}{k- \hat{k}_c}]\theta(h^b - h^a+2k)\big ] \
.\end{aligned}\ ] ] this function has a step - discontinuity at @xmath511 ( and also at @xmath512 ) and therefore one must fix the value of @xmath513 to lift the ambiguity when the sources are equal .
this is done by imposing the exact symmetry @xmath514 which yields @xmath515 . as a result
, one has @xmath516{\cal p}(h+k)+[1-\tanh \frac{k\hat{h}}{k-
\hat{k}_c}]{\cal p}(h - k)\big]\nonumber\\ & + \frac{k\:p(h+k)}{2(k- \hat{k}_c)}[\frac{k\hat{h}}{k-\hat{k}_c}+\ln \big(2\cosh \frac{k\hat{h}}{k-\hat{k}_c}\big)[-3+\tanh \frac{k\hat{h}}{k- \hat{k}_c}]\nonumber\\ & + \frac{k\:p(h - k)}{2(k- \hat{k}_c)}[\frac{k\hat{h}}{k- \hat{k}_c}-\ln \big(2\cosh \frac{k\hat{h}}{k-\hat{k}_c}\big)][3+\tanh \frac{k\hat{h}}{k- \hat{k}_c } ] \ , \end{aligned}\ ] ] so that @xmath517 \end{aligned}\ ] ] for @xmath518 . finally , we consider the two derivatives that are needed in eq .
( [ gd0_cusp_final ] ) to compute @xmath519 in the single - site effective model in the limit @xmath520 .
the first one is simply given by eq .
( [ eqcusprs ] ) , @xmath521 ^ 2}p(h - k)\ .\end{aligned}\ ] ] the second one is obtained from eq .
( [ eqgd0reg ] ) , @xmath522\\&=\frac{1 + 4k p(h - k)}{(k- \hat{k}_c)^2 } \ , \end{split}\ ] ] where we recall that the subscript @xmath380 indicates the limit of equal fields , @xmath511 .
j. p. sethna , k. a. dahmen , and o. perkovc in _ the science of hysteresis _ , edited by g. bertotti and i. mayergoyz , acedemic press , amsterdam ( 2006 ) .
j. p. sethna , k. a. dahmen , s. kartha , j. a. krumhansl , b. w. roberts , and j. d. shore , phys .
lett . * 70 * , 3347 ( 1993 ) .
k. a. dahmen and j. p. sethna , phys .
b * 53 * , 14872 ( 1996 ) .
f. detcheverry , e. kierlik , m. l. rosinberg , and g. tarjus , phys . rev . *
e 72 * , 051506 ( 2005 ) .
f. bonnet , t. lambert , b. cross , l. guyon , f. despetis , l. puech , and p. e. wolf , europhys .
* 82 * , 56003 ( 2008 ) .
f. detcheverry , e. kierlik , m. l. rosinberg , and g. tarjus , phys .
e * 73 * , 041511 ( 2006 ) .
d. dhar , p. shukla , and j. p. sethna , j. phys .
a : math . gen . * 30 * , 5259 ( 1997 ) .
s. sabhapandit , p. shukla , and d. dhar , j. stat . phys . * 98 * , 103 ( 2000 ) .
f. detcheverry f , m.l .
rosinberg , and g. tarjus , eur .
j. b * 44 * , 327 ( 2005 ) .
f. j. prez - reche , m. l. rosinberg , and g. tarjus , phys . rev .
b * 77 * , 064422 ( 2008 ) .
m. l. rosinberg , g. tarjus , and f.j .
perez - reche , j. stat .
p03003 ( 2009 ) .
m. l. rosinberg and t. munakata , phys .
b * 79 * , 174207 ( 2009 ) .
the fact that there are no metastable states outside the saturation hysteresis loop is a trivial consequence of the so - called ` no - passing ' rule for ferromagnetic systems@xcite . on the other hand , to the best of our knowledge , there is yet no general proof that the complexity in the large - disorder regime is strictly positive _ everywhere _ inside the loop in the large - disorder regime ( within the formalism developed in this work , it may very well be that the complexity vanishes at some finite value of the auxiliary field @xmath78 , say @xmath523 ; this would imply that the number of metastable states does not grow exponentially with the system size for @xmath524 and therefore that the hysteresis loop , obtained in the limit @xmath71 , is not the convex hull of the metastable states ) .
however , this seems quite unlikely , as illustrated by recent analytical and numerical calculations@xcite . on the other hand , in the low - disorder regime
, it is conjectured that the region where the complexity is positive is non - convex , and the loss of convexity results in the macroscopic jump in the non - equilibrium magnetization curve ( since the response function @xmath525 must always be convex ) .
moreover , in this case , it may be that the complexity does not vanish along the reentrant part of the envelope@xcite .
a. a. middleton , phys .
lett . * 68 * , 670 ( 1992 ) ; a. a. middleton and d. s. fisher , phys .
b * 47 * , 3530 ( 1993 ) . x. illa and m. l. rosinberg ( in preparation ) . since we consider a soft - spin model we can not use in the simulation the fast algorithms developed for the nonequilibrium @xmath2 rfim which allows one to study very large systems and to record all avalanches along the hysteresis loop@xcite . in the present case , the external field
is increased by a finite increment @xmath526 and in order not to miss small avalanches , which would bias the calculation of the second moment , @xmath526 must be very small ( _ e.g. _ @xmath527 for @xmath528 ) , forbidding the study of large systems .
m. c. kuntz , o. perkovic , k. a. dahmen , b. w. roberts , and j. p. sethna , computing in science and engineering 1 , * 73 * ( 1999 ) . as noted in ref.@xcite ( reference 100 )
@xmath529 is not exactly equal to @xmath530 where @xmath531 is the number of spins that flip in the avalanche , _
i.e. _ the number of spins that move from the ` down ' to the ` up ' potential well
. there is also a contribution to @xmath529 that comes from the harmonic response of the other spins through their coupling to the flipping spins .
p. le doussal and k. j. wiese , phys .
b * 68 * , 174202 ( 2003 ) ; nucl . phys .
b * 701 * , 409 ( 2004 ) .
g. tarjus and m. tissier , phys .
lett . * 93 * , 267008 ( 2004 ) ; phys .
b * 78 * , 024203 ( 2008 ) .
d. mouhanna and g. tarjus , phys .
e * 81 * , 051101 ( 2010 ) .
j. p. hansen and i. r. mcdonald , _ theory of simple liquids _
( elsevier , amsterdam,1986 ) .
p. le doussal and k. j. wiese , phys .
e * 79 * , 051106 ( 2009 ) .
g. s. joyce , j. phys .
a : math . gen .
* 37 * , 3645 ( 2004 ) .
e. kierlik , m. l. rosinberg , and g. tarjus , j. stat .
phys * 94 * , 805 ( 1999 ) .
m. feigelman and a. tsvelik , sov .
jetp * 50 * , 1222 ( 1979 ) ; a. j. bray and m. a. moore , j. phys .
c : solid state phys .
* 12 * , l441 ( 1979 ) ; a. v. lopatin and l. b. ioffe , phys . rev .
b * 66 * , 174202 ( 2002 ) . m. muller and l. b. ioffe , phys .
lett . * 93 * , 256403 ( 2004 ) .
m. muller and s. pankov , phys .
b * 75 * , 144201 ( 2007 ) .
a. georges , g. kotliar , w. krauth , and m. rozenberg , rev .
phys . * 68 * , 13 ( 1996 ) .
see _ e.g. _
j. m. cornwall , r. jackiw , and e. tomboulis , phys .
d * 10 * , 2428 ( 1974 ) .
j. k. percus , j. stat . phys . *
60 * , 221 ( 1990 ) . j. m. luttinger and j. c. ward , phys . rev . * 118 * , 1417 ( 1960 ) . g. baym and l. p. kadanoff , phys . rev . *
124 * , 287 ( 1961 ) .
note that the weiss fields @xmath532 s can be considered as sources acting on the composite variables @xmath533 ( and similarly with hatted quantities ) just as @xmath534 are sources acting on the fundamental variables @xmath535 .
this allows one to generate the ( two - particle irreducible ) luttinger - ward functional for the single - site effective model , which is then related to the local part of the luttinger - ward functional of the original model under the condition that the magnetizations @xmath536 and the local part of the green s functions are the same in the two models .
g. baym , phys . rev . * 127 * , 1391 ( 1962 ) . j. s. hoye and g. stell , j. chem
. phys . * 67 * , 439 ( 1977 ) ; mol . phys .
* 52 * , 1071 ( 1984 ) ; int .
j. thermophys .
* 6 * , 561 ( 1985 ) ; r. dickman and g. stell , phys .
lett . * 77 * , 996 ( 1996 ) .
the analytical expressions of the green function @xmath537 on the cubic lattice at the second and third neighbor distances can be found in the appendix of ref .
@xcite where the general recurrence relations derived in ref .
@xcite are used . note that @xmath538_{w=3/z}$ ] where @xmath539 is the green function for an arbitrary lattice site @xmath540 as defined in these references . g. s. joyce , j. phys
. a : math .
gen . * 35 * , 9811 ( 2002 ) .
t. horiguchi and t. morita , j. phys .
c : solid state phys . * 8 * , l232 ( 1975 ) .
a possible way out is to add a @xmath178-replica weiss field @xmath541 ( in the zeroth order of the expansion in number of free replica sums ) with @xmath542 a function to be adjusted so that the singular contribution present in @xmath313 when @xmath543 identically vanishes . at the same time
, a nonzero @xmath544 does not spoil the cusp - like behavior of the function @xmath412 . in this operational scheme
, one does _ not _ enforce the equality between @xmath545 in the original model and @xmath313 .
|
we present a formalism for computing the complexity of metastable states and the zero - temperature magnetic hysteresis loop in the soft - spin random - field model in finite dimensions .
the complexity is obtained as the legendre transform of the free - energy associated to a certain action in replica space and the hysteresis loop above the critical disorder is defined as the curve in the field - magnetization plane where the complexity vanishes ; the nonequilibrium magnetization is therefore obtained without having to follow the dynamical evolution .
we use approximations borrowed from condensed - matter theory and based on assumptions on the structure of the direct correlation functions ( or proper vertices ) , such as a local approximation for the self - energies , to calculate the hysteresis loop in three dimensions , the correlation functions along the loop , and the second moment of the avalanche - size distribution .
|
black holes are defined by their horizons .
there are different types of black hole horizons that can be used in different situations .
what type of horizon is relevant may depend on the question being asked . in black hole thermodynamics
the area of the black hole plays a role analogous to entropy via the bekenstein - hawking relation .
there has been a rigorous attempt to understand whether this entropy has an underlying microscopic explanation and it is hoped that the answer to this question will be an important clue to a theory of quantum gravity .
the event horizon of future null infinity is most often associated with the boundary of a black hole .
it has however been suggested that trapping horizons play an important role in black hole thermodynamics .
it is possible that it is the trapping horizon area and not the area of the event horizon that should be considered in black hole thermodynamics @xcite ( for a dissenting view see @xcite ) .
this raises a question of how different the areas of the event horizon and trapping horizon can be .
if the difference in areas is sufficiently large , much larger than the fundamental area in planck units , then a theory of quantum gravity should be able to tell us which area contains the black hole entropy .
furthermore in theories of semi - classical gravity with quantum fields on a classical spacetime background , one expects black holes to emit hawking radiation .
it has also been suggested that the trapping horizon plays a role in generating hawking radiation @xcite ( for a dissenting view see @xcite ) . in general numerical simulations
the trapping horizon always lies inside the event horizon .
if the radiation comes directly from the trapping horizon and the trapping horizon is hidden by the event horizon , then the radiation will not be visible from infinity .
in fact , no physical effects of the trapping horizon will be visible from outside .
a theorem of hawking and ellis @xcite implies that the apparent horizon always lies behind the event horizon . in such cases
the apparent horizon is never visible to asymptotic observers and its physical properties can not influence them .
the theorem depends on the validity of the null energy condition in the future of the region of the event horizon .
it is the same null energy condition that implies that the area of the event horizon can not decrease .
if the area of the event horizon is to decrease , a scenario that is envisioned in black hole evaporation through hawking radiation , then null energy condition must be violated somewhere along the event horizon . the quantum fields responsible for
the hawking radiation must be capable of violating the null energy condition .
in fact this scenario is borne out in explicit calculations in certain simplified models @xcite .
if hawking radiation violates the null energy condition then in certain stages of its lifetime the apparent horizon can appear outside the event horizon .
however , the null energy condition does not need to be violated locally in order for this to happen .
it is sufficient that it be violated somewhere in the future , reflecting the teleological nature of the event horizon . in this paper
we give an explicit model to show this effect .
although quantum effects are implied by both these motivations , the relation between the event horizon and trapping horizons can be studied in purely classical relativity on lorentzian manifolds . in principle the difference in areas can be arbitrarily large .
for example there are spacetimes that contain trapping horizons but not event horizons @xcite . in some other models @xcite
there are no event horizons but there is a region in which causally propagated signals can not avoid a quantum non - manifold region . in such cases a `` quantum horizon '' may replace the notion of an event horizon . in the standard semi - classical model of black hole formation and evaporation @xcite
there is both an apparent horizon and an event horizon .
causally propagated signals inside the event horizon can not reach future asymptotic observers . in such models one can ask how large can the difference between the area of the event horizon and the area of the apparent horizon be ? apparent horizons can also appear outside the event horizon in non - einstein models of gravity .
stationary solutions are the same in einstein and brans - dicke theory @xcite
. if the spacetime is globally static then the event horizon and apparent horizon will coincide .
however , in dynamical situations the apparent horizon can appear outside the event horizon and the event horizon area can decrease in the jordan - string frame in brans - dicke theory @xcite .
this is because the null energy condition can be violated in this frame even though it is not violated in the einstein frame .
the location of the apparent horizon changes under a conformal transformation of the metric .
the location of the event horizon does not change .
the change in area of the event horizon is related to the expansion .
the change of the expansion of the event horizon responds to the local geometry through the raychaudhuri equation and @xmath0 .
but the location of the event horizon is wholly non - local as is the requirement that it lie outside the apparent horizon .
but the area is decreasing when @xmath1 and this is also a condition for it to be inside the apparent horizon .
the null energy condition needs to be violated locally to drive the expansion of the event horizon down from a positive value to a negative value to allow it to pass inside the apparent horizon .
the area of the trapping horizon can also be increasing even when the local geometry is static vacuum .
this is because the horizon can acquire shear .
the shear obtains from the requirement that the surface be closed .
imagine the case where a thin , high speed , pencil of matter is sent into a black hole . the trapping horizon will increase in area even in regions that are initially causally disconnected from the incoming matter , since the horizon develops shear and shear is enough to increase the area of the trapping horizon .
shear is a property of the null rays normal to the surface and hence depends on the surface , not the local geometry . to study these issues
we will present a number of different cases .
firstly , we will examine the case of black holes that are accreting or evaporating at a constant rate . with the implicit assumption that an event horizon does exist in the future ,
the constant rate case allows certain conditions to be applied that enable us to solve for the location of the event horizon without needing to solve for the full spacetime . in these simple cases we can derive analytical formulae relating the areas of the event horizon and trapping horizons . to extend this analysis to more realistic situations
we will consider the case of a black hole that transitions from growing through matter accretion to evaporation .
this simple model is motivated by considering an isolated astrophysical black hole that is initially accreting energy from the cosmic microwave background ( cmb ) until the temperature of the cmb drops below that of the black hole and it starts to lose mass through hawking radiation .
in this situation we will solve for the location of the event horizon numerically assuming that the situation settles down in the far future to an almost constant rate evaporating black hole . during the transition from accretion to evaporation
we will be able to follow the evolution of the apparent horizon and event horizon and see that the apparent horizon moves outside the event horizon before the black hole finishes accreting . in the third part of the paper we will consider the issue of conformal transformations . for a specific example we will consider black holes in brans - dicke theory .
we will show how the definition of a trapping horizon can be modified to enable its location to be invariant under a conformal transformation .
interestingly , this condition can be related to black hole thermodynamics and the gravitational entropy of the black hole horizon .
we also show how this new definition guarantees the entropy increase theorem in brans - dicke theory .
the trapping horizons are three - dimensional surfaces , foliated by closed spacelike two surfaces for which the future directed null normals @xmath2 and @xmath3 satisfy _
l & = & 0 + _ n & < & 0 + _ n_l & < & 0 there is an implicit condition here in the choice of @xmath4 and @xmath5 .
each choice of @xmath4 and @xmath5 defines a set of spacelike surfaces normal to them .
different choices of @xmath4 and @xmath5 will lead to trapping horizons at different locations .
this issue has been examined in @xcite and can be formulated in terms of a choice of spacetime slicing into spacelike hypersurfaces .
a general spherically symmetric metric can be written in advanced null eddington - finkelstein coordinates ( @xmath6 ) as [ sphersymmetric ] s^2 = -e^-2(v , r)(v , r)v^2 + 2e^-(v , r)vr + r^2 ^ 2 in this general spherically symmetric spacetime , for spherically symmetric slicings the trapping horizon null normals will be radial null vectors and the location of the trapping horizon is just given by ( v , r ) = 0 this is a spacelike surface for @xmath7 , a null surface for @xmath8 and a timelike surface for @xmath9 .
choosing the conventional parametrization in terms of the misner - sharp mass , @xmath10 , the condition for the horizon is just r=2m(v , r ) [ r_is_two_m ] .this implicit equation can be solved explicitly in simple cases such as the vaidya spacetime where @xmath11 . for a linear mass function of the form @xmath12 , in terms of the timelike coordinate @xmath13 , the horizons will be located at r = [ r_is_not_two_m ] .
due to results in @xcite we expect that the non - spherically symmetric trapping horizons will intersect the spherically symmetric ones .
for further details see @xcite . since it is spherically symmetric the trapping horizon located at @xmath14 has area @xmath15 .
the event horizon is defined as a connected components of the past causal boundary of future null infinity and is generated by null geodesics that fail to reach infinity .
the event horizon is always a null surface since it is a causal boundary . in the above spacetime ( [ sphersymmetric ] )
the coordinate @xmath16 is constant on ingoing radial null geodesics .
any outgoing radial null geodesic must satisfy = for some parameter @xmath17 along the curve .
in particular , the null generators of the event horizon must satisfy this condition .
this first order ordinary differential equation generates the path of all outgoing radial null geodesics . in order to give the location of the event horizon it requires a boundary condition that corresponds to the known location of the event horizon at some particular point . in practice
this is usually given by the position of the event horizon at some future point , either when the black hole evaporates entirely or settles down to a stationary state .
if the black hole at some point settles down to a schwarzschild black hole with no further matter accreting , then the event horizon can be located by tracing back the null rays from the future schwarzschild radius .
the event horizon likewise has area @xmath18 on spherically symmetric slicings .
the difference in the areas @xmath19 is therefore a__diff = 4r__eh^2 - 16m(v , r)^2 for the case where the black hole is evolving ( either growing or shrinking ) at a constant rate we may guess that the difference in the areas of the trapping horizon and event horizon should be constant . with respect to a slicing with slices labelled by @xmath17 [ constantarea ] a__diff
= 0 this gives 2r__eh - 8 m__ah = 0 substituting in for @xmath20 [ ehrad ] r__eh = 2m__eh + 4r__ah__ahe^__eh where @xmath21 .
since we are comparing areas on a slice of constant @xmath17 , we can expand values at the event horizon in terms of values at the apparent horizon and @xmath22 m__eh = m__ah + m__ah(r__eh - r__ah ) + ( r)^2 + ... _ _ eh = _ _ ah + __ah(r__eh - r__ah ) + ( r)^2 + ... where @xmath23 .
substitute these expansions in to ( [ ehrad ] ) to get r__eh = r__ah + ( r)^2 + ... + ( 1+__ahr + ... ) which is just r = ( r)^2 + ( 1+__ahr + ( _ _ ah + ^2__ah)(r)^2 ) + ( ( r)^3 ) where @xmath24 can be interpreted as a dynamical surface gravity @xcite .
now solve this at linear order in @xmath25 r = ( 1 + ) + ( ^3 ) and thus , for @xmath26 to leading order in @xmath27 we expect that the event horizon will be outside the spherically symmetric trapping horizon for @xmath7 but inside for @xmath9 . in the case where the black hole is accreting matter at a steady rate and is a suitably long way from changing to a different state one can also find the approximate location of the event horizon by imposing the condition [ ehlocation ] = 0 .on the horizon generators .
this just reflects the fact that the event horizon is growing at a steady rate . while the condition ( [ constantarea ] ) compares values at two different points in the spacetime , this condition is purely local on the event horizon . in this case ,
equation ( [ ehlocation ] ) has the general solution [ steadyeh ] r = ( 1- ) .which holds at the event horizon . for @xmath28
this gives r = 2m(v)(1 + 4 + 32 ^ 2 + ( ^3 ) ) in terms of the coordinate @xmath13 for the linear mass function , equation ( [ steadyeh ] ) can be solved explicitly and the event horizon has radial coordinate r ~ + .this is just the location of the spherically symmetric trapping horizon with a constant offset of @xmath29 provided @xmath28 . in this approximation
the generators of both the trapping horizon and the event horizon have the same components but the norm of the generators is @xmath30 for the trapping horizon and zero for the event horizon .
the trapping horizon is spacelike for @xmath7 but the event horizon is still a null hypersurface .
in terms of the difference between the areal radius coordinates , @xmath25 , defined in the last section , the difference in the areas is a_diff = 4(2r_ahr - r^2 ) to get an idea of the numbers involved in the dynamical accretion and evaporation of mass by a black hole consider the following very approximate situations .
the eddington limit gives a good approximation of the rate at which black hole accretion disks can be supported by their own self - generated radiation pressure .
the eddington limit is given by 1.3 10 ^ 31 ( ) the dimensionless mass accretion rate is calculated by = for typically observed values of a black hole accreting at a tenth of the eddington rate @xcite the dimensionless accretion rate is approximately 10 ^ -22 ( ) .the matter falling into the black hole is ten times the energy being emitted as light . since this accretion is usually associated with a disk it will not be exactly spherically symmetric , but the approximation is still often applied . for a black hole accreting purely from the cosmic microwave background which is assumed to be isotropic , and using the stefan - boltzmann law
, we have approximately that the area , @xmath31 , is a 16m_^2 ( ) ^2 and so the dimensionless mass accretion rate is 10 ^ -50()^4()^2 .for a black hole whose dynamics are dominated by evaporation through hawking radiation with temperature t _ _ = we have a ( negative ) massless accretion rate of [ mdothawk ] -10 ^ -81()^2 .the difference in areas can be computed as [ adiff ] a__diff = a__eh - a__ah 10 ^ 78()^2l__p^2 the numerical factor is related to the solar schwarzschild area in planck units , @xmath32 .
the exact formula for the area difference for the case where the black hole is evaporating purely through hawking radiation ( ( [ mdothawk ] ) into ( [ adiff ] ) ) turns out to be a__diff = - = -l__p^2 thus in the case where an otherwise isolated black hole is evaporating at an almost constant rate due to hawking radiation , in the conventional picture of black hole evaporation @xcite , the difference in the area of the event horizon and trapping horizon is not resolvable at the planck scale ( although the trapping horizon is outside the event horizon ) . for any of the accreting black holes considered above ,
the difference is many orders of magnitude of the planck area . for a solar - mass black hole accreting at a tenth of the eddington rate
the difference in areas between the event horizon and the spherically symmetric trapping horizon will be around @xmath33 in units of planck area , while for a supermassive black hole of mass @xmath34 solar masses , accreting purely form the cmb , the difference in areas will be around @xmath35 in planck units .
to see the transition of a trapping horizon from inside the event horizon to outside , we need to look at black holes where the rate of change is changing .
one example of this is an otherwise isolated black hole that accretes matter from the cosmic microwave background that fills all of the universe .
a static solar - mass sized black hole has a hawking temperature of about @xmath36k .
the temperature of the cmb is currently about @xmath37k .
the temperature of the black hole changes as its mass changes and the temperature of the cmb changes as the universe expands .
roughly speaking , if the temperature of the cmb is larger than the temperature of the black hole , it will accumulate mass and grow .
if the temperature of the black hole is greater than the temperature of the cmb it will lose mass and shrink .
black holes of mass @xmath38 kg have a temperature today greater than the cmb .
primordial black holes of initial mass @xmath39 kg should be evaporating today @xcite .
their mass density however must be @xmath40 @xcite .
if we assume that the universe is dominated by a cosmological constant with @xmath41 the temperature of the cmb will equal @xmath42k in about @xmath43seconds or @xmath44years . during this time the black hole will accrete about @xmath45 kg of photonic matter from the cmb . since the mass of a solar - sized black hole is about @xmath46
kg it s percentage increase in mass will be tiny . in a similar length of time
the mass lost to hawking radiation will be even smaller , about @xmath47 kg . therefore ,
during the time that the cmb temperature falls to the black hole temperature , one can consider the black hole size as roughly constant .
consider a simple toy model consisting of a spherically symmetric black hole that absorbs radiation from the cmb and also emits hawking radiation .
one way of modelling the hawking radiation is to implement it as inflowing negative energy @xcite or with a region of outflowing positive energy but negative energy onto the black hole @xcite .
the vaidya solution for infalling radiation is s^2 = -(1-)v^2 + 2vr + r^2 ^ 2 for a black hole that accretes purely from the cmb and then evaporates via hawking radiation we can write a mass function as m(v ) = m_o - k_1(v - v_o ) - k_2 ( ) + k_2 where @xmath48 , @xmath49 and @xmath50 are constants .
the second term given the mass lost due to hawking radiation and the third term gives the mass gained by accretion from the cmb in a de sitter ( cosmological constant dominated ) universe .
interpreting the terms as accretion from the cmb and evaporation through hawking radiation we would have k_1 = k_2 = 4m_o^2t_o^4 = 4 where @xmath51 is the stefan - boltzmann constant , @xmath52 is the mass of the black hole at @xmath53 , @xmath54 is the temperature of the cmb at @xmath53 and @xmath55 is the cosmological constant .
since it is based on the vaidya solution this model does not embed the black hole in a true cosmological background either de sitter - like or otherwise and is only intended to hold in a small region near the horizons to illustrate the potential behaviour in more exact models . the accretion from cmb part acts like a heat bath at a given temperature rather than true accretion from a robertson - walker spacetime .
the trapping horizon reaches its maximum area where @xmath8 .
this corresponds to v__-v_o = ( ) the trapping horizon crosses the event horizon roughly when v _ _ - v_o = ( ) therefore the time ( elapse of the @xmath16 coordinate ) between the trapping horizon passing outside the event horizon and the area of the trapping horizon starting to decrease ( where the nec is locally violated ) , is v _ _ - v _ _ = ( ) @xmath56 , @xmath57 and @xmath58 .
the trapping horizon moves outside the event horizon while the trapping horizon is still growing . ] an example for the case @xmath59 @xmath56 , @xmath57 and @xmath58 is shown in fig.([fig : cosmohorizon ] ) .
the location of the event horizon is found by tracing back from the value given by ( [ steadyeh ] ) .
the difference in radial coordinate at @xmath60 is due to the fact that the black hole is still dynamical when it settles down to `` constant '' evaporation . in this example
the trapping horizon starts off inside the event horizon and grows rapidly as the black hole accretes matter .
the trapping horizon crosses the event horizon just as the event horizon starts to decrease in area . in the example
this occurs at @xmath61 and @xmath62 . at this point
the accreting `` cmb '' flux is still higher than the `` evaporation '' flux and we have @xmath7 implying that the null energy condition is still satisfied .
the apparent horizon grows to a maximum size of @xmath63 before the `` evaporation '' flux starts to dominate and the hole starts to shrink .
this example shows that it is possible , even likely , that the trapping horizon will appear outside the event horizon before the trapping horizon starts to shrink .
the crossing occurs precisely where the area of the event horizon starts to decrease .
this means that the trapping horizon becomes visible to asymptotic observers even before the energy conditions are violated and before its dynamics are dominated by hawking radiation .
if the trapping horizon starts to produce hawking radiation in its near vicinity after this time the hawking radiation can escape to infinity .
the event horizon s area starts to decrease , not because of some causal signal such as local violation of the energy conditions , but in anticipation of future violation .
this is then consistent with the violation of energy conditions being associated with a hawking flux from the trapping horizon and mass loss to infinity which causes the black hole to shrink .
under a conformal transformation the relative position of the event horizon and the trapping horizon can also change .
the use of conformal transformations is fairly common in looking at black holes solutions .
this is particularly true in string theory where conformal transformations are used to relate the string frame , with a non - minimally coupled dilaton field , to the einstein frame .
it has been argued in the literature that classically the two frames are physically equivalent @xcite .
the conformal transformation has the effect of rescaling lengths and areas as measured by the metric .
the physical effect of this rescaling is , for example , to change the meaning of mass since the norm of the four - momentum , @xmath64 , will no longer be constant from point to point or time to time .
it has been observed in numerical simulations of black hole collapse in brans - dicke theory , that the trapping horizon can appear outside the event horizon @xcite .
this occurs despite the fact that the string frame of brans - dicke theory can be related via a conformal transformation to einstein theory with a scalar field that obeys the null energy condition . in the einstein frame
the trapping horizon appears exclusively behind the event horizon , in accordance with the theorem of hawking and ellis @xcite .
two issues are involved here .
firstly , unlike the event horizon , the location of the trapping horizon changes under a conformal transformation @xcite . secondly , the trapping horizon can appear outside the event horizon in the string frame because the einstein equations do not hold in this frame .
to relate the two frames the metric is scaled by a conformal factor that can vary with spacetime point [ conformaltrans ] g_ab g_ab = w(x)g_ab the expansion of a null vector @xmath2 in any frame is given by _
l = q^ab_al_b = ( g^ab + + ) _ al_b where @xmath65 is a projection tensor onto the two - dimensional spacelike surface to which @xmath2 and @xmath3 are normal .
( if @xmath2 is defined as globally null then the third term vanishes identically . )
this is a result that holds for a lorentzian signature manifold independently of whether the einstein equations hold . in general
there is a freedom to rescale null vectors even without rescaling the metric .
the vanishing of the expansion does not depend on a pure rescaling of the null vector @xmath66 , although its value does since under this rescaling we have _
l w_l under a conformal transformation of the form ( [ conformaltrans ] ) we have @xmath67 and @xmath68 .
we can fix the normalization of @xmath2 by requiring @xmath69 with @xmath70 and thus _
a_b = w_al_b + l_b_aw - ( l_a_bw + l_b_aw - g_abl^c_cw ) therefore [ thetaltrans ] _
l + the vanishing of @xmath71 for a given surface is therefore not necessarily invariant under a conformal transformation .
the conformal transformation though does not change the coordinates of a given spacetime event nor the path of null rays .
the location of the event horizon is unchanged . in one frame
the solution of @xmath72 may lie inside the event horizon and in another frame outside .
brans - dicke theory is the prototype alternative to theory of gravity with scalar and tensor modes .
the action , in the string frame is given by = ( r - _ a^a ) + _ matter .variation of this action with respect to the metric gives the gravitational field equations
g_ab= 8t^matter_ab + ( _ a_b- g_ab_c^c ) + _ a_b-
g_ab_c^cthe proof of the apparent area theorem @xcite is a purely geometric proof that only relies on the validity of the null raychaudhuri equation and the geometrical condition @xmath73 , the null curvature condition .
this condition can be related via the einstein equations to the null energy condition , @xmath74 .
contracting the ricci tensor with @xmath2 gives r_abl^al^b = t_abl^al^b + ( l^a_a)^2 + where @xmath75 is the energy - momentum tensor of the non - gravitational matter fields . even if the matter obeys the null energy condition @xmath76
, the sign of the last term is indeterminate and therefore we may have a violation of the null curvature condition .
this is in fact required to happen for the surfaces found in @xcite . brans - dicke theory in the string frame can be recast in the einstein frame via the conformal transformation [ bdconftrans ] _ ab = g_ab in the einstein frame the null tangent vectors are unchanged @xmath69 and they are null with respect to the new metric as well as the old one . in the einstein frame
the gravitational field equations are _ ab = 8_ab + ( _ a_b- _ ab_c^c ) and thus _
abl^al^b = 8_abl^al^b + l^a_al^b_bprovided the matter obeys the null energy condition and @xmath77 the geometry will also obey the null curvature condition in the einstein frame .
the vanishing of the expansion is equivalent to the statement that the area is unchanged under translations along @xmath2 via the relation @xmath78 .
since the conformal factor changes how areas are measured this no longer selects out the same horizon .
the area two - form changes as @xmath79 .
the condition that the lie derivative of this `` conformally transformed area '' be zero is _
l ( w a ) = wa ( _ l + ) = 0 this is the same as the transformation in ( [ thetaltrans ] ) . in theories with gravitationally non - minimally coupled scalar fields , such as brans - dicke theory , the gravitational entropy , defined by the noether - charge entropy @xcite , is not always equal to the area .
thermodynamically , it has been argued that in brans - dicke theory the quantity @xmath80 is non - decreasing on the event horizon @xcite .
this suggests that in non - minimally coupled theories ( or in a frame where the metric tensor does nt satisfy the einstein equations ) the horizons should be defined not by the vanishing of the area along normal null directions , but by the vanishing of the gravitational entropy , @xmath81 , along null directions .
the three conditions for a trapping horizon would then be l^a_as_g & = & 0 + n^a_as_g & < & 0 + n^a_a(l^a_as_g ) & < & 0 this then provides a frame covariant way to identify the `` true '' trapping horizon . in the einstein frame
this would reduce to the ordinary requirements on the null expansions for a trapping horizon .
it has been claimed that if a field redefinition can be used to relate the actions of two theories then the black hole entropies in these theories are related by the same field redefinition , at least when applied to event horizons @xcite .
if this is also true of the gravitational entropies of trapping horizons then the above gives a frame independent way of identifying the physical trapping horizon in all theories for which a gravitational entropy can be defined . with these conditions
one can then examine how the generalised noether - charge entropy evolves along a horizon with tangent @xmath82 .
the variation of the generalised entropy is r^a_as = bl^a_as + cn^a_as the first term on the right hand side is zero by assumption . since we require the tangent @xmath83 to generate evolution along the generalised trapping horizon we have r^a_a ( l^a_as ) = 0 this can be rearranged to give c = - and so the change of the generalised entropy along the horizon can be written as r^a_as = -l^a_a(l^a_as ) with the sign of @xmath84 setting the orientation of @xmath83 and the sign of @xmath85 and @xmath86 both negative by assumption on the horizon , whether the generalised entropy is increasing or decreasing is just determined by the last term , @xmath87 . in the brans - dicke case
we have @xmath88 . for the entropy to be positive we require @xmath89 and since the area is positive , @xmath90 , we see that we require @xmath91 too ( note that if @xmath92 anywhere in the spacetime then the conformal transformation ( [ bdconftrans ] ) becomes illdefined . )
the variation of this entropy in the outgoing null direction is then l^a_as = a(_l + l^a_a ) and l^a_a(l^a_as ) = -a(_l^2 + ^2 + t_abl^al^b ) - ( l^a_a ) ^2 thus for @xmath77 and matter obeying the null energy condition , @xmath93 , the generalised entropy is guaranteed to increase along the generalised trapping horizon .
we have seen how the location of the trapping horizon can be related to the location of the event horizon for several different cases . for accreting black holes
the difference in areas can be very large relative to the planck scale .
this means that the microscopic degrees of freedom responsible for this entropy are likely to be very different in these cases . for evaporating black holes
the difference in area is not resolvable on the planck scale .
we have also presented a simple model of a black hole that transitions from accreting to evaporating .
the model is inspired by an otherwise isolated black hole accreting from the cmb and subsequently beginning to evaporate as the cmb temperature falls . in this case
the trapping horizon appears outside the event horizon before the evaporation becomes dominant and before the null energy condition is violated .
however , both these cases assume the standard picture of black hole evaporation that includes both an event horizon and a trapping horizon . in some models there is only a trapping horizon and no event horizon . whether there is an event horizon or not is virtually undecidable by experiment though , which is one of the key reasons why some authors have preferred to concentrate on trapping horizons . we have also suggested a redefinition for trapping horizons outside of the einstein conformal frame .
the definition relies on the notion of dynamical gravitational entropy . in non - einstein frames
the generalised trapping horizon will no longer be a marginally outer trapped surface .
w. a. hiscock , phys .
d * 40 * ( 1989 ) 1336 .
w.collins , phys .
d * 45 * ( 1992 ) 495 .
a. b. nielsen , gen .
* 41 * ( 2009 ) 1539 [ arxiv:0809.3850 [ hep - th ] ] .
r. d. sorkin , arxiv : gr - qc/9701056 .
a. corichi and d. sudarsky , mod .
a * 17 * ( 2002 ) 1431 [ arxiv : gr - qc/0010086 ] .
r. di criscienzo , m. nadalini , l. vanzo , s. zerbini and g. zoccatelli , phys .
b * 657 * ( 2007 ) 107 [ arxiv:0707.4425 [ hep - th ] ] .
a. b. nielsen and d. h. yeom , int .
j. mod .
a * 24 * ( 2009 ) 5261 [ arxiv:0804.4435 [ gr - qc ] ] . c. barcelo , s. liberati , s. sonego and m. visser , phys .
* 97 * ( 2006 )
171301 [ arxiv : gr - qc/0607008 ] . s. w. hawking and g. f. r. ellis , `` the large scale structure of space - time , '' _ cambridge university press , cambridge , 1973 _ m. visser , phys .
d * 54 * ( 1996 ) 5103 [ arxiv : gr - qc/9604007 ] . t. a. roman and p. g. bergmann , phys . rev .
d * 28 * ( 1983 ) 1265 .
s. a. hayward , phys .
lett . *
96 * , 031103 ( 2006 ) [ arxiv : gr - qc/0506126 ] .
a. ashtekar and m. bojowald , class .
* 23 * ( 2006 ) 391 [ arxiv : gr - qc/0509075 ] .
s. w. hawking , commun .
* 43 * ( 1975 ) 199 [ erratum - ibid .
* 46 * ( 1976 ) 206 ] .
s. w. hawking , commun
. math .
* 25 * ( 1972 ) 167 a. b. nielsen and j. h. yoon , class .
* 25 * ( 2008 ) 085010 [ arxiv:0711.1445 [ gr - qc ] ] .
a. ashtekar and g. j. galloway , adv .
* 9 * ( 2005 )
1 [ arxiv : gr - qc/0503109 ] . j. a. kollmeier _ et al .
_ , astrophys .
j. * 648 * ( 2006 ) 128 [ arxiv : astro - ph/0508657 ] . b. j. carr , arxiv : astro - ph/0511743 .
b. j. carr , k. kohri , y. sendouda and j. yokoyama , phys .
d * 81 * ( 2010 ) 104019 [ arxiv:0912.5297 [ astro-ph.co ] ] .
w. a. hiscock , phys .
d * 23 * ( 1981 ) 2813 .
v. faraoni and s. nadeau , phys .
d * 75 * ( 2007 ) 023501 [ arxiv : gr - qc/0612075 ] .
r. m. wald , phys .
d * 48 * ( 1993 ) 3427 [ arxiv : gr - qc/9307038 ] .
g. kang , phys .
d * 54 * ( 1996 ) 7483 [ arxiv : gr - qc/9606020 ] .
t. jacobson , g. kang and r. c. myers , phys .
d * 49 * ( 1994 ) 6587 [ arxiv : gr - qc/9312023 ] .
|
the relation between event horizons and trapping horizons is investigated in a number of different situations with emphasis on their role in thermodynamics . a notion of constant change
is introduced that in certain situations allows the location of the event horizon to be found locally .
when the black hole is accreting matter the difference in area between the two different horizons can be many orders of magnitude larger than the planck area .
when the black hole is evaporating the difference is small on the planck scale .
a model is introduced that shows how trapping horizons can be expected to appear outside the event horizon before the black hole starts to evaporate .
finally a modified definition is introduced to invariantly define the location of the trapping horizon under a conformal transformation . in this case
the trapping horizon is not always a marginally outer trapped surface .
|
early - type galaxies form the dominant population in the core of all galaxy clusters with few or no exceptions from low redshift to @xmath0 , from poor to rich clusters .
the uniformity of the properties of early - type galaxies provide important tools , such as the fundamental plane and the color - magnitude relation ( cmr ) , in the study of galaxy clusters and their evolution .
the cmr for early - type galaxies ( e / s0 s ) was first noted by baum ( 1959 ) in that the colors for elliptical galaxies become bluer as they become less luminous .
furthermore , e / s0 galaxies are also the reddest normal galaxies at any single given redshift .
hence , the e / s0 s form a sequence on the color - magnitude plane which is sometimes called the `` red sequence '' , as illustrated for the cluster abell 2256 using data from lpez - cruz & yee ( 1999a , also see lpez - cruz 1997 ) in figure 1 .
the existence of the cmr is in general attributed to a metallicity effect due to the varying ability of spheroids of different masses to retain metals in the presence of supernovae driven winds ( e.g. , see kodama & arimoto 1997 and references therein ) .
the dispersion , color , and slope of the red sequence in clusters have been used to study the evolution of cluster galaxies up to a redshift as high as @xmath1 ( e.g. , bower et al .
1992 ; aragn - salamanca et al .
1993 ; stanford et al . 1998 ; and gladders et al .
1998 ) . in this paper
, we demonstrate the efficiency of using the color - magnitude relation of clusters as a powerful tool for estimating the redshift of clusters photometrically and detecting clusters from two - filter imaging data .
the red sequence of early - type galaxies in clusters provides a powerful means for estimating the cluster redshift using only two filters .
the efficacy of using the red sequence to determine redshift arises from two factors .
first , early - type galaxy spectra , because of the large 4000 break , provide the largest signal for photometric redshift determination . by predetermining the spectral type of the galaxies , ( i.e. , picking the reddest galaxies available by virtue of the prominence of the red sequence ) , only a single color ( from two filters , assuming that they straddle the 4000 break ) is needed to provide the redshift signature .
second , the red sequence is composed of many such galaxies ; hence , if the whole sequence is used instead of a single galaxy , the measurement uncertainty would diminish significantly .
to test the efficiency and accuracy of this method , we use two - color ccd photometry from a sample of 45 abell clusters from the low - redshift cluster optical survey ( locos ) of lpez - cruz & yee ( 1999a ) .
the clusters have a redshift range between 0.025 to 0.18 . the locos data set consists of @xmath2@xmath3@xmath4 photometry obtained using the 2k@xmath52k t2ka ccd on the kpno 0.9 m over a field size of 23@xmath623@xmath7 to a depth of typically @xmath8 ( @xmath9 km / s / mpc ) .
the details of the analysis of the cmr of the clusters are presented in lpez - cruz & yee ( 1999b ) and gladders et al .
briefly , the cmr for each cluster is fitted with a linear function on the @xmath10 vs @xmath3 color - magnitude plane , using a robust regression method with outlier rejection .
the resultant slopes of the fits indicate that the cmr is consistent with being universal in the redshift range of the sample , reflecting the high @xmath11 formation epoch of the red galaxies in the clusters ( see also gladders et al .
we note that in order to apply the color of the red sequence to obtain a photometric redshift , we must assume that the red galaxies in different clusters at the same @xmath11 are coeval a condition that is approximately met if the cores of clusters are formed at high @xmath11 . from the fit of the cmr
, the @xmath10 color of a red sequence at any @xmath3 can be determined .
we have chosen a fiducial , though arbitrary , magnitude of @xmath12 for the purpose of obtaining an empirical calibration between @xmath11 and the @xmath10 color .
we note that an apparent magnitude is chosen since in determining a photometric redshift , no prior knowledge of the redshift can be assumed .
the correlation between @xmath11 and the @xmath10 color of the red sequence at @xmath12 is shown in figure 2 . a second degree polynomial fit to the data produces a residual in @xmath11 of 0.008 , or a @xmath13 . comparing to the residual of @xmath14 to 0.07 ( @xmath15 to 0.2 ) of typical empirical photometric redshift techniques using four colors ( see e.g. , brunner et al .
1997 , yee 1999 , and papers in these proceedings ) , the empirical calibration of the cluster redshifts using a single cmr color is remarkably tight and robust .
the ubiquitousness of the early - type galaxy red sequence can be used as a powerful tool for detecting high spatial density regions such as galaxy clusters and groups . selecting galaxies in color slices in the color - magnitude diagram
would preferentially increase the density enhancement of any grouping of early - type galaxies which are normally found in clusters or groups .
this technique can be used not only as a very efficient cluster finding algorithm , but also as a means of separating projected clusters and groups . cluster catalogs created by primarily locating surface density enhancements of galaxies , such as the abell and apm catalogs , suffer from projection effects .
estimates of such projection effects in the abell catalog range from 10 to 20% ( e.g. , olivier et al .
figure 1 demonstrates the ability of using the color - magnitude diagram of a cluster field to detect background clusters in an efficient manner . a second red sequence above the red sequence of a690 can be clearly seen .
the galaxies in this second red sequence , which corresponds to a redshift of @xmath160.25 , are found to be clustered to the nw of the main a690 cluster ( at @xmath17 ) , and coincident with a secondary x - ray peak . in the locos data set
, we identified 6 clusters out of the 45 as having a significant background cluster ( lpez - cruz & yee 1999b ) , giving a contamination rate of @xmath1610 to 15% for the abell clusters . a similar rate ( 2/9 )
is also obtained for the control fields .
we note that such contaminations are also present in the x - ray maps , although at a smaller rate of 1/2 to 1/3 of the optical images . a major problem in existing galaxy cluster catalogs from optical surveys ( e.g. , abell 1958 , postman et al .
1996 ) is that they are susceptible to projection effects .
an additional difficulty for any survey that attempts to find high - redshift clusters is the accumulated column density of galaxies in the foreground of a cluster , which decreases the surface density enhancement of galaxies in the cluster . the red sequence of early - type galaxies in clusters and groups , however , provides a crucial signature which allows us to identify such high - density regions .
we are in the process of carrying out a large optical imaging survey for galaxy clusters to redshifts @xmath18 , using the red sequence as a primary marker ( gladders & yee 1999 ) . the survey , conducted at the cfht 3.6 m and the ctio 4 m using large mosaic ccd imagers , will obtain images over 100 deg @xmath19 in area in the @xmath20 ( @xmath169200 ) and @xmath3 bands sufficiently deep to detect galaxy clusters to @xmath21 .
the main scientific goals are to obtain the mass spectrum of clusters of galaxies as a function of redshift in order to constrain the cosmological parameter pair : ( @xmath22 ) , and to study the evolution of clusters over a large redshift range with a sample of clusters covering a wide range of properties .
the cluster detection algorithm uses combined information on the color - magnitude plane and the x - y position plane .
galaxies are chosen based on slices in the @xmath23 vs @xmath20 plane which mimic the cmr of early - type galaxies at various redshifts .
model cmrs as a function of redshift can be obtained from synthesis models from , e.g. , kodama ( 1997 ) .
surface enhancements of galaxy density in the x - y plane in these slices are then cataloged as cluster candidates . for images with good seeing , morphological information in the form of a concentration parameter
can also be used to further isolate early - type galaxies . in short ,
clusters are identified in the 5-dimensional space of ra , dec , color , magnitude , and morphology .
the use of the red sequence minimizes projection effects from fore- and background clusters and groups , as the color allows a crude separation in redshift space .
furthermore , by isolating one color slice at a time ( and with added morphological selection when available ) , the excess signal from the cluster is greatly enhanced .
the detected red sequence can then be used to estimate a photometric redshift for the candidate , using either cmr models or an empirically calibrated relation . with a photometrically measured redshift
, global properties such as the luminosity function , richness , and blue fraction for the cluster can be estimated statistically from the photometric data .
_ 3.2.1 .
testing the red sequence algorithm with cnoc2 data _
we have tested the cluster detection algorithm using both real and simulated data .
the cnoc2 survey ( yee et al .
1998 ) provides the largest redshift / multi - color photometry database currently available for intermediate redshifts . using 3/4 of the data ( about 1.2 deg@xmath19 ) , we construct a catalog of groups / clusters using @xmath4 and @xmath24 photometry and the method outlined above . for each of the identified groups ,
a photometric redshift is determined by comparing with cmr models from kodama ( 1997 ) .
the groups are then verified using the redshift catalogs .
every identified group is found to have a corresponding group in redshift space .
the average richness of these groups is sub - abell richness class 0 , with a mean velocity dispersion of about 350 km / s .
figure 3 shows the comparison of the photometric redshifts as derived from the red sequence with the spectroscopic redshifts of the groups .
the agreement is excellent over the redshift range of 0.1 to 0.7 with a mean @xmath25 of only 0.028 , comparable to or better than the best photometric redshift results using 4 colors ( e.g. , brunner et al .
this demonstrates that the red sequence method is able to detect very poor clusters and groups and to estimate their redshifts with remarkably high accuracy using cmr models . _
3.2.2 . testing the red sequence algorithm with simulated data _ to test the efficiency , robustness , and systematic effects of the red sequence method
, we have also constructed detailed models of the observed sky projection of field galaxies .
the simulations of the background field galaxy distribution reproduce known galaxy count distributions ( to ab magnitude @xmath26 ) , galaxy luminosity functions , galaxy redshift distributions , color distribution , density morphology relation , and angular covariance function .
clusters of different properties , such as blue fraction , richness , ellipticity , and core radius etc .
, at different redshifts are then placed in the simulated field .
the cluster finding and photometric redshift procedures are then applied to the simulated data .
figure 4 shows the results of one set of simulations in which the blue fraction is varied for a sample of abell richness 1 clusters .
the simulated galaxy photometry catalogs have the same depth as the cluster survey with a 5@xmath27 detection limit of @xmath28 and @xmath29 .
it is found that the detection probability essentially stays constant at near 100% for clusters with blue fractions less than 0.8 up to @xmath30 ; and the detection probability remains significant at @xmath1650% at @xmath21 . at @xmath31 ,
the @xmath20 band passes blueward through the 4000 break , essentially erasing the advantages of using the red sequence method .
poorer clusters , of about abell richness class 0 , can be detected at near 100% detection probability up to @xmath320.9 .
other tests indicate that properties such as the ellipticity and core radius of the clusters have little or no effect on the detectability .
the cmr is a powerful tool for detecting galaxy clusters and estimating their redshifts .
we have demonstrated that at low - redshift where one can obtain an empirically calibrated relation between the color of the red sequence and redshift , cluster redshifts can be estimated to a high accuracy ( @xmath33 ) using just two filters .
the red sequence is also an extremely efficient method for identifying clusters over a large richness and redshift range .
we have begun a 100 deg@xmath19 survey using modern mosaic ccd imagers on 4 m class telescopes , from which we expect to find upward of 100 rich clusters at @xmath34 .
abell , g.o .
1958 , apjs , 3 , 211 aragn - salamanca , a. , et al .
1993 , mnras , 262 , 764 baum , w.a . 1959 , pasp , 71 106 bower , r.g . ,
lucey , j.r . & ellis , r.s .
1992 , mnras , 254 , 589 brunner , r. , connolly , a. , szalay , a. , & bershady , m. 1997 , apj , 482 , l21 gladders , m. , lpez - cruz , o. , yee , h.k.c .
, & kodama , t. 1998 , apj , 501 , 571 gladders , m. & yee , h.k.c . 1999 , in preparation kodama , t. 1997 , ph.d .
thesis , university of tokyo kodama , t. & arimoto , n. 1997 , a&a , 321 , 41 lpez - cruz , o. 1997 , ph.d .
thesis , university of toronto lpez - cruz , o. & yee , h.k.c .
1999a , to be submitted to rev .
a & a lpez - cruz , o. & yee , h.k.c .
1999b , to be submitted to apj olivier , s. , et al .
1990 , apj , 356 , 1 stanford , s.a . , eisenhardt , p.r . , &
dickinson , m. 1998 apj , 492 , 461 postman , m. , et al .
1996 , aj , 111 , 615 yee , h.k.c .
1999 , in the proceedings of the xth recontre de blois : _ the birth of galaxies _ , astro - ph/9809347 yee , h.k.c .
1998 , in _ highlights of astronomy _ , 11a , 460 , ( astro - ph/9710358 )
|
the color - magnitude relation ( cmr ) of early - type galaxies in clusters provides a powerful tool for finding clusters and determining photometric redshifts for clusters using two - filter imaging data .
we demonstrate the high accuracy of photometric redshifts attainable using the cmr by applying the technique to a sample of 45 abell clusters with known redshifts .
furthermore , by using the red sequence of galaxies from the cmr , we have developed an extremely efficient technique for detecting clusters and groups of galaxies .
we demonstrate this using both observed photometric catalogs with redshifts ( from the cnoc2 survey ) and simulations .
|
understanding the interaction among nucleons is one of the most important issues in particle and nuclear physics .
in particular , the properties of the three nucleon force ( tnf ) are the forefront of not only particle / nuclear physics but also astro physics . for the binding energies of light nuclei , the two nucleon ( 2n ) force alone underestimates the experimental data , and attractive tnf components are required @xcite . on the other hand , repulsive tnf components are also necessary in both of symmetric and asymmetric nuclear matters , in order to reproduce the empirical saturation density of nuclei in the former case , and to make the equation of state of asymmetric nuclear matter hard enough to sustain the observed massive neutron star masses in the latter case @xcite .
recently , it is argued that tnf is responsible for the anomaly in the drip line and the nontrivial magic number of neutron - rich nuclei @xcite , and could impact on the supernova nucleosynthesis .
pioneered by fujita - miyazawa @xcite , tnf have been commonly studied from the two - pion exchange models with the @xmath4-excitation . however , since tnf is originated by the fact that a nucleon is not a fundamental particle , it is desirable to study tnf from the fundamental degrees of freedom ( dof ) , i.e. , quarks and gluons . in this report , we carry out first - principle calculations of tnf using the lattice qcd technique . in the lattice qcd calculation of nuclear forces , it is recently proposed @xcite to use the nambu - bethe - salpeter ( nbs ) wave function so that the potential is faithful to the phase shift by construction .
the obtained ( parity - even ) 2n potentials are found to have desirable features , such as attractive well at long and medium distances , and the central repulsive core at short distance @xcite .
it is also examined that the non - locality of the obtained potential is small , and the leading terms in the derivative expansion dominate the potential at low energies @xcite .
the method has been successfully extended to the hyperon - nucleon ( yn ) and hyperon - hyperon ( yy ) interactions @xcite including h - dibaryon system @xcite .
further applications are given in refs .
@xcite . in this report
, we discuss the extension of this approach to the three nucleon ( 3n ) system , and lattice qcd calculation of tnf in the triton channel is carried out . a preliminary account of this study is also given in ref .
since the detailed formulation for the 2n force is given in ref .
@xcite , we discuss the extension to the 3n system .
we consider the equal - time nbs wave function @xmath5 , which can be obtained through the six point correlator , @xmath6 where @xmath7 , @xmath8 are the jacobi coordinates , and @xmath9 denotes either of @xmath10 or @xmath11 . with the derivative expansion of the potentials , the nbs wave function can be converted to the potentials through the following schrdinger equation , @xmath12 \psi(\vec{r ' } , \vec{\rho } ) = e \psi(\vec{r ' } , \vec{\rho } ) , \label{eq : sch_3n } \ ] ] where @xmath13 with @xmath14 denotes the potential between @xmath15-pair , @xmath16 the tnf , @xmath17 , @xmath18 the reduced masses .
if we calculate @xmath19 for all @xmath20 and @xmath21 , and if all @xmath13 are available by ( separate ) lattice calculations for the genuine 2n system , we can extract @xmath16 through eq .
( [ eq : sch_3n ] ) . in practice , however , the calculation cost of @xmath19 is prohibitively expensive .
note that the number of diagrams generated by wick contractions tends to diverge with a factor of @xmath22 ( @xmath23 are numbers of u , d quarks in the system ) .
we develop several techniques to reduce the calculation cost , e.g. , we take advantage of symmetries ( such as isospin symmetry ) , and employ the non - relativistic limit for the source nucleon operator .
similar techniques are ( independently ) developed in the calculations of binding energies of triton / helium nuclei @xcite . yet
, the calculation cost remains quite expensive , so we consider eq .
( [ eq : sch_3n ] ) with fixed three - dimensional ( 3d ) coordinates of @xmath20 and @xmath21 . to proceed , we find that the linear setup with @xmath24 is advantageous , where three nucleons are aligned linearly with equal spacings of @xmath25 .
in fact , because of @xmath24 , the third nucleon is attached to @xmath26-nucleon pair with only s - wave .
considering the total 3n quantum numbers of @xmath27 ( triton channel ) , the wave function can be completely spanned by only three bases , which can be labeled by the quantum numbers of @xmath26-pair as @xmath28 , @xmath29 , @xmath30 .
therefore , the schrdinger equation leads to the @xmath31 coupled channel equations with the bases of @xmath32 , @xmath33 , @xmath34 .
the reduction of the dimension of bases is expected to improve the s / n as well .
even when we calculate with the fixed 3d - configuration ( with the linear setup or whatever ) , the identification of genuine tnf remains nontrivial .
this is because the parity - odd 2n potentials are not available in lattice qcd at this moment , and we can not subtract the contribution of @xmath35 unambiguously . note that although the total parity of the 3n system can be projected , a 2n - pair inside the 3n system could be either of positive or negative parity .
the familiar procedure of partial wave expansion can not be performed here , since we calculate only fixed 3d - configurations to reduce the calculation cost . to overcome this problem , we find that the following channel in the triton is useful , @xmath36 .
\label{eq : psi_s } \ ] ] this wave function itself has been well known , but the point here is that it is anti - symmetric in spin / isospin spaces for any 2n - pair .
combined with the pauli - principle , it is automatically guaranteed that any 2n - pair couples with even parity only .
therefore , we can extract tnf unambiguously in this channel , without the information of parity - odd 2n potentials .
note that no assumption on the choice of 3d - configuration is imposed in this argument , and we can take advantage of this feature in future tnf calculations with 3d - configurations other than the linear setup .
we employ @xmath0 dynamical clover fermion configurations generated by cp - pacs collaboration @xcite .
we use 598 configurations at @xmath2 with the lattice size of @xmath37 , which corresponds to @xmath38 box in physical spacial size with the lattice spacing of @xmath39 .
we calculate at the hopping parameter of @xmath40 , @xmath41 quarks @xmath42 , which corresponds to @xmath3 gev , @xmath43 gev .
we use the wall quark source with coulomb gauge fixing . in order to enhance the statistics
, we perform the calculation for sources on 16 time slices for each configuration .
we perform the calculation with the linear setup at ten physical points of the distance @xmath44 .
since we have only one channel ( eq . ( [ eq : psi_s ] ) ) which is free from parity - odd 2n potentials , we can determine one type of tnf .
in this report , we consider the scalar / isoscalar type tnf . in the urbana ix model @xcite , which is an often - used phenomenological tnf ,
tnf consists of two parts , one is the two - pion exchange tnf and the other is phenomenologically introduced scalar / isoscalar repulsive tnf . since the two - pion exchange tnf is expected to be suppressed in the current lattice setup of @xmath3 gev , it is reasonable to assume the scalar / isoscalar nature for tnf here . in fig .
[ fig:3n ] ( left ) , we plot each wave function of @xmath45 , @xmath46 and @xmath34 in the triton channel at @xmath47 .
we observe that @xmath48 dominates the wave function .
this is because @xmath48 contains the component for which all three nucleons are in s - wave . by solving the @xmath31
coupled channels and subtracting the @xmath35 from the total potentials , we determine tnf . in fig .
[ fig:3n ] ( right ) , we plot the results for the scalar / isoscalar tnf . here
, the @xmath44-independent shift by energies is not included , and thus about @xmath49 mev systematic error is understood . at the long distance region of @xmath44 , tnf is small as is expected . at the short distance region , we observe the indication of repulsive tnf . recalling that the repulsive short - range tnf is phenomenologically required in high density matter , this is very encouraging result .
of course , we note that further researches are necessary , e.g. , the examination of the discretization error and the dependence on the quark masses . studies along this line is currently underway .
( color online ) .
( left ) the wave function with linear setup in the triton channel .
red , blue , brown points correspond to @xmath48 , @xmath50 , @xmath34 , respectively .
( right ) the scalar / isoscalar tnf in the triton channel , plotted against the distance @xmath44 in the linear setup .
, scaledwidth=80.0% ] ( color online ) .
( left ) the wave function with linear setup in the triton channel .
red , blue , brown points correspond to @xmath48 , @xmath50 , @xmath34 , respectively .
( right ) the scalar / isoscalar tnf in the triton channel , plotted against the distance @xmath44 in the linear setup .
, scaledwidth=80.0% ] we thank authors and maintainers of cps++@xcite , and cp - pacs collaboration and ildg / jldg @xcite for providing gauge configurations .
the numerical calculations have been performed on t2k at university of tsukuba ( 09a-11 , 10a-19 ) and blue
gene / l at kek ( 09 - 23 , 09/10 - 24 ) .
this research is supported in part by mext grant - in - aid , scientific research on innovative areas ( 20105001 , 20105003 , 21105515 ) .
t. d. is supported in part by grant - in - aid for jsps fellows 21@xmath515985 .
a. akmal , v. r. pandharipande and d. g. ravenhall , phys .
rev . * c58 * , 1804 ( 1998 ) . t. otsuka , t. suzuki , j. d. holt , a. schwenk and y. akaishi , phys . rev
. lett .
* 105 * , 032501 ( 2010 )
. j. d. holt , t. otsuka , a. schwenk and t. suzuki , [ arxiv:1009.5984 [ nucl - th ] ] .
n. ishii , s. aoki and t. hatsuda , phys .
lett . * 99 * , 022001 ( 2007 ) .
s. aoki , t. hatsuda and n. ishii , prog .
* 123 * , 89 ( 2010 ) .
k. murano , n. ishii , s. aoki and t. hatsuda , prog .
theor . phys . in press , [ arxiv:1103.0619 [ hep - lat ] ] .
h. nemura , n. ishii , s. aoki and t. hatsuda , phys .
lett . * b673 * , 136 ( 2009 ) . t. inoue _ et al . _
[ hal qcd collab .
] , prog .
phys . * 124 * , 591 ( 2010 ) .
k. sasaki [ hal qcd collab .
] , pos * lattice2010 * , 157 ( 2010 ) , [ arxiv:1012.5685 [ hep - lat ] ] . t. t. takahashi and y. kanada - enyo , phys . rev . *
d82 * , 094506 ( 2010 ) .
t. kawanai and s. sasaki , phys .
* d82 * , 091501 ( 2010 ) .
t. doi [ hal qcd collab .
] , pos * lattice2010 * , 136 ( 2010 ) , [ arxiv:1011.0657 [ hep - lat ] ] .
t. yamazaki , y. kuramashi and a. ukawa , [ pacs - cs collab .
] , phys .
rev . * d81 * , 111504 ( 2010 ) . a. ali khan _ et al . _
[ cp - pacs collab .
] , phys .
d * 65 * , 054505 ( 2002 ) [ e : d * 67 * , 059901 ( 2003 ) ] .
b. s. pudliner , v. r. pandharipande , j. carlson and r. b. wiringa , phys .
* 74 * , 4396 ( 1995 ) .
|
we study the three nucleon force in @xmath0 dynamical clover fermion lattice qcd , utilizing the nambu - bethe - salpeter wave function of the three nucleon system .
since parity - odd two nucleon potentials are not available in lattice qcd at this moment , we develop a new formulation to extract the genuine three nucleon force which requires only the information of parity - even two nucleon potentials . in order to handle the extremely expensive calculation cost
, we consider a specific three - dimensional coordinate configuration for the three nucleons .
we find that the linear setup is advantageous , where nucleons are aligned linearly with equal spacings .
the lattice calculation is performed with @xmath1 configurations at @xmath2 , @xmath3gev generated by cp - pacs collaboration , and the result of the three nucleon force in triton channel is presented .
address= graduate school of pure and applied sciences , university of tsukuba , tsukuba 305 - 8571 , japan
|
excitable media denotes a class of systems that share a set of features which make their dynamical behavior qualitatively similar .
these features include ( i ) the existence of two characteristic dynamical states , comprising a stable _ resting state _ and a metastable _ excited state _ , ( ii ) a _ threshold _ value associated with one of the dynamical variables characterising the system , on exceeding which , the system switches from the resting state to the excited state , and ( iii ) a _ recovery period _ following an excitation , during which the response of the system to a supra - threshold stimulus is diminished , if not completely absent @xcite .
natural systems which exhibit such features include , in biology , cells such as neurons , cardiac myocytes and pancreatic beta cells , all of which are vital to the function of a complex living organism .
other examples of dynamical phenomena associated with excitable media include camp waves observed during aggregation of slime mold , calcium waves observed in xenopus oocytes , muscle contractions during childbirth in uterine tissue , chemical waves observed in the belusov - zhabotinsky reaction and concentration patterns in co - oxidation reaction on pt(110 ) surface .
excitation in such systems is observed as the characteristic _ action potential _ , where a variable associated with the system ( e.g. , membrane potential , in the case of biological cells ) increases very fast from its resting value to the peak value corresponding to the excited state , followed by a slower process during which it gradually returns to the resting state .
the simplest model system capable of exhibiting all these features is the generic fitzhugh - nagumo set of coupled differential equations : @xmath0 which , having only two variables , is obviously incapable of exhibiting chaos .
however , when several such sets are coupled together diffusively to simulate a spatially extended media ( e.g. , a piece of biological tissue made up of a large number of cells ) , the resulting high - dimensional dynamical system can display chaotic behavior .
the genesis of this _ spatiotemporal chaos _ lies in the distinct property of interacting waves in excitable media , which mutually annihilate on colliding .
this is a result of the fact that an excitation wavefront is followed by a region whose cells are all in the recovery period , and which , therefore , can not be stimulated by another excitation wavefront , as for example when two waves cross each other @xcite .
interaction between such waves result in the creation of spatial patterns , referred to variously as _ reentrant excitations _ ( in 1d ) , _ vortices _ or _ spiral waves _ ( in 2d ) and _ scroll waves _ ( in 3d ) , which form when an excitation wavefront is broken as the wave propagates across partially recovered tissue or encounters an inexcitable obstacle @xcite .
the free ends of the wavefront gradually curl around to form spiral waves .
once formed , such waves become self - sustained sources of high - frequency excitation in the medium , and usually can only be terminated through external intervention .
the existence of nonlinear properties of wave propagation in several excitable media can lead to complex non - chaotic spatiotemporal rhythms .
thus , spiral waves are associated with periodic as well as quasiperiodic patterns of temporal activity . however , in this paper , we shall not be discussing the many schemes proposed to terminate single spiral waves , but instead , focus on the control of spatiotemporally chaotic patterns seen in excitable media ( in 2 or 3 dimensions ) , that occur when under certain conditions , spiral or scroll waves become unstable and break up . various mechanisms of such breakup have been identified @xcite , including meandering of the spiral focus .
if the meandering is sufficiently high , the spiral wave can collide with itself and break up spontaneously , resulting in the creation of multiple smaller spirals ( fig .
[ ss : fig0 ] ) .
the process continues until the spatial extent of the system is spanned by several coexisting spiral waves that activate different regions without any degree of coherence .
this state of _ spiral turbulence _ marks the onset of spatiotemporal chaos , as indicated by the lyapunov spectrum and kaplan - yorke dimension @xcite . .
the initial condition is a broken plane wave that is allowed to curl around into a spiral wave ( left ) .
meandering of the spiral focus causes wavebreaks to occur ( centre ) that eventually result in spiral turbulence , with multiple independent sources of high - frequency excitation ( right ) . ]
controlling spatiotemporal chaos in excitable media has certain special features .
unlike other chaotic systems , here the response to a control signal is not proportional to the signal strength because of the existence of a threshold . as a result ,
an excitable system shows discontinuous response to control .
for instance , regions that have not yet recovered from a previous excitation or where the control signal is below the threshold , will not be affected by the control stimulus at all .
also , the focus of control in excitable media is to eliminate all activity rather than to stabilize unstable periodic behavior .
this is because the problem of chaos termination has great practical importance in the clinical context , as the spatiotemporally chaotic state has been associated with the cardiac problem of ventricular fibrillation ( vf ) .
vf involves incoherent activation of the heart that results in the cessation of pumping of blood , and is fatal within minutes in the absence of external intervention . at present , the only effective treatment is electrical defibrillation , which involves applying very strong electrical shocks across the heart muscles , either externally using a defibrillator or internally through implanted devices .
the principle of operation for such devices is to overwhelm the natural cardiac dynamics , so as to drive all the different regions of the heart to rest simultaneously , at which time the cardiac pacemaker can take over once again .
although the exact mechanism by which this is achieved is still not completely understood , the danger of using such large amplitude control ( involving @xmath1 externally and @xmath2 internally ) is that , not only is it excruciatingly painful to the patient , but by causing damage to portions of cardiac tissue which subsequently result in scars , it can potentially increase the likelihood of future arrhythmias .
( i.e. , abnormalities in the heart s natural rhythm ) .
therefore , devising a low - power control method for spatiotemporal chaos in excitable media promises a safer treatment for people at risk from potentially fatal cardiac arrhythmias . in this paper , we have discussed most of the recent control methods that have been proposed for terminating spatiotemporal chaos in excitable media @xcite .
these methods are also often applicable to the related class of systems known as oscillatory media , described by complex landau - ginzburg equation @xcite , which also exhibit spiral waves and spatiotemporal chaos through spiral breakup .
we have broadly classified all control schemes into three types , depending on the nature of application of the control signal . if every region of the media is subjected to the signal ( which , in general , can differ from region to region ) it is termed as _ global control _ ; on the other hand , if the control signal is applied only at a small , localised region from which its effects spread throughout the media , this is called _
local control_. between these two extremes lie control schemes where perturbations are applied simultaneously to a number of spatially distant regions .
we have termed these methods as _ non - global , spatially extended control_. while global control may be the easiest to understand , involving as it does the principle of synchronizing the activity of all regions , it is also the most difficult to implement in any practical situation . on the other hand , local control will be the easiest to implement ( requiring a single control point ) but hardest to achieve . in the next section
we describe a few of the more commonly used models for studying control of spatiotemporal chaos in excitable media .
section 3 discusses proposed methods of global control , while section 4 discusses other spatially extended schemes .
the next section deals with local control methods , and we conclude with a brief section containing general discussions about chaos control and its implications .
( bottom ) the result of applying a positive ( `` + '' ) or negative ( `` @xmath3 '' ) additive perturbation of the same duration to the @xmath4 variable : `` + '' control decreases the threshold and makes excitation more likely , while `` @xmath3 '' control decreases the duration of the action potential and allows the system to recover faster . for the duration of the control signal , the @xmath4-nullcline shifts upward ( downward ) for positive ( negative ) perturbation as indicated by the dashed ( dash - dotted ) curve.,title="fig : " ] .
( bottom ) the result of applying a positive ( `` + '' ) or negative ( `` @xmath3 '' ) additive perturbation of the same duration to the @xmath4 variable : `` + '' control decreases the threshold and makes excitation more likely , while `` @xmath3 '' control decreases the duration of the action potential and allows the system to recover faster . for the duration of the control signal ,
the @xmath4-nullcline shifts upward ( downward ) for positive ( negative ) perturbation as indicated by the dashed ( dash - dotted ) curve.,title="fig : " ] the generic fitzhugh - nagumo model for excitable media ( eq . [ fhneq ] ) exhibits a structure that is common to most models used in the papers discussed here .
typically , the dynamics is described by a fast variable , @xmath5 , and a slow variable , @xmath6 , the ratio of timescales being given by @xmath7 . the resulting phase space behavior is shown in fig .
[ ss : fig1a ] ( left ) . for biological cells
, the fast variable is often associated with the transmembrane potential , while the slow ( recovery ) variable represents an effective membrane conductance that replaces the complexity of several different types of ion channels . for the spatially extended system ,
the fast variable of neighboring cells are coupled diffusively .
there are several models belonging to this general class of excitable media which display breakup of spiral waves ( in 2d ) and scroll waves ( in 3d ) , including the one proposed by panfilov @xcite @xmath8 here , @xmath9 is the function specifying the initiation of the action potential and is piecewise linear : @xmath10 , for @xmath11 , @xmath12 , for @xmath13 , and @xmath14 , for @xmath15 .
the physically appropriate parameters given in ref .
@xcite are @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 .
the function @xmath23 determines the time scale for the dynamics of the recovery variable : @xmath24 for @xmath25 , @xmath26 for @xmath15 , and @xmath27 for @xmath28 and @xmath29 with @xmath30 , @xmath31 , @xmath32 , and @xmath33 simpler variants that also display spiral wave breakup in 2d include ( i ) the barkley model @xcite : @xmath34 the appropriate parameter values being given in ref .
@xcite , and ( ii ) the br - eiswirth model @xcite , which differs from ( [ barkleyeq ] ) only in having @xmath35 , the functional form of @xmath9 and parameter values being as in ref .
the aliev - panfilov model @xcite is a modified form of the panfilov model , that takes into account nonlinear effects such as the dependence of the action potential duration on the distance of the wavefront to the preceding waveback .
it has been used for control in refs .
@xcite . all the preceding models tend to disregard several complex features of actual biological cells , e.g. , the different types of ion channels that allow passage of electrically charged ions across the cellular membrane .
there exists a class of models inspired by the hodgkin - huxley formulation describing action potential generation in the squid giant axon , which explicitly takes such details into account .
while the simple models described above do reproduce generic features of several excitable media seen in nature , the more realistic models describe many properties of specific systems , e.g. , cardiac tissue .
the general form of such models are described by a partial differential equation for the transmembrane potential @xmath36 , @xmath37 where @xmath38 is the membrane capacitance density and @xmath39 is the diffusion constant , which , if the medium is isotropic , is a scalar .
@xmath40 is the instantaneous total ionic - current density , and different realistic models essentially differ in its formulation . for example , in the luo - rudy i model @xcite of guinea pig ventricular cells , @xmath40 is assumed to be composed of six different ionic current densities , which are themselves determined by several time - dependent ion - channel gating variables whose time - evolution is governed by ordinary differential equations of the form : @xmath41 here , @xmath42 is the steady state value of @xmath43 and @xmath44 is its time constant . the voltage - dependent rate constants , @xmath45 and @xmath46 , are complicated functions of @xmath36 obtained by fitting experimental data .
the first attempt at controlling chaotic activity in excitable media dates back almost to the beginning of the field of chaos control itself , when proportional perturbation feedback ( ppf ) control was used to stabilize cardiac arrhythmia in a piece of tissue from rabbit heart @xcite . this method applied small electrical stimuli , at intervals calculated using a feedback protocol , to stabilize an unstable periodic rhythm . unlike in the original proposal for controlling chaos @xcite , where the location of the stable manifold of the desired unstable periodic orbit ( upo ) was moved using small perturbations , in the ppf method it is the state of the system that is moved onto the stable manifold .
however , it has been later pointed out that ppf does not necessarily require the existence of upos ( and , by extension , deterministic chaos ) and can be used even in systems with stochastic dynamics @xcite .
later , ppf method was used to control atrial fibrillation in human heart @xcite .
however , the effectiveness of such control in suppressing spatiotemporal chaos , when applied only at a local region , has been questioned , especially as other experimental attempts in feedback control have not been able to terminate fibrillation by applying control stimuli at a single spatial location @xcite .
more successful , at least in numerical simulations , have been schemes where control stimuli is applied throughout the system . such global control schemes either apply small perturbations to the dynamical variables ( @xmath4 or @xmath47 ) or one of the parameters ( usually the excitation threshold ) .
the general scheme involves introducing an external control signal @xmath48 into the model equations , e.g. , in the panfilov model [ eq . ( [ panfeq ] ) ] : @xmath49 for a control duration @xmath50 .
if @xmath48 is a small , positive perturbation , added to the fast variable , the result is an effective reduction of the threshold ( fig .
[ ss : fig1a ] , bottom ) , thereby making simultaneous excitation of different regions more likely . in general , @xmath48 can be periodic , consisting of a sequence of pulses .
[ ss : fig1 ] shows the results of applying a pulse of fixed amplitude but varying durations . while in general , increasing the amplitude , or the duration , increases the likelihood of suppressing spatiotemporal chaos , it is not a simple , monotonic relationship .
depending on the initial state at which the control signal is applied , even a high amplitude ( or long duration ) control signal may not be able to uniformly excite all regions simultaneously . as a result , when the control signal is withdrawn , the inhomogeneous activation results in a few regions becoming active again and restarting the spatiotemporal chaotic behavior .
starting from a spatiotemporally chaotic state ( top left ) .
pseudo - gray - scale plots of excitability @xmath4 show the result of applying a pulse of amplitude @xmath51 between @xmath52 11 ms and 27.5 ms ( top centre ) that eventually leads to elimination of all activity ( top right ) . applying the pulse between @xmath52 11 ms and
33 ms ( bottom left ) results in some regions becoming active again after the control pulse ends ( bottom centre ) eventually reinitiating spiral waves ( bottom right ) . ]
most global control schemes are variations or modifications of the above scheme .
osipov and collins @xcite have shown that a low - amplitude signal used to change the value of the slow variable at the front and back of an excitation wave can result in different wavefront and waveback velocities which destabilizes the traveling wave , eventually terminating all activity , and , hence , spatiotemporal chaos .
gray @xcite has investigated the termination of spiral wave breakup by using both short and long - duration pulses applied on the fast variable , in 2d and 3d systems .
this study concluded that while short duration pulses affected only the fast variable , long duration pulses affected both fast and slow variables and that the latter is more efficient , i.e. , uses less power , in terminating spatiotemporal chaos .
the external control signal can also be periodic [ @xmath53 , in which case the critical amplitude @xmath54 required for terminating activity has been found to be a function of the signal frequency @xmath55 @xcite .
other schemes have proposed applying perturbations to the parameter controlling the excitation threshold , @xmath56 . applying a control pulse on this parameter ( @xmath57 , during duration of control pulse;@xmath58 , otherwise ) has been shown to cause splitting of an excitation wave into a pair of forward and backward moving waves @xcite .
splitting of a spiral wave causes the two newly created spirals to annihilate each other on collision . for a spatiotemporally chaotic state ,
a sequence of such pulses may cause termination of all excitation , there being an optimal time interval between pulses that results in fastest control .
another control scheme that also applies perturbation to the threshold parameter is the uniform periodic forcing method suggested by alonso _
et al _ @xcite for controlling scroll wave turbulence in three - dimensional excitable media . such turbulence results from negative tension between scroll wave filaments , i.e. , the line joining the phase singularities about which the scroll wave rotates . in this control method ,
the threshold is varied in periodic manner [ @xmath59 and the result depends on the relation between the control frequency @xmath55 and the spiral rotation frequency .
if the former is higher than the latter , sufficiently strong forcing is seen to eliminate turbulence ; otherwise , turbulence suppression is not achieved .
the mechanism underlying termination has been suggested to be the effective increase of filament tension due to rapid forcing , such that , the originally negative tension between scroll wave filaments is changed to positive tension .
this results in expanding scroll wave filaments to instead shrink and collapse , eliminating spatiotemporal chaotic activity . in a variant method
, the threshold parameter has been perturbed by spatially uncorrelated gaussian noise , rather than a periodic signal , which also results in suppression of scroll wave turbulence @xcite . as already mentioned , global control , although easy to understand , is difficult to achieve in experimental systems .
a few cases in which such control could be implemented include the case of eliminating spiral wave patterns in populations of the dictyostelium amoebae by spraying a fine mist of camp onto the agar surface over which the amoebae cells grow @xcite .
another experimental system where global control has been implemented is the photosensitive belusov - zhabotinsky reaction , where a light pulse shining over the entire system is used as a control signal @xcite .
indeed , conventional defibrillation can be thought of as a kind of global control , where a large amplitude control signal is used to synchronize the phase of activity at all points by either exciting a previously unexcited region ( advancing the phase ) or slowing the recovery of an already excited region ( delaying the phase ) @xcite .
the control methods discussed so far apply control signal to all points in the system . as the chaotic activity
is spatially extended , one may naively expect that any control scheme also has to be global .
however , we will now discuss some schemes that , while being spatially extended , do not require the application of control stimuli at all points of the system .
the control method of sinha _ et al _ @xcite involving suprathreshold stimulation along a grid of points is based on the observation that spatiotemporal chaos in excitable media is a long - lived transient that lasts long enough to establish a non - equilibrium statistical steady state displaying spiral turbulence .
the lifetime of this transient , @xmath60 , increases rapidly with linear size of the system , @xmath61 , e.g. , increasing from 850 ms to 3200 ms as @xmath61 increases from 100 to 128 in the two - dimensional panfilov model .
this accords with the well - known observation that small mammals do not get life - threatening vf spontaneously whereas large mammals do @xcite and has been experimentally verified by trying to initiate vf in swine ventricular tissue while gradually reducing its mass @xcite .
a related observation is that non - conducting boundaries tend to absorb spiral excitations , which results in spiral waves not lasting for appreciable periods in small systems .
the essential idea of the control scheme is that a domain can be divided into electrically disconnected regions by creating boundaries composed of recovering cells between them .
these boundaries can be created by triggering excitation across a thin strip . for two - dimensional media , the simulation domain ( of size @xmath62 )
is divided into @xmath63 smaller blocks by a network of lines with the block size ( @xmath64 ) small enough so that spiral waves can not form . for control in a 3d system ,
the mesh is used only on one of the faces of the simulation box .
control is achieved by applying a suprathreshold stimulation via the mesh for a duration @xmath50 .
a network of excited and subsequently recovering cells then divides the simulation domain into square blocks whose length in each direction is fixed at a constant value @xmath65 for the duration of control .
the network effectively simulates non - conducting boundary conditions ( for the block bounded by the mesh ) for the duration of its recovery period , in so far as it absorbs spirals formed inside this block .
note that @xmath50 need not be large at all because the individual blocks into which the mesh divides the system ( of linear size @xmath65 ) are so small that they do not sustain long spatiotemporally chaotic transients . nor does @xmath66 , which is related to the mesh density , have to be very large since the transient lifetime , @xmath67 , decreases rapidly with decreasing @xmath61 .
the method has been applied to multiple excitable models , including the panfilov and luo - rudy models ( fig .
[ ss : fig3 ] ) .
an alternative method @xcite for controlling spiral turbulence that also uses a grid of control points has been demonstrated for the aliev - panfilov model .
two layers of excitable media are considered , where the first layer represents the two - dimensional excitable media exhibiting spatiotemporal chaos that is to be controlled , and the second layer is a grid structure also made up of excitable media .
the two layers are coupled using the fast variable but with asymmetric coupling constants , with excitation pulses travelling @xmath68 times faster in the second layer compared to the first . as the second layer consists only of grid lines , it is incapable of exhibiting chaotic behavior in the uncoupled state .
if the coupling from the second layer to the first layer is sufficiently stronger than the other way round , the stable dynamics of the second layer ( manifested as a single rotating spiral ) overcomes the spiral chaos in the first layer , and drives it to an ordered state characterized by mutually synchronized spiral waves .
an alternative method of spatially extended control is to apply perturbations at a series of points arranged in a regular array .
et al _ @xcite had proposed using such an arrangement for applying a time - delayed feedback control scheme . however
, this scheme does not control spatiotemporal chaos and is outside the scope of this review . ) using an array of control points with spacing @xmath69 and strength of control stimulus @xmath70 .
stimulation is started at the top left corner ( @xmath71 ms ) and lasts at each control point , as the wave reaches that point , for 17.9 ms . by 200 ms , the spatiotemporal chaos has disappeared . ] more recently , the authors @xcite have used an array of control points to terminate spatiotemporal chaos in the panfilov model .
[ ss : fig4 ] shows the result of applying a spatially non - uniform control scheme , which simulates an excitation wave traveling over the system , with the same wavefront velocity as in the actual medium .
the control points are placed distance @xmath72 apart along a regular array . at certain times , the control points at one corner of the system is stimulated , followed by the successive stimulation of the neighboring control points , such that a wave of stimulation is seen to move radially away from the site of original stimulation .
this process is repeated after suitable intervals .
note that , simulating a traveling wave using the array is found to be more effective at controlling spatiotemporal chaos than the simultaneous activation of all control points .
using a traveling wave allows the control signal to engage all high - frequency sources of excitation in the spiral turbulence regime , ultimately resulting in complete elimination of chaos .
if , however , the control had only been applied locally the resulting wave could only have interacted with neighboring spiral waves and the effects of such control would not have been felt throughout the system .
the efficacy of the control scheme depends upon the spacing between the control points , as well as the number of simulated traveling waves .
traveling waves have previously been used in ref .
@xcite to control spatiotemporal chaos , although in the global control context with a spatiotemporally periodic signal being applied continuously for a certain duration , over the entire system .
we now turn to the possibility of controlling spatiotemporal chaos by applying control at only a small localized region of the spatially extended system .
virtually all the proposed local control methods use _ overdrive pacing _ , generating a series of waves with frequency higher than any of the existing excitations in the spiral turbulent state . as low - frequency activity
is progressively invaded by faster excitation , the waves generated by the control stimulation gradually sweep the chaotic activity to the system boundary where they are absorbed .
although we can not speak of a single frequency source in the case of chaos , the relevant timescale is that of the spiral waves and is related to the recovery period of the medium .
control is manifested as a gradually growing region in which the waves generated by the control signal dominate , until the region expands to encompass the entire system .
the time required to achieve termination depends on the frequency difference between the control stimulation and that of the chaotic activity , with control being achieved faster when this difference is greater .
stamp _ et al _ @xcite has looked at the possibility of using low - amplitude , high - frequency pacing using a series of pulses to terminate spiral turbulence .
however , using a series of pulses ( having various waveform shapes ) has met with only limited success in suppressing spatiotemporal chaos . by contrast , a periodic stimulation protocol @xcite has successfully controlled chaos in the 2d panfilov model , as well as other models @xcite .
the key mechanism underlying such control is the periodic alternation between positive and negative stimulation .
a more general control scheme proposed in ref .
@xcite uses _ biphasic pacing _ , i.e. , applying a series of positive and negative pulses , that shortens the recovery period around the region of control stimulation , and thus allows the generation of very high - frequency waves than would have been possible using positive stimulation alone .
a simple argument shows why a negative rectangular pulse decreases the recovery period for an excitable system .
the stimulation vertically displaces the @xmath4-nullcline and therefore , the maximum value of @xmath47 that can be attained is reduced .
consequently , the system will recover faster from the recovery period ( fig .
[ ss : fig1a ] , bottom ) . to understand how negative stimulation affects the response behavior of the spatially extended system
, we can use _ pacing response diagrams _
[ ss : fig5 ] , left ) indicating the relation between the control stimulation frequency @xmath73 and the effective frequency @xmath74 , measured by applying a series of pulses at one site and then recording the number of pulses that reach another site located at a distance without being blocked by a region in the recovery period .
depending on the relative value of @xmath75 and the recovery period , we observe instances of @xmath76 response , i.e. , @xmath77 responses evoked by @xmath78 stimuli . if , for any range of @xmath73 , the corresponding @xmath74 is significantly higher than the effective frequency of spatiotemporal chaos , then termination of spiral turbulence is possible . as shown in ref .
@xcite , there are indeed ranges of stimulation frequencies that give rise to effective frequencies that dominate chaotic activity . as a result , the periodic waves emerging from the stimulation region gradually impose control over the regions exhibiting chaos ( fig .
[ ss : fig5 ] ) .
note that , there is a tradeoff involved here .
if @xmath74 is only slightly higher than the chaos frequency , control takes too long . on the other hand ,
if it is too high the waves suffer conduction block at inhomogeneities produced by chaotic activity which reduces the effective frequency , and therefore , control fails .
recently , another local control scheme has been proposed @xcite that periodically perturbs the model parameter governing the threshold .
in fact , it is the local control analog of the global control scheme proposed by alonso _ et al _ @xcite discussed in section [ ssch : sec3 ] . as in the other methods discussed here
, the local stimulation generates high - frequency waves that propagate into the medium and suppress spiral or scroll waves .
unlike the global control scheme , @xmath79 , so that the threshold can be negative for a part of the time .
this means that the regions in resting state can become spontaneously excited , which allow very high - frequency waves to be generated .
most of the methods proposed for controlling spatiotemporal chaos in excitable media involve applying perturbations either globally or over a spatially extended system of control points covering a significant proportion of the entire system .
however , in most practical situations this may not be a feasible option , either for issues of implementation , or because of the high power for the control signal that such methods would need .
moreover , if one is using such methods in the clinical context , e.g. , terminating fibrillation , a local control scheme has the advantage that it can be readily implemented with existing hardware of the implantable cardioverter defibrillator ( icd ) .
this is a device implanted into patients at high risk from fibrillation that monitors the heart rhythm and applies electrical treatment when necessary through electrodes placed on the heart wall .
a low - energy control method involving icds should therefore aim towards achieving control of spatiotemporal chaos by applying small perturbations from a few local sources . however , the problem with most local control schemes proposed so far is that they use very high - frequency waves to overdrive chaos .
such waves are themselves unstable and may break during propagation , resulting in reinitiation of spiral waves after the original chaotic activity has been terminated .
the problem is compounded by the existence of inhomogeneities in real excitable media .
recently , shajahan _ et al _ @xcite have found complicated dependence of spatiotemporal chaos on the presence of non - conducting regions and other types of inhomogeneities in an excitable system .
such inhomogeneities make the proposed local control schemes more vulnerable , as it is known that high - frequency pacing interacting with , e.g. , non - conducting obstacles , results in wave breaks and subsequent genesis of spatiotemporal chaos @xcite .
the search is still on for a control algorithm for terminating spatiotemporal chaos in excitable media , that can be implemented using low power , or , that need be applied in only a small , local region of the system , and which will yet be robust , capable of terminating spiral turbulence without the control stimulation itself breaking up subsequently . the payoffs for coming up with such a method are enormous , as the potential benefits include an efficient device for cardiac defibrillation .
* acknowledgements : * we would like to thank colleagues and collaborators with whom some of the work described above has been carried out , especially , rahul pandit , ashwin pande , avishek sen , t. k. shajahan , david j. christini , ken m. stein , johannes breuer and antina ghose .
we thank ifcpar and imsc complex systems project ( x plan ) for support .
|
excitable media are a generic class of models used to simulate a wide variety of natural systems including cardiac tissue .
propagation of excitation waves in this medium results in the formation of characteristic patterns such as rotating spiral waves .
instabilities in these structures may lead to spatiotemporal chaos through spiral turbulence , which has been linked to clinically diagnosed conditions such as cardiac fibrillation .
usual methods for controlling such phenomena involve very large amplitude perturbations and have several drawbacks .
there have been several recent attempts to develop low - amplitude control procedures for spatiotemporal chaos in excitable media which are reviewed in this paper .
the control schemes have been broadly classified by us into three types : ( i ) global , ( ii ) non - global spatially - extended and ( iii ) local , depending on the way the control signal is applied , and we discuss the merits and drawbacks for each .
|
compact groups ( hereafter cgs ) have been puzzling astronomers for a number of years . how can a few bright galaxies coexist within less than 100 kpc ?
cgs may have formed early , and have managed to survive the merging instability ( @xcite ; athanassoula , in these proceedings ) or else formed just recently @xcite . alternatively
, cgs may be not be truly dense in 3d , but caused instead by chance alignments of galaxies along the line of sight within larger loose groups @xcite , clusters @xcite and cosmological filaments ( @xcite , hereafter hkw ) . in this contribution
, we provide new light on this debate by studying the group velocity dispersions , x - ray , optical and continuum radio emission , and by predicting the frequency of dense groups as compact as @xcite s ( @xcite , hereafter hcgs ) appear to be , using the @xcite cosmological formalism .
we conclude on the nature of hcgs .
for near spherical virialized systems , mass increases with some power ( near 3 , as is easily shown by combining the virial theorem with a critical mean density for virialization ) of the velocity dispersion of virialized systems .
one therefore expects that _ there must be a minimum velocity dispersion for a virialized galaxy system to be more massive than the sum of the masses of its member galaxies_. systems near full collapse should have even larger velocity dispersions . for a tighter constraint , within a given radius @xmath1 , the sum of the masses of the galaxies _ before they they got close to one another _ must be smaller or equal to the mass within the same radius that the group would have _ once it virializes _ : @xmath2 .
assuming @xcite profiles for groups and halos of galaxies , one can show @xcite that the velocity dispersion @xmath3 of a given dense group must satisfy : @xmath4 where the sums are over the deprojected maximum rotation velocities of spirals and the internal velocity dispersions of ellipticals .
.hickson compact group minimum velocity dispersions [ cols="<,^,^,^,^,^,^,^ " , ] columns ( 2 ) and ( 4 ) : number of spirals and ellipticals used ; columns ( 3 ) and ( 5 ) : minimum group velocity dispersion ( in @xmath5 ) from spirals and from ellipticals ; column ( 6 ) : global minimum group velocity dispersion ( @xmath6 ) , _
@xmath7 ; column ( 7 ) number of accordant redshift galaxies in group ; column ( 8) measured group velocity dispersion ( @xmath6 ) .
although there is still little data on the internal kinematics of hcg galaxies , table 1 above shows that 5 hcgs out of 12 ( hcgs 16 , 40 , 44 , 88 and 89 ) have abnormally low velocity dispersions , and two others ( hcgs 57 and 90 ) have just marginal velocity dispersions .
we expect to measure by chance low velocity dispersions in roughly 20% of dense groups ( those with chance tangential velocity vectors ) .
still , most of the 5 hcgs mentioned above have too low velocity dispersions to be dense systems near virialization .
the simplest alternative is that they are chance alignments within loose groups near turnaround , since at turnaround the velocity dispersion of galaxy systems is expected to be small .
another possibility is that _ tidal friction _ has been effective in slowing down the galaxies , although this is only expected in much more compact groups near full coalescence ( @xcite ; temporin , in these proceedings ) .
the two groups with @xmath8 may simply have little intergalactic matter .
given that the internal kinematics data on hcg galaxies is very incomplete ( only one group has data for all its members , _
@xmath9 ) , some of the groups with high velocity dispersions may turn out to have only marginal velocity dispersions and thus possess little intergalactic matter , while some of the marginal ones may in fact be non - real . in any event , _ there seems to be 3 classes of compact groups , following decreasing velocity dispersion : groups with substantial intergalactic matter , groups will little intergalactic matter and chance alignments within loose groups ( or clusters or cosmological filaments)_. finally one may be tempted to secure more precise velocity dispersions by including the galaxies from the environment of hcgs with the data of @xcite and @xcite , but these spectroscopic surveys show that the velocity dispersion usually increases with inclusion of the environment galaxies , and it is not clear that one is not increasingly affected by interlopers .
we now focus on hcg 16 , the prime , example of a low velocity dispersion , spiral - rich compact group
. the x - ray properties of hcg 16 are controversial and possibly extreme . in their rosat / pspc
x - ray survey of hcgs , @xcite ( hereafter pbeb ) , hcg 16 was the coldest detected group ( @xmath10 ) , and there are no other spiral - only compact groups with diffuse x - ray emission ( mulchaey in these proceedings ; see also pbeb ) .
moreover , whereas diffuse x - rays were clearly detected by pbeb , @xcite failed to detect such diffuse emission at an upper limit 16 times lower , , their quoted upper limit for their luminosities were underestimated by a factor 2 for all undetected groups in their table 4 except hcg 3 . ] whereas only a factor 2.3 is attributable to the wider ( `` bolometric '' ) energy range in which pbeb compute their luminosities .
given the low temperature that pbeb derive for hcg 16 , their derived x - ray luminosity places it two orders of magnitude above their compact group luminosity - temperature relation and roughly a factor of two above the extrapolation of the cluster trend .
it thus seems difficult to reconcile hcg 16 with a low temperature extrapolation of regular x - ray emitting compact groups . @xcite
have re - analyzed the rosat / pspc observations of hcg 16 with the hopes of resolving the discrepancy between @xcite and pbeb and establishing if an irregular morphology is caused by clumpiness or fluctuations in signal - to - noise ratios .
figure [ h16]a below shows the x - ray emission of hcg 16 as contours overlayed on a greyscale map of the group .
the emission beyond the galaxies is indeed significant in a few compact regions : around the galaxies a&b , c and d , as well as a few clumpy regions outside the galaxies denoted c1 ( comprised of 4 sub - clumps ) , c4 and c5 .
figure [ h16]b shows the 20 cm continuum radio contours , measured with the nvss survey , and illustrates _ the similarity between the x - ray and 20 cm continuum radio morphologies of hcg 16_. a closer look @xcite reveals that c1a is connected with a radio - galaxy , which turns out to have a redshift @xcite that clearly places it in the background , c1b is connected with a radio - source , c1c is related to a foreground star , and c5 is connected to a radio - source and to a background group or cluster .
this reduces the x - ray luminosity of the diffuse hot gas connected with hcg 16 by 50% to @xmath11 . in the regions without significant x - ray emission , the upper limits to the counts correspond to at most 1/4 the the space density of hot gas in the detected regions .
hence , _ the hot gas in hcg 16 is clumpy_. if hcg 16 is part a cosmological filament viewed nearly end - on ( hkw ) , one should see this filament in its large - scale environment .
we have searched with ned a roughly cubical region around hcg 16 of @xmath12 size .
hcg 16 is close enough ( @xmath13 ) that ned should be fairly complete around it .
the projected environment of hcg 16 within @xmath14 from the group distance ( fig .
[ envmt]a ) suggests concentration of galaxies along 4 projection angles .
figure [ envmt]b shows the wedge diagrams in each of these position angles , and one can guess a filament at @xmath15 , stopping at hcg 16 and a wide sheet visible at @xmath16 and @xmath17 .
_ there are no filaments closely aligned to the line - of - sight _ as would have been favored by hkw .
although loose groups have generally not yet collapsed , one can apply the @xcite ( @xcite , hereafter , ps ) formalism to these systems , assuming that when they will collapse , their mass is what we infer today ( @xmath18 ) .
we then obtain @xmath19 where @xmath20 is the rate of formation of structures derived by @xcite from the ps formalism .
@xmath21-body simulations suggest that compact groups can not survive ( with at least 4 members ) for over @xmath22 ( @xcite ; athanassoula in these proceedings ) , the fraction increasing with mass .
assume therefore that compact groups must have undergone their cosmological collapse within that time .
one then obtains @xmath23 where @xmath24 is the probability that a dense group of mass @xmath25 exists today given that it collapsed with a mass @xmath26 at time @xmath27 ( given by @xcite ) .
figure [ ndglgps ] shows the resulting loose and dense group mass functions , both normalized to the ps cosmic mass function , for a @xmath28cdm cosmology ( scdm and ocdm with reasonable values for @xmath29 yield roughly similar curves ) .
massive dense groups survive longer simply because they have more galaxies , so the figure should be interpreted by adopting the recent collapse ( _ solid curve _ ) for the lower mass end and the earlier collapse for the high mass end .
the ratio of dense to loose groups then varies with increasing group mass from 12 to 20% .
if compact groups are replenished by infalling galaxies @xcite , they collapse even earlier and we would predict even more compact groups today . in contrast , the estimates of the ratio of hcgs to loose groups range from 0.58% ( @xcite ; see also @xcite ) .
comparison of the hcg sample with a similar compact group sample automatically selected from cosmos / edsgc galaxies ( @xcite ; iovino in these proceedings ) shows that the hcg sample is severely incomplete : inspection of @xcite s fig .
7 suggests that the incompleteness of the hcg sample is a factor of 3 at the bright - end and increasingly worse at fainter magnitudes .
now if dense groups represent only a few percent of the cosmic mass function , where are the remaining virialized cosmic structures with masses near @xmath30 ?
the answer is simple : _ the remaining structures collapsed too early to be visible as compact groups today , and must then be in different stages of the final coalescence of groups_. this includes quartets with massive dominant members ( thus failing @xcite s @xcite magnitude concordance criterion : @xmath31 ) , as well as triplets , and binaries .
but since the cosmic multiplicity function decreases with increasing galaxy number , one is led to conclude that most of the missing structures are fully coalesced , single galaxies , which harbor large x - ray halos , as discovered by @xcite ( @xcite , see mulchaey , in these proceedings ) and @xcite . therefore , we predict that _ x - ray over - luminous ellipticals should be more common than compact groups_.
it is very difficult to decide whether the numerous signatures of galaxy interactions in cgs are proofs of their reality or alternatively simply binary interactions within binary - rich chance alignments of galaxies within the line of sight @xcite .
the high fraction of hcgs with diffuse x - ray emission ( pbeb ) should be reduced when only considers groups with _ regular _ diffuse x - rays ( 3.1 ) .
however , the combination of hi ( verdes - montenegro , in these proceedings ) and optical ( mendes de oliveira , in these proceedings ) data on hcgs suggests that most of these are real interacting systems . in view of this beautiful data ,
the arguments spelled out by @xcite against the reality of hcgs must be reappraised . the large predicted cosmic rate of production of dense groups ( 4 above ) implies that the low survival time of dense groups is no longer a good argument against the reality of most compact groups .
moreover , the end products of dense groups isolated bright elliptical galaxies with huge x - ray halos are now being seen @xcite . whereas mergers tend to increase rapidly the difference in first and second ranked magnitudes , the low @xmath32 of hcgs @xcite is simply caused by @xcite s strong bias @xcite against groups with dominant members still satisfying the magnitude concordance criterion .
this bias probably also causes the absence of significant luminosity segregation in comparison with simulated groups @xcite .
the morphology - density relation of hcgs is offset @xcite relative to the cluster / loose group / field morphology - density relation ( measured by @xcite ) : at a given galaxy number density , hcgs have too many spirals .
however , hcgs have too _ few _ spirals given their velocity dispersions in comparison with the cluster morphology - velocity dispersion trend ( pbeb ) .
there happens to be virtually no low velocity dispersion hcgs beyond 10,@xmath33 , therefore the number of hcgs whose low velocity dispersion ( 2 ) suggests they are non - real is limited to roughly 10 out of 69 accordant redshift hcgs with at least 4 members .
therefore , _ there are few arguments left against the reality of hcgs_. however , the samples of automatically defined compact groups ( iovino , in these proceedings ) will be much looser on average ( given the same compactness criterion as originally used by @xcite ) , since the hcg sample is severely incomplete for marginally compact groups @xcite , hence one would expect that _ automatically selected compact groups will be more prone to chance alignments _ ( see @xcite ) .
|
a galaxy system must have a minimum velocity dispersion for its mass to be greater than the sum of the masses of its galaxies .
nearly half of the nearby hickson compact groups ( hcgs ) have too low a velocity dispersion in comparison with the rotational velocities of their spiral galaxies and internal velocity dispersions of their early types . a detailed study of the low velocity dispersion group , hcg 16 the only known group of late - type galaxies with diffuse intergalactic x - ray emitting hot gas
reveals that half of the diffuse x rays are associated with foreground / background sources and the remaining gas is clumpy and mostly associated with the bright galaxies of the group .
the large - scale environment of the group suggests that hcg 16 lies where a cosmological filament falls perpendicularly onto a large - scale sheet .
the observed frequency of compact groups is lower than predicted from the extended press - schechter formalism , which also predicts that most @xmath0 objects in the universe must be fairly old and hence have already coalesced into single objects , reminiscent of elliptical galaxies over - luminous in x - rays that are now being discovered .
thus , the low survival time of dense groups against the merging instability is no longer a worry for compact groups , as they form in large enough numbers .
i show why other arguments against the reality of hcgs no longer hold , partly because of the biases of hickson s sample .
-2 cm _ contribution to iau colloq .
174 on small galaxy groups , held in turku , finland , june 1318 , 1999 , ed .
m. valtonen & c. flynn , asp series _
|
given @xmath0 and @xmath1 , let @xmath2 be a fixed oriented , connected and compact surface of genus @xmath3 with @xmath4 boundary components .
we denote by @xmath5 the group of orientation preserving diffeomorphisms of @xmath2 which restrict to the identity on the boundary . the _ mapping class group _
@xmath6 is defined to be the set of isotopy classes of elements in @xmath7 , where the isotopies are understood to restrict to the identity on the boundary as well .
we refer to ( * ? ? ?
* section 2.1 ) for details .
it is well known that the mapping class group is perfect provided that @xmath8 @xcite ( e.g. , see ( * ? ? ? * theorem 5.1 ) ) and that mapping class groups are finitely presented @xcite ( e.g. , see ( * ? ? ? * section 5.2 ) ) . in this paper
we intend to study an enlargement of the mapping class group , namely the group of homology cobordism classes of homology cylinders .
a _ homology cylinder _ over
@xmath2 is roughly speaking a cobordism between surfaces equipped with a diffeomorphism to @xmath2 such that the cobordism is homologically a product . juxtaposing homology cylinders
gives rise to a monoid structure .
the notion of homology cylinder was first introduced by goussarov @xcite and habiro @xcite ( where it was referred to as a ` homology cobordism ' ) . by considering smooth ( respectively topological ) homology cobordism classes of homology cylinders
we obtain a group @xmath9 ( respectively @xmath10 ) .
these groups were introduced by garoufalidis and levine @xcite , @xcite .
we refer to section [ section : homcyl ] for the precise definitions of homology cylinder and homology cobordism .
henceforth , when a statement holds in both smooth and topological cases , we will drop the decoration in the notation and simply write @xmath11 instead of @xmath10 and @xmath9 .
it follows immediately from the definition that there exists a canonical epimorphism @xmath12 .
a consequence of work of fintushel - stern @xcite , furuta @xcite and freedman @xcite on smooth homology cobordism of homology @xmath13spheres is that this map is not an isomorphism .
in fact , using their results we can see the following : [ thm : topsmooth ] let @xmath14
. then the kernel of the epimorphism @xmath15 contains an abelian group of infinite rank .
if @xmath16 , then there exists in fact a homomorphism @xmath17 onto an abelian group of infinite rank such that the restriction of @xmath18 to the kernel of the projection map @xmath19 is also surjective . an argument of garoufalidis and
levine shows that the canonical map @xmath20 is injective .
( see also proposition [ prop : mgnembed ] . )
it is a natural question which properties of mapping class groups are carried over to @xmath11 . in particular in @xcite goda and sakasai ask whether @xmath21 is a perfect group and garoufalidis and levine ( * ? ? ?
* section 5 , question 9 ) ask whether @xmath21 is infinitely generated ( see also ( * ? ? ?
* problem 11.4 ) ) .
the following theorem answers both questions : [ mainthm ] if @xmath22 , then there exists an epimorphism @xmath23 which splits ( i.e. , there is a right inverse ) . in particular
, the abelianization of @xmath11 contains a direct summand isomorphic to @xmath24 .
note that theorem [ mainthm ] also implies that @xmath11 is not finitely related , since for a finitely related group its abelianization is also finitely related .
also refer to remark [ remark : proof of mainthm ] for a slightly more refined statement .
in many cases we can actually strengthen the result : [ mainthm2 ] if @xmath25 , then there exists an epimorphism @xmath26 furthermore , the abelianization of @xmath11 contains a direct summand isomorphic to @xmath27 .
we remark that for the special case of @xmath28 , this is a consequence of levine s work on knot concordance @xcite since one can easily see that @xmath29 maps onto levine s algebraic knot concordance group .
the general cases of theorem [ mainthm2 ] for @xmath30 and for @xmath31 and @xmath32 are new . in order to prove theorems [ mainthm ] and [ mainthm2 ] we will employ the torsion invariant of a homology cylinder , first introduced by sakasai ( e.g. , see ( * ? ? ?
* section 11.1.2 ) , ( * ? ? ?
* definition 6.5 ) and ( * ? ? ?
* definition 4.4 ) ) . in section [ section : torsion - invariant ] we recall the definition of the torsion of a homology cylinder and we study the behavior of torsion under stacking and homology cobordism .
the result can be summarized as a group homomorphism @xmath33 here @xmath34 and @xmath35 is the multiplicative group of nonzero elements in the quotient field of the group ring @xmath36 $ ] .
loosely speaking , @xmath37 reflects the action of surface automorphisms on @xmath38 , and @xmath39 is the subgroup of `` norms '' in @xmath35 .
( for details , see section [ section : torsion - invariant ] . )
an interesting point is that torsion invariants of homology cylinders may be _ asymmetric _ , in contrast to the symmetry of the alexander polynomial of knots .
indeed , in section [ section : homomorphisms ] , we extract infinitely many @xmath40valued and @xmath41valued homomorphisms of @xmath42 from symmetric and asymmetric irreducible factors , respectively
. theorems [ mainthm ] and [ mainthm2 ] now follow from the explicit construction of examples in section [ section : construction - computation ] .
we remark that theorem [ mainthm2 ] covers all the possibilities of the asymmetric case , since it can be seen that for either @xmath43 or @xmath28 the torsion of a homology cylinder over @xmath44 is always symmetric in an appropriate sense ( see section [ subsection : symmetry - asymmetry ] for details . )
we remark that our main results hold even modulo the mapping class group the essential reason is that our torsion invariant is trivial for homology cylinders associated to mapping cylinders .
more precisely , denoting by @xmath45 the normal subgroup of @xmath11 generated by @xmath6 , the torsion homomorphism @xmath46 actually factors as @xmath47 and theorems [ mainthm ] and [ mainthm2 ] hold for @xmath48 as well as @xmath11 . in section [ section : pretzel ]
, we study examples of homology cylinders which naturally arise from seifert surfaces of pretzel links .
we compute the torsion invariant and prove that these homology cylinders span a @xmath49 summand in the abelianization of @xmath50 . finally in section [ section : torelli ] , we consider the `` torelli subgroup '' @xmath51 of @xmath11 which is the homology cylinder analogue of the torelli subgroup of mapping cylinders .
we prove that the conclusions of theorems [ mainthm ] and [ mainthm2 ] hold for the torelli subgroup @xmath51 , but for a larger group of surfaces .
( see theorem [ thm : cyltorelli ] for details . )
this extends work of morita s ( * ? ? ?
* corollary 5.2 ) to a larger class of surfaces . in this paper ,
manifolds are assumed to be compact , connected , and oriented .
all homology groups are with respect to integral coefficients unless it says explicitly otherwise .
this project was initiated while the three authors visited the mathematics department of indiana university at bloomington in july 2009 .
we wish to express our gratitude for the hospitality and we are especially grateful to chuck livingston and kent orr .
the authors would like to thank irida altman , jake rasmussen , dan silver and susan williams for helpful conversations .
we are very grateful to takuya sakasai for suggesting a stronger version of theorem [ thm : topsmooth ] to us and for informing us of morita s work in @xcite .
the authors thank an anonymous referee for helpful comments , in particular the comment about proposition [ prop : mgnembed ] .
the first author was supported by the national research foundation of korea(nrf ) grant funded by the korea government(mest ) ( no .
2007 - 0054656 and 2009 - 0094069 ) .
the last author was supported by the national research foundation of korea(nrf ) grant funded by the korea government(mest ) ( no .
2009 - 0068877 and 2009 - 0086441 ) .
in this section we recall basic definitions and preliminaries on homology cylinders , the notion of which goes back to goussarov @xcite , habiro @xcite , and garoufalidis and levine @xcite , @xcite .
given @xmath52 and @xmath53 we fix , once and for all , a surface @xmath44 of genus @xmath3 with @xmath4 boundary components .
when @xmath3 and @xmath4 are obvious from the context , we often denote @xmath44 by @xmath54 .
a _ homology cylinder @xmath55 over @xmath54 _ is defined to be a @xmath13manifold @xmath56 together with injections @xmath57 , @xmath58 satisfying the following : 1 .
@xmath57 is orientation preserving and @xmath59 is orientation reversing . 2 .
@xmath60 and @xmath61 .
3 . @xmath62 .
4 . @xmath63 are isomorphisms .
[ example : homology - cylinder ] 1 .
let @xmath64 .
then @xmath65 gives rise to a homology cylinder @xmath66/\mathord{\sim},\ , i_+=\mathord{\id}\times 0 , \
, i_-=\varphi\times 1).\ ] ] where @xmath67 is given by @xmath68 for @xmath69 and @xmath70 $ ] .
if @xmath65 is the identity , then we will refer to the resulting homology cylinder as the _ product homology cylinder_. 2 .
let @xmath71 be a knot of genus @xmath3 such that the alexander polynomial @xmath72 is monic and of degree @xmath73 .
let @xmath54 be a minimal genus seifert surface in the exterior @xmath74 of @xmath71 .
then @xmath74 cut along @xmath54 is a homology cylinder over @xmath75 in a natural way ( e.g. , see ( * ? ? ?
* proposition 3.1 ) or @xcite for details ) .
two homology cylinders @xmath55 and @xmath76 over @xmath77 are called _ isomorphic _ if there exists an orientation - preserving diffeomorphism @xmath78 satisfying @xmath79 .
we denote by @xmath80 the set of all isomorphism classes of homology cylinders over @xmath2 .
a product operation on @xmath81 is given by stacking : @xmath82 this turns @xmath81 into a monoid .
the unit element is given by the product homology cylinder .
two homology cylinders @xmath55 and @xmath76 over @xmath54 are called _ smoothly homology cobordant _ if there exists a compact oriented _ smooth _
@xmath83manifold @xmath84 such that @xmath85 and such that the inclusion induced maps @xmath86 and @xmath87 are isomorphisms .
we denote by @xmath9 the set of smooth homology cobordism classes of elements in @xmath81 .
the monoid structure on @xmath81 descends to a group structure on @xmath9 , where the inverse of a mapping cylinder @xmath55 is given by @xmath88 .
( we refer to @xcite for details ) .
if there is a topological @xmath83manifold @xmath84 satisfying the above conditions , then we say that @xmath55 and @xmath76 are _ topologically homology cobordant_. we denote the resulting group of homology cobordism classes by @xmath10 .
note that there exists a canonical surjection @xmath12 .
we will see in the following sections that this map is in general not an isomorphism . in the original papers of garoufalidis and levine @xcite and in @xcite
the authors focus on the smooth case and denote the group @xmath89 by @xmath11 . in this section
we will discuss three types of examples : 1 .
surface automorphisms , 2 .
homology cobordism classes of integral homology spheres , and 3 .
concordance classes of ( framed ) knots in integral homology spheres .
first recall that @xmath90 gives rise to a homology cylinder @xmath91 .
note that if @xmath92 , then @xmath93 is isomorphic to @xmath94 .
in particular the map @xmath95 descends to a morphism of monoids . proposition [ prop : mgnembed ] below says that we can view the mapping class group @xmath6 as a subgroup of @xmath11 . to prove the proposition , we need the following folklore theorem : [ thm : folklore ] let @xmath96 be a surface with one boundary component , possibly with punctures .
let @xmath97 be a base point .
suppose @xmath98 is a homeomorphism with the following properties : 1 .
@xmath99 is the identity on @xmath100 , 2 .
@xmath101 is the identity , and 3 .
@xmath99 fixes each puncture .
then @xmath99 is isotopic to the identity where the isotopy restricts to the identity on @xmath100 .
the theorem is well - known , and we only give an outline of the proof .
we choose disjoint circles @xmath102 in @xmath96 based at @xmath103 such that @xmath96 cut along the @xmath102 is a punctured disk .
since @xmath104 is the identity , we may assume ( after applying an isotopy whose restriction to @xmath100 is the identity ) that @xmath99 fixes each @xmath102 .
now the map on the punctured disk induced from @xmath99 is isotopic to the identity with @xmath100 fixed pointwise .
this can be shown , for example , using the fact that the map from the pure braid group to @xmath105 is injective .
[ prop : mgnembed ] the map @xmath106 is injective .
the proposition was proved by garoufalidis and levine in the case that @xmath107 and in the smooth category .
( see ( * ? ? ?
* section 2.4 ) and ( * ? ? ?
* section 2.1 ) . ) using their arguments partially , we will show that the proposition holds for any @xmath1 ( and in the topological category as well ) .
first note that if @xmath108 , i.e. , if @xmath2 is a sphere or a disk , then it is well - known that any orientation preserving diffeomorphism is isotopic to the identity .
if @xmath28 , i.e. , if @xmath2 is an annulus , then it is known that @xmath109 and it injects into @xmath11 ( for example , this can be shown using the arguments in the subsection below entitled _ concordance of ( framed ) knots in homology @xmath13spheres _ ) .
therefore we henceforth assume that @xmath110 .
suppose @xmath111 is an orientation - preserving diffeomorphism such that 1 .
@xmath65 restricts to the identity on @xmath112 , and 2 .
@xmath113\in \ker\{\mgn\to \hgn\}$ ] . now fix @xmath114 and fix a base point @xmath103 lying in the @xmath115-th component , say @xmath116 , of @xmath112 .
we write @xmath117 .
we denote by @xmath118 the lower central series of @xmath119 defined inductively by @xmath120 and @xmath121 $ ] , @xmath122 .
the argument of garoufalidis and levine ( which builds on stallings theorem @xcite ) shows that if @xmath91 is homology cobordant to the product homology cylinder , then @xmath123 is the identity map for any @xmath124 .
since @xmath125 is trivial , this implies that @xmath126 is the identity .
we denote by @xmath127 the set of equivalence classes of orientation - preserving diffeomorphisms @xmath128 such that 1 .
@xmath99 is the identity on @xmath116 , and 2 .
@xmath99 fixes each boundary component setwise , where we say that two such maps are equivalent if they are related by an isotopy which fixes @xmath116 pointwise .
then @xmath113=[\mbox{identity}]$ ] in @xmath127 by theorem [ thm : folklore ] .
now consider the map @xmath129 .
it is known that the @xmath4 dehn twists along boundary components of @xmath2 generate a central subgroup isomorphic to @xmath130 in @xmath6 , and @xmath131 is the subgroup generated by the @xmath132 dehn twists along all , but the @xmath115-th boundary component .
( for example , see ( * ? ? ?
* section 4.2 ) . ) therefore , @xmath113\in
\bigcap_k \ker\{\mgn\to \pmkgn\ } = \{[\mbox{identity}]\}$ ] in @xmath6 . note that this shows that @xmath11 is non - abelian provided that @xmath32 or @xmath30 .
it is straightforward to see that in the remaining cases @xmath11 is abelian . in the following
we will see that the cobordism groups of homology cylinders over the surfaces @xmath133 with @xmath134 have been studied under different names for many years .
we first consider the case @xmath135 .
recall that oriented integral homology @xmath13spheres form a monoid under the connected sum operation .
two oriented integral homology @xmath13spheres @xmath136 and @xmath137 are called smoothly ( respectively topologically ) cobordant if there exists a smooth ( respectively topological ) @xmath83manifold cobounding @xmath136 and @xmath138 .
we denote by @xmath139 ( respectively @xmath140 ) the group of smooth ( respectively topological ) cobordism classes of integral homology @xmath13spheres . for @xmath135 the group @xmath141 ( respectively @xmath142 )
is naturally isomorphic to the group @xmath143 ( respectively @xmath140 ) ( e.g. , see@xcite ) .
furuta @xcite and fintushel - stern @xcite showed that @xmath139 has infinite rank .
( see also ( * ? ? ?
* section 7.2 ) . ) on the other hand it follows from the work of freedman @xcite that @xmath144 is the trivial group .
( see also ( * ? ? ?
* corollary 9.3c ) . )
this shows in particular that the homomorphism @xmath145 is not an isomorphism for @xmath135 .
we now turn to the case @xmath31 , i.e. , homology cylinders over the surface @xmath146 which we henceforth identify with the annulus @xmath147 $ ] .
let @xmath148 be an oriented knot in an integral homology @xmath13sphere .
let @xmath56 be the exterior of @xmath71 .
it is not difficult to see there are pairs
@xmath149 of maps @xmath150 satisfying ( 1)(4 ) in the definition of a homology cylinder and satisfying the condition that @xmath151 is a meridian of @xmath71 .
furthermore , the isotopy types of such @xmath152 are in 11 correspondence with framings on @xmath71 .
indeed , the linking number of @xmath71 and the closed curve @xmath153)\cup i_+(*\times[0,1])$ ] gives rise to a canonical 11 correspondence between the set of framings and @xmath41 .
conversely , a homology cylinder @xmath55 over @xmath146 determines an oriented knot endowed with a framing in the integral homology sphere , which is given by attaching a @xmath154handle along @xmath151 and then attaching a @xmath13handle .
we say that @xmath155 and @xmath156 are _ smoothly concordant _ if there exists a smooth cobordism @xmath74 of @xmath136 and @xmath137 such that @xmath157 is an integral homology @xmath158 , s^3\times 0 , s^3\times 1)$ ] and @xmath74 contains a smoothly embedded annulus @xmath159 cobounding @xmath160 and @xmath161 .
the set of smooth concordance classes of knots in integral homology @xmath13spheres form a group @xmath162 under connected sum .
we will now see that we can also think similarly of the concordance group of framed knots in integral homology @xmath13spheres .
note that a concordance @xmath163 as above determines a 11 correspondence between framings on @xmath160 and @xmath161 .
we say two framed knots in integral homology spheres are _ smoothly concordant _ if there is a concordance @xmath163 via which the given framings correspond to each other .
smooth concordance class of framed knots form a group as well , and it is easily seen that this framed analogue of @xmath164 is isomorphic to @xmath165 .
similarly , if we allow topologically locally flat annuli in topological cobordisms we obtain a group @xmath166 and its framed analogue @xmath167 . as usual
we adopt the convention that the group @xmath168 can mean either @xmath164 or @xmath166 .
it follows easily from the definitions that we have an isomorphism @xmath169 .
it follows from the work of levine @xcite that @xmath168 maps onto the algebraic knot concordance group which is isomorphic to @xmath170 , and furthermore the epimorphism from @xmath168 to @xmath171 splits .
this discussion in particular proves the following special case of theorem [ mainthm2 ] : [ mainthm2-special ] there exists a split surjection of @xmath172 onto @xmath173 .
note that the subgroup @xmath174 in @xmath175 is exactly the @xmath41 factor , so that @xmath176 .
in particular , theorem [ mainthm2-special ] holds for @xmath177 as well as @xmath172 .
before we turn to the proof of theorem [ thm : topsmooth ] we introduce a gluing operation on homology cylinders .
let @xmath56 be a homology cylinder over a surface @xmath178 and let @xmath179 be a homology cylinder over a surface @xmath180 .
assume that @xmath181 and fix a boundary component @xmath182 of @xmath96 and fix a boundary component @xmath183 of @xmath184 .
we can now glue @xmath96 and @xmath184 along @xmath182 and @xmath183 using an orientation reversing homeomorphism .
similarly we can glue @xmath56 and @xmath179 along neighborhoods of @xmath185 and @xmath186 .
this gives a homology cylinder over @xmath187 , which we denote by @xmath188 .
we refer to @xmath189 as the _ union of @xmath56 and @xmath179 along @xmath182 and @xmath183_. now let @xmath179 be the product homology cylinder over @xmath184 .
the association @xmath190 gives rise to a monoid homomorphism @xmath191 .
we refer to it as an _ expansion by @xmath184 along @xmath182_. note that the expansion map descends to a group homomorphism @xmath192 .
we are now in a position to prove theorem [ thm : topsmooth ] .
we adopt the convention that @xmath193 stands either for @xmath194 or @xmath140 .
recall that in section [ section : examples ] we saw that we can identify @xmath195 with @xmath193 . also recall that by section [ section : examples ] the group @xmath139 has infinite rank and that @xmath196 is the trivial group . as we saw above , an expansion by a surface of genus @xmath3 with @xmath197 punctures
gives rise to a homomorphism @xmath198 .
we also consider the composition @xmath199 of @xmath4 expansions by a disk along all boundary components of @xmath200 . loosely speaking
, @xmath201 is the homomorphism given by filling in the @xmath4 holes .
there exists a map @xmath202 such that the composition @xmath203 is the identity .
let @xmath204 .
let @xmath205 be a fixed handlebody of genus @xmath3 .
given @xmath206 we write @xmath207 .
since @xmath208 has a heegaard decomposition of genus @xmath3 there exist diffeomorphisms @xmath209 such that @xmath210 .
we write @xmath211 .
we will see in section [ section : actionh1 ] that the action of a homology cylinder @xmath56 over @xmath212 on @xmath213 gives rise to a homomorphism @xmath214 .
we will write @xmath215 .
we now pick a splitting map @xmath216 ,
i.e. a map such that @xmath217 is the identity on @xmath218 .
we pick @xmath219 such that @xmath220 .
note that we can not arrange that @xmath219 is a homomorphism .
now consider the following map @xmath221 note that this map is indeed well - defined , i.e. the right hand side is an integral homology 3sphere .
also note that this map descends to a map @xmath222 .
it is easy to verify that @xmath223 is indeed the identity map .
this concludes the proof of the claim .
before we continue we point out that the map @xmath224 is in general _ not _ a monoid morphism .
we now obtain the following commutative diagram : @xmath225^{=}\ar[d ] & \hh_{0,1}^\smooth\ar[r]^{\ee}\ar[d ] & \hh_{g , n}^\smooth\ar[d]\ar[r]^-{\ff } & \hh_{g,0}^\smooth\ar[d]\ar[r]^-{\gg } & \hh_{0,0}^\smooth\ar[d]\ar[r]^{= } & \theta^{\smooth}_3\ar[d]\\ \theta_3^{\top } \ar[r]^{= } & \hh_{0,1}^{\top}\ar[r]^{\ee } & \hh_{g , n}^{\top}\ar[r]^{\ff } & \hh_{g,0}^{\top}\ar[r]^{\gg } & \hh_{0,0}^{\top}\ar[r]^{= } & \theta^{\top}_3.}\ ] ] since @xmath226 we deduce that @xmath227 is injective .
furthermore it follows from @xmath228 and from the above diagram that @xmath229 is contained in the kernel of the projection map @xmath230 .
since @xmath194 has infinite rank this concludes the proof of the first part of theorem [ thm : topsmooth ] .
if @xmath16 , then we can take @xmath224 to be the identity map .
in particular all maps are homomorphisms , and the homomorphism @xmath231 has the desired properties .
let @xmath232 be a pair of manifolds .
let @xmath233 be a homomorphism to a free abelian group .
we denote the quotient field of @xmath36 $ ] by @xmath234 .
denote by @xmath235 the universal covering of @xmath56 and write @xmath236 .
then we can consider the chain complex @xmath237}q(h).\ ] ] let @xmath238 be a basis for the graded vector space @xmath239 , then we obtain the corresponding torsion @xmath240 this torsion invariant is well - defined up to multiplication by an element of the form @xmath241 ( @xmath242 .
if @xmath232 is @xmath234acyclic , then @xmath238 is the trivial basis and we just write @xmath243 .
we will not recall the definition of torsion but refer instead to the many excellent expositions , e.g. , @xcite , @xcite , @xcite and @xcite .
we will several times make use of the following well - known lemma ( e.g. , see ( * ? ? ? * proposition 2.3 ) for a proof ) .
[ lem : alsozero ] if @xmath244 and if @xmath245 is a homomorphism to a free abelian group , then @xmath246 .
we also adopt the following notation : given @xmath247 we write @xmath248 if @xmath249 for some @xmath250 and @xmath251 .
put differently , @xmath248 if and only if @xmath252 and @xmath253 agree up to multiplication by a unit in @xmath254 .
let @xmath55 be a homology cylinder over @xmath54 .
write @xmath255 .
we normally think of @xmath38 as a multiplicative group .
denote @xmath256 .
consider @xmath257 since @xmath258 it follows from lemma [ lem : alsozero ] that @xmath259 .
therefore we can define @xmath260 this is referred to as the _ torsion _ of the homology cylinder @xmath261 .
we will now show that the torsion can be defined in terms of homology .
first note that @xmath36 $ ] is a unique factorization domain , so that for any finitely generated @xmath36$]-module @xmath56 , the _ order _ of @xmath56 is defined as an element of @xmath36 $ ] .
we denote it by @xmath262 .
( for a precise definition , e.g. , see ( * ? ? ?
* section 4.2 ) . ) note that @xmath262 is well - defined up to multiplication by a unit in @xmath254 .
[ lemma : homology - and - torsion ] for any homology cylinder @xmath56 we have @xmath263)$ ] .
it is well known that @xmath264))^{(-1)^{i+1}}$ ] ( e.g. , see ( * ? ? ? * section 4.2 ) ) . since @xmath265 is an isomorphism it follows immediately that @xmath266)=0 $ ] .
one can check that @xmath267)=0 $ ] for all @xmath268 , by using poincar duality , universal coefficient spectral sequence , and the fact that @xmath269)$ ] is torsion .
[ remark : torsion ] 1 .
the torsion of a homology cylinder was first studied by sakasai @xcite and is closely related to the torsion invariants of sutured manifold introduced independently by benedetti and petronio @xcite and @xcite .
note that string links give rise to homology cylinders in a natural way . in this context
the torsion invariant was first studied by kirk , livingston and wang ( * ? ? ?
* definition 6.8 ) .
3 . by lemma
[ lemma : homology - and - torsion ] ( see also ( * ? ? ?
* lemma 3.5 ) ) the torsion @xmath270 is in fact an element in @xmath36 $ ] , furthermore , if @xmath271\to \z$ ] denotes the augmentation defined by @xmath272 for @xmath251 , then @xmath273 ( e.g. , see @xcite and ( * ? ?
* proposition 5 ) ) .
4 . by lemma
[ lemma : homology - and - torsion ] , if @xmath274 is the homology cylinder over an annulus corresponding to a knot @xmath275 as in section [ section : examples ] , then @xmath276 , the alexander polynomial of @xmath71 .
if @xmath90 , then @xmath277)=0 $ ] and therefore @xmath278 $ ] . before we can state the behavior of torsion under the product operation we have to study the action of a homology cylinder on the first homology of the surface .
our notation is as follows . throughout this paper
we write @xmath255 .
given a homology cylinder @xmath55 over @xmath54 we denote the automorphism @xmath280 by @xmath281 .
let @xmath282 be the image of @xmath283 in @xmath38 , and let @xmath284 we recall the following result : [ goda - sakasai ] [ prop : inauthdot ] let @xmath56 be a homology cylinder over @xmath54
. then @xmath285 .
let @xmath286 where @xmath287 denotes @xmath54 with capped boundary circles .
we write @xmath288 by choosing a splitting .
choosing an arbitrary basis of @xmath282 and a symplectic basis of @xmath289 , @xmath290 consists of all matrices @xmath291 of the form @xmath292 with @xmath293 ( * ? ? ?
* remark 2.4 ) .
note that any @xmath294 is indeed realized by a homology cylinder , in fact there exists @xmath90 such that the induced action on @xmath295 is given by @xmath291 ( e.g. , see ( * ? ? ?
* proposition 7.3 ) ) .
each @xmath296 induces an automorphism of @xmath254 , which we also denote by @xmath65 .
in particular , a homology cylinder @xmath56 over @xmath54 gives rise to @xmath297 .
we can now formulate the following proposition ( e.g. , see ( * ? ? ?
* section 7 ) and ( * ? ? ? * proposition 6.6 ) for related results ) .
[ prop : deltasum ] let @xmath261 and @xmath298 be homology cylinders over @xmath54 . then @xmath299 note that @xmath35 is a multiplicative abelian group .
the action of @xmath290 on @xmath38 extends to an action of @xmath290 on @xmath35 , and we can thus form the semidirect product @xmath300 . we can now formulate the following corollary : [ cor : metabmap ] the following is a well - defined homomorphism of monoids @xmath301 since we are mostly interested in abelian quotients of @xmath81 we will now show that @xmath46 also gives rise to a homomorphism to an abelian group .
we define @xmath37 to be the subgroup of @xmath35 generated by the following set @xmath302 we write @xmath303 when @xmath38 is clearly understood from the context .
the following is an immediate consequence of proposition [ prop : deltasum ] : [ corollary : monoid - homomorphism ] the torsion invariant gives rise to a monoid homomorphism @xmath304 we write @xmath305 and @xmath306 .
let @xmath307 , where @xmath308 is the gluing map .
we view @xmath56 , @xmath309 , @xmath310 ( = @xmath311 ) as subspaces of @xmath84 .
a mayer - vietoris argument shows that the inclusion map induces an isomorphism @xmath312 .
let @xmath313 .
we equip @xmath84 with the map @xmath314 induced by the inclusion .
restricting this , we equip @xmath310 ( @xmath315 ) , @xmath56 , @xmath309 with maps into @xmath316 . for notational convenience ,
we denote @xmath317 , the quotient field of @xmath318 $ ] .
we have the following short exact sequence of cellular homology : @xmath319 here the map @xmath320 is given by @xmath321 and the map @xmath322 is given by the inclusion map .
let @xmath238 be a basis for the graded @xmath323module @xmath324 .
it follows from the long exact homology sequence that the inclusion map induces an isomorphism @xmath325 .
we equip @xmath326 with the basis @xmath327 given by the image of @xmath238 .
we consider the resulting torsions @xmath328 , @xmath329 , @xmath330 , and @xmath331 . by applying ( * ? ? ?
* theorem 3.2 ) to the above short exact sequence , the torsions satisfy the following : @xmath332 we equip @xmath333 with the basis @xmath334 .
similarly we have @xmath335 from the above equations and the tautology @xmath336 , it follows that @xmath337 we now apply @xmath338 to the above equality . from the commutative diagram @xmath339^\cong & h_1(w ) & h_1(m)=f \ar[l]_\cong\\ & h \ar[ul]^{(j_+ ) _ * } \ar[ur]_{(i_- ) _ * } \ar[r]_{\varphi(m ) } & h \ar[u]_{(i_+ ) _ * } } \ ] ] it follows that our @xmath340 composed with @xmath341 gives @xmath342 .
therefore we have @xmath343 let @xmath38 be a finitely generated free abelian group . given a non - zero polynomial @xmath344 we denote by @xmath345 its mahler measure ( e.g. , see @xcite , @xcite , and @xcite ) .
it is well - known that given @xmath346 and @xmath251 we have @xmath347 and @xmath348 . furthermore given any @xmath349 we have @xmath350 ( e.g. , see ( * ? ? ?
* section 3 ) ) .
it follows that @xmath351 defines a monoid homomorphism @xmath352 .
using this homomorphism and using the examples of section [ section : tyingknot ] one can reprove theorem 2.4 in @xcite .
now let @xmath56 be a homology cylinder over a surface @xmath178 and let @xmath179 be a homology cylinder over a surface @xmath180 .
assume that @xmath181 and fix a boundary component @xmath182 in @xmath96 and fix a boundary component @xmath183 in @xmath184 .
recall that gluing now gives us a homology cylinder @xmath189 over @xmath187 .
[ prop : tauunion ] denote by @xmath353 and @xmath354 the inclusion induced maps of @xmath355 and @xmath356 into @xmath357 , respectively .
then the following holds : @xmath358 write @xmath359 . by the mayer - vietoris formula for torsion
we have @xmath360 it is seen easily that @xmath361 and @xmath362 from the definitions .
let @xmath38 be a free abelian multiplicative group .
we equip @xmath254 with the standard involution of a group ring , i.e. @xmath363 for @xmath251 and extend it to @xmath234 by setting @xmath364 .
the next theorem can be viewed as a generalization of the classical fox - milnor theorem @xcite that the alexander polynomial of a slice knot factors as @xmath365 for some @xmath366 .
[ thm : homcob ] let @xmath274 and @xmath367 be homology cylinders over a surface @xmath54 which are homology cobordant .
. then @xmath368 for some @xmath369 .
let @xmath84 be a homology cobordism between @xmath56 and @xmath309 .
view @xmath370 as a subspace of @xmath84 .
note that @xmath371 is an isomorphism .
we now equip @xmath84 with the homomorphism @xmath372 by the above we have @xmath373 .
therefore it follows from lemma [ lem : alsozero ] that @xmath374 and we can therefore consider @xmath375 .
also note that @xmath376 and @xmath377 .
we can hence also think of the torsion invariants @xmath378 and @xmath379 . from the following short exact sequence of acyclic chain complexes @xmath380
we obtain @xmath381 .
similarly , @xmath382 . by duality we have @xmath383 ( e.g. , see @xcite , @xcite and @xcite ) .
hence @xmath384 and the theorem follows .
define @xmath385 obviously @xmath39 is a subgroup of @xmath35 . as we do with @xmath303 , we often write @xmath386 .
the following corollary is now an immediate consequence of corollary [ cor : metabmap ] and theorem [ thm : homcob ] .
this corollary should be compared to ( * ? ? ? * theorem 5.1 ) .
[ cor : metabmap2 ] the map @xmath387 defined in corollary [ cor : metabmap ] descends to a group homomorphism @xmath388 since we are mostly interested in abelian quotients of @xmath11 we will for the most part work with the following corollary , which is an immediate consequence of theorem [ thm : homcob ] . [ corollary : group - homomorphism ] the torsion invariant gives rise to a group homomorphism @xmath389 where @xmath390 .
this descends to a homomorphism of the quotient of @xmath11 modulo the normal subgroup @xmath45 generated by the mapping class group @xmath6 . in later sections
, we will show that the image of this torsion homomorphism is large .
indeed , it has a quotient isomorphic to @xmath24 if @xmath391 , and has a quotient isomorphic to @xmath49 if either @xmath30 , or @xmath31 and @xmath32 .
it is well - known that the alexander polynomial @xmath392 of a knot @xmath71 is symmetric , i.e. @xmath393 .
furthermore , the alexander polynomial of any closed @xmath13manifold or any @xmath13manifold with toroidal boundary is symmetric ( e.g. , see @xcite ) . in this section
we will study the symmetry properties of the torsion of homology cylinders .
we start out with the following observation .
let @xmath96 be a surface which is either an annulus or a surface of genus one with at most one boundary component .
let @xmath56 be a homology cylinder over @xmath96 .
then @xmath394 .
first assume that @xmath96 is either an annulus or a torus .
we write @xmath395 , @xmath396 and consider the torsion @xmath329 corresponding to @xmath397 .
clearly it suffices to show that @xmath329 is symmetric .
first note that @xmath398 has euler characteristic zero .
it can be seen easily that @xmath399 , in particular its torsion @xmath400 is defined . also , from this it follows that @xmath56 is @xmath323-acyclic so that @xmath401 is defined .
it is well - known that the torsion of a torus is trivial and for an annulus it equals @xmath402 , where @xmath403 is a generator of the first homology group .
it follows in particular that @xmath404 . from the long exact sequence of torsion corresponding to the pair
@xmath405 it now follows that @xmath406 .
also , from duality for torsion we have @xmath407 . combining these equalities
, it follows that @xmath408 as desired .
now assume that @xmath96 is a surface of genus one with exactly one boundary component .
let @xmath184 be a disk and @xmath409 the result of gluing @xmath96 and @xmath184 along their boundary components .
let @xmath410 $ ] .
note that @xmath411 is an isomorphism .
it now follows immediately from proposition [ prop : tauunion ] that @xmath412 , in particular @xmath270 equals the torsion of a homology cylinder over a torus , which is symmetric as we saw above . in general torsion
is not symmetric though . to our knowledge
this phenomenon was first observed in ( * ? ? ?
* example 8.5 ) in the context of sutured manifolds and rational homology cylinders .
more examples will be given in section [ section : construction - computation ] . on the other hand recall that in order to define the torsion homomorphism in corollaries [ corollary : monoid - homomorphism ] and [ corollary : group - homomorphism
] , we thought of torsion invariants modulo @xmath303 , i.e. , up to the action of @xmath290 . when the number of boundary components @xmath413 , @xmath290 is simply the group of automorphisms on @xmath38 preserving the intersection pairing on @xmath54 . in particular
, the map @xmath414 is in @xmath290 .
therefore @xmath415 in @xmath416 for any @xmath252 .
in other words , modulo @xmath417 , everything is symmetric : [ lemma : symmetry - up - to - aut * ] suppose @xmath54 is a surface with at most one boundary components .
then for any homology cylinder @xmath56 over @xmath54 , we have @xmath418 in @xmath419 .
therefore the only remaining case is when either @xmath30 , or @xmath31 and @xmath32 .
we will show that in these cases the torsion invariant @xmath270 is in general asymmetric _ even modulo @xmath303_. in the next section we will consider general methods to construct homology cylinders and to compute their torsion , which will be used to illustrate the asymmetry of torsion .
in this section we will compute torsion for various homology cylinders .
the examples illustrate the computation of torsion and they will also be used later to prove theorems [ mainthm ] and [ mainthm2 ] . for any homology cylinder @xmath55 over @xmath54 ,
the pair @xmath420 admits a handle decomposition without @xmath421 and @xmath13handles . a handle decomposition of a pair @xmath420 is given as submanifolds @xmath422 = m_0 \subset m_1 \subset m_2 = m\ ] ] where @xmath423 is obtained by attaching @xmath353-handles to @xmath424 for @xmath206 .
@xmath56 is a homology cylinder if and only if the numbers of @xmath425-handles and @xmath154-handles are equal , say @xmath426 , and the boundary map @xmath427 is invertible
. from the definition , the torsion @xmath428 is equal to the determinant of the @xmath429 matrix @xmath430 over @xmath36 $ ] which represents the @xmath36$]-coefficient boundary map @xmath431)=\z[h]^r \to c_1(m,\sigma_+;\z[h ] ) = \z[h]^r\ ] ] with respect to the bases given by @xmath425 and @xmath154handles .
when a handle decomposition is explicitly given , the boundary map can be effectively computed using a standard method .
this is often useful in computing the torsion of a homology cylinder with a given handle decomposition . for readers who are not familiar with this type of computation
, it can be described best by a detailed example .
the construction and computation below will also be used later , to show the existence of nontrivial homomorphisms of @xmath11 onto @xmath41 .
let @xmath432 with @xmath32 and @xmath31 .
let @xmath433 be the @xmath13manifold obtained by attaching one @xmath425handle to @xmath434 $ ] .
the boundary of @xmath433 is the union of @xmath54 , @xmath435 $ ] , and a genus @xmath436 surface , say @xmath437 , which has the same boundary as @xmath54 .
we attach one @xmath154handle to @xmath433 along the simple closed curve @xmath438 on @xmath437 as illustrated in figure [ figure : example - attaching - curve ] .
the point @xmath103 is the basepoint and the gray band 0=0 represents @xmath439 parallel strands .
we denote the resulting @xmath13manifold by @xmath440 . at 145 63 at 272 15 at 272 95 at 261 228 at 120 287 at 164 239 at 103 194 at 102 169 on @xmath437,title="fig : " ] we have @xmath441 where @xmath41 is generated by the @xmath425handle .
recall that we can find a splitting @xmath442 where @xmath282 is the image of @xmath443 and @xmath444 with capped - off boundary@xmath445 . in our case
@xmath282 and @xmath446 have rank @xmath425 and @xmath73 , respectively .
we choose generators @xmath403 of @xmath41 , @xmath447 of @xmath282 , and @xmath448 ( @xmath449 ) of @xmath446 which lie on @xmath437 and which are dual to the curves @xmath450 , @xmath451 , @xmath452 illustrated in figure [ figure : example - attaching - curve ] with respect to the intersection pairing on @xmath437 .
that is , @xmath403 is represented by a loop which is disjoint to @xmath451 , @xmath452 and has intersection numbers @xmath453 with @xmath450 on @xmath437 , and similarly for other generators . given a curve on @xmath437 , its homology class in @xmath454 can be determined easily by counting the algebraic intersections with the dual curves . for the curve @xmath438 , we have @xmath455=(t , x^a_{\vphantom{1}}y_1^a ) \in \z\times h = h_1(m_1)$ ]
. thus the homology class of @xmath438 in @xmath456 is @xmath403 , and so the boundary map @xmath457 is the identity .
therefore our @xmath56 is a homology cylinder .
now we compute the @xmath36$]coefficient boundary map . since the attaching circle gives a relation @xmath458 , the generator @xmath403 is sent to @xmath459 under @xmath460 .
the @xmath36$]valued intersection number @xmath461 of @xmath438 with the belt circle @xmath462 of the @xmath425handle is given by the following formula : for a point @xmath463 on @xmath438 , let @xmath464 be the oriented initial segment of @xmath438 from the basepoint @xmath103 to @xmath463 . then @xmath465 here , @xmath466 is the usual sign of @xmath467 . precisely , @xmath468 if @xmath469 at @xmath463 agrees with the orientation of @xmath437 , and @xmath470 otherwise .
the numbers @xmath471 , @xmath472 are given similarly . from figure
[ figure : example - attaching - curve ] , it can be seen that @xmath473 where @xmath474 .
the @xmath36$]coefficient boundary map is the @xmath475 matrix @xmath476 $ ] .
therefore @xmath477 .
it is obvious that @xmath478 is asymmetric in the sense that @xmath479 . in example
[ example : non - self - dual - element ] , we will show that @xmath478 and @xmath480 are distinct even modulo @xmath303 .
in this subsection we describe an operation that modifies a homology cylinder by `` tying in '' a string link , and investigate its effect on the torsion invariant .
first we recall the definition of a string link .
fix @xmath481 points @xmath482 in the interior of the disk @xmath483 .
an @xmath481component string link @xmath462 is the disjoint union of @xmath481 properly embedded disjoint oriented arcs @xmath484 ( @xmath485 ) in @xmath486 $ ] from @xmath487 to @xmath488 .
we denote the exterior of a string link @xmath462 in @xmath486 $ ] by @xmath489 .
the string link @xmath490 \in d^2\times[0,1]$ ] is called the trivial string link .
the exterior of the trivial string link is denoted by @xmath491 .
note that if @xmath462 is framed , then there is a canonical identification of @xmath492 and @xmath493 which restricts to the identity on @xmath494 . including theorem [ theorem : string - link - sum - formula ]
, all results in this section hold for string links in homology @xmath486 $ ] as well .
we recall the definition of the torsion invariant for string links ( cf .
* definition 6.8 ) ) .
let @xmath495 , a subspace of @xmath496 $ ] .
suppose @xmath462 is a string link and @xmath497 is a homomorphism to a free abelian group @xmath38 .
since @xmath498 is @xmath41acyclic , it follows from lemma [ lem : alsozero ] that @xmath498 is @xmath36$]acyclic .
we denote by @xmath499 the torsion of the cellular chain complex @xmath500)$ ] .
when @xmath501 is clearly understood from the context we write @xmath502 . as usual , @xmath503 is well - defined up to multiplication by @xmath241 ( @xmath251 ) .
note that for any string link @xmath462 , @xmath489 can be viewed as a homology cylinder over the surface @xmath74 . now consider a homology cylinder @xmath55 on @xmath504 with @xmath255 and an embedding @xmath505 .
for any framed string link @xmath462 , define @xmath506 since @xmath489 has the same homology as @xmath491 , it follows that @xmath507 is a homology cylinder .
we say that @xmath508 is obtained from @xmath56 by tying the string link @xmath462 .
the map @xmath509 gives rise to a homomorphism @xmath510 induced by the inclusions .
we denote the resulting torsion invariant of @xmath462 by @xmath511 . [
theorem : string - link - sum - formula ] the torsion invariant of the homology cylinder @xmath508 is given by @xmath512 .
we write @xmath513 .
let @xmath514 .
choose bases of @xmath515 and @xmath516 . since @xmath498 is @xmath41acyclic ,
it is @xmath234acyclic , and therefore for any string link @xmath462 ( including the trivial one ) our basis of @xmath515 determines a basis of @xmath517 via the inclusion - induced map .
define the torsion invariants @xmath518 , @xmath519 , @xmath520 ( up to the usual ambiguity ) using these bases .
consider the following short exact sequence : @xmath521 since @xmath522 is @xmath234-acyclic , the bases we have chosen give rise to a basis of @xmath523 .
define @xmath524 using this basis . by (
* theorem 3.2 ) , we have @xmath525 considering the special case of a trivial string link @xmath462 , we obtain @xmath526 combining these , we have @xmath527 from the short exact sequence @xmath528 with @xmath498 acyclic , we obtain @xmath529 .
note that @xmath530 $ ] , hence @xmath531 .
the desired formula now follows immediately .
we will now study the case @xmath532 . note that a string link @xmath462 with one component gives us a knot @xmath71 in @xmath208 .
furthermore , for any embedding @xmath533 , it can be seen easily that the torsion @xmath511 is equal to @xmath534 , where @xmath392 is the alexander polynomial of @xmath71 and @xmath99 is the image of the generator of @xmath535 under the map @xmath536 induced by @xmath509 .
therefore , by theorem [ theorem : string - link - sum - formula ] , the torsion of @xmath508 can be described in terms of the alexander polynomial of @xmath71 .
the following is a special case of this , which gives a group homomorphism of the ( smooth or topological ) concordance group @xmath537 of knots in @xmath208 into @xmath11 .
recall that @xmath538 $ ] and @xmath539 .
[ prop : embedc ] let @xmath14 .
we write @xmath540 and @xmath541 $ ] .
let @xmath542 be an embedding , and let @xmath543 be the embedding @xmath544 .
then the assignment @xmath545 descends to a group homomorphism @xmath546 furthermore @xmath547 where @xmath99 is the image of the generator of @xmath548 under @xmath549 .
it is straightforward to verify that @xmath550 and that the assignment @xmath545 in fact descends to a group homomorphism @xmath551 .
the conclusion on torsion has already been proven in the paragraph above proposition [ prop : embedc ] .
in this section we will construct epimorphisms of @xmath11 onto nontrivial abelian groups .
this will give a proof of theorems [ mainthm ] and [ mainthm2 ] . throughout section [ section : homomorphisms ]
, we fix @xmath14 and write @xmath552 and @xmath553 . recall that our torsion invariant lives in the multiplicative abelian group @xmath554 . in this subsection
we investigate the algebraic structure of @xmath554 .
we will think of symmetric / asymmetric parts of @xmath554 , and define certain @xmath555 and @xmath41-valued homomorphisms which form sets of complete invariants of these parts , respectively . to be more precise ,
let @xmath556 note that @xmath557 .
there is an exact sequence @xmath558 which can be viewed as a decomposition of @xmath559 into symmetric and asymmetric parts .
recall that @xmath282 is the image of @xmath560 in @xmath38 and @xmath284 we define an equivalence relation @xmath67 on @xmath561 by @xmath562 if @xmath563 for some @xmath564 .
note that if @xmath565 then @xmath252 and @xmath253 represent the same element in @xmath419 . from now on we say @xmath252 is _ self - dual _ if @xmath566 .
recall that @xmath36 $ ] is a unique factorization domain , so that for each @xmath567 and irreducible @xmath568 $ ] , we can think of the exponent of @xmath569 in the factorization of @xmath252 .
( the exponent is an integer and may be negative . ) for an irreducible element @xmath569 in @xmath254 , we define a function @xmath570 as follows : given @xmath567 , @xmath571 is the sum of exponents of distinct irreducible factors @xmath572 of @xmath252 such that @xmath573 .
( as usual , irreducible factors are distinguished up to multiplication by a unit in @xmath254 . )
[ prop : complete ] 1 .
if @xmath569 is a self - dual irreducible element in @xmath254 , then the map @xmath574 defined by @xmath575 is a surjective group homomorphism .
furthermore , @xmath576}\psi_\lambda \colon q(h)^{sym}/\anh \to \bigoplus_{[\lambda ] } \z/2,\ ] ] is an isomorphism , where @xmath577 $ ] runs over the equivalence classes of self - dual irreducible @xmath569 .
if @xmath572 is a non - self - dual irreducible element in @xmath254 , then the map @xmath578 defined by @xmath579 is a surjective group homomorphism .
furthermore @xmath580,[\bar\mu]\ } }
\theta_\mu \colon q(h)^\times / q(h)^{sym } \to \bigoplus_{\{[\mu],[\bar\mu]\ } } \z,\ ] ] is an isomorphism , where @xmath581,[\bar\mu]\}$ ] runs over the unordered pairs of equivalence classes of non - self - dual irreducible @xmath572 and its involution @xmath582 .
consequently , @xmath554 is isomorphic to @xmath583 } \z/2\big)\oplus \big(\bigoplus_{\{[\mu],[\bar\mu]\ } } \z\big)$ ] . 1 . in the definition of @xmath584
, we have one summand for the two classes @xmath585 $ ] and @xmath586 $ ] . here
we have sign ambiguity since @xmath587 , but this does not cause any problems in our conclusions .
2 . the homomorphisms @xmath588 and @xmath589 extend to homomorphisms of @xmath554 , which will also be denoted by @xmath588 and @xmath589 . also , we denote by @xmath584 and @xmath590 the homomorphisms of @xmath554 induced by @xmath584 and @xmath590 .
then the isomorphism in the last sentence of theorem [ theorem : algebraic - classification ] can be written as @xmath591 .
3 . although @xmath590 could be defined for self - dual @xmath572 as well , it is not interesting since the resulting @xmath590 is always zero .
first we will observe that @xmath588 and @xmath589 are well - defined surjective homomorphisms .
since the factorization into irreducible factors is preserved by the @xmath290-action , @xmath571 is invariant under @xmath67 for any irreducible @xmath569 .
if @xmath569 is self - dual , then @xmath592 .
it follows that @xmath588 is a well - defined homomorphism .
the surjectivity of @xmath588 and @xmath589 follows from the observation that @xmath593 equals 1 if @xmath594 , and 0 otherwise . to see that @xmath590 and @xmath584 are well - defined homomorphisms , observe that for @xmath595 , @xmath596 .
their surjectivity now follows from the observation that for non - self - dual irreducible @xmath572 and @xmath597 , @xmath598 equals @xmath425 if @xmath599 , @xmath470 if @xmath600 , and 0 otherwise .
for the injectivity of @xmath589 , suppose @xmath601 represents an element in the kernel of @xmath589 .
we can rewrite the irreducible factorization of @xmath509 to obtain an expression @xmath602 , where the @xmath603 are mutually non - equivalent self - dual irreducible elements , @xmath604 , and @xmath605 . note that @xmath606 in @xmath607 since @xmath603 is self - dual . evaluating @xmath608 , we have that @xmath609 is even for each @xmath353 . from this
it follows that @xmath610 in @xmath607 .
the injectivity of @xmath584 is proved similarly : if @xmath611 represents an element in the kernel of @xmath589 , then from the irreducible factorization of @xmath509 we obtain an expression @xmath612 , where @xmath613 , @xmath614 , and the @xmath615 are non - self - dual irreducible elements such that @xmath616 whenever @xmath617 .
evaluating @xmath618 , we have @xmath619 for each @xmath353 .
it follows that @xmath610 in @xmath620 .
now , in order to understand the structure of @xmath554 , the only remaining part is to count the number of summands of @xmath589 and @xmath584 in proposition [ prop : complete ] . to state the result
, we introduce the following definition : we say that @xmath621 is _ small _ if either @xmath622 or @xmath31 and @xmath16 .
otherwise we say that @xmath621 is _
large_. [ theorem : algebraic - classification ] suppose @xmath38 is nontrivial , i.e. , @xmath2 is neither a sphere nor disk .
then the following hold : 1 .
@xmath589 is an isomorphism of @xmath607 onto @xmath24 .
2 . if @xmath621 is small , then @xmath623 .
if @xmath621 is large , then @xmath584 is an isomorphism of @xmath620 onto @xmath49 .
consequently , @xmath624 recall that in section [ subsection : symmetry - asymmetry ] we observed that there is no non - self - dual @xmath572 if @xmath621 is small .
the first sentence of theorem [ theorem : algebraic - classification ] ( 2 ) is an immediate consequence . in the following subsections , we will realize , as the values of torsion invariants of homology cylinders , infinitely many self - dual classes @xmath577 $ ] when @xmath38 is nontrivial ( see theorem [ theorem : realization ] ) , and infinitely many non - self - dual classes @xmath585 $ ] when @xmath621 is large ( see theorem [ theorem : realization - z ] ) . in sections [ subsection : proofs ] and [ subsection : invariants - under - aut*(h ) ] , we give proofs of theorems [ mainthm ] and [ mainthm2 ] . along the way we also conclude the proof of theorem [ theorem : algebraic - classification ] . [ theorem : realization ] if @xmath22 , then there exists a subgroup @xmath625 isomorphic to @xmath626 such that @xmath627 } \z/2\ ] ] is an injection whose image is the sum of a certain infinite set of @xmath555 summands of @xmath628 } \z/2 $ ] .
we pick non - trivial knots @xmath629 ( @xmath630 ) which are negative amphicheiral with irreducible alexander polynomials such that the multisets @xmath631 of nonzero coefficients of @xmath632 are mutually distinct up to sign .
for example , one can use the family of knots described in @xcite : their alexander polynomials are of the form @xmath633 .
it is well - known that the knots @xmath629 form a @xmath40-basis of a subgroup of the knot concordance group isomorphic to @xmath626 .
we write @xmath541 $ ] .
let @xmath533 be an embedding as in proposition [ prop : embedc ] , which is homologically essential .
denote by @xmath99 the image of the generator of @xmath634 under the homomorphism @xmath536 induced by the given embedding @xmath635 .
we now write @xmath636 . by proposition
[ prop : embedc ] , the @xmath423 span a subgroup of @xmath11 which is the homomorphic image of @xmath626 .
furthermore we have @xmath637 .
note that each @xmath638 is irreducible and self - dual since @xmath639 is irreducible and self - dual and since @xmath99 is easily seen to be an indivisible element in @xmath38 .
it is not difficult to see directly that the multiset @xmath631 is an invariant of @xmath638 under the equivalence relation @xmath67 .
( for a more general method from which this observation is derived as a special case , see section [ subsection : invariants - under - aut*(h ) ] . )
therefore , since the @xmath631 are all distinct , the equivalence classes of the @xmath638 are mutually distinct . from this
we obtain @xmath640 also , @xmath641 if @xmath642
. therefore the composition @xmath643 } \z/2 $ ] in the statement of this theorem is injective and has image @xmath644 } \z/2 \cong ( \z/2)^\infty$ ] .
we now obtain theorem [ mainthm ] as an immediate corollary . by theorem [ theorem :
realization ] , we have a subgroup @xmath645 of @xmath11 and a homomorphism @xmath646 whose restriction to @xmath645 is an isomorphism .
it follows that the homomorphism splits , and @xmath647 descends to a summand of the abelianization of @xmath11 . [ remark : proof of mainthm ] 1 .
using the full power of theorem [ theorem : string - link - sum - formula ] , i.e. , by tying in string links with several components , we can realize many more nontrivial values of the homomorphism @xmath589 .
2 . following the arguments in the proof of theorem [ mainthm ]
, one can easily show that if @xmath648 then there exists a commutative diagram @xmath649\ar[r]^{\id } & ( \z/2)^{\infty } \\ \hgnsmooth \ar[r ] & \hgn^{\top } \ar[u ] } \ ] ] such that the left hand map is injective and the right hand map is surjective .
we also have the following realization result : [ theorem : realization - z ] if @xmath621 is large , then the image of @xmath650,[\bar\mu]\ } } \z\ ] ] contains infinitely many summands of @xmath651,[\bar\mu]\}}\z$ ] .
the proof of theorem [ theorem : realization - z ] requires a more sophisticated method to detect non - equivalence of non - self - dual irreducible factors .
this will occupy all of section [ subsection : invariants - under - aut*(h ) ] .
assuming theorem [ theorem : realization - z ] we can now finally prove theorem [ mainthm2 ] : suppose that @xmath25 . by theorem [ mainthm2-special ]
, we may assume that @xmath621 is large .
let @xmath652 be the abelianization of @xmath11 . by theorem [ theorem : realization - z ]
, we have a surjection @xmath653 .
this induces a split surjection , say @xmath3 , of the abelianization @xmath652 of @xmath11 onto @xmath49 . also , by theorem
[ theorem : realization ] , we have a split surjection @xmath654 .
since the intersection of the images of the right inverses of @xmath509 and @xmath3 is automatically @xmath655 ( e.g. , compare the order ) , it follows that @xmath656 is a split surjection . in order to prove theorem [ theorem : realization - z ] we first will introduce a simple and practical method for distinguishing elements in @xmath36 $ ] up to the action of @xmath290 . as before ,
we denote @xmath255 and @xmath657 .
denote @xmath658 where @xmath287 is @xmath54 with boundary circles capped off , and write @xmath288 by choosing a splitting .
fix a basis @xmath659 of @xmath282 , so that each @xmath660 $ ] is viewed as a ( laurent ) polynomial in the variables @xmath661 . for @xmath662 $ ] ,
we write @xmath663 if @xmath664 for some @xmath665 .
this is an equivalence relation ; denote the equivalence class of @xmath660 $ ] by @xmath666 $ ] . since @xmath663
if and only if the polynomial @xmath463 is obtained from @xmath667 by shifting the exponents , it is very straightforward to check whether @xmath663 or not for two given polynomials @xmath463 and @xmath667 . given @xmath344 , write @xmath668 where @xmath669 $ ] , and define @xmath670 \mid g\in \widehat h\}.\ ] ] we view @xmath671 as a multiset , i.e. , repeated elements are allowed . for @xmath672\}$ ] , denote @xmath673\}$ ] .
[ lemma : aut - invariant ] @xmath671 is invariant up to sign under @xmath67 on @xmath36 $ ] , i.e. , if @xmath565 , @xmath671 is equal to either @xmath674 or @xmath675 .
note that @xmath564 fixes @xmath282 and sends @xmath676 to an element of the form @xmath677 for some @xmath678 , @xmath679 .
in addition the association @xmath680 is a bijection since @xmath65 is an isomorphism .
therefore , for @xmath681 $ ] , the classes @xmath682 $ ] are permuted by the action of @xmath65 .
it follows @xmath683 .
it is easily seen that @xmath684 for @xmath251 .
[ example : total - coefficients ] in the proof of theorem [ theorem : realization ] , we have observed that for @xmath685 $ ] , the multiset of all nonzero coefficients @xmath686 and @xmath687 is an invariant , up to sign , of @xmath252 under @xmath67 . this can be viewed as a consequence of lemma [ lemma : aut - invariant ] , since the multiset of nonzero coefficients of an element @xmath669 $ ] is invariant under @xmath688 .
we remark that if @xmath672\}$ ] , then @xmath689\}$ ] .
this combined with lemma [ lemma : aut - invariant ] often allows us to detect non - self - dual elements , as illustrated below .
[ example : non - self - dual - element ] fix @xmath690 . for a positive integer @xmath439 ,
let @xmath691 then @xmath692 , \ldots , [ x_i^{a-1}-x_i^a ] , [ x_i^a]\ } = \{[1-x_i ] , \ldots , [ 1-x_i ] , [ 1]\},\\ c(\bar p_a)&=\{[x_i-1 ] , \ldots , [ x_i-1 ] , [ 1]\}. \end{aligned}\ ] ] looking at the element @xmath693 $ ] we see that @xmath694 .
since @xmath695 we deduce that @xmath696 . therefore @xmath697 , i.e. , @xmath698 is non - self - dual .
in particular , the torsion of the homology cylinder @xmath699 in section [ subsection : exmaple - handle - decomp ] is non - self - dual . also , since @xmath700 , we have @xmath701 whenever @xmath702 .
we are now finally in a position to prove theorem [ theorem : realization - z ] .
first we consider the case @xmath31 and @xmath32 . for each positive integer @xmath439 , consider the homology cylinder @xmath699 constructed in section [ subsection : exmaple - handle - decomp ] .
let @xmath703 $ ] .
as we observed in example [ example : non - self - dual - element ] , @xmath698 is non - self - dual , and @xmath704 whenever @xmath702 .
applying the eisenstein criterion to @xmath698 , it can be seen that @xmath698 is irreducible .
therefore @xmath705 from this it follows that the image of the subgroup generated by the classes of the @xmath699 under the homomorphism @xmath706,[\bar\mu]\}}\z\ ] ] is equal to @xmath707,[\bar p_a]\ } } \z \cong \z^\infty$ ] .
this proves the theorem for this case . for @xmath708 ,
the theorem is proved by a similar construction of homology cylinders , using the generator of @xmath38 associated to a third boundary component of @xmath54 in place of the generator @xmath709 .
( in this case the torsion @xmath698 lies in @xmath710 $ ] and thus it is easier to see that @xmath704 whenever @xmath702 . )
in this section we study homology cylinders arising from pretzel links .
the pretzel link @xmath711 is a @xmath13component link with an obvious seifert surface @xmath54 , as shown in figure [ fig : pretzel ] .
we pick @xmath96 as a model surface for @xmath712 .
we then write @xmath713 cut along @xmath714 . at 68 140 at 178 140 at 28 98 at 130 98 at 204 98 at 115 14 with @xmath426 ,
@xmath715 , and @xmath403 full twists.,title="fig : " ] the main result of this section is the following .
[ prop : pretzels ] 1 .
@xmath716 defines a homology cylinder if and only if @xmath717 .
2 . the set of homology cylinders @xmath716 generates a @xmath49 subsummand of the abelianization of @xmath50
. the proof of the proposition will require the remainder of this section .
given @xmath426 , @xmath715 , @xmath403 , we write @xmath718 cut along @xmath54 .
note that @xmath56 is a handlebody and that @xmath719 is the free group on the generators @xmath439 and @xmath720 as shown in figure [ fig : pretzel ] .
let @xmath438 be a loop on @xmath96 which runs from down the left - hand strip , and back up via the middle strip .
similarly , let @xmath462 be a loop which runs down the right - hand strip and back up the middle , so that @xmath721 is generated by @xmath438 and @xmath462 .
we now denote by @xmath722 and @xmath723 the corresponding curves on @xmath724 .
then @xmath725 we find that @xmath726 is an isomorphism if and only if @xmath717 .
this concludes the proof of proposition [ prop : pretzels ] ( 1 ) .
we will now calculate the torsions for homology cylinders with @xmath727 . note that this condition on @xmath728 is equivalent to @xmath729 .
in particular one of @xmath730 or @xmath403 is necessarily negative . using the symmetries of pretzel links we can without loss of generality
assume that @xmath731 and @xmath732 .
note that the condition @xmath733 implies that @xmath734 and @xmath735 .
recall that we can view @xmath736 as the free abelian multiplicative group with basis @xmath737 . using the isomorphism @xmath738
we now identify @xmath739 with the multiplicative free abelian group generated by @xmath439 and @xmath720 . in particular we will view @xmath740 as an element in @xmath741 .
evaluating the @xmath742 determinant @xmath743 we obtain from ( * ? ? ?
* proposition 4.2 ) that @xmath740 equals @xmath744 this polynomial is always asymmetric .
in fact the support of the polynomial @xmath740 is given in figure [ fig : support ] . here
a point @xmath745 corresponds to @xmath746 .
furthermore a triangle ( @xmath747 ) indicates a coefficient @xmath470 and a big five - pointed star ( @xmath748 ) indicates a coefficient @xmath453 .
@xmath749 [ prop : irrfactor ] there exist infinitely many positive integers @xmath750 , negative integers @xmath751 , and irreducible polynomials @xmath752 with the following properties : 1 .
@xmath753 , 2 .
@xmath754 , and 3 .
@xmath755 divides @xmath756 if and only if @xmath757 .
pick distinct odd primes @xmath758 . furthermore pick @xmath759 and @xmath760 such that @xmath761 .
we then have @xmath762 we now set @xmath763 , it follows that @xmath753 . in order to show the existence of @xmath755 with the required properties we have to introduce various definitions . given @xmath764 we now denote by @xmath765 the coefficient of the lowest degree and by @xmath766 the coefficient of the highest degree .
note that for @xmath767 we have @xmath768 and @xmath769 .
we now write @xmath770 and we view @xmath771 as an element in @xmath772 . by equation ( [ comptau ] )
we have @xmath773 since @xmath661 is prime it follows that @xmath774 is irreducible .
in particular for any @xmath353 there exists an irreducible factor @xmath775 of @xmath771 with @xmath776 .
note that @xmath777 divides @xmath778 .
in particular @xmath779 and @xmath780 .
it follows easily that @xmath754 .
given @xmath781 we have @xmath782 and @xmath783 . in particular if @xmath617 , then @xmath784 and @xmath785 are distinct irreducible polynomials .
it follows that @xmath755 does not divide @xmath786 if @xmath617 .
write @xmath787 .
note that @xmath788 .
in particular the polynomials @xmath789 satisfy @xmath790 .
it now follows immediately that the homology cylinders @xmath716 span a subgroup of @xmath791 which surjects onto @xmath792 under the map @xmath793 .
this concludes the proof of proposition [ prop : pretzels ] .
in this section , we consider a subgroup of @xmath11 which generalizes the torelli group of the mapping class group .
we prove analogues of theorems [ mainthm ] and [ mainthm2 ] , but for a larger set of surfaces .
let @xmath14 and write @xmath390 . recall that each @xmath794 induces an action @xmath795 on @xmath38 and
the map @xmath796 sending @xmath65 to @xmath795 is an epimorphism .
the torelli group @xmath797 is defined to be the kernel of this map , i.e. @xmath798 is the subgroup of @xmath6 given by all elements which act as the identity on @xmath38 .
we refer to @xcite and @xcite for details on the torelli group .
the following theorem summarizes some of the key properties of the torelli group whose proofs can be found in ( * ? ? ?
* theorem 7.10 ) , @xcite , @xcite .
( see also @xcite ) , and @xcite . ) 1 .
the group @xmath799 is torsion - free , 2 .
the group @xmath799 is finitely generated for @xmath800 and @xmath135 , 3 .
the group @xmath801 is a free group on infinitely many generators , 4 . if @xmath800 , then the abelianization of @xmath802 is isomorphic to @xmath803 for some @xmath804 .
it is an open question though whether the torelli group @xmath802 is finitely presented for @xmath800 ( e.g. , see ( * ? ? ?
* section 7.3 ) ) .
now recall that the action of a homology cylinder on @xmath390 also gives rise to an epimorphism @xmath805 .
we now define the _ torelli group @xmath51 of homology cylinders over @xmath2 _ to be the kernel of @xmath65 . by proposition [ prop : mgnembed ]
we can view the torelli group @xmath798 as a subgroup of @xmath51 .
it is an immediate consequence of proposition [ prop : deltasum ] and theorem [ thm : homcob ] that the torsion function gives rise to a homomorphism @xmath806 note that in this case we do not need to divide @xmath35 out by @xmath303 since @xmath807 is the identity for any @xmath808 .
we can now prove the analogues of theorems [ mainthm ] and [ mainthm2 ] .
[ thm : cyltorelli ] 1 .
if @xmath22 , then there exists an epimorphism @xmath809 which splits .
in particular , the abelianization of @xmath51 contains a direct summand isomorphic to @xmath24 .
if @xmath810 or @xmath25 , then there exists an epimorphism @xmath811 furthermore , the abelianization of @xmath51 contains a direct summand isomorphic to @xmath27 . 1 .
note that provided the genus is at least two theorem [ thm : cyltorelli ] ( 2 ) holds also for closed surfaces and for surfaces with one boundary component .
this is in contrast to the situation in theorem [ mainthm2 ] .
2 . morita ( ( * ? ? ?
* corollary 5.2 ) ) used ` trace maps ' to show that the abelianization of @xmath812 has infinite rank .
theorem [ thm : cyltorelli ] ( 2 ) can therefore be seen as an extension of morita s theorem .
part ( 1 ) follows immediately from the proof of theorem [ mainthm ] since the examples @xmath813 , provided in the proof are easily seen to lie in @xmath51 .
we will now need the following claim : given any @xmath814 there exists @xmath815 with @xmath816 .
indeed , let @xmath261 be a homology cylinder over @xmath2 . as we saw in section [ section : actionh1 ] we have @xmath817 , and there exists @xmath818 such that the induced action on @xmath295 is given by @xmath807 .
now the homology cylinder @xmath819 acts as the identity on @xmath38 , i.e. @xmath179 defines an element in @xmath51 . on the other hand
it is clear that @xmath820 .
this concludes the proof of the claim .
we now turn to the proof of part ( 2 ) .
first suppose that @xmath25 and @xmath821 .
part ( 2 ) for this case follows from the proof of theorem [ mainthm2 ] since the examples @xmath822 , in the proof of theorem [ theorem : realization - z ] can be realized as elements in @xmath51 by the above claim .
when @xmath28 , following the arguments in section [ section : examples ] one can easily see that @xmath823 is isomorphic to @xmath824 and the desired result follows again from levine s work @xcite .
finally suppose that @xmath810 .
in this case , for @xmath825 , we consider the homology cylinder @xmath699 constructed using figure [ figure : example - attaching - curve ] modified in the following way : in figure [ figure : example - attaching - curve ] we have two boundary components , which are connected by @xmath451 , and one hole to which a generator @xmath826 is associated . now
remove the second boundary component and replace the first boundary component by a new hole , and denote by @xmath451 a closed curve which ` connects ' the two holes .
it follows from the discussion of section [ subsection : exmaple - handle - decomp ] and from the above claim that there exists a homology cylinder @xmath699 which lies in the torelli group @xmath51 and such that @xmath827 recall that the polynomials @xmath828 are irreducible .
also it is evident that the @xmath698 are non - symmetric ( i.e. , @xmath829 ) and @xmath830 for @xmath702 .
( note that @xmath831 , and therefore theorem [ mainthm2 ] can not be generalized for this case using our previous method . ) to detect @xmath832 , for each irreducible @xmath833 we use the homomorphism @xmath834 defined as in section [ section : algeraic structure ] with the following modification : @xmath835 the exponent of @xmath572 in @xmath252 .
then @xmath836 if @xmath837 and 0 otherwise . now following the lines of the proof of theorem [ mainthm2 ] , part ( 2 ) follows . note that theorem [ thm : cyltorelli ] ( 1 ) also implies that if @xmath838 , then @xmath51 is neither finitely generated nor finitely related , and it is not torsion - free . also for the structure of the group @xmath839 , redefining @xmath840 as we did for @xmath590 in the proof of theorem [ thm : cyltorelli ]
, one can easily obtain the following analogue of theorem [ theorem : algebraic - classification ] : [ theorem : algebraic - classification2 ] suppose @xmath38 is nontrivial , i.e. , @xmath2 is neither a sphere nor disk
. then @xmath841
we conclude with a short list of questions : 1 . study the structure of the kernel of @xmath842 2 .
does the abelianization of the group @xmath843 have infinite rank for @xmath32 and @xmath844 ? 3 .
characterize the image of the homomorphisms @xmath845 , @xmath846 and @xmath847 .
4 . can we realize any element in the image of @xmath589 by using theorem [ theorem : string - link - sum - formula ] ?
does the abelianization of @xmath11 contain @xmath83torsion ?
does it contain any other torsion ? 6 .
let @xmath32 and @xmath53 . does there exist a homomorphism @xmath848 onto an abelian group of infinite rank such that the restriction of @xmath18 to the kernel of the projection map @xmath15 is also surjective ?
does there exist a monoid homomorphism from @xmath81 onto a non - abelian monoid which vanishes on @xmath6 ? 8 .
is the group @xmath849 non - abelian ?
j. s. birman and h. m. hilden , _ on the mapping class groups of closed surfaces as covering spaces _
, advances in the theory of riemann surfaces ( proc .
stony brook , n. y. , 1969 ) , pp .
81115 . ann . of math .
studies , no .
princeton univ . press , princeton , n. j. , 1971 .
s. garoufalidis and j. levine , _ tree - level invariants of three - manifolds , massey products and the johnson homomorphism _ , graphs and patterns in mathematics and theorical physics , proc
pure math .
* 73 * ( 2005 ) , 173205 .
s. morita , _ cohomological structure of the mapping class group and beyond _ , in problems on mapping class groups and related topics ( edited by b. farb ) , proceedings of symposia in pure mathematics 74 , american mathematical society ( 2006 )
|
.7em garoufalidis and levine introduced the homology cobordism group of homology cylinders over a surface .
this group can be regarded as an enlargement of the mapping class group .
using torsion invariants , we show that the abelianization of this group is infinitely generated provided that the first betti number of the surface is positive .
in particular , this shows that the group is not perfect .
this answers questions of garoufalidis - levine and goda - sakasai .
furthermore we show that the abelianization of the group has infinite rank for the case that the surface has more than one boundary component .
these results hold for the homology cylinder analogue of the torelli group as well
. subjclassname@1991=subjclassname@2000=
|
the method of x - ray absorption spectroscopy ( xas ) is one of the widely used spectroscopic tools for electronic structure investigations in condensed matter physics @xcite .
it is based on the absorption of the x - ray photons when core shell electron is transferred into the unoccupied valence band states ( conduction band ) above the fermi level ( @xmath3 ) ( fig .
[ fig : scheme ] ) .
therefore one or more jumps ( absorption edges ) are usually observed in the absorption spectrum .
moreover the energy position of any edge is element specific since it coincides with the energy of the corresponding atomic core level . besides
, x - ray transitions are controlled by the dipolar selection rules and for well - defined atomic symmetry of core hole the final state angular momenta are selected in xas .
strong spatial localisation of the initial core shell state leads also to the site specific behaviour of xas .
generally , appearance of the fine structure of xas of solids can be described in the framework of photoelectron scattering model . near
the edge photoelectron energy is low and photoelectron de broglie wave undergoes multiple scattering on neighbouring atoms .
this leads to the origin of the reach and intense fine structure , or near edge x - ray absorption fine structure , nexafs . on the other hand , in the large photoelectron energies region ( far from the xas edge ) scattering of the photoelectron waves
gives rise to rather simple and weak variation of absorption coefficient , or extended x - ray absorption fine structure , exafs , well described by single scattering theory .
it is considered that the range of nexafs covers up to 50 ev above the absorption edge and the rest high photon energy part of xas extended up to 1000 ev above the threshold belongs to exafs .
the latter method is very effective tool for the short range atomic ordering determination in multiatomic systems including noncrystalline solids . in nexafs spectroscopy ,
the topic of the present paper , absorption spectra can be interpreted on the basis of the corresponding molecular orbital or band - structure calculations above @xmath3 .
this way is equivalent to the interpretation of xas data in the frame of photoelectron waves multiple scattering model , transparent and demonstrative to date . to measure spectral distribution of the x - ray absorption coefficient in massive ( bulk ) solids yield spectroscopy
is usually used the dependence on photon energy of the total photocurrent ( so called total electron yield , tey ) . since external photo - effect bunch , which accompanied x - ray absorption event ,
consists of primary fast photo- and auger - electrons together with the cloud of slow secondary electrons , the mean drift way of the whole electron bunch is govern by the relatively large mean free path of slow secondary electrons @xcite . in order to increase surface sensitivity of measurements ( for the better representation of the surface - related electronic structure )
the negative repulsive potential can be applied to the grid - electrode placed in front of the detector to suppress slow kinetic energy component in the electron bunch distribution .
this experimental mode gives so - called partial electron yield ( pey ) . in the case of pey nexafs with the linearly polarized light ,
the method can be used for the determination of the orbitals spatial orientation .
generally , one can observe the so - called _ search - light - like effect _
@xcite , which can be used for probing the quadrupole moment of the local charge around the absorbing atom . in such xas experiment ,
the absorption intensity associated with a specific molecular orbital final state has a maximum if the electric field vector is aligned parallel to the direction of maximum charge or hole density , i.e. along a molecular orbital , and the intensity vanishes if the electric field vector is perpendicular to the orbital axis .
a detailed description of the angular dependence of nexafs intensities can be found elsewhere @xcite . in case of magnetic samples and
circularly polarized light the absorption coefficient depends on the relative orientation of photon spin and magnetization direction of the sample .
the quantitative analysis of the absorption spectra obtained on the magnetic sample with the circularly polarized x - rays can be performed with the help of the so - called magneto - optical sum rules for evaluation of spin- and orbital - magnetic moments @xcite .
compared to the relative simplicity of the nexafs experiment , the theoretical description of x - ray absorption process in solid is not that easy .
the main problem here is that the excited system with the core hole has to be adequately described , which is always a difficult task for the standard density functional theory ( dft ) calculations .
particularly it is related to graphite and graphene - based systems .
the nexafs spectrum of graphite or graphene can be separated on two energy regions ( see fig . [
fig : scheme ] ) : ( i ) @xmath4 contribution originating from transition of @xmath5 core electron on c@xmath6 orbitals and ( ii ) @xmath7 one originating from transition of @xmath5 core electron on c@xmath8 hybrid orbitals @xcite .
as was recently shown @xcite , although the @xmath0 resonance was well reproduced in the framework of the core - hole final state approximation , the shape as well as the intensity of the @xmath1 resonance were not reproduced .
only the application of the mahan - noziers - de dominicis ( mnd ) theory of dynamical core - hole screening gives an adequate description of the nexafs spectrum of graphite @xcite .
it has to be noted here that in the double - peak structure of the @xmath1 resonance , the first peak ( @xmath9 ) , which reflects the excitonic final state can be reproduced in the final state and mnd approaches , and second @xmath10 peak is described as due to the quickly completely screened initial - state density of states .
the present manuscript is devoted to the spectroscopic studies of the graphene - metal systems .
the properties of these objects are the subject of many experimental and theoretical studies which were recently reviewed in several works @xcite .
as discussed the graphene / metal interface is an important object of the graphene - related fields of research . beyond the practical utilisation of these objects ( ideal ways to prepare the huge layers of graphene , graphene - based spin filters and gas sensors , etc . )
@xcite , there is also a fundamental interest to the nature of the interaction between graphene and metal @xcite .
as the nexafs spectroscopy can be used for the investigation of the empty electronic states of material in the vicinity of @xmath3 , it can be utilised as a complimentary to photoemission spectroscopy method , which can shed more light on the nature of interaction between graphene and a metal .
here we revisited the theoretical description of the absorption spectroscopy for graphene - based systems .
we used the scheme described in refs .
@xcite and our computer code for the modelling of nexafs spectra of the graphene - based systems : graphene single- and bilayer on ni(111 ) and graphene / al / ni(111 ) intercalation - like system .
our results , starting from the test example ( graphene ) and for more complex systems , are discussed on the basis of effects in the electronic structure of the studied objects and compared with available experimental results .
very good agreement between theory and experiment is found in our analysis .
the x - ray absorption spectroscopy , being one of the oldest and most efficient spectroscopic tools to investigate the electronic structure of materials , involves a very complicated dynamical many - body process , so that an adequate theoretical treatment of the experimental data appears to be a very complicated problem .
when the x - ray photon is absorbed , the electron from the inner shell excites into the unoccupied states of the conduction band where electronic states are described by wannier functions localized on the absorbed atom . the contribution of transitions into the states centred at neighbouring atoms is considered to be small enough to be neglected .
there are three main approaches for the description the x - ray absorption process .
generally accepted are first two one - electron ways of nexafs interpretation in terms of the initial state or the final state of system during x - ray absorption .
less common is the third way in which many - body processes tracked the x - ray transitions are taking into account .
these ways are different one of other sonly in approximations of core - hole screening : * the first approach assumes that the time of the interaction of the x - ray photon with the electronic subsystem is much shorter than any typical process in the system ( initial - state approximation ) . *
the second approach assumes that the photon - electron interaction is adiabatically slow that the system has a time for transfer from the ground state into a new one ( final - state approximation ) .
* the mnd theory assumes the creation of the core - hole as fast enough process , but takes into account the real dynamics of the response of the electronic subsystem to a core hole creation ( mnd approximation ) . in the initial - state approach ,
the first principle calculations of the electronic structure of material are performed with vasp code ( for details , see sec .
[ dft ] ) .
then , we obtain the projections of the bloch wave function onto the spherical harmonics centred on site @xmath11 ( atomic number @xmath12 ) and construct the matrix of initial - state green s functions as following : @xmath13 where @xmath14 , @xmath15 are angular momentum for ( @xmath16 , @xmath17 , @xmath18 , and @xmath19 ) orbitals , @xmath20 - wave vector , @xmath21 - band index , @xmath22 - energy , @xmath23 - chemical potential , and @xmath24 - infinitely small value
. then the initial - state nexafs spectrum is given by equation : @xmath25 where @xmath26 , @xmath27 is the dipole moment operator , and @xmath28 - x - ray electric field . in our work
we used these approximations as follows .
we model the studied system with a large supercell [ i.e. @xmath29 in the case of free - standing graphene ] .
then we replace one atom by the following atom in the periodic table - the so - called @xmath30 approximation .
the density of states ( doss ) are obtained by means of vasp calculations performed for the large supercell and the bloch wave functions are projected onto spherical harmonics centred on the @xmath30-atom site to get an expression for the final state , and on the @xmath12-atom site to get an expression for the initial state . the final - state nexafs spectrum is calculated according to the equation similar to eqn .
( [ i ] ) , where the bloch wave functions are projected onto atom @xmath31 with an atomic number @xmath32 , and the final - state green s function @xmath33 is written in the same manner as eqn .
( [ g - mn ] ) .
the mnd theory considers the creation of core - hole as an instant process taking into account real dynamics of the response of the electronic subsystems . as a result
many - electrons effects due to correlations of electrons with core - hole are included
. therefore we construct initial- and final - state green s functions as mentioned above and the so - called core - potential : @xmath34 the mnd spectrum is calculated according to the formula shown in the lower part of fig .
[ fig : xas - scheme ] .
the formulas for the calculation of the two - particles green s function @xmath35 $ ] and core - hole finite lifetime @xmath36 $ ] are presented in the left - hand side of the scheme in fig .
[ fig : xas - scheme ] . the full description of the calculation algorithm can be found in refs .
@xcite and in the supplementary material @xcite .
the dft calculations were carried out using the projector augmented wave method @xcite , a plane wave basis set and the generalized gradient approximation as parameterized by perdew _ et al . _
@xcite , as implemented in the vasp program @xcite .
the plane wave kinetic energy cutoff was set to @xmath37ev .
the long - range van der waals interactions were accounted for by means of a semiempirical dft - d2 approach proposed by grimme @xcite .
the corresponding structures of the graphene - metal based systems are shown in fig .
[ fig : structure ] and they are discussed in details further in the text . the supercell used to model the graphene - metal interface is constructed from a slab of @xmath38 layers of nickel atoms with a graphene layer ( or bilayer ) adsorbed at both sides and a vacuum region of approximately @xmath39 . in the case of graphene / al / ni(111 ) intercalation - like system , the supercell has @xmath40 lateral periodicity and contains by one al layer ( 3 atoms each ) introduced between graphene and ni(111 ) on both sides of the slab . in optimizing the geometry ,
the positions ( @xmath41-coordinates ) of the carbon atoms as well as those of the top two layers of metal atoms are allowed to relax . in the total energy calculations and during the structural relaxations the @xmath42-meshes for sampling the supercell brillouin zone are chosen to be as dense as @xmath43 and @xmath44 , respectively . to obtain an input for the nexafs spectra simulations , we employed large supercells of @xmath45 and @xmath46 periodicity when studying free - standing graphene and graphene / metal systems , respectively . in the latter case the number of nickel layers was reduced to @xmath47 . in order to have the same position for the fermi level , both , initial and final states doss ,
are taken from the same calculation sets .
in this part we present the results of the application of the computational procedure discussed in the previous section for the studies of a graphene layer on metallic support .
we consider several objects in which the strength of interaction of graphene with metal is ranged from `` strong '' [ graphene / ni(111 ) ] to `` weak '' [ graphene / al / ni(111 ) and graphene - bilayer / ni(111 ) ] . in order to prove our implementation of the algorithm discussed
earlier we performed the calculation of nexafs spectra for the free standing graphene ( figs .
[ fig : fsg ] and [ fig : theoryvsexp ] ) .
the main peaks of the graphene spectrum at @xmath48ev and @xmath49ev above @xmath3 correspond to transitions of the @xmath5 core electron onto the @xmath0 and @xmath1 states , respectively .
[ fig : fsg ] shows the theoretical spectra calculated according to the procedure described in sec .
[ comput ] for ten different incidence angles @xmath11 .
one can note a smooth decreasing ( increasing ) weights of @xmath50 ( @xmath51 ) transitions when going from @xmath52 to @xmath53 , that is in accordance with geometry of the absorption process .
this example is a nice demonstration of the _ search - light - like _ effect discussed earlier : for the grazing incident angles the maximum of absorption is observed for @xmath0 states and for angles close to the normal to the surface the maximum of absorption is detected for @xmath51 transitions .
furthermore , it is clearly seen that both the shape and positions as well as relative intensities of the peaks are well reproduced by theory [ for comparison between calculated and experimental spectra @xcite , see fig .
[ fig : theoryvsexp ] ( upper panel ) ] . as was discussed earlier in refs .
@xcite the shape as well as the position of @xmath54 and @xmath55 transitions are well reproduced by the mnd theory described in fig .
[ fig : xas - scheme ] . the second peak @xmath10 can be reproduced only if fully screened initial states density of states is taken into account and the corresponding initial states nexafs spectrum is summed with mnd spectrum in order to reproduce the experimental observations .
before investigating modifications in the absorption spectra due to the presence of the ni(111 ) substrate underneath graphene , let us consider briefly the atomic structure of the interface .
the surface lattice constant of ni(111 ) matches the lattice constant of graphene almost perfectly and therefore @xmath56 structure is formed .
the structure of graphene / ni(111 ) widely accepted in the literature has the carbon atoms arranged in the so - called _ top - fcc _ configuration on ni(111 ) : one carbon atom from the graphene unit cell is placed above the ni interface atom and the second one occupies the hollow _ fcc _ site of the ni(111 ) surface ( fig .
[ fig : structure ] ) .
our calculations support this model ( see supplementary tab .
s1 @xcite ) .
the equilibrium distance between graphene layer and ni(111 ) is relatively small , indicating `` strongly '' bonded system : significant hybridization of the valence band states of graphene and ni leads to the complete destruction of the graphene dirac cone @xcite .
[ this effect of hybridization is clearly seen when considering electronic band structure of the system ( see supplementary fig .
s1 @xcite ) . the modifications in the electronic structure are considered in details in ref .
@xcite ] .
consequently , the nexafs spectra in this case are modified with respect to those of free - standing graphene [ fig.[fig : theoryvsexp ] ( middle panel ) ] .
firstly , in the region of the @xmath50 transition the graphene / ni(111 ) absorption spectra have a double - peak structure compared to those of graphite that is explained by the transitions of c @xmath5 core electron into two unoccupied states ( interface states ) , which are the result of hybridization of the graphene @xmath57 and ni @xmath58 valence band states .
the first peak at about @xmath59ev corresponds to the interface state which is allocated in the range between @xmath60 and @xmath61ev above @xmath3 around @xmath2-point of the brillouin zone , having slightly different energy depending on the type of c - atom ( @xmath62 or @xmath63 ) and on the spin - orientation .
the second peak , near @xmath64ev , comes mainly from the unoccupied interface state at the @xmath65 point ( see supplementary figs . s1 and s2 @xcite ) . from this analysis one
can clearly see that both peaks belong to the hybrid states where both carbon atoms are involved .
therefore the previous analysis of nexafs spectra for the graphene / ni(111 ) system performed in refs .
@xcite on the basis of calculations of the @xmath2-edge electron loss spectra @xcite is not fully correct ( the double peak structure was assigned to different carbon atoms in the unit cell ) .
secondly , the visible reduction in the energy separation between the @xmath0 and @xmath1 features are observed for all graphene / ni(111 ) spectra compared to that in the spectra of free - standing graphene , in accordance with the shift of the @xmath1 state for the carbon atoms at the interface by ca.@xmath61ev to lower energies that can be explained by the lateral bond softening within the adsorbed graphene monolayer and can be connected with the downward shift of @xmath1 states for graphene / ni(111 ) compared to free - standing graphene . in the graphene / al / ni(111 ) system ,
the _ fcc _ al(111 ) monolayer is placed in the space between graphene and ni(111 ) ( fig .
[ fig : structure ] ) .
there is only one possible arrangement of aluminium atoms in the high - symmetry positions as shown in the figure : the al(111 ) lattice plane is rotated by @xmath66 with respect to the graphene / n(111 ) lattice such that the al atoms occupy all three kind of high - symmetry adsorption sites ( _ top _ , hollow _ fcc _ ,
hcp _ ) in the space between the graphene layer and the ni(111 ) surface .
in this structure there are four different occupation sites for carbon atoms in the unit cell with respect to the adsorption sites of the ni(111 ) surface when they are placed above a ni or al atom .
they are labeled in fig .
[ fig : structure ] as @xmath62 , @xmath63 , @xmath67 , and @xmath68 , respectively .
a mean distance between the graphene and al layer is @xmath69 , indicating `` weak '' bonded graphene - metal system @xcite .
the latter fact can be evaluated when considering the electronic band structure of the system under study : all electronic bands of graphene are shifted to lower binding energies compared to graphene / ni(111 ) .
furthermore , the electronic structure of the graphene layer as well as the dirac cone in the vicinity of @xmath3 are fully restored ( there is a small electron doping of graphene leading to a shift of the dirac point below @xmath3 by ca.@xmath70ev ) .
thus , intercalation of al decouples the electronic structure of graphene from the substrate , preventing thereby hybridization between graphene @xmath57 and ni @xmath58 valence band states . in the present system
there is no strong hybridization between graphene and al valence band states conserving the dirac cone .
there is only one energy region where such hybridization is visible : around the @xmath65 point at approximately @xmath71ev ( see ref .
@xcite ) .
the decoupling process in this case is also easily visible when considering the nexafs spectra obtained for the system of interest .
intercalation of a thin al layer underneath a graphene layer on ni(111 ) leads to drastic changes in the absorption spectrum compared to the one for graphene / ni(111 ) : the shape of the spectrum , positions of main spectroscopic features as well as the energy separation between @xmath0 and @xmath1 features become similar to those in the spectrum of free - standing graphene [ fig .
[ fig : theoryvsexp ] ( lower panel ) ] . due to the avoiding hybridisation between the valence band states of graphene and metal in this case ,
the individual spectra evaluated for the c - atoms of the four types are indistinguishable from each other ( see supplementary fig .
s3 @xcite ) .
weak hybridisation between graphene @xmath0 and al @xmath72 states does not affect the result .
note : agreement between theory and experiment is again excellent .
the energetically most stable arrangement of the graphene layers above ni(111 ) in this case corresponds to the situation when carbon atoms of the first layer are adopting the _ top - fcc _ configuration [ as it is in the case of the single graphene layer deposited on ni(111 ) ] and the second graphene layer is displaced horizontally with respect to the first graphene layer as it is in bulk graphite in the way , that carbon atoms from the graphene unit cell are occupying the both hollow sites with respect to the ni(111 ) surface [ fig .
[ fig : structure ] ( right - hand side figure ) ] .
the resulting configuration can therefore be named as _ top - fcc / fcc - hcp _
( see supplementary tab .
s1 @xcite ) . in the studied system , the interaction energy between the top ( second ) graphene layer and graphene / ni(111 ) is @xmath73mev per graphene unit cell that is comparable with the value known for the interlayer interaction in graphite ( @xmath74mev @xcite ) .
therefore we can expect , that bilayer - graphene / ni(111 ) should display properties of the both systems , free - standing graphene and graphene / ni(111 ) .
when looking at the results of the electronic band structure calculations , one can see that , indeed , while valence band states of the first graphene layer and ni are strongly hybridized and dirac cone is destroyed ( see supplementary fig .
s4 ) , picture observed for the second graphene layer is much more similar to that of the free - standing graphene : there is no hybridisation visible and dirac cone is preserved ( see supplementary fig .
s5 @xcite ) . at the same time
, a band gap of ca.@xmath75ev is opened at the @xmath2-point at @xmath76ev .
there is , however , no significant difference between pictures observed for the empty states of the second graphene layer and those of the free - standing graphene .
the same is true for the empty states of the first graphene layer in this system and graphene in graphene / ni(111 ) .
this is reflected in the nexafs spectra calculated for the bilayer : while results obtained for the first graphene layer are identical with those of graphene / ni(111 ) , the nexafs spectra of the second layer are similar to those obtained for the free - standing graphene as well as graphene in the graphene / al / ni(111 ) trilayer [ fig .
[ fig : bilayer ] ( left panel ) ] . as discussed earlier
the nexafs spectra can be collected in two modes , more bulk - sensitive tey and more surface - sensitive pey modes . here , the variation of the repulsive potential ( which suppress the secondary electron tail ) applied to the electrode before electron detector allows for the gradual variation of the sensitivity ( surface vs bulk ) .
[ fig : bilayer ] ( right panel ) shows the nexafs spectra of graphene - bilayer / ni(111 ) where the sensitivity is varied from tey mode ( lower spectrum ) to pey mode ( upper spectra ) .
the clear change of the shape of the spectra to the one characteristic for the free - standing graphene is visible .
these simulations allow to discriminate between single and bilayer graphene in the investigated system .
here we presented the theoretical description of the spectra of the graphene layers on metallic substrate .
the interaction in these systems is varied from `` strong '' ( when the electronic structure of graphene in the vicinity of @xmath3 is strongly modified ) to `` weak '' ( when the electronic structure of the graphene layer is a free - standing - graphene like ) .
all these variations in interaction and in electronic structure are consequently reflected in the simulated nexafs spectra which can be used as a fingerprint for the description of interaction at the graphene / metal interface .
the presented results are compared with the available experimental data and they show the excellent agreement between theory and experiment .
we appreciate the support from the german academic exchange service ( daad ) through the center of excellence : `` german - russian interdisciplinary science center ( g - risc ) '' .
the high performance computing network of northern germany ( hlrn ) is acknowledged for computer time .
s. bae , h. kim , y. lee , x. xu , j .- s .
park , y. zheng , j. balakrishnan , t. lei , h. r. kim , y. i. song , y .- j .
kim , k. s. kim , b. ozyilmaz , j .- h .
ahn , b. h. hong and s. iijima , nature nanotech . * 5 * , 574578 ( 2010 ) .
the scheme of the x - ray light absorption process .
the light from the synchrotron light source is scanned around the energy corresponding to the binding energy of the particular core level and in this case the absorption coefficient is proportional to the density of the valence band states above @xmath3 .
the orbital selectivity is reached via relative orientation of the linearly polarised light and the sample.,scaledwidth=42.0% ] nexafs spectra for a free - standing graphene as a function of the angle , @xmath11 , between electric field vector of x - ray and the normal to the surface .
the experimental geometry is shown as an inset.,scaledwidth=48.0% ] nexafs spectra for @xmath77 for the graphene - bilayer / ni(111 ) system .
spectra are separated on the contributions from the 1st and 2nd layer , respectively .
( right panel ) sequence of the simulated c @xmath2 nexafs spectra for graphene - bilayer / ni(111 ) , where the contribution from both layers is increased / decreased from top to bottom in order to model the change of surface - vs - bulk sensitivity in experimental data.,scaledwidth=96.0% ] supplementary material for the manuscript : + * theoretical description of x - ray absorption spectroscopy of the graphene - based systems * + elena voloshina,_@xmath78
_ roman ovcharenko,_@xmath79 _ alexander shulakov,_@xmath79 _ and yuriy dedkov__@xmath80 _ _ + _ @xmath78humboldt - universitt zu berlin , institut fr chemie , 12489 berlin , germany _ + _ @xmath79 _ +
_ @xmath80specs surface nano analysis gmbh , voltastrae 5 , 13355 berlin , germany _ * content : * * description of the calculation algorithm of the nexafs spectra . * table s1 : results for the graphene / ni(111 ) and graphene - bilayer / ni(111 ) . all adsorption energies ( ae ) are given per graphene unit cell .
* figure s1 : calculated majority and minority spin band structures ( black ) for a slab terminated by graphene / ni(111 ) interface ( _ top - fcc _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character as a weighting factor are superimposed .
contributions from the @xmath81 and @xmath82 are separated .
* figure s2 : calculated c @xmath2 nexafs spectra for @xmath77 for the graphene / ni(111 ) system .
spectra are separated on the contributions from the @xmath81 and @xmath82 , respectively .
* figure s3 : calculated c @xmath2 nexafs spectra for @xmath77 for the graphene / al / ni(111 ) system .
spectra are separated on the contributions from the @xmath83 , @xmath84 , @xmath85 , and @xmath86 , respectively .
* figure s4 : calculated majority and minority spin band structures ( black ) for a slab terminated by graphene - bilayer / ni(111 ) interface ( _ top - fcc / fcc - hcp _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character of the first graphene layer as a weighting factor are superimposed .
contributions from the @xmath81 and @xmath82 are separated .
* figure s5 : calculated majority and minority spin band structures ( black ) for a slab terminated by graphene - bilayer / ni(111 ) interface ( _ top - fcc / fcc - hcp _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character of the second graphene layer as a weighting factor are superimposed .
contributions from the @xmath87 and @xmath82 are separated . * figure s6 : calculated c @xmath2 nexafs spectra for @xmath77 for the free - standing graphene as obtained by means of different approximations ( sk = slater - koster method ) . * figure s7 : calculated c @xmath2 nexafs spectra for @xmath77 for the graphene / ni(111 ) system as obtained by means of different approximations ( sk = slater - koster method ) . * figure s8 : calculated c @xmath2 nexafs spectra for @xmath77 for the graphene / al / ni(111 ) system as obtained by means of different approximations ( sk = slater - koster method ) .
* figure s9 : calculated c @xmath2 nexafs spectra for @xmath77 for the graphene - bilayer / ni(111 ) system as obtained by means of different approximations ( sk = slater - koster method ) . * description of the calculation algorithm of the nexafs spectra .
* let us consider the radiative x - ray transitions of an electron from the core level @xmath88 on the band states @xmath89 localized on the atom @xmath90 .
in the tight - binding approximation , neglecting the `` cross '' transition contribution to the intensity of radiation , the nexafs intensity @xmath91 can be represented as the sum of the products of the electric dipole probabilities of transition @xmath92 from the core level on the atom - like valence band state @xmath23 with partial density of states @xmath93 @xmath94 @xmath95 where the amplitude @xmath96 of transition is given by @xmath97 here @xmath98 and @xmath99 are the atomic wave functions of valence electron and core hole , respectively . in the framework of the one - particle approximation , the partial density of states ( pdos ) can be expressed via the imaginary part of the advanced green s function projected onto the atomic orbitals @xmath100 @xmath101 completely ignoring the influence of the core hole , the green s function can be written as @xmath102 with @xmath103 the fermi energy , @xmath104 and @xmath105 the unperturbed band energy and wave function , respectively .
it may be considered as an assumption that the lifetime of the core hole is much smaller than the relaxation time of the system ( an approximation of the final state ) .
another extreme case takes place when the band wave functions are completely changed because of the core hole coulomb potential ( initial state approximation ) .
then the expression for the green s function takes the form @xmath106 where @xmath107 and @xmath108 are perturbed band energies and wave functions , respectively . for pdos in the presence of the core hole one
gets @xmath109 mnd theory considers the core hole filling completely in the dynamical way .
the expression for the nexafs intensity within the mnd theory in the form @xmath110 where the matrix @xmath111 satisfies the system of integral equations @xmath112 the kernel of the integral operator @xmath113 is given by @xmath114 where @xmath116 and @xmath117 have the form @xmath118 note that the diagonal elements of matrixes @xmath119 and @xmath117 coincide with the pdos defined above in eqs.([dens1 ] ) and ( [ dens2 ] ) in the final and initial state approximations .
there are two ways to define input data for the mnd calculations expressed by eqs.([i_mnd])([i_fun ] ) .
the first approach is to use the expressions ( [ g_final ] ) , ( [ g_initial ] ) , and ( [ core_potential ] ) to introduce the green functions @xmath120 and the core potential @xmath121 .
it works well if the band wave functions basis set is large enough .
if it is not the case , the core hole potential @xmath121 depends on energy . to minimize the error caused by the energy dependence of the potential
one can use a second approach .
the unperturbed green function @xmath122 and core hole potential @xmath121 defined from equations ( [ g_initial ] ) and ( [ core_potential ] ) , respectively , whereas the perturbed green function @xmath123 is obtained from the slater - koster equation @xmath124_{\nu ' \nu},\ ] ] where @xmath125 although the potential still depends on energy , the perturbed green function now takes it into account . llll system & @xmath126 ( ev)&@xmath127 ( ev)&@xmath128 ( ev ) + + _ top - fcc _ & & @xmath129 & + _ top - hcp _ & & @xmath130 & + _ fcc - hcp _ & & @xmath131 & + & & & + + _ top - fcc / top - hcp _ & @xmath132&@xmath133&@xmath134 + _ top - fcc / fcc - hcp _ &
@xmath135&@xmath136&@xmath137 + _ top - fcc / top - fcc _ & @xmath138&@xmath139&@xmath140 + & & & + _ top - hcp / top - fcc _ & @xmath141&@xmath142&@xmath143 + _ top - hcp / fcc - hcp _ & @xmath135&@xmath144&@xmath145 + _ top - hcp / top - hcp _ &
@xmath138&@xmath146&@xmath147 + & & & + _ fcc - hcp / top - hcp _ &
@xmath148&@xmath149&@xmath150 + _ fcc - hcp / top - fcc _ & @xmath151&@xmath152&@xmath153 + _ fcc - hcp / fcc - hcp _ & @xmath153&@xmath154&@xmath155 + * figures1 .
* calculated majority and minority spin band structures ( black ) for a slab terminated by graphene / ni(111 ) interface ( _ top - fcc _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character as a weighting factor are superimposed .
contributions from the @xmath81 and @xmath82 are separated .
* figures3 . * calculated c @xmath2 nexafs spectra for @xmath77 for the graphene / al / ni(111 ) system .
spectra are separated on the contributions from the @xmath83 , @xmath84 , @xmath85 , and @xmath86 , respectively .
* figures4 .
* calculated majority and minority spin band structures ( black ) for a slab terminated by graphene - bilayer / ni(111 ) interface ( _ top - fcc / fcc - hcp _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character of the first graphene layer as a weighting factor are superimposed .
contributions from the @xmath81 and @xmath82 are separated .
* figures5 .
* calculated majority and minority spin band structures ( black ) for a slab terminated by graphene - bilayer / ni(111 ) interface ( _ top - fcc / fcc - hcp _ arrangement ) .
the bands replotted in red ( blue ) using the carbon @xmath6 ( @xmath8 ) character of the second graphene layer as a weighting factor are superimposed .
contributions from the @xmath87 and @xmath82 are separated .
|
the present manuscript considers the application of the method of the near - edge x - ray absorption spectroscopy ( nexafs ) for the investigation of the graphene - based systems ( from free - standing graphene to the metal - intercalation - like systems ) .
the spectra for the selected systems are calculated in the framework of the approach , which includes the effects of the dynamic core - hole screening .
the presented spectral changes from system to system are analysed with the help of the corresponding band - structure calculations .
the obtained results are compared with available experimental data demonstrating the excellent agreement between theory and experiment .
the direct correlation between the strength of the graphene interaction with the metallic substrate and the spectral distributions ( shape and intensities of @xmath0 and @xmath1 features in the c @xmath2 nexafs spectra ) is found that can be taken as a fingerprint for the description of interaction at the graphene / metal interface .
nexafs / xes / xmcd project :
|
it is well known that the electroweak standard model and many of its extensions have the potential to support string like configurations that are the particle physics analogs of vortices or magnetic flux tubes in condensed matter physics .
such objects are usually called _ cosmic strings _ to distinguish them from the fundamental variables in string theory , and also to indicate that they typically stretch over cosmic length scales .
they are also called electroweak strings or @xmath0strings@xcite@xcite because the @xmath0component of the electroweak gauge boson acquires the structure of the abelian nielsen olesen vortex@xcite .
such strings may have emerged copiously in the early universe at interfaces between regions of different vacuum expectation values of the higgs field(s ) in several stages after the big bang until electroweak symmetry breaking was reached . if strings are absolutely stable they should have survived and we should be able to observe them today . in the electroweak standard model
topologically stable strings are ruled out , but the absence of this particular stabilizing mechanism does not imply that electroweak strings are unstable or irrelevant for particle physics . while their direct gravitational effects are negligible , @xmath0strings can still be relevant for cosmology at a sub dominant level@xcite .
their most interesting consequences originate , however , from their coupling to the standard model fields .
@xmath0strings provide a source for primordial magnetic fields@xcite and they also offer a scenario for baryogenesis with a second order phase transition@xcite . in contrast , a strong first order transition as required by the usual bubble nucleation scenario is unlikely in the electroweak standard model@xcite without non - standard additions such as supersymmetry or higher dimensional operators@xcite
. these interesting effects are only viable if @xmath0strings are energetically stabilized by their coupling to the remaining quantum fields .
the most important contributions are expected to come from ( heavy ) fermions , since their quantum energy dominates in the limit @xmath1 , where @xmath2 is the number of qcd colors or of any other internal degree of freedom .
the dirac spectrum in typical string backgrounds is deformed to contain either an exact or near zero mode , so that fermions can substantially lower their energy by binding to the string .
this binding effect can overcome the classical energy required to form the string background .
however , the remaining spectrum of modes is also deformed and for consistency its contribution ( the vacuum polarization energy ) must be taken into account as well .
heavier fermions are expected to provide more binding since the energy gain per fermion charge is higher ; a similar conclusion can also be obtained from decoupling arguments@xcite .
a number of previous studies have investigated quantum properties of string configurations .
naculich@xcite has shown that in the limit of weak coupling , fermion fluctuations destabilize the string .
the quantum properties of @xmath0strings have also been connected to non perturbative anomalies@xcite .
the emergence or absence of exact neutrino zero modes in a @xmath0string background and the possible consequences for the string topology were investigated in ref . .
a first attempt at a full calculation of the fermionic quantum corrections to the @xmath0string energy was carried out in ref . .
those authors were only able to compare the energies of two string configurations , rather than comparing a single string configuration to the vacuum because of limitations arising from the non trivial behavior at spatial infinity .
we will discuss a solution to this obstacle in detail below .
the fermionic vacuum polarization energy of the abelian nielsen
olesen vortex has been estimated in ref .
with regularization limited to the subtraction of the divergences in the heat kernel expansion .
quantum energies of bosonic fluctuations in string backgrounds were calculated in ref . .
finally , the dynamical fields coupled to the string can also result in ( abelian or non abelian ) currents running along the core of the string .
the time evolution of such structured strings was studied in ref .
, where the current was induced by the coupling to an extra scalar field .
this presentation is based on a number of publications@xcite@xcite regarding the contribution of fermion quantum corrections to the vacuum polarization energy of a straight and infinitely long cosmic string that appeared over the last year or so .
technical details underlying the results given here can be learned from those publications , _ cf . _ in particular the appendices of ref . .
for the current investigation the fermion doublet will be considered degenerate so that the introduction of a matrix notation for the higgs field is appropriate
. then the string configuration reads @xmath3 for the higgs field and @xmath4 for the gauge boson ( in temporal gauge ) .
the variables @xmath5 and @xmath6 are polar coordinates in the plane perpendicular to the string , while the higgs vacuum expectation value @xmath7 and the gauge coupling constant @xmath8 are model parameters .
the string configuration involves profile functions @xmath9 and @xmath10 which are the analogs of the nielsen olesen vortex profiles with boundary conditions @xmath11 the integer @xmath12 is the winding of the string , for which we will typically take @xmath13 in numerical calculations . finally , the ( variational ) parameter @xmath14 measures the relative strength of the higgs and gauge boson components of the string .
we will consider a modified version of the electroweak standard model which only has a @xmath15 gauge symmetry .
this modification is equivalent to a vanishing weinberg angle .
then the ( classical ) boson part of the lagrangian reads @xmath16 with the covariant derivative @xmath17 and the @xmath18 field strength tensor @xmath19\ , .
\label{fieldtensor}\ ] ] the boson masses are determined from @xmath8 and @xmath7 and the higgs self
coupling @xmath20 as @xmath21 and @xmath22 for the gauge and higgs bosons , respectively .
the interaction with the fermion doublet is described by the lagrangian @xmath23 upon spontaneous symmetry breaking the yukawa coupling @xmath24 induces a fermion mass @xmath25 .
the similarity with the standard model of particle physics suggests the model parameters @xmath26 where we have taken the fermion doublet to have the mass of the top quark . in our numerical search for a stable string
, we will also study other model parameters , _ cf .
_ section [ sec : charged ] . the classical energy per unit length of the string is solely governed by @xmath27 : @xmath28 \cr\cr & & \hspace{3 cm } + \frac{f_h^{\prime2}}{f^2 } + \frac{\mu_h^2}{4f^2}\left(1-f_h^2\right)^2\biggr\}\ , , \label{eq : ecl}\end{aligned}\ ] ] where the dimensionless radial integration variable is related to the physical radius by @xmath29 and @xmath30 .
the central object of the present investigation is the fermion contribution to the energy .
it will obtained from the solutions to the dirac equation . ] in the two dimensional plane perpendicular to the string @xmath31 where the single particle hamiltonian , @xmath32 is extracted from @xmath33 .
the interaction part , @xmath34 depends on the chosen gauge .
a specific choice of gauge , which accommodates the subtleties at large @xmath5 , will be discussed in sec .
[ subsec : gauge ] .
the spectrum of the dirac operator consists of bound state solutions with discrete eigenvalues @xmath35 and continuous scattering solutions whose eigenvalues @xmath36 are labeled by momentum @xmath37 , i.e. @xmath38 . finally
, the dirac hamiltonian for the string background anti commutes with @xmath39 .
thus the spectrum is charge conjugation invariant and it suffices to consider the non negative eigenvalues .
the vacuum polarization energy is the renormalized sum of the changes of the zero point energies of fermions in the background of a static configuration @xmath40 here @xmath41 are the energy eigenvalues in the presence of the string as obtained from the dirac equation ( [ eq : dirac ] ) , and @xmath42 are their free counterparts .
the overall sign has been chosen to describe fermionic vacuum fluctuations ; for bosons it must be altered . in the second part of equation ( [ eq : def ] )
the changes of the single particle energies are re written as the contribution from distinct bound states ( @xmath43 ) and the ( renormalized ) change of the density of scattering states , @xmath44 .
these states obey the standard dispersion relation @xmath45 with the mass @xmath46 of the quantum fluctuations . in the above
, the factor @xmath47 has been made explicit to stress that @xmath48 is a quantum effect .
as in the previous section natural units ( @xmath49 and @xmath50 ) will be adopted in the remainder of this paper . the computation of the fermion contribution to the vacuum polarization energy of a cosmic string proceeds in three stages
first , spectral methods are employed to express this energy in form of scattering data@xcite .
the equivalence of terms in the born series and feynman diagrams makes it possible to impose standard renormalization conditions . in a second step
this approach is extended to accommodate configurations that are translationally invariant in one or more spatial directions , using the so called interface formalism@xcite .
third , we have to cope with the fact that the cosmic string configuration has a non trivial structure at spatial infinity . because the fields approach a pure gauge rather than zero ,
the direct application of spectral methods is impossible .
rather a particular subset of gauges has to be adopted to define a well behaved scattering problem@xcite .
the basic idea of the spectral approach is to express @xmath44 as the momentum derivative of the phase shifts of the scattering states@xcite .
then @xmath51_{\rm ren}\ , .
\label{eq : attempt1}\ ] ] it is implicitly assumed that the system allows for a partial wave decomposition with degeneracy factor @xmath52 for the partial wave of ( generalized ) angular momentum @xmath53 . the major concern in eq .
( [ eq : attempt1 ] ) is renormalization . as it stands , in three spatial dimensions the vacuum polarization energy , eq .
( [ eq : attempt1 ] ) , is quadratically divergent in the ultra violet . for large momenta
the born series adequately represents the phase shift .
we therefore consider @xmath54_n\ , , \label{eq : attempt2}\ ] ] where the subscript on the phase shift indicates the subtraction of the @xmath55 leading terms of the born series to the phase shift : @xmath56_n=\delta_\ell-\delta_\ell^{(1)}-\delta_\ell^{(1 ) } -\ldots-\delta_\ell^{(n)}$ ] . choosing @xmath55 sufficiently large renders @xmath57 finite .
technically the born series is an expansion in powers of @xmath34 .
the contribution of any such power to the vacuum polarization energy can be associated with a feynman diagram @xmath58{3.5cm}{\vspace{-0.4 cm } \psfig{file = loop1.eps , height=2cm , width=3cm}}}\ , , \label{eq : ebornfd}\ ] ] where the loop corresponds to the quantum fluctuation and the external lines represent insertions of the background field , _ i.e. _ the interaction of the cosmic string with the fermions .
therefore , the artificial subtraction in equation ( [ eq : attempt2 ] ) is the sum over all feynman diagrams with up to @xmath55 insertions of the background interaction : @xmath59 .
these diagrams can be computed with standard techniques , in particular dimensional regularization can be implemented to handle the ultra violet divergences .
the most important advantage of shuffling the divergences into feynman diagrams is , however , that they can be straightforwardly combined with the counterterm contribution to the energy , @xmath60 .
the latter is found by substituting the background configuration into the counterterm lagrangian in just the same way as the classical energy , eq .
( [ eq : ecl ] ) is obtained from @xmath61 . in any ( multiplicatively ) renormalizable theory
the counterterm lagrangian has the same structure as the classical one and a suitable choice of its coupling constants cancels all ultra violet divergences completely .
then @xmath62 is free of any divergences .
the finite pieces of the counterterm coupling constants will be uniquely determined from appropriate conditions describing properties of the particles that are associated with the fields .
this procedure will be briefly discussed in section .
[ subsec : onshell ] . in total
, the sum @xmath63 gives an unambiguous result for the vacuum polarization energy once the renormalization conditions are fixed .
a first principle derivation of this result based on a quantum field theoretic formulation of the energy momentum momentum tensor rather and the analytic properties of the greens function for scattering boundary conditions is presented in ref . .
equation ( [ eq : efinite ] ) is perfectly suited for a computation of the vacuum polarization energy for static configurations that allow for a full decomposition into partial waves . however , this is not the case for the cosmic string which is translationally invariant along the @xmath64axis in coordinate space . in this scenario
the wave function of the quantum fluctuation factorizes into @xmath65 so that we have a scattering problem in the plane perpendicular to the string , and the total dispersion relation @xmath66 .
this implies the replacements @xmath67 and @xmath68 in equation ( [ eq : attempt2 ] ) so that an integration over @xmath69 ( with measure @xmath70 ) yields the vacuum polarization energy per unit length of the string . in doing
so an immediate obstacle arises .
the scattering data do not depend on the momentum @xmath69 along the symmetry axis .
hence for any @xmath55 , this @xmath69integral will not be finite .
a careful analysis treats the @xmath69integral in dimensional regularization@xcite @xmath71_{n}\big]\bigg\}\ , , \label{eq : inface2}\ ] ] where @xmath72 is the analytic dimension of the subspace in which the configuration is translationally invariant .
the divergence now manifests itself via the singularity of the @xmath73function coefficient as @xmath74 . due to sum rules for scattering data@xcite , which represent generalizations of levinson s theorem , the expression in curly brackets in eq .
( [ eq : inface2 ] ) vanishes as @xmath74 .
hence this limit can indeed be taken@xcite .
finally , the fermion spectrum in the string background is charge conjugation invariant .
thus @xmath57 is twice its contribution from the non negative part of the spectrum , @xmath75\ , \frac{d}{dk}\left[\delta_\ell(k)\right]_n \cr & & \hspace{4 cm } + \sum_j\left[\epsilon_j^2\ , { \rm ln}\frac{\epsilon_j^2}{\mu_r^2 } -\epsilon_j^2+m^2\right]\bigg\}\ , .
\label{eq : ereg}\end{aligned}\ ] ] here @xmath76 is an arbitrary renormalization scale that has no effect on @xmath57 by exactly the same sum rules .
the expression , eq .
( [ eq : ereg ] ) , for @xmath57 replaces the analog in equation ( [ eq : efinite ] ) .
note that the function multiplying the ( born subtracted ) phase shift is of higher power in @xmath37 than its counterpart before integrating over the momentum @xmath69 conjugate to the coordinate of translational invariance .
hence @xmath55 must be increased if directions are added in which the configuration is translationally invariant .
this , of course , merely reflects the fact that ultra
violet divergences turn more severe in higher dimensions .
we stress that the expressions obtained so far for the vacuum polarization energy strongly rely on the analytic properties of the scattering data .
furthermore , in numerical calculations , we mostly consider them for purely imaginary momenta .
this has ( at least ) two advantages : ( i ) the oscillating phase shifts turn into exponentially decaying ( logarithms of the ) jost functions , and ( ii ) the momentum integral and the sum over angular momenta may be exchanged . while ( i ) drastically improves numerical stability , ( ii ) significantly simplifies the treatment of the logarithmic divergences that emerge at third and fourth order of the born and feynman expansions .
these technical details are discussed at length in ref . and briefly addressed in section [ sec_fake ] .
we have now established a formalism for computing the vacuum polarization energy of background fields in the string geometry .
however , we still have the problem that the string does not induce a well behaved scattering problem because of its non trivial structure at spatial infinity
. though gauge invariant combinations of the higgs and gauge bosons are trivial at spatial infinity , the individual terms in the born and feynman series are not gauge invariant and therefore ill defined .
this ambiguity appears because the dirac hamiltonian that is obtained by straightforward substitution of the field configuration , eqs .
( [ eq : higgs ] ) and ( [ eq : gauge ] ) does not turn into the free dirac hamiltonian as @xmath77 , but instead becomes @xmath78 .
this local gauge transformation acts only on the left
handed fermions , @xmath79 with @xmath80 .
unfortunately , the gauge transformation @xmath81 does not solve the problem for all @xmath82 $ ] : although it would generate vanishing interactions at infinity , it also induces a @xmath83 potential at the center of the string , @xmath84 .
this might still yield well defined phase shifts , but the conditions underlying the analyticity of the scattering data are certainly violated by this singular behavior . as argued at the end of the previous section , analyticity is central for numerical feasibility of our approach . as a solution , we can define a radially extended gauge transformation @xmath85 this transformation fixes the gauge and in equation ( [ eq : dirac ] ) it yields the interaction term [ subsec : gauge ] @xmath86 + \frac{1}{2}\frac{\partial \xi}{\partial \rho } \begin{pmatrix}-\vec{\sigma}\cdot\hat{\rho } & \vec{\sigma}\cdot\hat{\rho } \cr \vec{\sigma}\cdot\hat{\rho } & -\vec{\sigma}\cdot\hat{\rho}\end{pmatrix}\vec{n}\cdot\vec{\tau } \cr\cr & & \hspace{0.5 cm } + \frac{n}{2\rho}\ , \begin{pmatrix}-\vec{\sigma}\cdot\hat{\varphi } & \vec{\sigma}\cdot\hat{\varphi } \cr \vec{\sigma}\cdot\hat{\varphi } & -\vec{\sigma}\cdot\hat{\varphi}\end{pmatrix } \big[f_g\,{\rm sin}(\delta)i_g(\delta)+(f_g-1)\,{\rm sin}(\xi)i_g(-\xi)\big]\ , .
\label{eq : diracint}\end{aligned}\ ] ] the new gauge function @xmath87 is hidden in the difference @xmath88 which appears both explicitly and as the argument of the space
dependent weak isospin matrix @xmath89 i\,{\rm cos}(x)\,{\rm e}^{-in\varphi } & { \rm sin}(x ) \end{pmatrix } \ , .
\label{defig}\ ] ] imposing the boundary conditions @xmath90 and @xmath91 for the new gauge function @xmath87 defines a well behaved scattering problem .
otherwise , the specific form of @xmath87 is irrelevant .
this property allows us to verify our numerical results by modifying its shape while keeping the boundary conditions fixed .
all explicit matrices in eq .
( [ eq : diracint ] ) act in spinor space . together with the boundary conditions , eq .
( [ eq : bc ] ) a well behaved scattering problem is obtained . with this choice of gauge a scattering matrix and ,
more generally , a jost function can be straightforwardly computed .
moreover , the born series to these scattering data can be constructed simply by iterating @xmath34 .
note that the gauge transformation is single valued at spatial infinity , @xmath92 . in this respect
it differs from the analogous problem of fractional fluxes in qed . in that case
a similar choice of gauge is hence not a remedy ; rather the calculation of the vacuum polarization energy requires the introduction of a return flux to arrive at a well behaved scattering problem@xcite .
the return - flux approach can also be used for the present calculation , but it is much more laborious numerically@xcite .
the idea of utilizing a fake boson field to simplify the treatment of higher order divergences was first implemented in ref . .
as mentioned above , the continuation to imaginary momenta @xmath93 and @xmath94 , where @xmath95 is the logarithm of the jost function , allows the exchange of the momentum integral with the angular momentum sum . then the third and fourth order contribution from the born series produce logarithmic divergences .
these divergences are similar to the ones found in the second order vacuum polarization energy of a boson field fluctuating about a scalar potential .
matching its strength appropriately allows to replace @xmath96_n\,\longrightarrow\ , \frac{d}{dt}\left[\sum_\ell d_\ell\left ( \nu_\ell(t)-\nu^{(1)}_\ell(t)-\nu^{(2)}_\ell(t)\right ) -\sum_{\ell}\bar{d}_{\ell}\bar{\nu}^{(2)}_\ell(t)\right ] \label{eq : fakebos}\ ] ] under the integral in eq .
( [ eq : ereg ] ) .
the quantity @xmath57 with this replacement will be called @xmath97 .
the over bared quantities refer to the bosonic scattering data .
the replacement eq .
( [ eq : fakebos ] ) must , of course , be accompanied by the boson feynman diagram @xmath98 so that the total vacuum polarization energy becomes @xmath99 where the superscript indicates the inclusion of the counterterm contributions .
each of the three terms on the right hand side of equation ( [ eq : master ] ) is ultra violet finite by itself . the advantage of eq .
( [ eq : master ] ) and the replacement eq .
( [ eq : fakebos ] ) is now obvious : instead of fermionic contributions up to order @xmath100 , we only need to compute second order fermionic and bosonic feynman diagrams and terms in the corresponding born series .
numerical results as well as dimensionfull parameters are measured in appropriate units of the ( perturbative ) fermion mass @xmath46 .
we focus on the contribution of fermion fluctuations to the vacuum polarization energy because it dominates the boson counterpart by a factor proportional to the number of internal degrees of freedom , _ e.g. _
@xmath2 , the number of colors . in this scenario
we are now prepared to compute the vacuum polarization energy of a prescribed string configuration . despite of the simplification in eq .
( [ eq : master ] ) , the numerical computation is still expensive .
the scattering data are extracted from a multi channel problem and for the final result to be reliable a huge number of partial wave must be included .
this numerical effort restricts the number of variational parameters that can be used to characterize the profile functions .
we have already introduced the strength parameter @xmath14 .
in addition , we introduce three scale parameters @xmath101 , @xmath102 and @xmath103 via the anstze @xmath104\ , .
\label{eq : ansaetze}\ ] ] the scale @xmath103 parameterizes the shape of the gauge profile .
as explained above , this shape and thus @xmath103 should not be observable .
the other specifics of the profiles are chosen to keep @xmath105 regular .
we have also considered an exponential parameterization for the gauge field @xmath106 which yields a slightly better agreement with the original nielsen
olesen profiles that minimize @xmath105 for @xmath107 .
no significant difference in @xmath48 was found between these anstze .
we check gauge invariance by varying the shape of the gauge profile , @xmath87 .
a typical result is shown in table [ tab : res1 ] .
[ tab : res1 ] as expected , the individual contributions to @xmath48 depend strongly on @xmath103 .
however , these changes essentially compensate each other
. numerically the most cumbersome part of the calculation is @xmath97 . from various numerical considerations ( change of extrapolation scheme for partial wave sum , modification of momentum integration grid , etc . )
its numerical accuracy is estimated to be at the 1% level . within that range
@xmath48 is independent of @xmath103 , thus verifying gauge invariance .
the above results are obtained in the @xmath108 renormalization scheme , which essentially omits the non divergent parts of the feynman diagrams .
any other scheme merely differs by manifestly gauge invariant ( finite ) counterterms . with the above mentioned choice of units , the dependence of @xmath48 on the model parameters factorizes in the @xmath108 scheme which simplifies the computation because this dependence can easily be traced from @xmath105 .
however , for physically meaningful results we need to impose renormalization conditions that correspond to a particle interpretation , inducing a mild parameter dependence in @xmath60 . to be specific
we consider the so called _ on shell _ scheme in which the coefficients of the four allowed counterterms are determined such that * the tadpole graph vanishes * the higgs mass remains unchanged * the normalization of higgs particle remains unchanged * and the normalization of vector meson remains unchanged in the presence of fermionic quantum corrections .
note that the vector meson mass @xmath109 is not fixed by these conditions and thus will be a prediction that includes quantum corrections .
hence we tune the gauge coupling to reproduce the physical value for @xmath109 .
typical results for the vacuum polarization energy per unit length of the string are shown in figure [ fig_onshell ] , as functions of the variational parameters . except for narrow string configurations dominated by the higgs field
, the vacuum polarization energy turns out to be positive .
therefore , fermionic quantum corrections do not provide any sensible binding and no stable uncharged string is found for the physically motivated parameters , eq .
( [ eq : parameters ] ) , for which @xmath105 dominates the total energy . yet
, @xmath105 deceases quickly with increasing yukawa coupling @xmath24 and some stability is indeed seen for large @xmath24 and narrow strings .
unfortunately , in this regime the restriction to one fermion loop in the vacuum polarization energy is unreliable because of the occurrence of the landau ghost@xcite .
cosmic strings induce many fermionic bound state levels ( whose energies are denoted by @xmath43 ) for the two dimensional scattering problem . for @xmath107 there even exists an exact zero mode@xcite . in the three dimensional problem
these bound states acquire a longitudinal momentum for the motion along the symmetry axis and their energies become @xmath110 here @xmath111 is the length of the string . in leading order of the @xmath112 limit the sum over the discrete longitudinal momentum turns into a continuum integral , @xmath113 . to minimize the bound state contribution a chemical potential @xmath114 with @xmath115
is introduced and all levels with @xmath116 are populated .
this procedure defines a fermi momentum for each level , @xmath117 which enters the total charge per unit length of the string @xmath118 this relation can be inverted to give @xmath119 and thus @xmath120 . from this
the binding energy ( per unit length ) for a prescribed charge @xmath121 \label{eq : ebind}\ ] ] is computed relative to an equal number of free fermions that have energy @xmath46 each .
figure [ fig_ebline ] shows the fermion contribution to the binding energy , @xmath122 . for a given configuration the graph terminates at the point when all available bound state levels are occupied . for small charges narrow strings
are favorable while the binding energy of strings with larger widths decreases more quickly as @xmath123 increases .
surprisingly , the envelope along which @xmath124 is minimal forms a straight line . extrapolating this line
to @xmath125 indicates that the fermion vacuum polarization energy should ( approximately ) vanish .
this extrapolation circumvents the landau ghost problem . to finally decide on dynamical stability
, the classical energy must be included . to this end
we scan through several hundred configurations characterized by the variational anstze , eq .
( [ eq : ansaetze ] ) .
we label them by @xmath126 and compute their total binding energy @xmath127 \label{eq : ebindform}\ ] ] for a given charge . if @xmath128<0 \label{eq : stable}\ ] ] a stable configuration is constructed .
figure [ fig_etot1 ] shows @xmath129 as a function of charge for various values of the yukawa coupling constant , _
i.e. _ the mass of a non interacting fermion . for @xmath130
the classical and fermion energies essentially cancel each other and leave @xmath129 roughly charge independent is artificial because very narrow strings have not been considered to avoid the landau ghost inconsistency . ] .
bound objects are observed by further increasing the yukawa coupling to about @xmath131 , which corresponds to a heavy fermion mass which is less than twice the top quark mass .
we find that the minimizing configurations have @xmath132 , _
i.e. _ they are dominated by the higgs field .
we have presented and discussed the formalism to compute the fermion contribution to the vacuum polarization energy per unit length of an infinitely long straight string in a simplified version of the electroweak standard model .
our approach is based on the interface formalism , for which the analytical properties of scattering data are essential .
we have also seen that a particular subset of gauge choices circumvents obstacles that in a nave treatment arise from the non trivial structure of the string configuration at spatial infinity .
numerically we have found that the vacuum polarization is small and positive in the regime in which the one fermion loop approximation is reliable .
hence , there is no quantum stabilization of the string .
however , we have seen that a heavy fermion doublet can stabilize a nontrivial string background for a non zero fixed charge per unit length .
the resulting configuration is dominated by the higgs field . since any additional variational degree of freedom
can only lower the total energy , the embedding of this configuration in the full standard model , with the @xmath133 gauge field included , will also yield a bound object as long as mixing between this heavy and the standard model fermions can be ignored .
we see binding set in at @xmath134 , which is still within the range of energy scales at which the standard model is expected to provide an effective description of the relevant physics , and also within the range to be probed at the lhc .
light fermions would contribute only weakly to the binding of the string , since their yukawa couplings are small . as a result
, we can add them to our model , _
e.g. _ to accommodate anomaly cancellation , without significantly changing the result .
one of us ( hw ) is grateful to the organizers for putting together this enjoyable conference and providing the opportunity to present this project .
00 t. vachaspati , phys .
lett . * 68 * , 1977 ( 1992 ) [ erratum - ibid . *
69 * , 216 ( 1992 ) ] .
a. achucarro and t. vachaspati , phys .
rept . * 327 * , 347 ( 2000 ) .
y. nambu , nucl .
b * 130 * , 505 ( 1977 ) .
h. b. nielsen and p. olesen , nucl .
b * 61 * , 45 ( 1973 ) .
a. achucarro and c. j. a. martins , arxiv:0811.1277 [ astro - ph ] .
e. j. copeland and t. w. b. kibble , proc .
a * 466 * , 623 ( 2010 ) . r.h .
brandenberger and a. davis , phys .
b * 308 * , 79 ( 1993 ) ; r.h .
brandenberger , a. davis , and m. trodden , phys . lett . b * 335 * , 123 ( 1994 ) .
k. kajantie , m. laine , k. rummukainen , and m. e. shaposhnikov , phys .
lett . * 77 * , 2887 ( 1996 ) ; k. rummukainen , m. tsypin , k. kajantie , m. laine , and m. e. shaposhnikov , nucl .
b * 532 * , 283 ( 1998 ) ; f. csikor , z. fodor , and j. heitger , phys .
lett . * 82 * , 21 ( 1999 ) . c. grojean , g. servant , and j. d. wells , phys .
d * 71 * , 036001 ( 2005 ) ; a. menon , d. e. morrissey , and c. e. m. wagner , phys .
d * 70 * , 035005 ( 2004 ) .
e. dhoker and e. farhi , nucl . phys
. b * 248 * , 59 ( 1984 ) ; 77 ( 1984 ) . s. g. naculich , phys .
* 75 * , 998 ( 1995 ) .
f. r. klinkhamer and c. rupp , j. math .
phys . * 44 * , 3619 ( 2003 ) .
g. starkman , d. stojkovic , and t. vachaspati , phys .
d * 65 * , 065003 ( 2002 ) ; g. starkman , d. stojkovic , and t. vachaspati , phys .
d * 63 * , 085011 ( 2001 ) ; d. stojkovic , int .
j. mod .
a * 16s1c * , 1034 ( 2001 ) .
m. groves and w. b. perkins , nucl .
b * 573 * , 449 ( 2000 ) .
m. bordag and i. drozdov , phys .
d * 68 * , 065026 ( 2003 ) .
j. baacke and n. kevlishvili , phys . rev .
d * 78 * , 085008 ( 2008 ) .
m. lilley , f. di marco , j. martin , and p. peter , phys .
d * 82 * , 023510 ( 2010 ) .
h. weigel , m. quandt , n. graham , and o. schrder , nucl .
b * 831 * , 306 ( 2010 ) .
h. weigel and m. quandt , phys .
b * 690 * , 514 ( 2010 ) .
h. weigel , m. quandt , and n. graham , phys . rev .
lett . * 106 * , 101601 ( 2011 ) .
n. graham , m. quandt , and h. weigel , phys .
d * 84 * , 025017 ( 2011 ) .
n. graham , m. quandt , and h. weigel , lect .
notes phys .
* 777 * , 1 ( 2009 )
. n. graham , r. l. jaffe , m. quandt , and h. weigel , phys . rev .
lett . * 87 * , 131601 ( 2001 ) .
m. g. krein , mat . sborn .
( n.s . ) * 33 * 597 ( 1953 ) ; m. g. krein , sov .
math.dokl . * 3 * 707 ( 1962 ) ; m. sh . birman and m. g. krein , sov .
math.dokl . * 3 * 740 ( 1962 ) .
n. graham , r. l. jaffe , v. khemani , m. quandt , m. scandurra , and h. weigel , nucl .
b * 645 * , 49 ( 2002 ) .
r. d. puff , phys .
a * 11 * , 154 ( 1975 ) ; n. graham , r. l. jaffe , m. quandt , and h. weigel , annals phys . * 293 * , 240 ( 2001 ) . n. graham , v. khemani , m. quandt , o. schrder , and h. weigel , nucl .
b * 707 * , 233 ( 2005 ) ; h. weigel , j. phys .
a * 39 * , 6799 ( 2006 ) .
e. farhi , n. graham , r. l. jaffe , and h. weigel , nucl .
b * 630 * , 241 ( 2002 ) .
g. ripka and s. kahana , phys .
d * 36 * , 1233 ( 1987 ) ; j. hartmann , f. beck , and w. bentz , phys .
c * 50 * , 3088 ( 1994 ) .
|
we provide a thorough exposition of recent results on the quantum stabilization of cosmic strings .
stabilization occurs through the coupling to a heavy fermion doublet in a reduced version of the standard model .
the study combines the vacuum polarization energy of fermion zero - point fluctuations and the binding energy of occupied energy levels , which are of the same order in a semi - classical expansion . populating these bound states
assigns a charge to the string .
strings carrying fermion charge become stable if the higgs and gauge fields are coupled to a fermion that is less than twice as heavy as the top quark .
the vacuum remains stable in the model , because neutral strings are not energetically favored .
these findings suggest that extraordinarily large fermion masses or unrealistic couplings are not required to bind a cosmic string in the standard model .
|
ultraluminous x - ray sources ( ulxs ) are nonnuclear , point - like x - ray sources in external galaxies with an apparent luminosity over the eddington limit of a stellar - mass black hole , which is typically @xmath3 ergs s@xmath1 for a 20 @xmath4 black hole .
those with variability on time scales from minutes to months are accreting compact objects .
the nature of ulxs is still unclear .
if the emission is ( roughly ) isotropic and under the eddington limit , they may harbor black holes up to @xmath5 @xmath4 , belonging to the missing population of intermediate - mass black holes ( imbhs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , they could also be stellar - mass black holes , with beamed or super - eddington emission @xcite .
ulxs with relatively low luminosities ( @xmath6 ergs s@xmath1 ) may be more likely to be stellar - mass black holes of about 10 @xmath4 like galactic bhbs or up to a couple tens of @xmath4 like ic 10 x-1 @xcite .
galactic bhbs have been observed in a few cases with luminosities up to @xmath7 ergs s@xmath1 @xcite , although there are large uncertainties on their distances .
@xcite proposed that most ulxs are stellar - mass black holes in an ` ultraluminous state ' with super - eddington accretion based on analysis of high quality x - ray spectra .
a major difference between the two populations is that galactic bhbs spend most of their lifetime in the quiescent or low state , while most ulxs have persistent emission and have been active since their discovery at time scales up to decades in some cases .
the non - transient behavior of the majority of ulxs implies that they do not contain black holes significantly more massive than @xmath8 @xmath4 @xcite . despite these hypotheses ,
the connection between ulxs and galactic bhbs has not been well addressed .
if they contain black holes of similar mass , we may expect to see transitions from the normal states to the ultraluminous state in an individual source .
multicolor blackbody emission from an optically thick accretion disk could shed light on the nature of the central black hole . at a fixed fraction of eddington luminosity , the disk inner temperature scales with the black hole mass to the @xmath9 power @xcite , indicating that accretion disks around imbhs are cooler and brighter than those around stellar - mass black holes .
soft excesses are detected in ulxs and can be modeled as cool disk emission from imbhs @xcite .
however , the cool disk model is not the unique interpretation of the soft excess ( for a brief review see the introduction of * ? ? ?
massive outflows due to near or super - eddington accretion may produce blackbody emission at the similar temperature @xcite
. the disk model can be tested by determining if the disk luminosity varies with the 4th power of the inner disk temperature , i.e. @xmath10 .
this relation has been robustly verified for stellar - mass black holes when the disk emission dominates the x - ray spectrum @xcite .
although the x - ray spectra of some ulxs can be modeled by dominant disk emission @xcite , the observed luminosity variation and number of observations have been inadequate for proper testing of the @xmath10 relation .
a possible @xmath11 relation was found in ngc 5204 x-1 , but the disk emission was not the dominant component in the spectrum and a large correction for comptonization was required @xcite .
m82 is a nearby starburst galaxy in which two ulxs have been found @xcite .
it has been extensively studied in x - rays with the chandra x - ray observatory . in our recent chandra observation on 2008
october 4 , the x - ray source cxom82 j095546.6 + 694041 was 5 - 10 times brighter than its normal flux level and was identified as the third ulx in m82 . according to the naming convention suggested by @xcite
, we refer to this source as x37.8 + 54 . in this paper , we investigated its spectral and timing behavior using all available chandra observations , and discussed its possible nature .
the distance to the host galaxy is adopted as 3.63 mpc @xcite .
so far there have been fourteen observations of m82 with the chandra advanced ccd imaging spectrometer ( acis ) , including the most recent three that we proposed to monitor the brightest ulx in it ( the results will be reported separately ) . among them , eleven observations with an exposure longer than 5 ks are used for spectral analysis ( table [ tab : obs ] ) .
the other three short acis observations ( obs i d 378 , 380 - 1 , and 380 - 2 ; 380 has two observations ) as well as the four high resolution camera ( hrc ) observations ( obs i d 1411 - 1 , 1411 - 2 , 8505 and 8189 ; 1411 has two observations ) are examined for long term variability .
all data are processed using ciao 4.1.2 with caldb 4.1.3 ; new bad pixel files and level 2 events files are recreated when necessary .
cccc 361 & 1999 - 09 - 20 & acis - i & 33.3 + 1302 & 1999 - 09 - 20 & acis - i & 15.5 + 379 & 2000 - 03 - 11 & acis - i & 8.9 + 2933 & 2002 - 06 - 18 & acis - s & 18.0 + 6097 & 2005 - 02 - 04 & acis - s & 52.8 + 5644 & 2005 - 08 - 17 & acis - s & 68.1 + 6361 & 2005 - 08 - 18 & acis - s & 17.5 + 8190 & 2007 - 06 - 02 & acis - s & 52.8 + 10027 & 2008 - 10 - 04 & acis - s & 18.3 + 10025 & 2009 - 04 - 17 & acis - s & 17.4 + 10026 & 2009 - 04 - 29 & acis - s & 16.9 + two images of the central region of m82 in the 0.3 - 8 kev energy range taken from chandra acis are displayed in figure [ fig : img ] .
the top panel shows the image from observation 10027 when x37.8 + 54 was the brightest ever seen , and the bottom one is from observation 5644 in which the second brightest flux was recorded .
the source in 10027 spreads on multiple pixels because it lies at a large off - axis angle . aligning the two images using isolated point - like sources suggests that the brightening in observation 10027 around the source region is indeed from x37.8 + 54
. a faint object to the south of x37.8 + 54 is seen in observation 5644 , and is partly blended with x37.8 + 54 only in observation 10027 .
it displays a variability by a factor of 2 . for observation 10027
, we subtracted the contribution of this dim source assuming a spectrum from its highest state and found no significant change in the spectral fitting for x37.8 + 54 , indicative of negligible contamination .
lllllll 361 + 1302 & @xmath12 & @xmath13 & @xmath14 & @xmath15 & @xmath16 & 60.9/54 + 379 & @xmath17 & @xmath18 & @xmath19 & @xmath20 & @xmath21 & 6.4/12 + 2933 & @xmath22 & @xmath23 & @xmath24 & @xmath25 & @xmath26 & 19.9/28 + 6097 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & @xmath31 & 37.9/50 + 5644 & @xmath32 & @xmath33 & @xmath34 & @xmath35 & @xmath36 & 161.2/149 + 6361 & @xmath37 & @xmath38 & @xmath39 & @xmath40 & @xmath41 & 20.3/25 + 8190 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & @xmath46 & 92.6/82 + 10027 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 & 122.4/115 + 10025 & @xmath52 & @xmath53 & @xmath54 & @xmath55 & @xmath56 & 33.3/26 + 10026 & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 & 19.2/23 lllllllll + 379 & @xmath62 & @xmath63 & @xmath64 & & @xmath65 & @xmath66 & & 6.8/12 + 2933 & @xmath67 & @xmath68 & @xmath69 & & @xmath70 & @xmath71 & & 27.0/28 + * 10027 * & @xmath72 & @xmath73 & @xmath74 & & @xmath75 & @xmath76 & & 123.5/115 + 10025 & @xmath77 & @xmath78 & @xmath79 & & @xmath80 & @xmath81 & & 37.4/26 + 10026 & @xmath82 & @xmath83 & @xmath84 & & @xmath85 & @xmath86 & & 23.8/23 + & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & 51.6/52 + * 6097 * & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 & 36.5/48 + * 5644 * & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & 151.4/147 + * 6361 * & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & 17.5/23 + 8190 & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & @xmath121 & 80.8/80 the energy spectra of x37.8 + 54 are extracted from events in a 3 @xmath122 elliptical region around the source found by wavdetect .
the background is subtracted from a nearby circular region with a radius of about 4 arcsec on the same ccd and off the readout column of the source .
background lightcurves are checked using 10 - 15 kev events on the whole ccd ; intervals with background flares were excluded in observations 5644 and 6361 .
spectral fitting was done using xspec 12.5.0 and the errors are quoted at 68% confidence level if not specified otherwise .
observations 361 and 1302 were performed in the same day and the spectral shapes are consistent .
these two spectra are fitted together to improve the signal to noise ratio .
the galactic absorption column density along the line of sight to the source is @xmath123 @xmath124 @xcite , and is set as the lower boundary of the total absorption column density during the fits . using the pileup_map tool available in ciao
, we obtained that the count rate on the brightest 3 by 3 pixels is @xmath125 counts per frame .
for the two observations 5644 and 10027 when the source was brightest , the highest count rate in a 3 by 3 pixel island is about 0.03 counts per frame due to a large off - axis angle and/or a small readout time .
the derived spectral parameters do not change within errors when a pileup model is applied .
therefore , the effects of pileup are insignificant and , thus , not taken into account . first , we tried a power - law model subject to interstellar absorption on each spectrum from observations listed in table [ tab : obs ] .
the best - fit parameters are listed in table [ tab : pl ] .
the source luminosity was found at two different levels , a relatively low level of about @xmath126 ergs s@xmath1 and a high level above @xmath127 ergs s@xmath1 .
the long time scale lightcurve from all of the eighteen chandra observations are shown in figure [ fig : longlc ] .
for the three short acis and the four hrc observations , which are inadequate for spectral fitting , their 0.3 - 8.0 kev luminosities are calculated using pimms from count rates , assuming a power - law spectrum with the average absorption column density and power - law photon index .
they are found to be consistent with the relatively low flux level .
we then tried a multicolor accretion disk ( mcd ) model . a single mcd model with absorption
did not improve the fits with respect to the power - law model . for five out of the ten spectra , an additional soft blackbody component to the mcd
was justified at a significance of 3 @xmath122 or more .
spectral parameters derived from the mcd model or the mcd plus blackbody are listed in table [ tab : mcd ] .
the bi - thermal model produced comparable or better fits than the power - law model .
the disk luminosity versus disk inner temperature is plotted in the top panel of figure [ fig : mcd ] .
the soft blackbody component mainly lies in the low energy band , and the diskbb component contributes more than 90% of the unabsorbed flux in the 2.0 - 8.0 kev band .
we note that four of the five brightest observations require the soft component , but the brightest observation , 10027 , does not .
if we add a blackbody component into the 10027 spectrum and fix the temperature at its mean value of 0.2 kev , the blackbody luminosity is @xmath128 ergs s@xmath1 , similar to those from other observations , and it contributes only 2.3% , with a up limit of 5% , to the total observed flux .
the mcd parameters are consistent with those from a single mcd model . in the other four spectra
, the fraction of the soft blackbody is larger than 18% .
the non - detection of the cool blackbody component in 10027 could be due to its relative weakness when the mcd component became much stronger than in others .
cc @xmath129 ( @xmath130 @xmath124 ) & @xmath131 + @xmath132 ( @xmath4 ) & @xmath133 + @xmath134 ( @xmath135 g s@xmath1 ) & @xmath136 + @xmath137 & @xmath138 @xmath139 + @xmath140 ( degree ) & @xmath141 + @xmath142 & 118.6/113 except for observations in which the source spectra are hard and the luminosities are the lowest , the disk luminosity seems to be correlated with the disk inner temperature . in the bottom panel of figure [ fig : mcd ] , we removed points from observations when the source was in the low / hard state , specifically when @xmath143 ergs s@xmath1 and @xmath144 , and then fitted the remaining observations to a power - law function
. there appears to be a robust correlation and the best - fit relation between the luminosity and temperature is @xmath145 with 1 @xmath122 error .
for the five observations ( 379 , 2933 , 8190 , 10025 , 10026 ) in which the spectrum is hard and the luminosity is relatively low , the source may be in the hard state with a power - law spectrum .
due to low statistics , these spectra can also be fitted by thermal disk radiation , but it is not surprising that they do not vary in a @xmath10 pattern . the mcd model has been widely used as a standard indicator of the thermal dominant state . however , it does not include relativistic effects .
we thus tried a more physical multicolor disk blackbody model with relativistic effects fully taken into account ( kerrbb in xspec ; * ? ? ?
* ) , to the spectrum from observation 10027 when the source was brightest . in the model
, we set zero torque at the inner boundary of the disk , a hardening correction of 1.7 , a distance of 3.63 mpc , and turned on self - irradiation and limb - darkening .
the kerrbb model provides adequate fits to the spectrum with parameters ( absorption column density , black hole mass , mass accretion rate , spin , and inclination ) listed in table [ tab : kb ] .
models consisting of a power - law with an additional soft blackbody or cool mcd component were also fitted to the spectra .
neither showed obvious improvement relative to a single power - law model for most observations . for each observation
, the short - term variability was examined using the kolmogorov - smirnov test .
variability in the 0.3 - 8 kev range was found only in observations 5644 and 6361 with a significance of 9 and 5 @xmath122 , respectively . in all other observations ,
the evidence for variability is at a confidence level below 3 @xmath122 .
we found that the source lies near the readout node boundary in observations 5644 and 6361 , causing periodic oscillations when the source position on the detector plane moved across the node boundary due to dithering .
the power spectrum shows a peak at 1000 s which is the dithering period along the direction perpendicular to the boundary .
we therefore created a lightcurve from these two observations with time steps of about 1000 s selected from intervals when the source was at least two pixels away from the readout boundary .
interestingly , there are dip - like variations on time scales of several ks
. simulations using marx 4.4.0 applying the aspect solution file and bad pixel map obtained from the observation indicates that dithering is not responsible for the variations shown in figure [ fig : lc ] .
we extracted energy spectra from intervals during the dip ( 58.171 - 65.171 ks ) and the nondip ( 28.171 - 56.371 ks ) , which are marked in figure [ fig : lc ] .
we found that the dip spectrum is softer , with a power - law photon index of @xmath146 versus @xmath147 from the nondip spectrum .
the spectrum during dip is also softer than in the low state , which has a power - law index less than 2.3 .
fitting the nondip spectrum with a blackbody plus mcd model modified by a shared absorption component leads to consistent results with from the entire spectrum .
we then use two absorption components applied onto each emission component individually . for the nondip spectrum ,
the absorption column density @xmath129 on the mcd component is the same as on the blackbody component and is consistent with that derived using a single absorption component . for the dip spectrum ,
the fitting is made by fixing the temperature and @xmath129 of the blackbody component to those of the nondip spectrum , due to limited photon number ; other parameters , the normalization of the blackbody , the mcd and its absorption are free in the fit .
this leads to a best - fit @xmath129 of @xmath148 @xmath124 on the mcd component , which is significantly larger than the @xmath129 from the nondip spectrum , and suggests that the dip may be caused by extra absorption of the mcd component .
the dip and nondip spectra with this blackbody plus mcd model are shown in figure [ fig : spe ] for comparison .
observation 6361 was performed 62 ks after the end of 5644 .
the source flux in 6361 seems to drop from the nondip level of 5644 to the relatively low level found in other observations .
we searched in the literature and found no infrared or radio counterpart for the x - ray source .
the nearest radio source to x37.8 + 54 is 37.54 + 53.2 @xcite at an angular distance of @xmath149 calculated using the x - ray position of r.a.=@xmath150 and decl.=+@xmath151 ( j2000.0 ) @xcite .
we identify x37.8 + 54 as the third ulx in the starburst galaxy m82 based on its brightening in a recent chandra observation to an observed flux 5 - 10 times higher than previously observed .
the peak isotropic luminosity in the 0.3 - 8 kev band corrected for interstellar absorption is @xmath0 ergs s@xmath1 inferred from fitting with a power - law model .
the source was often found at relatively low luminosity , near @xmath126 ergs s@xmath1 . at the lowest luminosities , @xmath152 ergs s@xmath1 , the source appears to have hard spectra with the photon index less than 2.3 .
as the luminosity increases above this value , it shows dramatic spectral variation with the power - law photon index @xmath153 varying from 1.9 to 3.0 .
this suggests a change in spectral state , and , indeed , most of the higher luminosity spectra are better fitted with an mcd spectral model .
the lower panel of figure [ fig : mcd ] shows that when observations in the low / hard state are excluded , the remaining observations exhibit a robust correlation between @xmath154 and @xmath155 .
the correlation coefficient between @xmath156 and @xmath157 is 0.997 with a chance probability of @xmath158 .
fitting the data to a power - law relation , we obtained a power - law slope of @xmath2 , consistent with the 4th - power relation predicted for the mcd model in the thermal dominant state .
we note that the disk luminosity varies by about an order of magnitude .
thus , the data provide an adequate test of this relation .
we conclude that x37.8 + 54 exhibits two spectral states : a low / hard state and a thermal dominant state .
following the recipe in @xcite and adopting canonical correction coefficients with @xmath159 and @xmath160 , where @xmath161 is the hardening correction factor and @xmath162 is to correct for the radius of maximum temperature , the disk luminosity can be expressed as a function of the black hole mass ( @xmath163 in solar mass ) and the disk inner temperature ( @xmath155 in kev ) as @xmath164 ergs s@xmath1 , where @xmath165 is related to the black hole spin with @xmath166 for a non - spinning black hole and @xmath167 for maximal spinning . using a 4th power - law relation to fit the data points in the bottom panel of figure [ fig : mcd ] , we obtain @xmath168 ergs s@xmath1 .
thus , we have @xmath169 , suggesting the ulx contains a black hole of about 10 solar masses if it is a schwartzchild black hole with a face - on disk , or a few tens solar masses if it is fast spinning or viewed at a high inclination .
the best - fit compact object mass is @xmath133 @xmath4 , derived from fitting with the fully relativistic disk model , indicative of a stellar black hole slightly more massive than galactic ones .
we also tried different hardening correction factors from 1.5 to 1.9 .
the fitted mass varies by 20% with respect to that from @xmath159 .
interestingly , the specific angular momentum converges to its maximally allowed value , @xmath170 with a lower limit of 0.80 , suggesting the source contains a fast spinning black hole .
the mass accretion rate in eddington unit is about 1.0 , which is larger than that found in the thermal dominant state of galactic bhbs of no more than 0.6 @xcite , but is under the theoretical upper limit of about 10 for a stellar - mass black hole with a radiation dominated accretion disk @xcite .
@xcite investigated the spectra of six disk - dominated ulxs and found five sources had best - fit masses between 23 and 73 @xmath4 .
the three ulxs that have a derived mass @xmath171 @xmath172 at a confidence @xmath173 also show rapid spin , which is quite similar to x37.8 + 54 .
the best - fit black hole mass of x37.8 + 54 is also similar with that of ic 10 x-1 @xcite .
it has been suggested that ic 10 x-1 has a persistently high accretion rate @xcite ; a rapidly spinning black hole is naturally expected since accretion would bring a fair amount of angular momentum to the compact object .
we conclude that stellar kerr black holes with a mass of several tens @xmath4 could appear as ulxs with a thermal spectrum of a few kev in the 1 - 10 kev band , as having been suggested by @xcite .
such emission could also appear as a hard power - law spectrum if the statistics is inadequate to detect the spectral curvature .
these black holes are more massive than galactic black holes , which are mostly around 10 @xmath4 , and may be formed by the collapse of massive stars in a low - metallicity environment and account for a portion of ulxs @xcite .
similar sources , suzaku j1305 - 4931 in ngc 4945 and ngc 253 x-2 , have been reported in the literature , which show a @xmath10 correlation as their luminosities varied by a factor of about 2 - 3 @xcite . from surveys of bright ulxs in nearby galaxies ,
many hard ulxs were found and interpreted as imbhs in the low / hard state @xcite .
for those sources , most of them present a constant spectral index or hardness despite large variation of the luminosity .
therefore , those sources may have a different nature from x37.8 + 54 .
it has been suggested that ulxs are a special case ( face - on ) of the galactic microquasar ss 433 , where super - eddington accretion drives a powerful wind off the accretion disk @xcite .
the wind could be optically thick and produce a blackbody continuum @xcite .
this may be responsible for the soft excess observed in many ulxs with a temperature of about 0.1 - 0.3 kev @xcite . when fitting spectra not from the low / hard state with an mcd component , all observations except 10027 need an additional blackbody component at a significance of at least 3 @xmath122 .
no obvious correlation was found between the luminosity and temperature of the soft blackbody .
however , the uncertainties are large and this result is not very constraining .
the calculated surface area of the soft blackbody emission is about @xmath174 @xmath175 for all observations , corresponding to a spheric radius of @xmath176 cm , about @xmath177 for a 10 solar mass black hole , where @xmath178 is the gravitational radius equal to @xmath179 .
the absorption column density varies significantly between observations no matter what models are used .
such large variation is likely internal to the binary , implying the outflow is inhomogeneous and time variable .
we note that the soft blackbody appears at energies where the absorption is important .
thus caution is warranted when interpreting it .
interesting timing behavior is seen in one of the major outbursts caught by the observations 5644 and 6361 ( figure [ fig : lc ] ) .
the source shows soft dips on time scales of several ks .
soft dips could be caused by absorption of the inner accretion disk , like in 4u 1630@xmath18047 @xcite .
this interpretation is also consistent with our spectral analysis results in the penultimate paragraph of section 2 .
we note that the soft blackbody component is assumed not to suffer extra absorption
. a dense massive outflow from the disk could be the origin of the extra absorption and also the source of the soft blackbody .
other mechanisms could also produce soft dips , e.g. , the possible ejection of the inner part of accretion disk into relativistic jets in grs 1915@xmath181105 @xcite .
the time scale of ks of the dip in x37.8 + 54 is longer than that in 4u 1630@xmath18047 and in grs 1915@xmath181105 by about an order of magnitude
. more detailed study of the soft dip requires data with better quality .
to summarize , x37.8 + 54 provides us a good candidate of a stellar kerr black hole being ultraluminous with a near eddington accretion rate
. this ulx could be representative of a class of ulxs with relatively low luminosity , i.e. a few @xmath182 ergs s@xmath1 .
the luminosity of x37.8 + 54 is usually not high enough to be classified as a ulx , while in the outburst it truly enters the ultralumious regime .
the x - ray spectrum at peak luminosity is dominated by a disk component , which is unlike many brighter ulxs at their maximum luminosity that they enter the ` ultraluminous state ' characterized by a strong comptonizing component @xcite .
therefore , this source may be in between the low luminosity tail of ulxs and high luminosity tail of galactic bhbs .
we note that the hard state of x37.8 + 54 is more luminous than the canonical hard state found in galactic bhbs .
this is consistent with the correlation found in outbursts of bhbs that the peak luminosity of the hard state is scaled with the peak luminosity of the thermal state in the following outburst @xcite .
we thank the referee for useful comments .
hf acknowledges funding support from the national natural science foundation of china under grant no .
10903004 and 10978001 , the 973 program of china under grant 2009cb824800 , and the foundation for the author of national excellent doctoral dissertation of china under grant 200935 .
pk acknowledges support from chandra grant go9 - 0034x .
|
we report on the identification of a third , new ultraluminous x - ray source ( ulx ) in the starburst galaxy m82 . previously , the source was observed at fluxes consistent with the high state of galactic black hole binaries ( bhbs ) .
we observe fluxes up to @xmath0 ergs s@xmath1 in the ultraluminous regime .
when the source is not in the the low / hard state , spectral fitting using a multicolor disk model shows that the disk luminosity varies as the disk inner temperature raised to the power @xmath2 , consistent with the behavior of galactic bhbs in the thermal dominant state .
fitting the spectrum with a multicolor disk model with general relativistic corrections suggests that the source harbors a rapidly spinning black hole with a mass less than 100 solar masses .
a soft excess was found in the source spectrum that could be blackbody emission from a photosphere created by a massive outflow .
the source also showed soft dips during a flare .
|
this study is the fourth in a series of papers exploring the structure of the region kr 140 and its associated star - forming activity . in @xcite ves 735 , the o - star powering the region was examined .
a second paper , @xcite , was a multiwavelength study of the structure , energetics , and kinematics of the region .
finally , @xcite presented an analysis of submillimetre ( submm hereafter ) observations of the region at 450 and 850 @xmath1 m .
the main result of @xcite was the discovery of numerous submm dust cores located within the molecular gas surrounding the region , including a large number of cores that were clearly located at the interface between ionized and molecular gas , a likely location for star - formation induced or `` triggered '' by the expansion of the region @xcite .
two of the more isolated cores were clearly associated with _ iras _ sources , but the low resolution of _ iras _ combined with the extensive diffuse emission of dust associated with the region made it impossible to determine if any of the other cores were associated with star - formation activity .
the goal of this study is to determine the distribution of young stellar objects ( ysos ) throughout the kr 140 molecular cloud .
we focus especially on the embedded stellar content of the submm cores as these are potentially the youngest star - forming regions .
we use the spatial distribution of the ysos relative to the photodissociation region ( pdr ) surrounding the region as a means of gauging the relative importance of spontaneous and sequential , or triggered , star formation in this region . to achieve this
we have analyzed a combination of submm data from @xcite and @xcite , 2mass ( 2 micron all sky survey ) and midcourse space experiment ( _ msx _ ) infrared data , and newly acquired @xmath2co , @xmath3co and c@xmath0o molecular line data from the five college radio astronomy observatory ( fcrao ) . in the next section we review the pertinent parameters of kr 140 and provide information on the new observations of the region . in [ sec:2mass ] and
[ sec : dist ] we first show how 2mass data can be utilized to identify the yso population and then examine its spatial distribution .
our results are discussed in [ sec : discuss ] and conclusions are presented in [ sec : conclude ] .
kr 140 ( @xmath4 , @xmath5 ; * ? ? ?
* ) is a 5.7 pc diameter region located at a distance of @xmath6 kpc .
it is close to , but apparently isolated from , the large w3/w4/w5 star - formation complex .
the region is ionized by a single o8.5 v(e ) star ves 735 .
foreground extinction is @xmath7 magnitudes toward ves 735 and ranges between @xmath8 over the region .
molecular line observations ( @xmath2co j@xmath91@xmath100 ) indicate that the region is associated with a molecular cloud with a mass in the range 10@xmath11
10@xmath12 m@xmath13 .
a detailed discussion and derivation of these properties can be found in @xcite and @xcite . for this study we used a combined sample of submm clumps identified in the @xcite and @xcite 850 @xmath1 m scuba scan - maps of the kr 140 region .
there are only minor disagreements between the two samples where they overlap due to the different ways in which structures were identified in the two studies and the correspondence between the two samples is shown in table [ tbl : subcomp ] . in the area covered by our molecular - line observations ( @xmath14 and @xmath15 , see
[ sec : co ] ) there are a total of 39 submm clumps : kmjb 1 - kmjb 20 from @xcite and clumps 7 , 9 , 12 - 13 , 21 , 39 - 40 , 42 - 45 and 47 - 54 from @xcite . .spatial correspondence between @xcite and @xcite submm clumps ( kmjb and m07 respectively ) . [ cols="^,^,^,^,^",options="header " , ] @xmath16 @xcite i d numbers have a trailing m ( e.g. 7 m ) , other i d numbers from @xcite + @xmath17 2mass 02210449 + 6106045 = msx6c g133.5492 + 00.0900 + @xmath18 _ iras _ 02171 + 6058 + @xmath19 _ iras _ 02186 + 6033 = msx6c g133.6890 + 00.1643 + but showing the position of ysos spatially associated with submm clumps .
the size of the symbols indicate the uncertainty in colour .
ysos are identified using the nomenclature from table [ tbl : ysosubmm ] , e.g. , the yso from the `` p2 '' sample associated with clump 13 is labeled `` 13.p2 '' . a lower limit to the _ ( j@xmath10h ) _ colour is indicated by an arrow symbol .
[ fig : ccyso],width=317 ] but showing the position of ysos spatially associated with submm clumps .
the size of the symbols indicate the uncertainty in magnitude and colour .
ysos are identified using the nomenclature from table [ tbl : ysosubmm ] , e.g. , the yso from the `` p2 '' sample associated with clump 13 is labeled `` 13.p2 '' . a lower limit to the _ ( h@xmath10k ) _ colour is indicated by an arrow symbol .
[ fig : cmyso],width=317 ] of which 2 have associated ysos and 3 do not .
the rightmost bin includes all clumps with @xmath20 .
[ fig : av_combi],width=317 ] visually there appear to be four concentrations of ysos in the kr 140 molecular cloud ( for discussion we will denote these regions a through d ) along with a noticeable lack of ysos in the northeast portion of the molecular cloud ( see fig . [ fig : p123dist ] ) .
region a is at the eastern end of the molecular filament near @xmath21 , region b is associated with submm clump 1 at @xmath22 , region c is near the central pdr at @xmath23 , and finally region d is associated with clumps 17 , 18 and 19 near @xmath24 . to quantify this visual impression we divided the molecular cloud area into @xmath25 boxes and counted the number of stars falling within each box as a rough measure of the yso surface density .
the average number of ysos / box is @xmath26 ( @xmath27 ) and boxes near the four regions in question had surface densities of 7 , 7 , 6 and 11 for a to d respectively .
there were no other regions in the molecular cloud with similar @xmath28 deviations from the average value .
region a consists of 17 ysos , four of which are associated with submm clumps .
all of the ysos in this region are t tauri - like with an average _
( h@xmath10k ) _ value of @xmath29 .
none of the ysos are coincident with _ iras
_ 02186 + 6033 ( msx6c g133.6890 + 00.1643 ) , strengthening its identification as a class 0/i yso . the focal point of region
b is the submm clump kmjb 1 which was identified as a class i yso ( _ iras _ 02174 + 6052 ) in @xcite .
the higher resolution of the _ msx _ and 2mass data reveal that clump 1 contains two ysos , including an intermediate - mass haebe which is also a _ msx _
point source with a very red mid - infrared spectrum ( f@xmath30 jy , f@xmath31 jy ) .
just to the east of the clump there are two very red ysos .
2mass 02210851 + 6105582 is detected only in one 2mass band with _
_ k__@xmath32 and a lower limit colour of _
( h@xmath10k)@xmath33 _ @xmath34 .
it is also a bright _ msx _ point source ( msx6c g133.5561 + 00.0919 ) with an extremely red spectrum ( f@xmath35 jy , f@xmath36 jy ) .
the other yso , 2mass 02210610 + 6106043 , is detected in both the _ h _ and _ k _ bands and has _ ( h@xmath10k ) _ @xmath37 and _ k _ @xmath38 . both of these ysos are probably more highly embedded versions of 1.p1 .
in addition , just to the west of kmjb 1 , there is another highly embedded yso with _
k _ @xmath39 and _
( h@xmath10k)@xmath33 _ @xmath40 .
it is quite likely that these objects have formed together as a single grouping of stars .
region c is interesting because of its association with the pdr ridge seen in the _ msx _ images .
the majority of the stars are found to the northeast of the pdr , while 5 of the remaining 6 stars are found along the pdr with one of the ysos being coincident with a submm clump .
this pattern suggests that star - formation is progressing from northeast to southwest with the submm clumps found along the pdr being the latest round of star formation .
region d contains the submm clumps 17 , 18 , and 19 .
intermediate - mass and solar - mass ysos are associated with clump 19 and , just to the north of this clump , there are an additional six t tauri ysos .
all three of these clumps make up the infrared source _ iras _ 02157 + 6053 which _ msx _ and
2mass now clearly show is actually a concentration of ysos combined with extended emission from the pdr .
when examining star formation processes within a molecular cloud containing an region , a key distinction is between sequential star formation , where the trigger for star formation is directly related to the presence of the region ( e.g. , * ? ? ?
* ) , and spontaneous star formation , which occurs with no apparent causal relationship with the region . it is well known that for both classical density - bounded regions and blister / champagne - flow regions , the ionization front ( if ) is preceded by a shock front as it propagates into the molecular cloud @xcite .
time - dependent models of pdr / if development around ob stars @xcite show that the if and pd front merge rapidly , i.e. , on time scales less than the main - sequence lifetime of the exciting star , for typical molecular cloud densities . with this in mind
it is useful to consider two distinct environments within the molecular cloud ; the region beyond the direct influence of the region , and the pdr , which is the region directly influenced by the expansion and uv flux of the region and which can be clearly traced by pah emission .
a comparison of the yso population found in each of these regions then gives a measure of the relative importance of sequential and spontaneous star formation within the molecular cloud .
to delineate these two regions we use the @xmath41 w m@xmath42 sr@xmath43 contour in the _ msx _ band a image to ( generously ) define the extent of the pdr . we find that 42% of the full yso sample ( p1 , p1 + , p2 and p3 ) lie within the pdr as defined and 58% lie outside the pdr .
the exact percentages will vary somewhat depending on how the pdr is defined .
for example , using a more conservative definition for the pdr region ( @xmath44 w m@xmath42 sr@xmath43 ) results in 31% of the ysos within the pdr and 69% outside the pdr . the percentages , in both cases , are the same for the yso population both with and without including the p1 + sample . in all cases
it is clear that a substantial fraction of the yso population lies beyond the pdr in this star - forming region and is thus _ not _ the result of sequential star formation associated with the region .
the exciting star of the region , ves 735 , is about 2 myr old @xcite , and kinematic models of the region set expansion timescales of 1 - 2 myr @xcite .
our observations and analysis of kr 140 are consistent with a molecular cloud undergoing continuous star formation over this period .
there is no clear signature of a significant older population of stars seen in the 2mass photometry of the region ( see figs .
[ fig : ccd_p123 ] and [ fig : cmd_p123 ] ) .
lower mass ysos that formed at the same time as ves 735 have had time to disperse through the molecular cloud and may form some of the more spatially distributed population of ysos seen in our sample .
the expansion of the region has swept up material which has subsequently collapsed to form the submm clumps that we observe in the pdr . at the same time
the elongated morphology of the densest portion of the molecular cloud has resulted in ongoing star formation in regions well away from the direct influence of the region . as discussed earlier , the association of haebe and t tauri stars with dense regions traced in the submm and c@xmath0o support ages of @xmath45 years .
the presence of the class 0/i ysos , _ iras _ 02171 + 6058 and _ iras _ 02186 + 6033 shows there continues to be very recent star formation activity on the timescale of @xmath46 years @xcite . finally , many of the massive , apparently starless , submm clumps associated with the pdr may represent the next sites of star formation in the molecular cloud .
deeper images of the region at longer wavelengths would be able to determine if these clumps are starless and test the scenario outlined above by being able to distinguish between different yso evolutionary stages @xcite .
while much of the focus of studies of star formation in and around regions is on the interface regions , our study of kr 140 illustrates that spontaneous star formation occurring away from the interface regions in these molecular clouds can be as important as sequential or triggered star formation related to the expansion of the region .
our analysis shows that the molecular cloud associated with kr 140 contains an extensive population of ysos that is not associated with the pdr .
we have also shown that this yso population is not uniformly distributed though the molecular cloud but is concentrated in four distinct regions .
many of the ysos are also clearly associated with a filamentary structure within the molecular cloud which is traced by c@xmath0o emission .
finally , we see that both low - mass and intermediate - mass ysos are forming throughout the molecular cloud .
particularly interesting is the region around kmjb 1 ( _ iras _ 02174 + 6052 ) which contains a tight grouping of a number of embedded intermediate - mass and solar - mass ysos including the very highly embedded intermediate - mass yso 2mass 02210610 + 6106043 .
this research has made extensive use of the nasa / ipac infrared science archive ( irsa ) , the canadian astronomy data centre ( cadc ) , and data from the two micron all sky survey ( 2mass ) .
irsa is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
2mass is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by nasa and the nsf .
the cadc is operated by the national research council of canada with the support of the canadian space agency .
fcrao was supported by nsf grant ast 0540852 .
cb is supported by an rcuk fellowship at the university of exeter , uk .
the authors would like to acknowledge the work done by caroline pomerleau in the initial stages of this project supported by the women in engineering and science program of nrc canada .
ballantyne , d.r . ,
kerton , c.r . ,
martin , p.g . , 2000 , apj , 539 , 283 bedijn , p.j , tenorio - tagle , g. , 1981 , a&a , 98 , 85 bessell , m.s . , brett , j.m . , 1988 ,
pasp , 100 , 1134 cohen , m. , green , a.j . , 2001 ,
mnras , 325 , 531 cutri , r.m .
et al . , 2003 ,
explanatory supplement to the 2mass all sky data release ( washington : nasa ) , http://www.ipac.caltech.edu/2mass/releases/allsky/doc/explsup.html diaz - miller , r.i . , franco , j. , shore , s.n . , 1998 ,
apj , 501 , 192 deharveng , l. , zavangno , a. , salas , l. , porras , a. , caplan , j. , cruz - gonzalez , i. , 2003 , a&a , 399 , 1135 deharveng , l. , zavagno , a. , caplan , j. , 2005 , a&a , 433 , 565 draine , b.t . , li , a. , 2007 , apj , 657 , 810 elmegreen , b.g . , 1998 , in woodward , c.e . ,
shull , j.m .
, thronson jr . , h.a . , eds , asp conf . ser . , vol .
148 , origins , astron .
, san francisco , p. 150
erickson , n.r . ,
grosslein , r.m . ,
erickson , r.b .
, weinreb , s. , 1999 , ieee trans .
microwave theory tech . , 47 , 2212 finkenzeller , u. , mundt , r. , 1984 , a&as , 55 , 109 fuente , a. , martin - pintado , j. , bachiller , r. , neri , r. , palla , f. , 1998 , a&a , 334 , 253 giard , m. , bernard , j.p . ,
lacombe , f. , normand , p. , rouan , d. , 1994 , a&a , 291 , 239 hatchell , j. , richer , j.s . , fuller , g.a . ,
qualtrough , c.j . ,
ladd , e.f . ,
chandler , c.j .
, 2005 , a&a , 440 , 151 johnstone , d. , di francesco , j. , kirk , h. , 2004 , apj , 611 , l45 kallas , e. , reich , w. , 1980 , a&as , 42 , 227 kenyon , s.j , hartmann , l. , 1995 , apjs , 101 , 117 kerton , c.r .
, ballantyne , d. , martin , p.g . , 1999 , aj , 117 , 2485 kerton , c.r . , martin , p.g
, johnstone , d. , ballantyne , d.r . , 2001 ,
apj , 552 , 601 kirk , h. , johnstone , d. , di francesco , j. , 2006 , apj , 646 , 1009 koornneef , j. , a&a , 1983,128 , 84 mckee , c. , 1999 , in lada , c.j . , kylafis , n.d .
, eds , the origin of stars and planetary systems . kluwer , dordrecht , p. 29
moore , t.j.t . , bretherton , d.e . ,
fujiyoshi , t. , ridge , n.a .
, allsopp , j. , hoare , m.g . , lumsden , s.l .
, richer , j.s
. , 2007 , mnras , 379 , 663 neuhasuer , r. , frink , s. , torres , g. , sterzik , m.f . , roeser , s. , randich , s. , 1998 in donahue , r.a .
bookbinder , j.a .
, eds , asp conf . ser .
154 , the tenth cambridge workshop on cool stars , stellar systems and the sun .
, san francisco , p. 1748 onishi , t. , mizuno , a. , kawamura , a. , ogawa , h. , fukui , y. , 1998 , apj , 502 , 296 osterbrock , d.e .
, 1985 , astrophysics of gaseous nebulae and active galactic nuclei , university science books , sausalito , ca price , s.d . ,
egan , m.p . , carey , s. , mizuno , d.r . , kuchar , t.a .
, 2001 , aj , 121 , 2819 rieke , g.h . , lebofsky , m.j . , 1985 ,
apj , 288 , 618 robitaille , t.p . ,
whitney , b.a . , indebetouw , r. , wood , k. , denzmore , p. , 2006 ,
apjs , 167 , 256 roger , r.s . ,
dewdney , p.e . , 1992 , apj , 385 , 536 ruffle , d.p . ,
hartquist , t.w . ,
rawlings , j.m.c.,williams , d.a . , 1998 , a&a , 334 , 678 rydgren , a.e . ,
vrba , f.j . , 1983 , aj , 88 , 1017 skrutskie , m.f .
et al . , 2006 ,
aj , 131 , 1163 strong , a.w . , mattox , j.r .
, 1996 , a&a , 308 , 21 th , p.s . , de winter , d. , prez , m.r . , 1994 , a&as , 104 , 315
|
we use 2mass and _ msx _ infrared observations , along with new molecular line ( co ) observations , to examine the distribution of young stellar objects ( ysos ) in the molecular cloud surrounding the halo region kr 140 in order to determine if the ongoing star - formation activity in this region is dominated by sequential star formation within the photodissociation region ( pdr ) surrounding the region .
we find that kr 140 has an extensive population of ysos that have `` spontaneously '' formed due to processes not related to the expansion of the region .
much of the yso population in the molecular cloud is concentrated along a dense filamentary molecular structure , traced by c@xmath0o , that has not been erased by the formation of the exciting o star .
some of the previously observed submillimetre clumps surrounding the region are shown to be sites of recent intermediate and low - mass star formation while other massive starless clumps clearly associated with the pdr may be the next sites of sequential star formation .
stars : pre - main - sequence stars : formation regions infrared : stars
|
in recent years fractional calculus ( fc ) and fractional differential equations ( fde ) became very popular in many areas of science .
the books dedicated to the applications of fdes published in 2010 - 2011 include @xcite in physics in general , @xcite in modeling and control , @xcite in viscoelasticity , @xcite in systems with long - range interaction .
a good review of applications of fc to chaos in hamiltonian systems is given in @xcite . because fractional derivatives are integro - differential operators , they are used to describe systems distributed in time and/or space : systems with long range interaction @xcite , non - markovian systems with memory ( @xcite ch.10 , @xcite ) , fractal media @xcite , etc .
biological systems are probably the best examples of systems with memory . as it has been shown recently @xcite , even processing of external stimuli by individual neurons can be described by fractional differentiation . in some cases @xcite fdes
are equivalent to the volterra integral equations of the second kind .
this kind of equations ( non necessarily fdes ) is used in nonlinear viscoelasticity ( see for example @xcite ) and in biology ( for applications to mathematical models in population biology and epidemiology see @xcite ) . as in regular dynamics ( in application to various areas in science ) , the non - linearity plays a significant role in fractional dynamics . chaos and order in nonlinear systems with long range interactions
were considered in @xcite , nonlinear fdes in application to the control were reviewed in books @xcite , nonlinear fractional reaction - diffusion systems considered in gafiychuk2010,guf1,guf2,guf3 . corresponding to the fact that in physical systems the transition from integer order time derivatives to fractional ( of a lesser order ) introduces additional damping similar in appearance to additional friction @xcite , phase spaces of the systems with fractional time derivatives demonstrate different kinds of structures similar to the attractors of the dissipative dynamical systems controlbook2010,zse , th . as in the case of the systems which can be described by regular differential equations ,
general properties of systems described by fdes can be demonstrated on the examples of maps which can be derived by integrating the fdes ( if it is possible ) over a period of perturbation .
equations of some fractional maps ( fm ) , which include among others fractional standard maps ( fsm ) and fractional dissipative standard maps ( fdsm ) , were derived in a series of recent publications @xcite from the corresponding fdes .
two of the most studied examples in the regular case are the chirikov standard map ( sm ) @xcite and the zaslavsky dissipative standard map ( dsm ) @xcite .
the sm provides the simplest model of the universal generic area preserving map and the description of many physical systems and effects ( fermi acceleration , comet dynamics , etc . )
can be reduced to the studying of the sm .
the topics examined include fixed points , elementary structures of islands and a chaotic sea , and fractional kinetics @xcite .
different properties of the dsm were discussed in @xcite , and a rigorous proof of the existence of chaotic attractor in such type of systems was obtained in @xcite . it was also proved in @xcite that the system of this type exhibits quasi - periodic attractors , periodic sinks , transient chaos with the dynamics attracting to the sink , and the existence of the srb measure for the purely chaotic dynamics .
all this stimulated the study of the fsms @xcite and fdsm @xcite in hope to reveal the most general properties of the fractional dynamics .
two - dimensional maps investigated in @xcite were derived from the fdes ( see section [ be ] ) and , as a result , they are discrete maps with memory in which the present state of evolution depends on all past states
. the earlier studies of maps with memory were done on one - dimensional maps which were not derived from differential equations @xcite .
most results were obtained for the generalizations of the logistic map and the main general result is that the presence of memory makes systems more stable .
the initial study of the fsm in @xcite was concentrated on the investigation of the fixed / periodic points stability of the fsm , new properties and new types of attractors .
the results obtained in @xcite revealed new unusual properties of the fractional attractors of the fsm , which are different not only from the properties of the fixed / periodic points of the non - dissipative systems ( the sm in our case ) , but also from the properties of attractors of the regular ( not fractional ) dissipative systems ( e.g. the dsm ) .
the most unusual observed property of the fsms is the existence ( and persistence ) of the new type of attractors - cascade of bifurcations type trajectories ( cbtt ) . a cascade of bifurcations , when with the change in the value of a parameter a system undergoes a sequence of period - doubling bifurcations , is a well known pathway of a transition form order to chaos . in the case of the cbtt
the period doubling occurs without a change in a system parameter and is an internal property of a system . in section [ r ]
we summarize results of the extensive numerical investigation of three fractional maps to demonstrate the properties of fractional attractors .
an analysis of the conditions for the cbtts appearance will be presented in the following publications .
the standard map in the form @xmath0 can be derived from the differential equation @xmath1 where @xmath2 , following the steps proposed by tarasov in @xcite . the equations for the riemann - liouville fsm ( fsmrl )
were obtained in @xcite and @xcite . following the steps proposed @xcite , the fsmrl in the form which converges to the sm as @xmath3 can be derived from the differential equation with the riemann - liouville fractional derivative describing a kicked system @xmath4 where @xmath2 , with the initial conditions @xmath5 where @xmath6 @xmath7 @xmath8 , and @xmath9 is a fractional integral .
after integration of equation ( [ diffsmrl ] ) the fsmrl can be written in the form @xmath10 @xmath11 where @xmath12 and momentum @xmath13 is defined as @xmath14 here it is assumed that @xmath15 and @xmath16 .
the condition @xmath17 is required in order to have solutions bounded at @xmath18 for @xmath19 @xcite . in this form the fsmrl equations in the limiting case @xmath20 coincide with the equations for the standard map under the condition @xmath21 . for consistency and in order to compare corresponding results for all three maps ( the sm , the fsmrl , and the caputo fsm ( fsmc ) ) all trajectories considered in this article have the initial condition @xmath21 . following the steps proposed @xcite , the fsmc in the form which converges to the sm as @xmath3 can be derived from the differential equation similar to ( [ diffsmrl ] ) but with the caputo fractional derivative @xmath22 where @xmath2 , with the initial conditions @xmath23 where @xmath24 @xmath25 integrating equation ( [ diffsmc ] ) with the momentum defined as @xmath26 and assuming @xmath15 and @xmath16
, one can derive the fsmc in the form @xmath27 @xmath28,\ \ ( { \rm mod } \
2\pi ) , % \eqno ( 2)\ ] ] @xmath29 @xmath30 it is important to note that the fsmc ( ( [ fsmcp ] ) , ( [ fsmcx ] ) ) can be considered on a torus ( @xmath31 and @xmath32 mod @xmath33 ) , a cylinder ( @xmath31 mod @xmath33 ) , or in an unbounded phase space , whereas the fsmrl ( ( [ fsmrlp ] ) , ( [ fsmrlx ] ) ) can be considered only in a cylindrical or an unbounded phase space .
the fsmrl has no periodicity in @xmath32 and can not be considered on a torus .
this fact is related to the definition of momentum ( [ momrl ] ) and initial conditions ( [ fsmrlic ] ) .
the comparison of the phase portraits of two fsms is still possible if we compare the values of the @xmath31 coordinates on the trajectories corresponding to the same values of the maps parameters .
the dsm in the form @xmath34 @xmath35 can be derived integrating the differential equation of the the kicked damped rotator ( see for example @xcite ) @xmath36 two forms of the fdsm were derived in @xcite .
the form which has been investigated numerically in @xcite @xmath37 @xmath38 where functions @xmath39 are defined by @xmath40\ ] ] and @xmath41 is the incomplete gamma function @xmath42 has been derived from the following fractional generalization of ( [ dr-0b ] ) @xmath43 where @xmath44 here @xmath45 further in this paper we will also use @xmath46 and @xmath47 .
it appears that the dependence of the sm s fixed point ( 0,0 ) stability properties on the map parameter @xmath48 plays an important role when transition to the fsms is considered .
( 0,0 ) sm fixed point is stable for @xmath49 . at @xmath50
it becomes unstable but period two ( @xmath51 ) antisymmetric trajectory @xmath52 appears which is stable for @xmath53 . at @xmath54
this t=2 trajectory becomes unstable but it gives birth to two @xmath51 trajectories @xcite with @xmath55 these @xmath51 trajectories become unstable when @xmath56 ( see figure [ figkc]a ) , at the point where @xmath57 stable trajectories are born .
this period doubling cascade of bifurcations sequence continues with @xmath58 and @xmath59 stable trajectories appearing at @xmath60 and @xmath61 correspondingly .
it continues until at @xmath62 all periodic points become unstable and corresponding islands disappear .
this scenario of the elliptic - hyperbolic point transitions with the births of the double periodicity islands inside the original island has been investigated in @xcite and applied to investigate the sm stochasticity at low values of the parameter @xmath63 .
stability of the fixed point ( 0,0 ) has been considered in @xcite .
it has been shown that for fractional values of @xmath64 ( we consider only @xmath65 ) this point turns from elliptic ( for @xmath66 ) into a sink which is stable for @xmath67 where @xmath68 and can be calculated numerically ( see figure [ figkc]a ) . it can be shown that @xmath69 , which corresponds to the sm case .
the structure of the fsm phase space for @xmath70 preserves some features which exist in the @xmath66 case .
namely , stable higher period points , which exist in the sm case , still exist in the fsm , but they exist in the asymptotic sense and they transform into sinks or , in the fsmrl case , into attracting slow ( @xmath71 ) diverging trajectories ( asdt ) ( see figure [ figlowk ] )
. pure chaotic trajectories disappear .
each attractor has its own basin of attraction .
the traces of the sm chaotic sea exist in the following sense : initially close trajectories , which do not start from any basin of attraction , may fall into absolutely different attractors . in most of the cases attracting points
themselves do not belong to their own basins of attraction : trajectory which starts from an attracting point may fall into a different attractor .
the fsmrl trajectories which converge to the fixed point follow two different routs .
trajectories which start from the basin of attraction converge according to the power law @xmath72 and @xmath73 , while those starting from the outside of the basin of attraction are attracting slow converging trajectories ( asct ) with @xmath74 and @xmath75 .
all fsmc trajectories converging to @xmath76 fixed point follow the slow power law @xmath77 and @xmath75 ( see figure [ figlowkstab ] ) .
this convergence may correspond to the recently introduced for the fractional dynamic systems notion of the generalized mittag - leffler stability @xcite , for which the power - law stability is a special case .
numerical simulations confirm that , as in the sm case , in the fsms a period @xmath51 antisymmetric trajectory ( sink ) exists ( asymptotically ) for @xmath79 ( @xmath80 in the sm case ) with @xmath81 asymptotic existence and stability of this sink is a result of the gradual transformation of the sm s elliptic point with the decrease in the order of derivative from @xmath82 ( see figure [ figt2as ] ) . as in the fixed point case , there are two types of convergence of the trajectories to the fsmrl @xmath51 sink : fast with @xmath83 and slow with @xmath84 and @xmath85 convergence in the fsmc case @xcite .
let s consider the fsmrl and assume the asymptotic existence of the @xmath51 sink @xmath86 and @xmath87 . in this case
the limit @xmath88 in ( [ fsmrlp ] ) and ( [ fsmrlx ] ) gives @xmath89 @xmath90 the equation for @xmath91 takes the form @xmath92 from which the condition of the existence of the @xmath51 sink is @xmath93 from ( [ kcr1 ] ) it follows that @xmath94 .
* * remark 1 : * the @xmath51 sink exists only in the asymptotic sense . on a trajectory which starts from point @xmath95=@xmath96 @xmath97 and @xmath98 . * * remark 2 : * in spite of the fact that in the derivation of ( [ rlplim ] ) and ( [ rlxlim ] ) we used the limit @xmath99 , computer simulations demonstrate convergence of all trajectories to the limiting values in the very good agreement with ( [ fastconv ] ) , ( [ slowconv ] ) , and ( [ caputoconv ] ) . * * remark 3 : * direct computations show that @xmath94 is valid for the fsmc too .
the sm s @xmath51 antisymmetric trajectory becomes unstable when @xmath101 , at the point in phase space where a pair of @xmath51 trajectories with @xmath102 , @xmath103 appears .
numerical simulations show ( see figure [ figt2nonas ] ) that the fsms demonstrate similar behavior .
let s assume that the fsmrl equations ( [ fsmrlp ] ) and ( [ fsmrlx ] ) have an asymptotic solution @xmath104.\ ] ] then it follows from ( [ fsmrlp ] ) that relationship ( [ rlplim ] ) @xmath105 holds in this case too .
simulations similar to those presented in figure [ figlowkstab]b @xcite show that for @xmath100 ( see figure [ figkc]a ) the fsmrl has an asymptotic solution @xmath106 with the same @xmath107 for both even and odd values of @xmath108 . substituting ( [ t2nonlimas ] ) in ( [ fsmrlx ] ) and considering limit @xmath88 one can derive ( see @xcite ) @xmath109 which has solutions for @xmath110 ( see figure [ figkc ] ) and the value of @xmath107 can also be calculated : @xmath111 these results are in a good agreement with the direct fsmrl numerical simulations and hold for the fsmc
. in the sm further increase in @xmath48 at @xmath112 causes an elliptic - hyperbolic point transition when @xmath51 points become unstable and stable @xmath57 elliptic points appear .
a period doubling cascade of bifurcations leads to the disappearance of the corresponding islands of stability in the chaotic sea at @xmath61 .
the cusp in figure [ figkc]a points to approximately this spot ( @xmath20 , @xmath113 ) .
inside of the band leading to the cusp a new type of the fsm s attractors appears : cascade of bifurcations type trajectories ( cbtt ) ( see figure [ figcbtt ] ) . in cbtts period doubling cascade of bifurcations
occurs on a single trajectory without any change in the map parameter . in some cases cbtts
behave like sticky islands of the hamiltonian dynamics : occasionally a trajectory enters a cbtt and then leaves it entering the chaotic sea . near
the cusp cbtts are barely distinguishable - the relative time trajectories spend in cbtts is small . with decrease in @xmath64
the relative time trajectories spend in cbtts increases and at @xmath64 close to one a trajectory enters a cbtt after a few iterations and stays there over the longest computational time we were able to run our codes - 500000 iterations .
a typical fsmc s cbtt of this kind is presented in the figure [ figmix]c .
cbtts , easily recognizable in the phase space of the fsmrl ( figure [ figcbtt]a ) , often almost impossible to distinguish in the phase space of the fsmc ( figure [ figcbtt]b ) .
but they clearly reveal themselves on a coordinate versus time ( step of iteration @xmath108 ) plot ( figure [ figcbtt]c ) .
the main feature of the sm s phase space at @xmath114 , which defines dynamics and transport and is well investigated ( see for example @xcite ) , is the presence of the sticky accelerator mode islands .
slight decrease in the order of derivative turns accelerator mode islands into attracting accelerator mode trajectories ( amt ) ( figure [ figmix]a ) .
depending on the values of @xmath48 and @xmath64 trajectories either permanantly sink into this attractor , or intermittently stick to it and return to the chaotic sea .
amts do not exist ( we were unable to find them ) for @xmath115 .
in the ( @xmath116)-plot ( figure [ figkc]a ) the area of the amt existence is a part ( @xmath114 ) of a very narrow strip near @xmath20 line . in the area of the @xmath116-plot above the cbtt strip ( we considered only @xmath117 ) different kinds of chaotic attractors ( or possibly pure chaotic behavior ) can be found .
two examples are presented in figure [ figmix ] .
figure [ figmix]c contains two overlapping attractors one of which is a cbtt .
existence of the overlapping attractors and intersecting trajectories is a feature of maps with memory which is impossible in a regular dynamical systems .
the simplest example of intersecting trajectories can be constructed by starting one trajectory at an arbitrary point @xmath95 and then the second at @xmath118 .
consideration of the fdsm in @xcite demonstrated that already a small deviation of @xmath64 from @xmath119 leads to significant changes in properties of the map .
for example , a window of ballistic motion which exists in zaslavsky map near @xmath120 is closing when @xmath121 .
further decrease in @xmath64 at @xmath122 produces different kinds of chaotic attractors and sinks @xcite .
some of the new features we observed , which are presented in figure [ figfdsm ] , include inverse cascade of bifurcations and trajectories which start as a cbtt and then converge as an inverse cascade of bifurcations . in phase space
those trajectories may look like chaotic attractors .
we studied three different fractional maps which describe nonlinear systems with memory under periodic perturbations ( kicks in our case ) .
the types of solutions which we found in all maps include periodic sinks , attracting slow diverging trajectories , attracting accelerator mode trajectories , chaotic attractors , and cascade of bifurcations type trajectories .
new features discovered include attractors which overlap , trajectories which intersect , and cbtts .
the models with a similar behavior may include periodically kicked media which can be described by fdes : viscoelastic materials @xcite or dielectrics @xcite .
experiments can be proposed to observe different kinds of solution , for example cbtts , in those media .
nonlinear models with memory which have chaotic and periodic solutions , where cascades of bifurcations considered as a result of a change in a system parameter , are used in population biology and epidemiology @xcite .
our calculations show that such behavior can be a consequence of the essential internal properties of the systems with memory and suggest construction of new fractional models of biological systems .
our computational results are based on the runs over more that @xmath123 periods of perturbations . due to the integro - differential nature of the fractional derivatives , at present time it is impossible to solve fdes over more than @xmath124 periods of oscillations .
our results suggest a new way of looking for a particular solutions of the fdes .
for example , we may suggest that cbtts can be found at the values of system parameters , where in corresponding integer system a period doubling cascade of bifurcations leads to the disappearance of a system of islands .
b. n. lundstrom , a. l. fairhall , m. maravall , _ multiple time scale encoding of slowly varying whisker stimulus envelope in cortical and thalamic neurons in vivo .
_ j. neuroscience , * 30 * ( 2010 ) , 50715077 .
a. a. kilbas , b. bonilla , j. j. trujillo , _ nonlinear differential equations of fractional order is space of integrable functions .
_ doklady mathematics * 62 * ( 2000 ) , 222226 , translated from doklady akademii nauk * 374 * ( 2000 ) , 445 - 449 .
( in russian ) .
g. m. zaslavsky , a. a. stanislavsky , m. edelman , _ chaotic and pseudochaotic attractors of perturbed fractional oscillator .
_ chaos * 16 * ( 2006 ) , 013102 .
m. s. tavazoei , m. haeri , _ chaotic attractors in incommensurate fractional order systems .
_ physica d * 237 * ( 2008 ) , 26282637 .
|
using the riemann - liouville and caputo fractional standard maps ( fsm ) and the fractional dissipative standard map ( fdsm ) as examples , we investigate types of solutions of non - linear fractional differential equations .
they include periodic sinks , attracting slow diverging trajectories ( asdt ) , attracting accelerator mode trajectories ( amt ) , chaotic attractors , and cascade of bifurcations type trajectories ( cbtt ) .
new features discovered include attractors which overlap , trajectories which intersect , and cbtts .
|
let @xmath1 be a closed simply connected symplectic @xmath0-manifold and let @xmath2 be a symplectic form on @xmath1 , i.e. a closed , nondegenerate @xmath3-form .
a closed ( oriented ) embedded surface @xmath4 is called + - _ symplectic _ if @xmath5 restricts to a symplectic form on @xmath6 , i.e. @xmath7 ; + - _ lagrangian _ if @xmath5 restricts to the trivial form on @xmath6 , i.e. @xmath8 . given two submanifolds representing the same homology class , we say that they are isotopic if they can be joined by a family of embedded submanifolds . by the isotopy extension theorem ( see e.g. @xcite ) , this is equivalent to the existence of an ambient isotopy of @xmath1 connecting them .
isotopic submanifolds are essentially the same , from the point of view of the smooth topology of @xmath1 , so in order to study the topological aspects of symplectic and lagrangian submanifolds of @xmath1 we need to take into account the equivalence relation above .
this leads to the following + * problem * ( p ) : fix a homology class @xmath9 ; classify all the symplectic ( lagrangian ) representatives of @xmath10 up to the equivalence relation above . +
this problem was first posed , in the study of lagrangian @xmath11 s linear outside a ball in @xmath12 ( where the homology condition becomes irrelevant ) by arnold in @xcite , and further generalized , also in the lagrangian case , by eliashberg and polterovich ( see @xcite ) .
the interest in the symplectic version of this problem was triggered more recently by siebert and tian ( see @xcite ) .
there are several ways in which we can interpret ( p ) .
the first starts from the observation that two submanifolds of a simply connected @xmath0-manifold are homologous if and only if they are images of homotopic embeddings ( this follows from a quite straightforward exercise in algebraic topology ) .
therefore the study of the representatives of a given homology class , up to isotopy , is a suitable generalization of classical knot theory .
problem ( p ) amounts therefore to symplectic ( lagrangian ) _ surface - knot theory _ for @xmath13 .
the second interpretation stems from the classical observation that if we consider the case when @xmath1 is a khler surface , and we look at complex representatives of the same homology class , these are always isotopic ( even in a stronger sense , i.e. through complex representatives ) .
problem ( p ) can be interpreted therefore as the investigation of the natural counterpart of that property in the symplectic setup , with the added condition that , in a symplectic manifold , we have two kinds of natural submanifolds .
( observe that , while symplectic submanifolds generalize the notion of complex submanifolds , they are defined with an open condition , while lagrangian submanifolds are defined , like the complex ones , with a closed one ; in this sense , both kinds of submanifolds share some of the properties of complex ones . ) with this interpretation , problem ( p ) enters in the larger question of understanding whether the class of symplectic manifolds , that lies somewhere between the class of khler surfaces and general @xmath0-manifolds , exhibits features typical of the former or of the latter . to clarify this comment , we want to point out that for a smooth @xmath0-manifold @xmath1 there is a rather general construction ( see @xcite ) of infinitely many nonisotopic representatives of a homology class @xmath10 , so that the smooth counterpart of problem ( p ) ( i.e. surface - knot theory for @xmath13 ) can be extremely rich , as opposed to complex " knot theory for khler manifolds which is , as noted , trivial . problem ( p ) , in the way it is stated ,
is clearly very ambitious , and what we can realistically hope at the moment is to prove that for a pair @xmath13 there is a unique symplectic ( lagrangian ) representative , or else build infinitely many nonisotopic representatives . the first results around ( p )
have all been in the direction of uniqueness .
just to mention few , eliashberg and polterovich proved in @xcite uniqueness of lagrangian representatives for @xmath11 in @xmath12 , answering arnold s question ; also , they proved in @xcite an analogous result for the class of the section in @xmath14 and @xmath15 . in the symplectic case , siebert and tian proved in @xcite that a symplectic surface in @xmath16 homologous to an algebraic curve of degree @xmath17 is in fact isotopic to it ( at least for @xmath18 ) . such results , and the nature of their proofs ,
might have suggested the possibility of obtaining a uniqueness result holding for all pairs @xmath13 , as in the khler case .
instead , fintushel and stern have shown in @xcite that this is not the case , exhibiting the first example of homologous , nonisotopic symplectic tori , in a large class of symplectic manifolds ( including , for example , elliptic surfaces ) .
the first examples of nonuniqueness for the lagrangian case have appeared in @xcite .
the aim of this paper is to present and discuss several examples where we can partially answer ( p ) showing nonuniqueness of symplectic or lagrangian representatives for some pair @xmath13 .
all these constructions , in a more or less direct way , start from the ideas contained in @xcite , and we will try to point out the several common aspects . in section [ const ] we are going to discuss the actual construction of infinite families of symplectic ( lagrangian ) representatives for suitable pairs @xmath13 , while in section [ noniso ] we will discuss how ( and to which level ) we can distinguish the elements of these families .
in these two sections we will restrict the discussion to genus one surfaces . in section
[ last ] , instead , we are going to show a more recent construction of examples for genus greater than one .
the results presented below , when not explicitly attributed , can be considered to be a sort of common folklore " , or can be easily obtained by looking at related papers ( of the author or others ) , quoted in the references .
it should be clear , however , that much credit is due to the ideas contained in @xcite and @xcite .
finally , for obvious reasons of brevity , this paper does not aspire to be a complete survey on the topic , and follows instead tastes and preferences of the author , even at the price of omitting significant results .
it is quite clear that describing the symplectic ( lagrangian ) submanifolds of a symplectic @xmath0-manifold starting from the definition is a rather difficult task , as the definition itself does not shed too much light on global properties of such submanifolds .
the basic idea underlying this section will be to use a class of symplectic @xmath0-manifolds where we are allowed to reduce the problem , at least in part , to a @xmath19-dimensional problem , and we can translate the property of being symplectic or lagrangian into a quite simpler problem of curves in a fibered @xmath19-manifold . the class of @xmath0-manifolds that we are going to consider , and that has saved the day in many problems of symplectic topology , is the class of link surgery manifolds .
precisely , we are going to exploit the following dictionary ( inspired from @xcite ) : [ cols="^,^",options="header " , ] to explain the dictionary , we start with the definition of link surgery manifolds ( see @xcite ) for a fibered link @xmath20 .
consider an @xmath21-component oriented link @xmath22 whose exterior fibers over @xmath23 with fiber @xmath6 , i.e. @xmath24 note that we do not require @xmath6 to span the link @xmath20 , namely the ( poincar dual of the ) class of @xmath6 in @xmath25 may be different from @xmath26 .
in particular , we will consider the case where @xmath6 spans one link component , e.g. @xmath27 = ( 1,0, ... ,0)$ ] . if @xmath27 = ( m_{1}, ... ,m_{n})$ ] , the fiber @xmath6 intersects the @xmath28-th boundary component @xmath29 , up to isotopy , in the curve @xmath30 where @xmath31 is a simple closed curve and @xmath32 is some integer coefficient determined by the equation above ( see , e.g. , @xcite ) .
complete @xmath31 to a basis @xmath33 for @xmath29 .
for example , if @xmath34 , @xmath35 and we can choose @xmath36 .
next , choose a family @xmath37 of symplectic @xmath0-manifolds , each containing a framed , symplectic , self - intersection @xmath38 torus @xmath39 .
we can now define the @xmath0-manifold @xmath40 the gluing map on the boundary @xmath19-tori is defined in such a way to identify @xmath39 with @xmath41 and , reversing the orientation , the meridian to the torus @xmath39 with @xmath31 .
the manifold above depends on various choices , besides the link @xmath20 and the pairs @xmath42 , but we are going to omit such dependence from the notation as this will not affect our discussion . the homotopy type of @xmath43 can be easily determined from the @xmath37 s and the gluing data . in particular , in the cases we will consider @xmath43 will be simply connected
. an important feature of this manifold is that we can explicitly compute the seiberg - witten polynomial of @xmath43 ( and , more generally , of any link surgery manifold ) in terms of the relative seiberg - witten invariants of @xmath42 and the symmetrized multivariable alexander polynomial of the link @xmath20 .
the work of @xcite and @xcite shows that , for @xmath44 ( the case we will be interested in ) , @xmath45 in this equation @xmath46 represents the ( poincar dual to the ) homology class of @xmath47 , and we are identifying in the product the homology classes according to the gluing of ( [ glue ] ) .
remember that the relative seiberg - witten polynomial of @xmath42 is defined as the seiberg - witten polynomial of @xmath48 , where @xmath49 is the simply connected elliptic surface without singular fibers of euler characteristic @xmath50 ( in particular @xmath51 , and we will be mainly interested in this case ) . to show that @xmath43 admits a symplectic structure ,
we need a second presentation of it as symplectic fiber sum of symplectic @xmath0-manifolds . in order to do so ,
perform dehn surgery on @xmath52 along @xmath20 , using as slopes on each boundary component the curves @xmath31 described in ( [ slope ] ) : this gives a closed three manifold @xmath53 . by extending the fibration of @xmath54 with the fibration of @xmath55 of degree @xmath56 on each dehn filling , @xmath53 fibers over @xmath23 with fiber @xmath57 , the closed surface obtained by capping off each boundary component of @xmath6 with a disk ( contained in the suitable dehn filling ) .
the cores @xmath58 of the dehn fillings ( i.e. the so called dual link ) are transverse , up to isotopy , to @xmath59 .
( note that , unless all divisibilities @xmath56 are equal to one , these are not sections of the fibration . )
now we can associate to a fibration of @xmath53 a closed nondegenerate @xmath60-form @xmath61 , defined up to isotopy .
if we denote by @xmath62 a closed two - form on @xmath53 that restricts to a volume form on the fiber @xmath59 , the @xmath0-manifold @xmath63 admits a symplectic @xmath3-form @xmath64 , where @xmath65 is a sufficiently small constant .
consider the cores @xmath58 ; because of the transversality condition we have @xmath66 , which implies that @xmath67 i.e. the tori @xmath68 are symplectic .
moreover , they are naturally endowed with a framing inherited from the framing of the @xmath58 s , that we can assume to be the one identified by @xmath69 . by suitably scaling the symplectic form on the @xmath37 s
, we can arrange that the symplectic volume on @xmath68 coincides with the one of @xmath70 .
we can then write @xmath43 as symplectic fiber sum : @xmath71
gompf s theory of symplectic fiber sums ( @xcite ) guarantees then that @xmath43 admits a symplectic form which restricts , outside the gluing locus , to the ones of the factors .
there is a second result ( see @xcite ) that allows us to control some of the homology classes of @xmath43 , namely the fact that the natural map @xmath72 obtained by composing knneth isomorphism and inclusion is injective . as a consequence of the previous formula
if @xmath73 is a curve in @xmath54 the homology class of the torus @xmath74 in @xmath43 is given by @xmath75 = \sum_{i=1}^{n } \mbox{lk}(\gp , k_{j } ) [ s^{1 } \times \mu(k_{j})].\ ] ] if we apply the construction above choosing as link @xmath20 the trivial knot @xmath76 and a symplectic manifold @xmath1 , we obtain of course @xmath77 ; the advantage of the presentation ( [ glue ] ) is that we have an explicit control of the symplectic topology of the neighborhood of @xmath78 .
when @xmath20 is a nontrivial fibered knot , there is only one choice for @xmath6 , and @xmath79 differs from @xmath1 by the fact that we have substituted a neighborhood of @xmath78 with a homology @xmath80 that is actually a @xmath6-bundle over @xmath81 , and has a richer topology than the product bundle .
the strategy to obtain families of symplectic tori in @xmath43 follows directly from the observations above .
namely , if we can identify a family @xmath82 of closed curves in @xmath54 that are transverse to the fibration , these will define symplectic tori @xmath74 in @xmath43 , as the steps leading to ( [ tra ] ) follow _
verbatim_. moreover , if the @xmath73 s represent the same homology class of @xmath54 , the symplectic tori @xmath74 will represent the same homology class of @xmath43 , according to ( [ hom ] ) .
if the curves @xmath73 are not isotopic ( in @xmath54 ) , it is reasonable to expect that the tori @xmath74 are not isotopic in @xmath43 either .
the symplectic case of problem ( p ) for the link surgery manifold @xmath43 is transformed this way into the problem of finding homologous , nonisotopic curves in @xmath54 transverse to the fibration .
* example : braiding construction 1 .
* the first example we are going to consider is when @xmath20 is the unknot @xmath76 .
in this case , a curve transverse to the trivial fibration of @xmath83 is just a ordinary braid ( closing to a knot @xmath73 ) having the unknot @xmath76 as axis .
the number of strands of the braid determines the linking number @xmath84 and therefore the homology class of @xmath74 . by choosing @xmath1 to be the elliptic surface @xmath49
, we can produce examples of symplectic tori representing some multiple of the homology class of the fiber @xmath78 , identified in ( [ glue ] ) with @xmath85 .
the original examples of @xcite belong to this class .
a collection of braids with @xmath86 strands which satisfies the description above is the braid presenting the @xmath87 torus knot ; figure [ tipiqu ] illustrates this braid when @xmath88 . here
@xmath89 is coprime to @xmath86 , to ensure that by closing the braid we obtain a knot . as a braid with axis the unknot @xmath76 . ]
it is clear from this construction that ( at least using this approach ) we can not construct an interesting family of tori that represent the homology class of @xmath85 i.e. the class of the fiber in @xmath49 , for the simple fact that the only braid with one strand is the trivial one .
this prevents us , in particular , from constructing nonisotopic symplectic representatives of a primitive homology class .
* example : braiding construction 2 .
* there is a way , presented in @xcite , to circumvent the difficulty of representing the fiber class , at least for @xmath90 , based on the fact that we can represent @xmath49 as link surgery manifold of the @xmath19-component link @xmath91 given by the hopf link @xmath92 plus a third component @xmath93 obtained by pushing a copy of @xmath94 along its @xmath38-framing ( a necklace with two rings ) .
the observation that @xmath95 shows that , by choosing @xmath96 , @xmath97 , and as @xmath6 the spanning surface @xmath98 of @xmath99 pierced by @xmath100 we have @xmath101 , for @xmath21 at least @xmath19 . at this point
we consider the family @xmath73 presented in figure [ itborr ] .
to motivate such a choice , observe that each braided block " is given by the braid producing the borromean rings .
the linking matrix of @xmath102 therefore does not depend on @xmath89 and this gives @xmath103 = \sum_{i=1}^{3}lk(\gamma_{p},h_{i } ) [ \mu(h_{i } ) ] = [ \mu(h_{1})].\ ] ] the tori @xmath74 represent therefore the homology class of @xmath104 , identified in ( [ glue ] ) with the fiber of @xmath49 .
although the @xmath73 s are homologous , the nontriviality of the borromean braid translates into the fact that they are not isotopic in @xmath54 , and we can reasonably expect that the homologous tori @xmath74 are not isotopic in @xmath49 .
other constructions based on braiding , that produce symplectic representatives in various other classes of elliptic surfaces , are contained in @xcite and @xcite . to complete the dictionary in the introduction
, we have to discuss lagrangian tori .
we start from the following observation .
consider a curve @xmath105 , lying on a fiber @xmath106 .
after dehn surgery , @xmath107 is contained in @xmath108 and satisfies therefore @xmath109 .
the torus @xmath110 is therefore lagrangian , as @xmath111 and so is its image in @xmath43 after the gluing .
for example , it is clear from this that the rim torus " @xmath112 , obtained from the trivial link surgery construction with the unknot , is lagrangian . to produce infinitely many lagrangian tori we have to look for a family @xmath73 of curves lying on a fiber of @xmath54 that are not isotopic .
it is rather clear that in order to obtain such a family we need to use a knot or a link with a fiber having a richer topology than a punctured sphere .
we will discuss two examples .
* example : toy model .
* the first example we consider is a finite toy model ( that has anyhow , as we will see later , something to teach us ) , that works for any nontrivial fibered knot @xmath76 ; in the fiber @xmath6 we consider the two homologous curves @xmath113 , a contractible curve , and a copy of the longitude of @xmath76 itself , that we denote as @xmath114 .
these two curves are illustrated in figure [ general ] .
as long as @xmath76 is nontrivial , @xmath113 and @xmath114 are not isotopic , as they are respectively unknotted and knotted in @xmath52 .
* example : torus knots on the trefoil fiber . *
the second example we consider , taken from @xcite , is an infinite family @xmath82 of curves in the fiber of the trefoil @xmath76 ( or , by simple modification , the figure eight knot or any knot containing one of these knots as murasugi - summands ) .
the family is presented in the l.h.s . of figure [ torusfive ] ; it is not too hard to verify that , as a knot in @xmath52 , @xmath73 is the torus knot @xmath115 .
note that in both examples above the linking number of @xmath73 and @xmath76 is @xmath38 , as @xmath73 lies in the spanning surface of @xmath76 .
the lagrangian tori @xmath74 of @xmath79 are therefore nullhomologous .
to obtain lagrangian tori with nontrivial homology class we can consider a @xmath3-component link given by @xmath76 and a copy @xmath116 of its meridian ; such a link admits an obvious fibration , the fiber being the spanning surface of @xmath76 pierced once by @xmath116 .
( note that this is not the spanning surface of the link ; its dual cohomology class is in fact @xmath117 . ) by suitably positioning the curves @xmath73 above
, we can get @xmath118 for the toy model , and @xmath119 for the trefoil case , where @xmath86 is any positive number and we require @xmath120 . the r.h.s .
of figure [ torusfive ] illustrates such a case , with @xmath121 . on a fiber of the trefoil @xmath76 ( l.h.s . ) and with linking number @xmath60 with the meridian @xmath116 ( r.h.s . ) ; here , @xmath122.,title="fig : " ] on a fiber of the trefoil @xmath76 ( l.h.s . ) and with linking number @xmath60 with the meridian @xmath116 ( r.h.s . )
; here , @xmath122.,title="fig : " ] in the latter construction the lagrangian tori @xmath74 satisfy @xmath123 = \mbox{lk}(\gp , m)[s^{1 } \times \mu(m)]\ ] ] in @xmath43 , where @xmath124 and the gluing conditions of ( [ glue ] ) determined by the fiber spanning @xmath76 pierced once by @xmath116 .
it is not difficult to verify that , by choosing @xmath125 , the manifold @xmath43 actually coincides with the knot surgery manifold @xmath126 .
we can return now to the symplectic case to try to take advantage of the constructions of lagrangian tori .
first , we want to point out that whenever we can construct a homologically essential lagrangian torus , we can make it symplectic by a perturbation of the symplectic form ( see @xcite ) .
therefore , the constructions of nonisotopic lagrangian tori immediately provide examples of nonisotopic symplectic tori whenever these tori are essential ( the proof of nonisotopy depends in fact only on _ smooth _ properties of the tori ) .
second , we want to take advantage of the fact , observed above , that we can find nullhomologous curves , on a fiber @xmath6 , that are nonisotopic . if we can sum " these curves to a fixed transverse curve , we can hope that the resulting symplectic tori inherit the nonisotopy .
this is the key point of the next example .
* example : cabling construction .
* we can apply the idea mentioned above in the following way ( see @xcite , @xcite ) .
take again a fibered knot @xmath76 and consider a copy of the longitude and a contractible curve on a fiber , as in the toy model discussed above .
we denote by @xmath73 the @xmath127-cable of @xmath76 .
it is easy to see that such cables , up to isotopy , are transverse to the fibration of @xmath83 , so that , in @xmath79 , the tori @xmath74 are symplectic .
their homology class is as usual determined by @xmath128 , so that @xmath129 = [ s^{1 } \times \mu(k ) ] = [ f]$ ] .
we can say that different values of @xmath89 correspond to the different number of times we have _ circle summed _ the longitude ( the @xmath130-cable ) to the meridian ( the @xmath131-cable ) .
note that , also in this case , this construction makes sense only for a nontrivial @xmath76 . in a more general vein
, we can consider curves , transverse to the fibration , obtained by circle summing a copy ( or @xmath86 copies ) of the meridian to nullhomologous , nonisotopic curves as the ones presented in figure [ torusfive ] .
an example of that is given in figure [ gamma ] .
this construction can be applied , for example , when @xmath132 ; in this case the tori represent the homology class @xmath133 $ ] of @xmath134 . in particular , this is a primitive homology class . .
the common feature of the constructions of the previous section is a family of homologous curves , transverse to , or contained in , a fiber of @xmath54 , and that are ( or at least appear ) nonisotopic .
in fact , the proof of nonisotopy in @xmath54 for the examples of section [ const ] reduces to an obvious observation ( as in the case where the @xmath73 s are different torus knots ) or to the computation of the alexander polynomial of the link @xmath135 , that depends on the isotopy class of @xmath73 . however
, the fact that the @xmath73 s are not isotopic does not imply that the tori @xmath136 are not isotopic , and this requires us to use a direct method to decide this question .
several methods can be used .
a useful approach to this question , suggested in @xcite , is based on the observation that , if two tori @xmath137 , @xmath138 are isotopic , there exist a diffeomorphism of their exteriors @xmath139 .
such a diffeomorphism implies , on its own , that there exist a diffeomorphism of the manifolds that we obtain by fiber summing @xmath140 identifying @xmath137 and the fiber @xmath78 of @xmath140 , i.e. @xmath141 . a word of caution is needed here .
in general , the definition of fiber sum requires a choice of a diffeomorphism between @xmath78 and @xmath137 , and a lifting of such a diffeomorphism to the ( topologically trivial ) circle bundle over these tori . in general ,
such choices affect the smooth type of the resulting manifold .
in particular , it is _ a priori _ impossible to establish if an isotopy between @xmath142 and @xmath143 preserves the choice of a homology basis for these tori ( as we are not assuming the stronger condition that the marked tori are images of _ isotopic embeddings _ ) , or their framing .
however , the choice of @xmath140 makes the situation much simpler , as any orientation preserving diffeomorphism of @xmath144 extends to an orientation preserving diffeomorphism of @xmath145 ( see e.g. @xcite ) , so that the smooth type of @xmath141 is in fact unaffected by the choice of the gluing map , in particular it depends only on the isotopy type of @xmath137 .
more is actually true , namely it depends only on the diffeomorphism type of the pair @xmath146 , or even of the exteriors ; the smooth type of @xmath141 will therefore distinguish the tori @xmath137 up to that equivalence
. therefore we can try to detect the isotopy type of @xmath137 by computing the seiberg - witten invariants of @xmath141 ( i.e. the relative seiberg - witten invariants of the pair @xmath146 ) .
this technique can be applied ( with various degrees of difficulty ) to all the constructions of section [ const ] , but to make the discussion more concrete , we are going to consider two explicit cases , namely the family of symplectic tori arising from the family of transverse curves @xmath73 of figure [ gamma ] , and the family of lagrangian tori arising from the family @xmath73 in the r.h.s .
of figure [ torusfive ] , for the case where @xmath121 .
remember than in the first case the manifold @xmath43 is the knot surgery manifold obtained from ( say ) @xmath140 and the trefoil knot @xmath76 , while in the second case @xmath43 is obtained by using two copies of @xmath140 and the link @xmath124 .
let s proceed with the first case : the manifold @xmath147 , then , is itself a link surgery manifold , as it is obtained as in ( [ glue ] ) from the @xmath3-component link @xmath148 summing copies of @xmath140 along @xmath85 and @xmath74 ( as remarked above , the choice of the gluing diffeomorphism is immaterial ) .
we can compute the seiberg - witten polynomial as in ( [ alex ] ) , ( using the fact that @xmath149 ) to get @xmath150 the task of computing the alexander polynomial of the link @xmath148 of figure [ gamma ] , or of other classes of links , appears discouraging .
moreover , to prove that the seiberg - witten polynomial of ( [ mypol ] ) effectively distinguishes the smooth type of @xmath151 for different values of @xmath89 , we need to show that there is no automorphism of the second homology group transforming one polynomial into the other .
those are not simple problems , at least in the general case .
there is however a simple shortcut that we can take in order to obtain , essentially , the same kind of answer to problem ( p ) , namely exhibit infinitely many nonequivalent tori .
in fact , if we are able to show that the number @xmath152 of seiberg - witten basic classes of @xmath147 satifies @xmath153 , we are granted that infinitely many among the @xmath147 are non - diffeomorphic , as @xmath154 . to prove
this is a rather simpler task : in fact this amounts to showing that the number of nonzero terms of @xmath155 satisfies that property .
we can simplify this task even further : using torres formula we can observe that @xmath156 as a knot in @xmath52 , @xmath73 is the torus knot @xmath115 : using the formula for the alexander polynomial of torus knots , we obtain that @xmath152 is bounded below by the number of nonzero terms in the alexander polynomial of @xmath115 , namely @xmath157 but it is at this point an easy exercise ( see @xcite ) to verify that such number grows with no bound with @xmath89 .
the second case , for the lagrangian tori arising from figure [ torusfive ] , follows almost _ verbatim _ : here , the manifold @xmath147 is itself a link surgery manifold obtained as in ( [ glue ] ) from the @xmath19-component link @xmath158 summing copies of @xmath140 along @xmath85 , @xmath159 and @xmath74 .
equation ( [ alex ] ) gives now @xmath160 and , again by application of torres formula , the number of basic classes of that manifold is bounded below by the number of nonzero terms in the polynomial @xmath161 to obtain the last equality we have used the fact that the pair @xmath162 is given by the torus knot @xmath115 and its meridian @xmath116 , so that @xmath163 .
it is clear at this point that also in this case the number of basic classes is bounded below by ( twice ) the number of nonzero terms of the alexander polynomial of @xmath115 .
we want to point out the existence of another way to use seiberg - witten theory for detecting nonisotopy of lagrangian tori , namely the lagrangian framing defect .
such invariant , whose definition depends _ a priori _ on symplectic properties of a lagrangian torus , can be shown under favorable circumstances to be a smooth invariant , relating it to seiberg - witten invariants . a detailed description of this invariant appears in @xcite .
finally , we discuss the simplest invariant that can detect nonisotopy of tori ( or other submanifolds ) , namely the fundamental group of the exterior . in this case ,
different fundamental groups imply even the absence of pair homeomorphism .
for some of the constructions of section [ const ] ( for example , the braiding constructions of @xcite and @xcite ) this invariant gives no information , while it is sufficient in other cases ( for example , the cabling construction , see @xcite ) to distinguish the tori even up to pair homeomorphism .
we will discuss its use in the following section .
all the examples discussed in the previous sections concerned the case of tori
. examples of nonisotopic symplectic surfaces of genus greater than @xmath60 have been somewhat elusive ( at least in our setup ; otherwise , examples in non - simply connected manifolds have been presented in @xcite , and singular examples in @xcite ) .
one of the reason for this is the difficulty in using seiberg - witten poynomials , as in section [ noniso ] , to detect potential higher genus examples built by doubling " the braiding constructions considered in section [ const ] . in this section we are going to discuss how we can instead double the cabling construction and retain enough information from the fundamental group of the exterior to be able to distinguish higher genus examples . for sake of simplicity we will consider just the case of genus @xmath3 , but the general case follows along the same lines . for details ,
see @xcite . start by considering the knot surgery manifold @xmath126 obtained from a nontrivial fibered knot @xmath76 choosing @xmath164 .
in such a manifold we can identify a symplectic surface of genus @xmath165 and self - intersection @xmath166 obtained by capping off the spanning surface @xmath6 of the knot with a @xmath166-disk section in @xmath167 , according to the gluing prescription of ( [ glue ] ) .
summing such a surface with a copy of the elliptic fiber , and suitably resolving , we get a symplectic surface @xmath168 of genus @xmath169 and self - intersection @xmath38 . following
@xcite we can define a symplectic , simply connected @xmath0-manifold by fiber summing two copies of @xmath126 along @xmath168 , i.e. @xmath170 identifying the two boundary components @xmath171 with a diffeomorphism that is the identity on @xmath168 and reverses the orientation on @xmath23 . in @xmath126
we have the family of homologous symplectic tori @xmath74 that we introduced in section [ const ] through the cabling construction .
these tori intersect the surface @xmath168 in a single transverse point , inherited from the intersection point between @xmath73 and the spanning surface of @xmath76 .
it follows that in @xmath172 the two copies of @xmath74 located in each half give , by connected sum , a genus @xmath3 surface @xmath173 .
we can attempt to distinguish these surfaces using the fundamental group of the exteriors @xmath174 .
a rather lengthy exercise in van kampen theorem ( see @xcite ) shows that this group is given by @xmath175 i.e. the fundamental group of the homology sphere obtained by applying dehn surgery to @xmath52 along @xmath76 with slope @xmath176 .
the problem of distinguishing the surfaces @xmath173 boils down therefore to showing that , for a given knot @xmath76 , infinitely many among these homology spheres are distinguished by means of their fundamental group . before commenting on that , note that if @xmath76 is the unknot , this fundamental group is always trivial : this is exactly what we expect , as it follows immediately from the construction of section [ const ] that in that case the tori @xmath74 are isotopic ( the @xmath127-cable of the unknot is isotopic to the meridian ! ) .
let us return to the problem of distinguishing up to isomorphism the groups of ( [ group ] ) .
we will consider here the case where @xmath76 is a fibered hyperbolic knot . in this case , with a finite number of exception , the homology spheres @xmath177 are hyperbolic @xmath19-manifolds .
mostow rigidity implies in that case that these manifolds are homeomorphic if and only if their fundamental groups coincide .
therefore , we are left with the problem of showing that infinitely many of the @xmath177 are not homeomorphic .
but at this point we can appeal to theorem 1 of @xcite , that states that hyperbolic fillings of a hyperbolic knot exterior can be homeomorphic only if the knot is amphicheiral and the slopes are opposite . restricting ourselves to positive @xmath89 for sake of simplicity , we deduce that for all but finitely many choices of @xmath89 the fundamental group of ( [ group ] ) distinguishes ( up to homeomorphism the exteriors , hence isotopy ) the genus @xmath3 surfaces @xmath173 .
an analogous result can be proven for all nontrivial fibered knots .
99 v. arnold , _ first steps in symplectic topology _
, russian math . surveys * 41*(6 ) , 121 ( 1986 ) .
d. auroux , s. donaldson , l. katzarkov , _ luttinger surgery along lagrangian tori and non - isotopy for singular symplectic plane curves _
ann . * 326 * , 185203 ( 2003 ) .
s. a. bleiler , c. d. hodgson , j. r. weeks , _ cosmetic surgery on knots _ , geometry & topology monographs , volume 2 : proceedings of the kirbyfest , 2334 ( 1999 ) .
d. eisenbud , w. neumann , _ three dimensional link theory and invariants of plane curve singularities _ , annals of mathematics studies * 110 * ( 1985 ) .
y. eliashberg , l. polterovich , _ unknottedness of lagrangian surfaces in symplectic @xmath0-manifolds _ , internat .
notices * 11 * , 295301 ( 1993 ) .
y. eliashberg , l. polterovich , _ local lagrangian @xmath3-knots are trivial _ , ann . of math .
( 2 ) * 144 * , 6176 ( 1996 ) .
y. eliashberg , l. polterovich , _ the problem of lagrangian knots in four - manifolds _ , geometric topology .
proceedings of the 1993 georgia international topology conference ( w. h. kazez , ed . ) , international press , 313327 ( 1997 ) .
t. etg , b. d. park , _ non - isotopic symplectic tori in the same homology class _ , trans .
soc . * 356 * , 37393750 ( 2004 ) .
t. etg , b. d. park , _ homologous non - isotopic symplectic tori in a k3-surface _ , commun .
math . , to appear ( 2003 ) .
t. etg , b. d. park , _ homologous non - isotopic symplectic tori in homotopy rational elliptic surfaces _ , math .
cambridge philos .
soc . , to appear ( 2003 ) .
r. fintushel , r. stern , _ knots , links and @xmath0-manifolds _ , invent . math . * 134 * , 363400 ( 1998 ) .
r. fintushel , r. stern , _ surfaces in @xmath0-manifolds _ , math .
lett . * 4 * , 907914 ( 1997 ) .
r. fintushel , r. stern , _
symplectic surfaces in a fixed homology class _ , j. differential geom . *
52 * , 203222 ( 1999 ) .
r. fintushel , r. stern , _
invariants for lagrangian tori _ , geom .
* 8 * , 947968 ( 2004 ) .
r. gompf , _ a new construction of symplectic manifolds _ , ann . of math .
( 2 ) * 142 * , 527595 ( 1995 ) .
r. gompf , a. stipsicz , _
4-manifolds and kirby calculus
_ , graduate studies in mathematics ( vol .
20 ) ams ( 1999 ) .
c. mcmullen , c. taubes , _ 4-manifolds with inequivalent symplectic forms and 3-manifolds with inequivalent fibrations _ , math
* 6 * , 681696 ( 1999 ) .
b. d. park , _ doubling homotopy @xmath178 surfaces _ ,
j. knot theory ramifications * 12 * , 347354 ( 2003 ) .
b. d. park , m. poddar , s. vidussi , _ homologous non - isotopic symplectic surfaces of higher genus _ , preprint ( 2004 ) .
i. smith , _ symplectic submanifolds from surface fibrations _ , pacif . j. math . * 198 * , 197206 ( 2001 ) . c. taubes ,
_ the seiberg - witten invariants and 4-manifolds with essential tori _
, geom . topol .
* 5 * , 441519 ( 2002 ) .
b. siebert , g. tian , _ on the holomorphicity of genus two lefschetz fibrations _ , ann . of math .
, to appear ( 2003 ) .
g. tian , _
symplectic isotopy in four dimensions _ , first international congress of chinese mathematicians ( beijing , 1998 ) , ams / ip stud .
, * 20 * , 143147 ( 2001 ) .
s. vidussi , _ nonisotopic symplectic tori in the fiber class of elliptic surfaces _ , j. symplectic geom . * 2 * , 207218 ( 2004 ) s. vidussi , _
lagrangian surfaces in a fixed homology class : existence of knotted lagrangian tori _ , j. differential geom .
, to appear ( 2003 ) .
s. vidussi , _ symplectic tori in homotopy @xmath140 s _ , proc .
soc . , to appear ( 2003 ) .
|
the purpose of this paper is to present some results on the existence of homologous , nonisotopic symplectic or lagrangian surfaces embedded in a simply connected symplectic @xmath0-dimensional manifold .
|
type ia supernovae ( sne ia ) have been widely believed to be a thermonuclear explosion of a mass - accreting white dwarf ( wd ) ( e.g. , @xcite for a recent review ) .
however , the immediate progenitor binary systems have not been identified yet ( @xcite ) .
there exist two models discussed frequently as progenitors of sne ia : 1 ) the chandrasekhar ( ch ) mass model , in which a mass - accreting carbon - oxygen ( c+o ) wd grows in mass up to the ch mass and explodes as an sn ia , and 2 ) the sub - chandrasekhar ( sub - ch ) mass model , in which an accreted layer of helium atop a c+o wd ignites off - center for a wd mass well below the ch mass .
the early time spectra of the majority of sne ia are in excellent agreement with the synthetic spectra of the ch mass models , while the spectra of the sub - ch mass models are too blue to be comparable with the observations ( @xcite ; @xcite ) . for the evolution of accreting wds toward the ch mass ,
two scenarios have been proposed : 1 ) a double degenerate ( dd ) scenario , i.e. , merging of double c+o wds with a combined mass surpassing the ch mass limit ( @xcite ; @xcite ) , and 2 ) a single degenerate ( sd ) scenario , i.e. , accretion of hydrogen - rich matter via mass transfer from a binary companion ( e.g. , @xcite ; @xcite ) . the issue of dd vs. sd
is still debated ( e.g. , @xcite ) , although theoretical modeling has indicated that the merging of wds leads to the accretion - induced collapse rather than sn ia explosion ( saio & nomoto 1985 , 1998 ; @xcite ) . for the ch / sd scenario , a new evolutionary model has been proposed by hachisu , kato , & nomoto ( 1996 ; hereafter hkn96 ) . hkn96
have shown that if the accretion rate exceeds a critical rate , the wd blows a strong wind and burns hydrogen steadily at this critical rate and expels excess matter in the wind .
the wd increases its mass up to the ch mass avoiding formation of a common envelope .
li & van den heuvel ( 1997 ) have extended hkn96 s model to a system consisting of a mass - accreting wd and a lobe - filling , more massive , main - sequence ( ms ) or sub - giant star ( hereafter `` wd+ms system '' ) , identified with luminous supersoft x - ray sources , and found that such a system is one of the main progenitors of sne ia as well as a system consisting of a wd and a lobe - filling , less massive , red - giant ( hereafter `` wd+rg system '' ) proposed by hkn96 .
recently , yungelson & livio ( 1998 ) have reanalyzed the models by hkn96 and li & van den heuvel based on their population synthesis code and concluded that both hkn96 s wd+rg and li & van den heuvel s wd+ms systems can account for only ( at most ) 10% of the inferred rate of sne ia in our galaxy .
however , yungelson & livio ( 1998 ) overlooked important evolutionary processes both in the wd+ms and wd+rg systems . in this paper
, we first describe an important evolutionary process to form the wd+ms system , which has been overlooked in previous works ( e.g. , @xcite ; @xcite ; @xcite ) . another evolutionary process leading to the wd+rg system
is discussed elsewhere ( hachisu , kato , & nomoto 1999 , hereafter hkn99 ) . in 2 , we describe the new evolutionary process to form the wd+ms system . including this new evolutionary path
, we will show that the secondary ( slightly evolved ms star ) becomes a helium - rich star like in u sco ( e.g. , @xcite ) , which is transferring helium - rich matter onto the primary ( wd ) .
we have reanalyzed such a helium - rich matter accretion onto the wd based on the optically thick wind theory developed by kato & hachisu ( 1994 ) .
if the secondary ms star has a mass of @xmath2 , a wd with an initial mass of @xmath3 grows in mass to the ch mass and explodes as an sn ia .
we describe the evolution of such a wd+ms system in 3 .
the new parameter region thus obtained is much wider than that by li & van den heuvel ( 1997 ) and yungelson & livio ( 1998 ) .
discussions follow in 4 , in which we have estimated the realization frequency of our wd+ms systems that accounts for about a third of the inferred rate of sne ia in our galaxy .
in this section , we describe the evolutionary path to form the wd+ms system where the secondary ( ms star ) has a helium - rich envelope .
this important evolutionary path shown in figure [ hemtsn1a ] has been overlooked in previous work .
yungelson & livio ( 1998 ) applied their population synthesis code to the wd+ms systems and found that the realization frequency of the wd+ms systems is at most one tenth of the inferred rate of sne ia in our galaxy . in their population synthesis code , they consider only initial systems consisting of a more massive agb star with a c+o core and a less massive main - sequence star .
this system undergoes a common envelope evolution and finally yields a binary system of a mass - accreting c+o wd and a lobe - filling ms or sub - giant star .
their code does not include another important evolutionary path , in which a more massive component fills up its inner critical roche lobe when it develops _ a helium core _ of @xmath0 in _ its red - giant phase_. this evolutionary path from _ stage a _ to _ f _ in figure [ hemtsn1a ] has been first introduced by hachisu & kato ( 1999 ) to explain the helium - rich companion of the recurrent nova u sco .
we consider , for example , a close binary at a separation @xmath4 with the primary of mass @xmath5 and the secondary of @xmath6 ( _ stage a _ ) .
when the primary has evolved to a red - giant of the radius @xmath7 forming a helium core of mass @xmath8 ( _ stage b _ ) , it fills up its inner critical roche lobe , i.e. , @xmath9 . here , @xmath10 is the effective radius of the inner critical roche lobe of the primary , which is approximated by eggleton s ( 1983 ) formula , @xmath11 for the mass ratio @xmath12 .
the long dashed line in figure [ m7m2com2 ] shows the primary radius @xmath7 against the helium core mass @xmath8 ( @xcite ) . the radius of the primary increases with the helium core mass from @xmath13 to @xmath14 until helium burning ignites at the center of the helium star .
if the primary fills its inner critical roche lobe at a certain @xmath8 , i.e. , @xmath15 , the separation of the binary , @xmath4 , is given by @xmath16 using this relation , we can plot the initial separation @xmath17 against @xmath8 as shown by the thick solid line in figure [ m7m2com2 ]
. then the mass transfer begins .
this mass transfer is dynamically unstable because the primary red - giant star has a convective envelope .
the binary undergoes common envelope ( ce ) evolution ( _ stage c _ ) , which yields a much more compact close binary consisting of a naked helium star of @xmath18 and a main - sequence star of @xmath19 ( _ stage d _ ) . figure [ m7m2com2 ] shows the separation @xmath20 and the inner critical roche lobe radius of the secondary @xmath21 after the common envelope evolution . here , we assume the relation @xmath22 with the efficiency @xmath23 for the common envelope evolution ( e.g. , @xcite ; @xcite ; @xcite ) . after the common envelope evolution ,
the radii of the inner critical roche lobes of the primary and the secondary become @xmath24 and @xmath25 , respectively .
since the secondary radius ( shown by the dashed line in figure [ m7m2com2 ] for the @xmath26 zams ) should be smaller than its inner critical roche lobe , i.e. , @xmath27 , the initial separation @xmath28 should exceed @xmath29 as shown in figure [ m7m2com2 ] .
the upper bound of the initial separation is obtained from the maximum radius of the @xmath30 star which has formed a helium core , i.e. , @xmath31 .
thus , the allowable range of the initial separations for our model is @xmath32 , so that @xmath33 and @xmath34 for the case of figure [ m7m2com2 ] . after the hydrogen - rich envelope
is stripped away and hydrogen shell burning vanishes , the naked helium core contracts to ignite central helium burning and becomes a helium main sequence star of mass @xmath8 ( _ stage d _ ) . for @xmath35 ,
its c+o core mass is less than @xmath36 which is the lower mass limit to the non - degenerate carbon ignition ( e.g. , @xcite ; @xcite for a review )
. then the helium star forms a degenerate c+o core , whose mass @xmath37 grows by helium shell burning .
when @xmath37 becomes @xmath38 and the core becomes strongly degenerate , its helium envelope expands to @xmath39 ( e.g. , @xcite ; @xcite ) to fill its inner critical roche lobe again ( _ stage e _ ) .
helium is transferred to the secondary stably on an _ evolutionary time scale _ of @xmath40 yr because the mass ratio is smaller than @xmath41 ( @xmath42 ) .
the resultant mass transfer ( mt ) rate is @xmath43 yr@xmath44 , which is too low to form a common envelope ( e.g. , @xcite ; @xcite ) .
after the helium envelope is lost , the primary becomes a c+o wd of @xmath45 ( _ stage f _ ) . in figure
[ m7m2hemt ] , @xmath46 is plotted as a function of @xmath8 . here , we assume the relation of ( in solar mass units of @xmath47 ) @xmath48 for the final degenerate c+o wd mass vs. the initial helium star mass relation ( reduced from the evolutionary paths by @xcite and @xcite ) .
helium stars with @xmath49 do not expand to @xmath50 , thus burning most of helium to c+o without transferring helium to the secondary . for @xmath51 ,
the c+o core has a mass larger than @xmath36 before becoming degenerate , thus igniting carbon to form an o+ne+mg core ( e.g. , @xcite ) , thus we do not include their mass range . for @xmath52 , helium is not ignited to form a c+o core .
the secondary increases its mass @xmath53 by receiving almost pure helium matter of @xmath54 , which is a function of @xmath8 as plotted in figure [ m7m2hemt ] .
then a helium - enriched envelope is formed as illustrated in figure [ hemtsn1a ] ( _ stage f _ ) . after the helium mass transfer ( mt ) , the separation increases by 10%@xmath5540% , i.e. , @xmath56 . here , we assume the conservation of the total mass and angular momentum during the helium mass transfer , which leads to the relation of @xmath57 and we use this to obtain @xmath58 and then the orbital period @xmath59 in figure [ m7m2hemt ] . since the secondary receives @xmath60 helium , its hydrogen content in the envelope decreases to @xmath61 if helium is completely mixed into the central part of the star . however , the envelope of the mass - receiving star is not convective but radiative so that the helium content may be higher in the outer part of the star .
we have examined total @xmath62 cases , @xmath63 , 5 , 6 , 7 , @xmath64 and @xmath65 , 1.5 , 2.0 , 2.5 , @xmath66 , and have found the possible progenitors to be in the range of @xmath67 and @xmath68 with the separation of @xmath69 . in these cases , the secondary forms a helium - enriched envelope for the primary mass of @xmath70 ( but not for @xmath71 ) .
we assume , in this paper , that the average mass fractions of hydrogen and helium in the envelope are @xmath72 and @xmath73 respectively .
( assuming the solar abundance of heavy elements @xmath74 ) .
for the @xmath75 case , much more helium is transferred , while much less helium is transferred for the @xmath76 case ( see table [ tbl-1 ] ) .
starting from a close binary of @xmath77 and @xmath78 with the separation of @xmath79 , we have followed a growth of the wd component to examine whether the wd reaches @xmath80 and explodes as an sn ia ( from _ stage f _ to _ j _ in fig .
[ windsn1a ] ) .
the initial secondary now has a helium - rich envelope ( _ stage f _ ) .
it evolves to expand and fills its inner critical roche lobe near the end of main - sequence ( ms ) phase .
then , it starts mass transfer ( _ stage g _ ) .
this is a _ case a mass transfer _
after kippenhahn & weigert ( 1967 , also @xcite ) or a _ cataclysmic - like mass transfer _ after iben & tutukov ( 1984 ) .
since the donor is more massive than the accretor ( wd component ) , the separation decreases and the inner critical roche lobe decreases even to scrape the envelope mass off the donor star .
thus , the mass transfer proceeds on a _ thermal time scale _ rather than an evolutionary time scale ( _ stage h _ ) .
the transferred matter is helium - rich as observed in the recurrent nova u sco .
hachisu , kato , & nomoto ( 1996 , hkn96 ) have shown that optically thick winds blow from the white dwarf when the mass accretion rate exceeds a critical value ( _ stage h _ ) . in the present case ,
the accreted matter to form the white dwarf envelope is helium - rich , which is different from the solar abundance in hkn96 .
assuming @xmath72 , @xmath73 , and @xmath74 , we have calculated the envelope models of accreting white dwarfs for various accretion rates and white dwarf masses , i.e. , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath85 , @xmath86 , @xmath87 , @xmath88 , @xmath89 , and @xmath90 .
optically thick winds occur for all these ten cases of @xmath91 as shown for five cases of @xmath92 , @xmath85 , @xmath87 , @xmath88 , and @xmath90 in figures [ dmdtenvx50z02 ] and [ dmdttrvx50z02 ] .
our numerical methods have been described in kato & hachisu ( 1994 ) .
the envelope solution is uniquely determined if the envelope mass @xmath93 is given , where @xmath93 is the mass above the base of the hydrogen - burning shell .
therefore , the wind mass loss rate @xmath94 and the nuclear burning rate @xmath95 are obtained as a function of the envelope mass @xmath93 , i.e. , @xmath96 and @xmath97 as shown in figure [ dmdtenvx50z02 ] .
the envelope mass of the white dwarf is determined by @xmath98 if the mass transfer rate does not change much in a thermal time scale of the wd envelope , the steady - state @xmath99 , i.e. , @xmath100 is a good approximation .
in such a steady - state approximation , the ordinates in figures [ dmdtenvx50z02 ] and [ dmdttrvx50z02 ] , @xmath101 , are regarded as the mass transfer rate from the secondary @xmath102 .
thus , the envelope solution is determined from the relation in figure [ dmdtenvx50z02 ] for the given mass transfer rate @xmath102 .
the photospheric radius , temperature , and velocity are also obtained from the relations in figure [ dmdttrvx50z02 ] .
optically thick winds blow when the mass transfer rate exceeds the critical rate , which corresponds to the break of each solid line in figure [ dmdtenvx50z02 ] .
there exists only a static ( no wind ) envelope solution for the mass transfer rate below this break .
its critical accretion rate is approximated as @xmath103 for @xmath104 , @xmath105 and @xmath106 . if the mass accretion rate exceeds the critical rate , i.e. , @xmath107 , the strong wind blows from the white dwarf . the white dwarf accretes helium almost at the critical rate , i.e. , @xmath108 , and expels the excess matter in the wind at a rate of @xmath109
here , we assume that the white dwarf accretes helium - rich matter from the equator via the accretion disk and blows winds to the pole or off the equator . steady hydrogen shell burning converts hydrogen into helium atop the c+o core , which can be regarded as a helium matter accretion onto the c+o wd . to estimate whether or not the white dwarf mass grows to @xmath80
, we must calculate the mass accumulation efficiency , that is , the ratio between the mass accumulated in the wd after h / he - burning and the mass transferred from the companion .
we denote the efficiency by @xmath110 and @xmath111 for hydrogen shell - burning and helium shell - burning , respectively .
as shown in the previous subsection , the excess matter is blown in the wind when the mass transfer rate exceeds the critical rate , which leads to @xmath112 , for the wind phase . during the evolution of mass - accreting white dwarfs ,
the accretion rate becomes lower than @xmath113 in some cases .
then the wind stops ( _ stage i _ in fig . [ windsn1a ] ) .
hydrogen steadily burns for @xmath114 .
then , we have @xmath115 . for @xmath116 ,
hydrogen shell - burning becomes unstable to trigger weak shell flashes .
once a weak hydrogen shell flash occurs , a part of the envelope mass of the white dwarf may be lost from the system ( e.g. , @xcite ) . in the present study ,
no mass loss is assumed during the weak hydrogen shell flashes until the mass transfer rate becomes lower than @xmath117 , i.e. , @xmath115 for @xmath118 .
when the mass transfer rate becomes lower than @xmath119 , however , no mass accumulation is expected from such relatively strong hydrogen shell - flashes ( e.g. , @xcite ) , i.e. , @xmath120 for @xmath121 .
therefore , we stop calculating the binary evolution either when the primary reaches @xmath80 ( _ stage j _ in fig . [ windsn1a ] ) , i.e. , @xmath122 or the mass transfer rate becomes lower than @xmath119 . to summarize , we have used the following simplified relation @xmath123 for the mass accumulation efficiency of hydrogen shell burning . for @xmath124 , steady hydrogen burning
is equivalent to the helium accretion at the critical rate of @xmath113 given by equation ( [ critical ] ) . in this case ,
weak helium shell flashes are triggered ( e.g. , @xcite ) and almost all processed matter is accumulated on the c+o wd .
recently , kato & hachisu ( 1999 ) have recalculated the helium wind model after the helium shell flashes and estimated the mass accumulation efficiency with the updated opal opacity ( @xcite ) . here
, we adopt their new results in a simple analytic form , i.e. , @xmath125 where the helium mass accretion rate , @xmath126 , is in units of @xmath47 yr@xmath44 .
we use this formula for various white dwarf masses and accretion rates , although their results are given only for the @xmath127 white dwarf ( @xcite ) .
the wind velocity in helium shell flashes reaches as high as @xmath128 km s@xmath44 ( @xcite ) , which is much faster than the orbital velocities of our wd+ms binary systems @xmath129 km s@xmath44 .
it should be noted that either a roche lobe overflow or a common envelope does not play a role as a mass ejection mechanism because the envelope matter goes away quickly from the system without interacting with the orbital motion ( see @xcite for more details ) .
we have followed binary evolutions from the initial state of @xmath130 or @xmath131 , where @xmath132 is the initial orbital period . here
, the subscript naught ( 0 ) denotes _ stage f _ in figure [ windsn1a ] , that is , before the mass transfer from the secondary starts .
the radius and luminosity of slightly evolved main - sequence stars are calculated from the analytic form given by tout et al .
the mass transfer proceeds on a thermal time scale for the mass ratio of @xmath133 .
we approximate the mass transfer rate as @xmath134 where @xmath135 is the kelvin - helmholtz timescale ( e.g. , @xcite ) , and @xmath136 and @xmath137 are the mass - radius exponents of the inner critical roche lobe and the main sequence component , respectively ( e.g. , @xcite ) . the effective radius of the inner critical roche lobe , @xmath138 , is calculated from equation ( [ inner_critical_roche_lobe ] ) .
the separation is determined by @xmath139 we estimate the total mass and angular momentum losses by the winds as @xmath140 and @xmath141 where @xmath142 is a numeric factor expressing the specific angular momentum of the wind , i.e. , @xmath143 for the very fast winds such as @xmath144 , the wind has , on average , the same specific angular momentum as that of the wd component , i.e. , @xmath145 because the wind is too fast to interact with the orbital motion ( see also hkn99 ) . in equations ( [ separation_change])@xmath55([specific_angular_momentum_wind ] ) , we must take into account the sign of the mass loss rates , i.e. , @xmath146 , @xmath147 , @xmath148 , and so on . figure [ evlx50z02 ] shows an example of such close binary evolutions which lead to the sn ia explosion .
here , we start the calculation when the secondary fills its inner critical roche lobe .
the initial parameters are @xmath149 , @xmath150 , and @xmath151 ( @xmath152 d ) .
the mass transfer begins at a rate as high as @xmath153
. the wd burns hydrogen to form a helium layer at the critical rate of @xmath154 and the wind mass loss rate is @xmath155 .
thus a large part of the transferred matter is blown off in the wind .
since the mass ratio @xmath156 decreases , the mass transfer rate determined by equation([secondary_mass_transfer ] ) gradually decreases below @xmath113 .
the wind stops at @xmath157 . the mass transfer rate becomes lower than @xmath158 yr@xmath44 ( for @xmath159 ) at @xmath160 and very weak shell flashes may occur .
the wd mass gradually grows to reach @xmath80 at @xmath161 yr . at this time , the mass transfer rate is still as high as @xmath162 , because the mass ratio @xmath156 is still larger than @xmath41 , implying thermally unstable mass transfer .
this wd+ms system may not be observed in x - rays during the strong wind phase due to the self - absorption of x - rays .
however , it is certainly identified as a luminous supersoft x - ray source from @xmath157 to @xmath163 , because it is in a steady hydrogen shell burning phase without a strong wind . just before the explosion
, it may be observed as a recurrent nova like u sco , which indicates a helium - rich accretion in quiescence ( @xcite ) .
thus we have obtained the final outcome of close binary evolutions for various sets of @xmath130 or @xmath164 ) .
figure [ zams11ms ] depicts the final outcomes in the @xmath165 plane for @xmath166 .
final outcome is either 1 . forming a common envelope ( denoted by @xmath167 ) because the mass transfer rate at the beginning is large enough to form a common envelope , i.e. , @xmath168 for @xmath169 yr@xmath44 as seen in figure [ dmdttrvx50z02 ] , 2 .
triggering an sn ia explosion ( denoted by @xmath170 , @xmath171 , or @xmath172 ) when @xmath173 , or 3 . triggering repeated nova cycles ( denoted by @xmath174 ) , i.e. , @xmath175 when @xmath176 . among the sn ia cases , the wind status at the explosion depends on @xmath102 as follows . 1 .
wind continues at the sn ia explosion for @xmath177 yr@xmath44 ( denoted by @xmath170 ) .
wind stops before the sn ia explosion but the mass transfer rate is still high enough to keep steady hydrogen shell burning for @xmath178 ( @xmath171 ) .
wind stops before the sn ia explosion and the mass transfer rate decreases to between @xmath179 at the sn ia explosion ( @xmath172 ) .
the region producing an sn ia is bounded by @xmath180 and @xmath181 d as shown by the solid line . 1 .
the left bound is determined by @xmath182 in figure [ m7m2com2 ] , where @xmath183 is the minimum radius of the secondary at the zero age main - sequence ( zams ) .
the right bound corresponds to the maximum radius at the end of main sequence , after which central hydrogen burning vanishes and the star shrinks .
the lower bound is determined by the decrease in the mass transfer rate mainly because the secondary mass decreases to reach the mass ratio @xmath156 below unity , i.e. , @xmath175 .
the upper bound is limited by the formation of a common envelope . when the mass transfer rate is as high as a few times @xmath184 yr@xmath44 or more , the photosphere of the white dwarf envelope reaches the secondary and then swallows it , i.e. , @xmath185 ( see fig . [ dmdttrvx50z02 ] ) .
it may be regarded as the formation of a common envelope .
the binary will undergo a common envelope evolution and will not become an sn ia . the final outcome of the evolutions is also plotted in figures [ zams10ms]-[zams08ms ] for other initial white dwarf masses of @xmath149 , @xmath186 , and @xmath187 , respectively .
figure [ zamstot ] shows the regions that lead to an sn ia for all the white dwarf masses of @xmath188 , @xmath83 , @xmath84 , and @xmath189 ( thin solid lines ) together with @xmath149 ( thick solid line ) .
the region for @xmath190 vanishes .
the shrinking of the upper bound for smaller @xmath191 is due to larger initial mass ratio of @xmath192 , which enhances the mass transfer rate at the beginning of mass transfer ( _ stage g _ ) , thus resulting in the formation of a common envelope .
the shrinking of the lower bound can be understood as follows : the white dwarf with smaller @xmath191 needs to accrete more mass from the slightly evolved main - sequence companion .
thus the companion s mass near the _ stage i _ or _
j _ ( just before sn ia explosion ) is smaller after a considerable part of the hydrogen - rich envelope is transferred to supply hydrogen - rich matter to the white dwarf .
the thermal time scale of the companion is longer for smaller masses , thereby decreasing the mass transfer rate down to the nova region .
we have estimated the rate of sne ia originating from our wd+ms systems in our galaxy by using equation ( 1 ) of iben & tutukov ( 1984 ) , i.e. , @xmath193 where @xmath194 and @xmath195 denote the appropriate ranges of the initial mass ratio and the initial separation , respectively , and @xmath196 , and @xmath197 are the lower and the upper limits of the primary mass that leads to sn ia explosions , respectively . for the wd+ms progenitors , we assume that @xmath198 in order to obtain a relatively compact condition of @xmath20 after the common envelope evolution . if the @xmath199 c+o wd is a _ descendant from an agb star _ , its zero - age main sequence mass is @xmath200 ( see , e.g. , eq.(11 ) of @xcite ) and the binary separation is larger than @xmath201 ( e.g. , @xcite ) .
its separation shrinks to @xmath202 after the common envelope evolution for the case of @xmath203 and @xmath204 secondary .
then the orbital period becomes @xmath205 d , which is too long to become an sn ia ( e.g. , @xcite ; see also fig .
[ zams10ms ] ) .
therefore , the wd+ms systems descending from an agb star may be rare as pointed out by yungelson & livio ( 1998 ) and may not be a main channel to sne ia . to obtain the realization frequency of our wd+ms system _ descending from a red - giant with a helium core _
, we have followed total @xmath206 evolutions with the different initial set of @xmath207 and estimated the appropriate range for the initial separation of @xmath208 for total @xmath209 cases of @xmath210 , each for the primary mass of @xmath211 , 5 , 6 , 7 , and @xmath64 and the secondary mass of @xmath212 , 1.5 , 2 , 2.5 , and @xmath213 ( see table [ tbl-1 ] ) . in table
[ tbl-1 ] , we omit the case of @xmath214 because it never leads to sn ia explosions .
we find that sn ia explosions occur for the ranges of @xmath215 , @xmath216 , and @xmath217 .
we thus obtain the realization frequency of sne ia from the wd+ms systems as @xmath218 yr@xmath44 for @xmath219 , by substituting @xmath220 , @xmath221 , @xmath222 , and @xmath217 into equation ( [ realization_frequence ] ) . for comparison
, we have obtained a realization frequency of sne ia for @xmath223 .
it is still as high as @xmath224 yr@xmath44 , which is about one fourth of the inferred rate .
our new rate of @xmath218 yr@xmath44 is about a third of the inferred rate of sne ia in our galaxy and much higher than @xmath225 yr@xmath44 ( as an upper limit ) obtained by yungelson & livio ( 1998 ) .
the reason of their low frequency is probably due to the absence of the path through the primary s helium star phase in their scenarios .
however , it should be noted here that yungelson & livio ( 1998 ) have also obtained a realization frequency of @xmath226 yr@xmath44 under the assumption of no restrictions in their binary evolutions , the conditions of which are unlikely .
a part of our wd+ms systems are identified as the luminous supersoft x - ray sources ( ssss ) ( @xcite ) .
ssss are characterized by a luminosity of @xmath227 erg s@xmath44 and a temperature of @xmath228 k @xmath229 ev ) , which have been established as a new class of x - ray sources through rosat observations ( e.g. , @xcite for a recent review ) . a population synthesis for ssss
has first been done by rappaport , di stefano , & smith ( 1994 ) , followed by a more complete population synthesis ( @xcite ) .
these calculations predict the total number of the galactic ssss of @xmath128 and led to the conclusion that the sss birth rate is roughly consistent with the observation ( e.g. , @xcite for a review ) .
our sn ia progenitors should be observed as an sss during the steady hydrogen shell burning phase without winds , which is about a few times @xmath230 yr as shown in figure [ evlx50z02 ] .
then the number of ssss from our scenario is roughly estimated to be at least about @xmath231 , which should be added to 1000 by yungelson et al .
the total number is still consistent with observations .
our wd+ms progenitor model predicts helium - enriched matter accretion onto a wd .
strong he ii @xmath232 lines are prominent in the luminous supersoft x - ray sources ( e.g. , @xcite for a recent review ) as well as in the recurrent novae like u sco ( @xcite ; @xcite ) and v394 cra ( @xcite ) .
thus the weakness of the hydrogen emission lines relative to the he ii and cno lines is very consistent with the requirement that the accreted matter and hence the envelope of the secondary have a hydrogen - poor ( helium - rich ) composition . for sne ia
, several attempts have been made to detect signature of circumstellar matter .
there has been no radio detections so far .
radio observations of sn 1986 g have provided the most stringent upper limit to the circumstellar density as @xmath233 yr@xmath44 ( @xcite ) , where @xmath234 means @xmath235 km s@xmath44 .
this is still @xmath236 times higher than the density predicted for the white dwarf winds , because the wd wind velocity is as fast as @xmath128 km s@xmath44 .
further attempts to detect high velocity hydrogen signature are encouraged .
we thank the anonymous referee for helpful comments to improve the manuscript . this research has been supported in part by the grant - in - aid for scientific research ( 05242102 , 06233101 , 08640321 , 09640325 ) and coe research ( 07ce2002 ) of the japanese ministry of education , science , culture , and sports .
ccccccccc @xmath237 & & & 0.0,0.0 & 0.0,0.2 & 0.1,0.2 & 0.37,0.37 & 0.5,0.6 & 0.6,0.6 @xmath238 & & & & & & & & @xmath239 & & & & & & & & @xmath240 & & & 1.9,2.4 & 2.4,2.6 & 1.8,2.3 & 2.3,2.8 & & @xmath241 & & & 1.8,2.4 & 2.3,2.7 & 1.8,2.3 & 2.4,2.8 & 1.8,2.3 & 2.4,2.5 @xmath66 & & 2.3,2.5 & 2.1,2.4 & 2.4,2.7 & 2.0,2.3 & & &
|
we have found a new evolutionary path to type ia supernovae ( sne ia ) which has been overlooked in previous work . in this scenario , a carbon - oxygen white dwarf ( c+o wd ) is originated , not from an asymptotic giant branch star with a c+o core , but from a red - giant star with a helium core of @xmath0 .
the helium star , which is formed after the first common envelope evolution , evolves to form a c+o wd of @xmath1 with transferring a part of the helium envelope onto the secondary main - sequence star .
this new evolutionary path , together with the optically thick wind from mass - accreting white dwarf , provides a much wider channel to sne ia than previous scenarios .
a part of the progenitor systems are identified as the luminous supersoft x - ray sources or the recurrent novae like u sco , which are characterized by the accretion of helium - rich matter .
the white dwarf accretes hydrogen - rich , helium - enhanced matter from a lobe - filling , slightly evolved companion at a critical rate and blows excess matter in the wind .
the white dwarf grows in mass to the chandrasekhar mass limit and explodes as an sn ia .
a theoretical estimate indicates that this channel contributes a considerable part of the inferred rate of sne ia in our galaxy ,
i.e. , the rate is about ten times larger than the previous theoretical estimates for white dwarfs with slightly evolved companions .
|
although we have yet to obtain a microscopic solution of the kondo lattice problem that serves as a model for heavy electron materials , thanks to a recent phenomenological analysis @xcite of decades of experimental efforts on heavy electron behavior , we now have an experiment - based model for their behavior that explains existing experiments and has predictive power in the exploration of new families of heavy electron materials .
our model provides a unified explanation of the emergence of the heavy electron kondo liquid that is responsible for the anomalies seen in the knight shift , magneto - transport , and other experiments , leads to a new phase diagram that replaces the doniach picture , and has the potential for becoming the standard model of heavy electron physics .
it contains several basic aspects : * there exists a unique temperature scale @xmath0 for each heavy electron material ; it marks the emergence of a new state of matter , a coherent heavy electron ( kondo ) liquid that is produced by the collective hybridization of the local moments with the conduction electron sea , and exhibits universal behavior that scales with @xmath0@xcite .
* @xmath0 is the nearest neighbor rkky ( ruderman - kittel - kasuya - yosida ) local moment interaction ; for the local moments it separates their high-@xmath4 weakly interacting phase from their low-@xmath4 entangled phase .
it is given by @xmath5@xmath6 , where @xmath2 is the single - ion kondo coupling to the background conduction electron density of states @xmath3 @xcite .
* the entangled phase below @xmath0 is described by a two - fluid model @xcite in which an order parameter , @xmath7 , characterizes the reduction of the overall system entropy brought about by the transfer of spectral weight from the f - electron local moments to the itinerant heavy electron kondo liquid formed out of the conduction electron sea . * the quasiparticles of the kondo liquid exhibit coherent non - landau fermi liquid behavior ; their spin susceptibility and specific heat vary as @xmath8 down to a crossover temperature @xmath9 .
* their behavior below @xmath9 can lead to unconventional superconductivity produced by quantum critical fluctuations @xcite , or an antiferromagntic ordering of the local moments in which kondo liquid relocalization occurs as a precursor @xcite .
the position of a number of heavy electron materials on our proposed new phase diagram is given in fig .
[ fig1 ] , where we see that @xmath100.15 roughly separates the antiferromagnets and superconductors . in striking contrast to the conventional doniach diagram which postulates a competition between collective behavior produced by rkky interaction and single - ion behavior characterized by the single - ion kondo temperature , @xmath11
, the latter quantity is found to be much smaller than @xmath0 for almost all the materials considered there . in fig .
[ fig2 ] we plot the variation of @xmath0 , @xmath9 , @xmath11 , and the ordering temperatures , @xmath12 and @xmath13 as a function of pressure for the heavy electron material cerhin@xmath14 .
there one can see how the competition between localization ( antiferromagnetic order ) and itinerancy ( superconductivity ) for kondo liquid quasiparticles plays out as @xmath15 increases .
new phase diagram of cerhin@xmath14 @xcite .
@xmath0 is from resistivity peak and hall anomaly .
@xmath9 is from hall anomaly @xcite .
@xmath11 is estimated from ce@xmath16la@xmath17rhin@xmath14 . ]
new phase diagram of cerhin@xmath14 @xcite .
@xmath0 is from resistivity peak and hall anomaly .
@xmath9 is from hall anomaly @xcite .
@xmath11 is estimated from ce@xmath16la@xmath17rhin@xmath14 . ]
the kondo coupling @xmath2 between a localized f - moment and conduction electron spins gives rise to two competing characteristic temperature scales in a kondo lattice : @xmath18 where @xmath3 , the density of states of conduction electrons , can be determined by the sommerfeld coefficient @xmath19 of the corresponding nonmagnetic compound ( e.g. , lacoin@xmath14 for cecoin@xmath14 ) , @xmath20 .
after a correction to account for change of chemical pressure upon doping , the value of @xmath2 for the kondo lattice material can be determined by measurements of @xmath11 in the diluted compound ( e.g. , ce@xmath16la@xmath17coin@xmath14 ) , where it is given by @xmath21 .
an analysis @xcite of dozens of heavy electron materials and the pressure data of cerhin@xmath14 showed that in all cases of interest , @xmath11 was quite small compared to @xmath0 , so that it was natural to inquire whether @xmath0 might be given by the rkky interaction , eq .
( [ rkky ] ) , with a near - universal value of @xmath22 . as may be seen in fig .
[ fig1 ] , this is indeed the case , with a best fit analysis yielding @xmath230.45 , so that @xmath24
although @xmath0 marks the onset of coherent collective hybridization , the process is not instantaneous ; the heavy electron kondo liquid emerges over a broad temperature range from @xmath0 to a much lower cut - off temperature , @xmath9 , and throughout this temperature region , one therefore expects two components : the emergent kondo non - landau fermi liquid composed of conduction electrons made heavy by their collective hybridization with the local moments and a local moment spin liquid formed by the substantial fraction of local moments that have not hybridized .
one is thus led to postulate a two - fluid model @xcite with an order parameter @xmath7 that denotes the fraction of the f - moments that participate in the collective hybridization ; both @xmath7 and the scaling properties of the kondo liquid can be determined directly from experiments displaying the various heavy electron anomalies .
consider , for example , knight shift measurements @xcite .
since the f - moments are fully localized above @xmath0 , the knight shift of a given probe nucleus measures directly their susceptibility . below @xmath0 , it probes both the local moment and kondo liquid contributions to the total magnetic susceptibility , @xmath25 , but since it couples differently to each , it will no longer be proportional to the total susceptibility , hence the appearance of the anomaly seen beginning at @xmath0 .
as shown in @xcite , the two - fluid model expressions are thus : @xmath26\chi_l,\\ k&=&k_0+af(t)\chi_h + b[1-f(t)]\chi_l,\end{aligned}\ ] ] where @xmath27 and @xmath28 are the intrinsic susceptibility of the heavy electrons and the unhybridized local moments , respectively . @xmath29 and @xmath30 are their hyperfine couplings .
since the knight shift anomaly below @xmath0 is given by , @xmath31 we see that it acts as a direct probe of the emergent kondo liquid susceptibility .
a careful analysis of the knight shift and other experiments on cecoin@xmath14 shows that @xmath32 while if one requires that the intrinsic entropy of the kondo liquid approaches the free moment entropy @xmath332 at @xmath0 , @xmath34 where @xmath35 is the wilson ratio .
knight shift measurements on other heavy electron materials and fits to the anomalous behavior seen in magneto - transport , point contact spectroscopy , and raman spectroscopy show that these expressions are universal @xcite .
a comparison to recent state - of - the - art numerical dmft calculations @xcite shows good agreement with the kondo liquid quasi - particle density of states determined from eq .
( [ chih ] ) , and demonstrates that that these contain the feedback needed to achieve collective hybridization .
the low temperature cut - off at @xmath9 in the measured scaling behavior of the kondo liquid marks the onset of new physical processes that take precedence over its growth due to collective hybridization .
the two - fluid model continues to apply in this region , and provides important information about both the unconventional superconducting states and the onset of antiferromagnetism observed in heavy electron materials .
knight shift measurements are especially informative , as these provide direct information on changes in the kondo liquid as the material becomes superconducting or antiferromagnetic .
an analysis of cecoin@xmath14 confirms the universal behavior of the kondo liquid @xcite : it follows two - fluid prediction above @xmath9 ( @xmath36 ) and , upon becoming superconducting , follows the bcs d - wave prediction below @xmath13 , independent of the probe nuclei .
important information on the pairing mechanism for the measured d - wave superconductivity can be obtained from the nqr spin - lattice relaxation rate @xmath37 . by using the two - fluid formula @xmath38 where the local moment @xmath39 is given by the high temperature fit .
a simple two - fluid analysis reveals quantum critical fluctuations in the kondo liquid , and yields @xcite @xmath40 where @xmath41 is the distance from the quantum critical point .
a similar two - fluid analysis applied to heavy electron antiferromagnets cerhin@xmath14 and cept@xmath42in@xmath43 has recently led to the discovery that at a temperature @xmath9 well above the nel temperature , the kondo liquid begins to relocalize , with a corresponding decrease in @xmath44 , confirming the important role played by the kondo liquid in bringing about the antiferromagnetic order of both it and the residual local moments present at @xmath12 .
by analogy to the idea of a standard model based on a well - defined microscopic theory , we have proposed here an experiment - based standard phenomenological model for heavy electron materials that derives from the successful systematic examination of existing experiments on a broad spectrum of heavy electron materials .
its predictions for heavy electron emergence at @xmath0 accompanied by anomalous scaling behavior can easily be checked for any newly - discovered heavy electron family , and have recently been confirmed by experiments on the newly - discovered 127 family . of particular interest are the correlations in our model between measured values of @xmath15 and the appearance of antiferromagnetism and superconductivity at low temperatures .
as we have noted , materials with @xmath45 all appear to be antiferromagnets , but can easily be made into superconductors by applying pressure to increase @xmath15 , as fig .
[ fig2 ] demonstrates vividly for the rh-115 family .
moreover , since the superconducting transition temperature seems to scale with @xmath0 , a strategy for finding a higher @xmath13 family of heavy electron materials might begin by searching for materials with large values of @xmath0 , a strategy that proved successful in the discovery of the puga members of the 115 family .
we note this can be achieved either by finding materials with values of @xmath15 that significantly exceed 0.15 , or , with @xmath46 , finding materials that possess a comparatively low value of @xmath3 , or equivalently a large conduction electron bandwidth , and a correspondingly high value of @xmath2 .
moreover , to the extent that @xmath13 scales with @xmath0 , for a fixed value of @xmath15 , one would have @xmath13 scaling with @xmath2 or the conduction electron bandwidth @xmath47 .
thanks h .- o . lee and t. park for experimental data of cerhin@xmath14 .
z.f . acknowledges support through nsf grant no .
nsf - dmr-0801253 .
work at los alamos was performed under the auspices of the us department of energy through the ldrd program .
9 yang y - f and pines d 2008 _ phys . rev .
lett . _ * 100 * 096404 yang y - f , fisk z , lee h - o , thompson j d and pines d 2008 _ nature _ * 454 * 611 yang y - f 2009 _ phys . rev . b _ * 79 * 241107(r ) yang y - f , urbano r r , curro n j , pines d and bauer e d 2009 _ phys .
lett . _ * 103 * 197004 nakatsuji s , pines d and fisk z 2004 _ phys .
lett . _ * 92 * 016401 curro n j ,
young b - l , schmalian j and pines d 2004 _ phys .
b _ * 70 * 235117 aproberts - warren n , dioguardi a p , shockley a c , lin c h , crocker j , klavins p , pines d , yang y - f and curro n j , unpublished .
park t and thompson j d 2009 _ new j. phys . _ * 11 * 055062 nakajima y , shishido h , nakai h , shibauchi t , behnia k , izawa k , hedo m , uwatoko y , matsumoto t , settai r , onuki y , kontani h and matsuda y 2007 _ j. phys .
jpn . _ * 76 * 024703 shim j h , haule k and kotliar g 2007 _ science _ * 318 * 1615
|
we propose a predictive standard model for heavy electron systems based on a detailed phenomenological two - fluid description of existing experimental data .
it leads to a new phase diagram that replaces the doniach picture , describes the emergent anomalous scaling behavior of the heavy electron ( kondo ) liquid measured below the lattice coherence temperature , @xmath0 , seen by many different experimental probes , that marks the onset of collective hybridization , and enables one to obtain important information on quantum criticality and the superconducting / antiferromagnetic states at low temperatures . because @xmath0 is @xmath1 , the nearest neighbor rkky interaction , a knowledge of the single - ion kondo coupling , @xmath2 , to the background conduction electron density of states , @xmath3 , makes it possible to predict kondo liquid behavior , and to estimate its maximum superconducting transition temperature in both existing and newly discovered heavy electron families .
|
the collapse of a wave function is an undeniable feature of individual quantum mechanical systems .
however , the collapse of a state along a @xmath0 = constant surface of an arbitrary coordinate system is unbelievable , inasmuch as nature does not recognize a surface that is so obviously constructed by humans . for this reason , a foundation theory must provide for the collapse of a wave along an invariant surface that is independent of coordinate representations .
the collapse of a wave is not just invariant , it is trans - representational ; that is , it is independent of any choice of basis states .
furthermore , it will collapse along the surface of the backward time cone that is here called a _ conic _ surface .
this is the hellwig and kraus state reduction @xcite that has been widely dismissed as being causally problematic @xcite .
the supposed paradox can not be resolved by relativity and quantum mechanics alone because a foundation theory is also required .
this will govern the collapse and impose constraints that establish an unambiguous temporal and causal order between reductions .
when that is done , it is shown below that a ` later ' collapse can not causally penetrate the backward time cone of a ` former ' collapse . and
since a collapse occurs against a background of countless widely disbursed prior collapses , the spatial - temporal extent of any collapse in minkowski space is limited .
one consequence of a hellwig - kraus reduction is that the newly collapsed state must also be the initial state of the next phase of evolution .
so initial states as well as collapsed states are conically defined .
therefore the most natural quantum mechanical state is a function mapped onto the conic surface of the backward time cone of an event in space - time .
it is shown that the dynamic principle projects these functions forward to successor conic states whose vertices lie along a specified world line .
claims regarding a hellwig - kraus causal ambiguity are usually advanced by referring to non - local correlations . consider a pair of particles @xmath1 and @xmath2 that are created by a decay to become correlated in the spin zero state .
@xmath3 the first particle moves to the left in fig . 1 and the second particle moves to the right .
initially they both have an uncertain spin direction as indicated by the double vertical arrow along each path in that figure .
imagine that the first particle is measured to have spin - up at an event * a * in fig . 1
, causing a hellwig - kraus state reduction . in that case
@xmath2 will go spin - down when it intercepts the backward light cone of event * a * at event * b*. if @xmath2 is measured later at event * b * it will of course record spin - down .
the result is the state @xmath4 .
capital bold face letters indicate events that ` cause ' state reduction like measurements ; and lower case bold face letters identify events that are not vertices of reduction sites .
event * b * also results in a state reduction .
the effect of * b * on @xmath1 might conceivably be the same as the effect of * a * on @xmath2 , where @xmath1 goes spin - up the moment it intercepts the backward light cone of event * b * at event * a * as shown in .
so when @xmath1 is later measured at * a * it will record spin - up .
the final results are then the same as those shown in , but this possibility leads to an odd circularity that is characteristic of a causal loop : * a * causes * b * causes * b * , and * b * causes * a * causes * a*. these two particles are shown to follow straight lines even though they are waves spreading out over the indicated paths .
this does not matter because the rules developed here are not concerned with paths .
we are concerned only with the location of reduction events like * a * and * b * and their associated conic states .
the straight lines in these figures may be thought of as having heuristic value only .
the _ qrules _ are three rules given in another paper that govern the collapse of a wave function @xcite . these rules generate _ qrule equations _ that are of the form @xmath5 where @xmath0 is _ conic time _ as opposed to conventional _
lorentz time_. the lorentz time at an event * x * is defined over a horizontal plane that passes through * x * , whereas conic time is defined over the conic surface with its vertex at * x*. it is shown in appendix i how a conventional lorentz wave function @xmath6 with its origin at * x * can be mapped into a conic wave function @xmath7 with its vertex at * x * , where * r * represents a variable that is defined ( in appendix i ) on the conic surface in 3 + 1 space .
we say that the conic wave function @xmath7 mirrors the lorentz wave function @xmath6 in a flat minkowki space .
the qrule component @xmath8 is then given by the square modulus of @xmath7 with its * r * variable integrated out , making a qrule component a function of @xmath0 only .
it is also shown in that a conic dynamic principle that mirrors the standard dynamic principle will project the conic function @xmath7 into successor vertices along the indicated world line .
a qrule equation is always given by a capital @xmath9 as a function of the components @xmath8 or @xmath10 may be expressed differently .
for instance , the component @xmath8 consisting of an atom @xmath11 and a molecule @xmath12 can be written .
each component in a qrule equation is ` complete ' in that it implicitly contains all the particles in the universe ( ref.3 ) .
the plus sign in a qrule equation always indicates a _ discontinuous _ quantum jump that is also _
the particle pair @xmath1 and @xmath2 in eq . 1 is assumed to be created at an event * 0 * that results from of a decay that begins at a prior conic time @xmath13 along some preferred world line . if the pair is created by the decay of a composite particle @xmath14 , the corresponding qrule equation is given by @xmath15 where the @xmath16 s in this equation are the _ qrule values _ of the designated particles .
they represent the square modulus of the conic function of each particle with its independent variables integrated out . in particular , the component @xmath17 is the qrule value of the zero spin wave function @xmath18 in eq . 1 .
the fact that the first particle goes to the left ( in fig . 1 ) and
the second particle goes to the right is a distinction that is lost to this component because it has been integrated out along with the spin distinction .
in addition , the component @xmath19 in eq . 2 is the qrule value of the composite particle that produces the decay particles .
both of these are trans - representational components that are derived from wave functions on conic surfaces in minkowski space , where @xmath0 is the conic time referring to these surfaces .
their evolution is governed by the dynamic principle operating on the underlying wave function .
each component is multiplied by an environmental term representing the rest of the universe , thereby satisfying the requirement that it is ` complete ' .
the choice of @xmath0 in these equations does not affect the outcome as will be demonstrated . the second component in eq . 2
is underlined , meaning that it is a _
ready _ component .
ready components are not empirically real , so the component @xmath17 in eq
. 2 refers to particles that do not yet exist at time @xmath0 in that equation .
the non - underlined component @xmath19 is called a _ realized _ component .
it does exist at time @xmath0 in that equation .
the second component @xmath17 is zero at time @xmath13 and increases in time as probability current flows to it from the first component @xmath19 in eq .
the gap between them represented by the + sign is the decay interaction that is discontinuous and irreversible .
the qrules tell us that the probability current gives the probability per unit time that the second component will experience a stochastic hit .
only ready components can be stochastically chosen according to the qrules .
if a hit occurs at a time @xmath20 ( the time of event * 0 * ) , then the rules tell us that there will be a collapse of to @xmath21 where the first component in eq .
2 goes to zero . the wave function associated with eq .
3 is also mapped onto backward time - cone surfaces in this case on vertices of the successors of event * 0 * , where the time variable is the conic time referring to these vertices .
equations 2 and 3 reflect the fact that the collapse of a wave function in quantum mechanics is a trans - representational affair .
the dynamic principle directs the evolution of a qrule equation , but the qrules govern the stochastic process that interrupts its continuous flow .
the foundation theory of ghirardi et .
al . incorporates a collapse directly into the dynamic principle @xcite , but that theory is not trans - representational .
when the spin - measuring devices @xmath22 and @xmath23 of particle @xmath1 and @xmath2 are later introduced , eq . 3 becomes the qrule equation @xmath24p_2(\downarrow)\otimes m_2(t ) \\ & + & \underline{p}_1(\uparrow)[p_2(\downarrow)m_2]\otimes m_1(t ) \nonumber \\ & + & [ \underline{p}_1(\downarrow)m_1]p_2(\uparrow)\otimes m_2(t ) \nonumber \\ & + & \underline{p}_1(\downarrow)[p_2(\uparrow)m_2]\otimes m_1(t ) \nonumber\end{aligned}\ ] ] where both measuring devices are on standby in the first component of eq . 4 , and the four ` ready ' components ( on the right ) are zero at @xmath20 . in the ready component of the first row
the first particle engages the spin - measuring device @xmath22 ( in square brackets ) , and in the second row the second particle engages @xmath23 .
the third and fourth rows are similar except that they provide for the reverse spin measurements .
probability current begins to flow from the first component to the ready components in the first and third rows when the first particle interacts with the measuring device @xmath22 sometime after @xmath20 .
current will begin to flow to the ready components in the second and fourth rows when the second particle interacts with @xmath23 .
so all four ready components are exposed to the possibility of a stochastic hit .
figure 3a is a graphic description of .
the gray area in that figure represents the minkowski region of interaction where the ready components in eq .
4 become non - zero ( i.e. , where the particle interacts with the detector ) .
neither one of the world lines in fig .
3a defines the time @xmath0 in eq .
that time specifies the chosen orientation ( i.e. , the chosen lorentz observer ) in the minkowski space . since the probability of any one of these ready states being stochastically chosen is independent of the choice of _ t _ , eq .
4 is invariant under _
t_. if another @xmath0 is chosen , then the world lines in fig .
3 will be rotated to the left or to the right ( as with a lorentz transformation ) , but the orientation of the conic surfaces will be unchanged . the graphic relationships in fig .
3 will be unchanged , except event simultaneity .
suppose the first row in eq .
4 is stochastically chosen at a time @xmath25 .
the system is then reduced to @xmath26p_2(\downarrow)\otimes m_2(t)\ ] ] where @xmath25 is the time of the stochastic hit at event * a * that appears along the world line of @xmath1 ( see fig . 3b representing eq . 5 ) .
it is again understood that the wave functions representing @xmath27 , and @xmath23 in this figure are spread out over the conic surface of the vertex of event * a * in fig .
3b . equation 5 not only represents event * a * , it also represents the cut - off event * b * that is simultaneous with * a * in conic time .
we make no assumption as to the world lines of the measuring devices .
further evolution carries the resulting conic state forward following event * a * until @xmath2 interacts with @xmath23 .
a new ready state then emerges giving @xmath28p_2({\downarrow})\otimes m_2(t ) \\ & + & [ \underline{p}_1({\uparrow})m_1][p_2({\downarrow})m_2](t ) \nonumber\end{aligned}\ ] ] which is graphically represented in fig .
3c in which the shaded area is the region of the new interaction that gives rise to the ready component in the second row of eq .
6 when @xmath2 encounters @xmath23 . the evolution of the second particle after event * b * in fig .
3c repeats part of the evolution that has already occurred in fig . 3a .
the latter part of this evolution involves the interaction @xmath29 $ ] between the second particle and its measuring device .
this part is not empirically significant because its repetition consists in its appearance in both ` ready ' components of both eqs .
4 . an interaction repetition like this would occur even if the time in eqs .
4 referred to lorentz time , so it is not a characteristic of a hellwig - kraus reduction .
it is a feature of _ any _ correlated state reduction .
however , the free particle part of the repetition certainly does involve the realized component in eqs . 4 ,
so it does ( in principle ) have empirical significance .
it is also characteristic of a hellwig - kraus collapse .
but this free particle part is not really measurable because it exists independent of a measurement interaction .
let a second state reduction occur at event * b * at time @xmath30 , giving the final result @xmath31[p_2(\downarrow)m_2](t)\ ] ] equation 7 appears in fig .
since the qrules are trans - representational they give us an objective account of what happens .
equation 4 causally precedes eq . 5 , and that causally precedes eq . 6 , and that causally precedes eq . 7 .
causal priority is established by the _ formal _
priority among these equations .
a collapse equation comes into existence only when its predecessor disappears .
we state the corollary .
* corollary : * _ the evolution of a qrule equation can not affect the evolution of a previously ` collapsed ' qrule equation since the latter no longer exists . _
this means that the evolution described in eq .
6 can have no influence on the evolution described in eq . 4 ;
for when eq .
6 is actively evolving , eq .
4 _ no longer exists_. therefore , event * b * can not influence event * a*. this is indicated in fig .
3d where the backward time cone of event * b * is not allowed to ` penetrate ' the backward time cone of event * a*. it is the qrules that govern the causal order of reduction , not relativity or standard quantum mechanics , and these rules are perfectly clear concerning priority .
this is shown graphically in the minkowski space of fig .
the more darkly shaded area ( below * a * in fig .
4 ) indicates the region of minkowski space that evolves according to eq . 4 ; and the more lightly shaded area ( below * b * ) indicates the region that evolves according to eq . 6 , where the lighter area can have no influence on the darker area because the latter is causally prior . the reductions in fig .
4 go to infinity on the left and on the right as expected of a hellwig - kraus collapse .
however , the event * b * collapse is limited in that it can not penetrate the event * a * reduction . this will generally be the case .
the universe is full of prior reductions that can not be penetrated by either events * a * or * b * ; and as a result , both reductions will be limited in the space - time extent of their influence .
this is not just true of correlated reductions , but will also be true of independent reductions ( see appendix ii ) .
following event * b * , the measuring device @xmath22 ( that now includes particle @xmath1 ) evolves _ independent _ of the measuring device @xmath23 ( that now includes @xmath2 ) .
these two systems can be described along two different conic world lines designated @xmath32 and @xmath33 , where @xmath32 begins with the vertex at event * a * , and a conic time @xmath33 begins with the vertex at event * b*. it is not necessary to introduce these dual times , but doing so helps to emphasize the spatial separation as well as the independence of the two systems .
the underlying conic function is then of the form @xmath34 which comes about by using two different origins ( on the horizontal @xmath35-axis ) to describe two widely separated systems . as a result ,
a single cone spreading over two systems of this kind is replaced by an envelope that drapes over both like the lightly shaded area over events * a * and * b * in fig.4 . the qrule equation following event * b * is then written @xmath36[p_2(\downarrow)m_1(t_2 ) + \underline{i}_2(t_2)]\ ] ] where @xmath37 is an interaction that is located in some part of the lightly shaded envelope above event * a * in fig .
4 , and @xmath38 is another interaction that is located in the lightly shaded envelope above event * b*. we simplify the equation by dropping the time dependence in each component @xmath39[p_2(\downarrow)m_1 + \underline{i}_2]\ ] ] where subscript-1 states are understood to be on the @xmath32 conic surface , and subscript-2 states are understood to be on the @xmath33 conic surface .
equation 8 might not seem entirely correct according to the qrules .
the recipient state @xmath40 in each bracket of the equation does not appear to be a complete component ; and if that were true , it would not be a ready component .
however , the equation can be written in the form @xmath41 + \underline{i}_1[p_2(\downarrow)m_2 + \underline{i}_2]\ ] ] the brackets in this equation have a square modulus that is constant in time , so the current from the first component comes solely from @xmath42 and goes exclusively to @xmath43 .
since both components are clearly complete , the first must be a realized component and the second is a ready component . both _
i _ s in eq .
8 can be understood in this way .
we therefore keep the form of eq .
it can be generalized to any number of independent systems in a way that satisfies the qrules . in 2 + 1 space the mountaintops in fig . 4
do not appear in front or in back of one another .
rather , they are superimposed on one another like mountains on a flat terrain .
a new peak will not penetrate an old peak on that terrain because the above corollary holds for all completed state reductions including independent as well as correlated reductions as will be shown in appendix ii .
these mountaintops all build on top of one another without penetration .
it is important to understand how eq . 4 and fig . 3
play out relative to a lorentz frame in which event * a * occurs ( causally ) before event * b * but appears to follow in lorentz time .
figure 5a is identical with fig.3a where the regions of measurement interaction for @xmath1 and @xmath2 are indicated by the darkened world lines .
figure 5b looks at the same interactions relative to a lorentz frame in which the second particle is at rest .
this is called the ` primed ' frame .
equation 4 has the same form relative to the primed frame as it does relative to the unprimed frame .
the difference is only in the way that current flows from the realized component to the four ready components . in the unprimed frame ( fig .
5a ) current flows simultaneously into all four components for as long as it takes for a stochastic hit on the first component ( in our example ) . in the primed frame ( fig .
5b ) current begins to flow into the second particle _ before _
it begins to flow into the first particle in lorentz time .
however , this current will run out without the second particle being stochastically chosen , while current continues to flow into the first particle . as a result ,
the first particle is the first to be stochastically chosen at event * a * ( in fig .
5c ) , thereby cutting off the second particle at , which is again _ simultaneous _ with event * a * in conic time .
the second particle is subsequently revived at a time that allows it to again interact with the measuring device @xmath23 , thereby allowing it to be stochastically chosen at event * b * ( fig .
5d ) _ after _ event * a * in conic time
although it is _ before _ in lorentz time .
the second particle does not exactly repeat its prior history .
the first time that it interacts with the measuring device it is not stochastically chosen , but the second time ( following event * b * ) it is stochastically chosen . the first time the probability of a hit on event * b * is 0.25 , the second time it is 1.0 .
this repetition has no empirical significance because both interactions appear inside ready components . in general ,
stochastic interactions are not empirically real in qrule theory .
stochastic hits are empirically real . figures 3 and 5 represent two different lorentz observers , where the causal and the temporal order of events are the same in both frames . only in lorentz time
does there appear to be some disagreement between the causal and temporal orders .
we attribute this to fact that lorentz time is given along artificiality defined horizontal planar surfaces .
the structure in figs . 3 and 5 should be the same for any foundational theory that is viewed in an invariant way .
if the theory in question provides for the collapse of a wave like eq . 1 , and if the collapse travels backward over a conic surface in a way that destroys the possibility of any further influence on itself , then it should produce the same minkowki architecture that appears in .
for any such theory , in any lorentz frame , the measurement interaction of the second particle will run its course without result ( in our example ) allowing to be chosen first .
the resulting collapse will re - start the second particle at an event like * b * in fig .
3b that is conically _ simultaneous _ with * a * , so event * b * will occur _ after _ event * a * in conic time . as in our qrule analysis
, it is the reduction theory that governs the casual order , not relativity or standard quantum mechanics .
the result will be the succession of mountain peaks like those shown in fig . 4 , where a peak in background occurs causally and temporally _ after _ a peak in foreground , and where both are limited in space and time by other reductions .
a paradoxical causal loop seems to appear when the collapse of a quantum mechanical wave function is viewed as a hellwig - kraus reduction in minkowski space .
this paradox can not be removed by special relativity and/or standard quantum mechanics alone , but requires the constraints imposed by a suitable foundation theory governing the collapse . in this paper
the qrule foundation theory is used to resolve this causal difficulty .
we define conic time @xmath0 over the surface of a backward time cone of some event * x*. a quantum mechanical wave function @xmath44 is also mapped onto that surface .
this function is used instead of the usual quantum mechanical wave function @xmath45 that is defined on a horizontal surface going through * x*. a hellwig - kraus reduction describes two correlated space - like reduction ev - ents * a * and * b * , and it is found that the qrules establish a definite causal priority between them .
there is no _ causal loop _ or temporal ambiguity between these two events when the qrules govern the collapse .
the causal and conic temporal orders of events * a * and * b * are the _ same_. the resulting landscape in minkowski space is shown in fig . 4 , where each mountaintop in that figure is a separate reduction .
the peak in the background of fig . 4 will occur causally and temporally _ after _ the peak in the foreground .
the background reduction can not ` penetrate ' or causally influence the interior of the foreground reduction , as shown in the figure .
any reduction will take place against a background of countless prior reductions that can not be penetrated , thereby limiting its range in space and time .
this will be true even if the reduction is independent ( i.e. , not a correlation ) as is shown in appendix ii .
it is my belief that the structure of fig
. 4 will be similar for any foundation theory that is viewed in an invariant way , so long as the theory requires that a collapse destroys the possibility of any further influence on itself as do the qrules .
the dynamic principle is understood to influence the temporal development of a qrule component through the underlying function @xmath46 .
this function is defined in this paper and evolves in time as described in appendix i. however , the qrules govern state reduction and therefore interrupt the dynamic process under the right conditions .
more will be said in a subsequent paper about the relationship between the dynamic principle , the qrule equations , and the architecture of state reduction in minkowski space .
a particle is initially located at @xmath20 between @xmath47 and @xmath48 in fig .
. its wave function @xmath49 advances along the @xmath50-axis occupying the shaded area in flat minkowski space .
this function is mapped onto the dark line on the surface of the conic section with its vertex at @xmath20 . each event on that surface
has assigned coordinates @xmath51 ( perpendicular to the @xmath0-axis ) and @xmath20 , so we project @xmath52 onto the conic surface giving @xmath53 where @xmath54 holds for all @xmath35 and @xmath51 . if @xmath47 is negative , this function can be extended over the vertex to the rising side of the conic surface without difficultly .
more generally we say that any wave function in flat space that is mapped onto a conic section with its vertex at @xmath0 is given by @xmath46 .
we want to know how to write the dynamic principle for @xmath46 .
assume that it is analogous to the schrdinger equation in the non - relativistic case , with a dynamic operator @xmath55 in place of the hamiltonian .
@xmath56 this equation is correct if we let @xmath57 the function that is mapped onto successive conic surfaces is therefore propagated in time by a dynamic principle that mirrors the schrdinger equation for a non - relativistic system .
equations 9 and 10 applied to @xmath46 will advance it from one conic surface to another .
the same construction can be applied to each of the four components of a dirac wave function , in which case the dynamic operator @xmath55 mirrors the dirac hamiltonian .
the amplitude in the above equations then has the required four components . in 2 + 1 space
the surface is a cone with a vertex time @xmath0 .
a function @xmath45 defined on a horizontal plane with its origin at the vertex is projected onto the cone s surface .
its @xmath58 coordinates are mapped onto @xmath59 coordinates that intercept the conic surface and are perpendicular to the @xmath0-axis .
the dynamic principle causes these surfaces to evolve along the specified world line . in 3 + 1
space the incoming ` spherical ' conic surface will converge on a vertex time @xmath0 .
the particle volume that is simultaneous with @xmath0 contains the initial condition @xmath60 , and each of these values can be mapped onto the incoming spherical surface when ( prior to @xmath0 ) the surface passes through each particle part at its time @xmath0 locaton .
our claim is that the @xmath35-axis is superfluous , and that the only coordinates we need are those associated with a conic surface . if given the wave function @xmath46 of a particle along the surface with vertex @xmath20 in fig . 7
, the dynamic principle will carry it into all succeeding surfaces like that of @xmath32 in that figure .
the ` real ' location of the particle in the shaded area of fig . 6 or in the shaded area of fig .
7 is our choice .
if we say the particle is initially specified at time @xmath20 , then it will occupy the shaded area of fig .
6 when we understand ` simultaneous ' to be the horizontal surface through @xmath20 .
but it will occupy the shaded area of fig .
7 when we understand ` simultaneous ' to be the conic surface with @xmath20 at its vertex .
it is possible to draw a minkowski diagram using one vertex time @xmath0 , with all the particles in the system mapped onto the surface of its backward time cone .
we have seen that it is also possible to use a diagram with two or more vertices as will be done in the present case of two independent particles .
consider a nucleus @xmath61 with a vertex @xmath32 as shown in the diagram of , together with a second nucleus @xmath62 with a vertex @xmath33 .
both nuclei are radioactive , where the first decays at event * a * in fig .
multiple time descriptions like this are possible only when the vertex events so identified have a space - like relationship to each other . in this case
the vertices are chosen to _ follow _ the separate particles , since there is no reason why the time scale must be the same as that of the minkowski observer . the total wave function in fig .
8a will be spread over both peaks .
it will be mapped onto the dashed lines joined at their intersection as shown in the figure .
the qrule equation follows eq . 8 for these independent nuclei .
@xmath63 [ n_2 + \underline{p}_2p_2']\ ] ] where @xmath61 and @xmath64 are specified relative to the @xmath32 vertex , and @xmath62 and @xmath65 are specified relative to the @xmath33 vertex .
the component @xmath64 represents the decay products of @xmath61 , and @xmath65 represents the decay products of @xmath62 .
the two ready components are initially zero .
probability current will flow from both @xmath61 and @xmath62 in this equation to the corresponding ready components , marking both candidates for a possible stochastic choice .
the first ready component is stochastically chosen at event * a * , after which the qrule equation is @xmath66 as shown in fig .
8b . at event
* a * the first nucleus is reduced to its decay products , although they do not yet appear in fig .
the second nucleus in eq .
11 continues unaffected in its unstable state , and is represented by the line that extends beyond the shaded area . that line is the same length in fig .
8b as it is in because it is not reduced by the first decay
. further evolution of the system is described by same equation @xmath67 except that the magnitude of the component @xmath68 increases in time as represented by the longer line in fig .
9a . the decay particles @xmath1 and @xmath2 are also shown in fig .
they are actually waves spreading out in all directions from the vertex , but they are shown as two distinct world lines as are all other particle paths in these diagrams .
the qrules do not specify paths , only reduction events and their associated states . with a stochastic hit on @xmath68 in eq .
12 , the reduction at event * b * will yield @xmath69 this equation is shown in the minkowski diagram of fig . 9b , where the decay products @xmath70 do not yet appear .
as before , the lighter shaded reduction under event * b * in fig . 9b does not penetrate the darker shaded reduction defined by event * a*. this is because the * b * reduction in eq .
13 occurs causally _
after _ the * a * reduction in eq . 11 ,
so it can have no influence on the evolution leading to eq .
it was shown in the text that correlated reductions are causally sequenced so that a later reduction will not penetrate a former reduction .
we see here that the same is true when the reductions are independent of one another .
a distinct causal order is therefore characteristic of _ all _ reductions .
as before , the reduction in may appear to go back infinitely far in time and extend infinitely far in space , but that will not happen . every reduction will find a floor of other mountaintops that will support it and limit its range in space and time .
|
it is claimed in another paper that the collapse of a quantum mechanical wave function is more than invariant , it is trans - representational . it must occur along a fully invariant surface .
the obvious surface available for this purpose is that of the backward time cone of the collapse event as proposed by hellwig and kraus . this collapse is widely believed to result in paradoxical causal loops that can not be removed by special relativistic and/or standard quantum mechanical considerations alone .
however , the paradox is resolved when we apply the qrule foundation theory that is developed in the other paper .
the causal and temporal orders of state reduction are then found to be in agreement with one another , and the resulting boundaries in minkowski space are shown to have a novel architecture that limits the range of a hellwig - kraus reduction in space and time . although these boundaries have been worked out using the qrules , they should be the same for any foundation theory that treats the collapse of a wave in an invariant way , and requires that a collapse destroys the possibility of any further influence on itself as do the qrules .
keywords : measurement , state reduction , wave collapse .
|
recently , there has been an increasing interest in using magnetic beads in separation of , say , biochemical species in microfluidic systems @xcite . the principle is to have biochemically functionalized polymer beads with inclusions of superparamagnetic nanometersize particles of , for example , magnetite or maghemite .
they attach to particular biochemical species and can be separated out from solution by applying external magnetic fields . as most biological material is either diamagnetic or weakly paramagnetic , this separation is specific .
paramagnetic particles in fluids are also used to measure the susceptibility of , for example , magnetically labelled cells by measuring particle capture or motion in a known field @xcite . in this paper
we study microfluidic capturing of paramagnetic beads from suspension by modeling the beads as a continuous distribution @xcite .
the separation of suspended paramagnetic beads from their host fluid is an important process as it decides operating characteristics for practical microfluidic devices .
it involves an interplay between forces of several kinds governing the dynamics of the process : ( a ) magnetic forces from the application of strong magnetic fields and field gradients .
( b ) drag forces due to the motion of the beads with respect to the host fluid .
( c ) the trivial effect of gravity , which we ignore in the following .
we emphasize the effects of bead motion on the fluid flow as this gives rise to a hydrodynamic interaction between the beads .
as we have noted in a previous few - bead study , this interaction is more important than the magnetic bead - bead interactions @xcite .
it is created by drag forces in two steps : first , drag transfers momentum to the fluid from the beads moving under the influence of external forces .
second , the modified flow changes the drag on and thus motion of other beads .
sketch of the microfluidic system with @xmath0 m and @xmath1 m .
a suspension of paramagnetic beads enters at @xmath2 with a parabolic poiseuille flow profile , @xmath3 , and leaves at @xmath4 .
beads are caught by the pair of wires placed @xmath5 m from the outlet at the top and carrying currents @xmath6 .
( b)(d ) simulated stationary density of the beads ranging from zero ( white ) to @xmath7 ( black ) for increasing values of the current - distance product @xmath8 as indicated . at @xmath2
the concentration is @xmath9 m@xmath10 and the maximum flow speed is @xmath11 m/s . ]
as sketched in fig . [ fig : geometryconc](a ) we consider a viscous fluid ( water ) flowing in the @xmath12 direction between a pair of parallel , infinite , planar walls .
the walls are placed parallel to the @xmath13 plane at @xmath14 and @xmath15 , respectively .
a steep magnetic field gradient is generated by a parallel pair of closely spaced , infinitely long , thin wires along the @xmath16 direction separated by @xmath17 and carrying opposite currents @xmath6 . the system is translation invariant in the @xmath16 direction thereby reducing the simulation to a tractable problem in 2d .
the simulation domain is defined by @xmath18 and @xmath19 with @xmath20 m and @xmath21 m .
the wires intersect the @xmath22 plane near @xmath23 just above the top plate .
paramagnetic beads in suspension are injected into the microfluidic channel by the fluid flow at @xmath2 .
they are either exiting the channel at @xmath4 or getting collected at the channel wall near the wires .
when a suspension of beads is viewed on a sufficiently large scale compared to the single bead radius @xmath24 but on a scale comparable to density variations , we can describe the distribution of beads in terms of a continuous , spatially varying bead number density @xmath25 .
we consider a suspension of beads with radius @xmath26 m and denote the initial number density at @xmath2 by @xmath27 .
the four basic constituents of the model are described in the following .
_ magnetic force .
_ the beads are paramagnetic with a magnetic susceptibility @xmath28 . in an external magnetic field @xmath29 the force on such a bead is @xmath30 assuming that the bead is so small that we can take the external field @xmath31 to be approximately constant across the bead , i.e. , @xmath32 when determining the magnetization @xmath33 .
as mentioned , @xmath31 in this study arises from a pair of current carrying wires .
it is determined in the following manner . from ampre s law
, we readily find the magnetic field , @xmath34 , around a straight circular wire , @xmath35 , where the electrical current vector @xmath36 is along the wire orthogonal to the position vector @xmath37 which is in the @xmath22 plane .
the magnetic field from the two closely spaced anti - parallel wires is found by decreasing the separation @xmath17 and increasing the current , @xmath38 , while keeping the product @xmath39 constant , @xmath40 this together with eq .
( [ eq : magforce ] ) yields @xmath41 a manifestly attractive central force ( from the mid - point of the wires ) , independent of the direction of @xmath42 .
_ fluid flow .
_ [ sec : flow ] the beads are suspended in a fluid of viscosity @xmath43 and density @xmath44 that is launched at @xmath2 with a parabolic velocity profile , @xmath3 , and flows past the wires . in microfluidics
inertial effects are unimportant compared to drag , so the small beads in suspension almost always move with constant velocity relative to the fluid . except for acceleration phases shorter than microseconds the external forces
are exactly balanced by drag @xcite . the momentum transfer from beads to fluid
is included by adding a bulk force term , @xmath45 to the navier stokes equation .
this bulk force term is proportional to the number density @xmath25 of beads and the magnetic external force @xmath46 on an individual bead at position @xmath37 .
the velocity @xmath47 of the fluid is given by @xmath48 along with the incompressibility condition @xmath49 .
_ bead motion . _
[ sec : beads ] to complete the set of equations , it is necessary to have an equation of motion for the bead number density @xmath25 .
as the beads neither appear nor disappear in the bulk , @xmath25 must obey a continuity equation @xmath50 where the bead current * j * is defined by the nernst
planck equation @xcite @xmath51 with diffusivity @xmath52 and bead mobility @xmath53 . for our spherical beads
the diffusivity is given by the einstein expression @xmath54 which for water at room temperature equals @xmath55m@xmath56/s .
in the simulations below , however , we artificially increase the magnitude of @xmath52 in order to stabilize the computations and to use a coarser mesh and thus save computation time .
the fraction @xmath57 of beads caught as function of the current - distance product @xmath8 for twenty different flow speeds ( 50 1000 @xmath58 m/s ; indicated by the arrows ) .
larger current leads to higher @xmath57 ; faster flow to smaller @xmath57 . in this simulation
the initial concentration is low , @xmath59 m@xmath10 .
_ inset : _ rate @xmath60 of bead capture as function of @xmath8 , for the flows above .
the faster flow or the larger current , the higher @xmath60 . ] _ boundary conditions .
_ in addition to the bulk equations ( [ eq : ns ] ) , ( [ eq : particlecontinuity ] ) , and ( [ eq : np ] ) , we need appropriate boundary conditions . as the beads move out to the walls of the domain and settle there , merely demanding that the normal component of the bead current vector @xmath61 vanishes is not correct , rather , it must be free to take on any value as long as it is directed into the wall .
as beads do not enter the bulk from the walls ( by assumption once settled , beads stick ) we demand that the normal current component is never directed into the liquid . for the fluid we demand the usual no - slip condition at the walls . at the inlet @xmath2 of the microfluidic channel
we assume that the fluid comes in with the constant initial number density @xmath27 and with a parabolic fluid velocity profile @xmath3 . at the outlet @xmath4 we
let the bead current take on any value , while the fluid pressure is zero .
having set up the equations for bead and fluid motion , they are solved with the finite element method on a mesh with @xmath62 elements refined in the vicinity of the wires . to this end
we employ the finite element solver software package femlab @xcite . the parameter values for the fluid
are those of water , @xmath63 mpa@xmath64s and @xmath65 kg / m@xmath66 , while for the beads @xmath26 m and @xmath67 to @xmath68 m@xmath10 . to study capturing
we must keep track of which beads are captured and which are flushed through the channel with the flow .
this is done by calculating the rates @xmath69 by which the beads are either captured or transported in / out at each of the four boundary segments @xmath70 of the channel ( inlet , outlet , upper wall , and lower wall ) . by integration of the normal components of the bead currents along each segment @xmath70 ,
we find @xmath71 the rate of capture is @xmath72 . in steady state
the conservation of beads enforces @xmath73 , which provides a useful check of the simulation results .
the primary control parameters are the current - wire distance product @xmath8 , the maximum fluid in - flow speed @xmath74 , and the bead number density , @xmath27 .
the product @xmath8 decides the magnetic force which captures the beads against the fluid flow . as we are investigating effects of bead - bead interaction ,
our interest is properties that depend on the bead number density , in particular those that do so nonlinearly .
the fraction @xmath57 of beads caught versus @xmath75 , the ratio of the current - distance product squared and the fluid flow velocity .
this demonstrates scaling in the competition between capturing and flushing : the twenty curves from fig .
[ fig : betagamma ] approximately collapse to one single . ] _ electrical current and fluid flow . _
the effects of having electrical wires near , and thus a magnetic field gradient in , the channel is illustrated in fig . [
fig : geometryconc](b)(d ) for three values of the current - distance product @xmath8 . at small values of @xmath8 only a narrow region
is emptied but increasing the current the region expands until it covers the width of the channel .
a simple measure of the capturing is the ratio @xmath57 of the bead capture rate @xmath60 to the bead in - flow rate @xmath76 , @xmath77 if capturing dominates @xmath57 tends to unity , if flushing dominates @xmath57 tends to zero .
[ fig : betagamma ] shows this in that slow flow and strong current leads to a high @xmath57 whereas fast flow and weak current leads to a small value .
the rate @xmath60 of bead capture as function of wire current and flow velocity is illustrated in the inset of fig . [
fig : betagamma ] .
if there is a competition between magnetic capturing and flushing , then we expect that the data can be described essentially by the ratio of the magnetic forces to the inlet fluid flow speed @xmath74 .
the force is proportional to the square of the current - distance product @xmath8 . in fig .
[ fig : collapsed ] , we plot the data from fig . [
fig : betagamma ] as function of @xmath75 and see that the data mostly collapses onto a single curve .
the collapse is not perfect and is not expected to be as the underlying flow and bead distribution patterns ( see fig . [
fig : geometryconc ] ) are different for different flows and magnetic fields .
fraction @xmath57 of beads caught as function of initial bead number density with and without the bulk force term @xmath78 in eq .
( [ eq : ns ] ) .
the fixed values for the current - distance product @xmath8 and the maximum in - flow speed @xmath74 are shown . at low densities
less than 50%are caught ; at high densities the collective motion of the beads leads to 100% capture . ] _ interactions and concentration . _
the second point we wish to make is that modification of the overall flow , and the effective bead - bead interaction this entails , is significant for bead capturing .
we can study the effect by excluding momentum transfer to the fluid flow due to the bulk force term @xmath78 in the navier stokes equation ( [ eq : ns ] ) . at
high bead number densities the force acting on the beads contributes a significant force on the fluid affecting fluid flow and spawning the effective interaction .
the strength of this interaction must thus depend on the density of particles .
this is illustrated in fig .
[ fig : captureconc ] ; capturing was simulated at fixed in - flow speed , @xmath79 m/s , and a fixed value of the current - wire distance product , @xmath80am , but for varying bead number densities @xmath81 ranging from @xmath82 to @xmath68 m@xmath10 . at low densities
we find that capturing is roughly independent of density and the fraction @xmath57 of beads captured has some intermediate value , however , for high densities all beads are caught .
in contrast , leaving out the bulk force term @xmath83 in the navier stokes equation , i.e. , the force acting on the fluid , gives concentration independence as shown in fig .
[ fig : captureconc ] . as can be seen in fig .
[ fig : deltabeta ] , a complementary way of exhibiting the importance of the bulk force term is to plot the difference @xmath84 between including and excluding @xmath83 as a function of concentration and the current - distance product .
this shows that interactions makes an appreciable difference at high concentrations and intermediate magnetic fields .
_ diffusion constant .
_ even for the small beads of radius @xmath26 m , the diffusion constant given by the einstein relation is small compared to the dimensions entering the problem .
the time - scale for a bead to diffuse across the channel is @xmath85 .
if we are to see the influence of diffusion competing with bead advection , then the relevant quantity is the pclet number @xmath86 which is advection time - scale @xmath87 over the diffusion time - scale .
when this number is larger than unity , which it is except for artificially large diffusion constants , then convection dominates . in the simulations the diffusion constant
is increased artificially up to @xmath88 m@xmath56/s in order to help numerical convergence .
but we have verified that values smaller than @xmath89 m@xmath56/s have no influence on the results .
the difference @xmath90 in captured bead fractions between two situations : including and excluding the bulk force term @xmath83 in the navier stokes equation ( [ eq : ns ] ) . at high concentrations
( @xmath91 m@xmath10 ) there is an appreciable difference between including and excluding the bulk force term , corresponding to hydrodynamic bead - bead interactions . ]
we have studied microfluidic capture of paramagnetic beads in suspension .
the three main findings of work are : the approximate scaling shown in fig . [
fig : collapsed ] , the existence of a critical bead density for capture shown in fig .
[ fig : captureconc ] , and the qualitative difference for capturing between models with and without the hydrodynamic bead - bead interaction shown in figs .
[ fig : captureconc ] and [ fig : deltabeta ] . clearly , it is very important for the capture process to include the action of the beads on the host fluid medium .
simply leaving it out can give qualitatively wrong results for high concentrations of beads .
this casts some doubt on the measurement of cell susceptibility through capturing as it depends on cell concentration @xcite .
deduction of susceptibilities from single bead or cell considerations together with measurements at high bead or cell concentration is suspect .
care must be taken to compare with standards of known and similar susceptibility , size , and concentration .
the effective bead - bead interaction greatly helps capturing .
it should make detectable differences depending on whether there are a few or hundreds of particles in a channel at a time in actual experiments especially when the flow and magnetic field are such that the beads are barely caught one by one .
this interaction should be considered when choosing operating conditions for microfluidic devices based on capturing of beads as higher bead number densities potentially eases requirements for external magnets and allows faster flushing .
we hope that experimental studies will be initiated to verify this prediction of our work . by newton s second law and @xmath92 , a spherical bead with radius @xmath24 approaches terminal velocity exponentially with a time constant @xmath93 , where @xmath43 is the fluid viscosity . in this work @xmath94s .
|
we study theoretically the capturing of paramagnetic beads by a magnetic field gradient in a microfluidic channel treating the beads as a continuum .
bead motion is affected by both fluidic and magnetic forces .
the transfer of momentum from beads to the fluid creates an effective bead - bead interaction that greatly aids capturing .
we demonstrate that for a given inlet flow speed a critical density of beads exists above which complete capturing takes place .
|
in the standard cosmological model , the large - scale structure in the density field grows from small initial perturbations through the mechanism of gravitational instability .
statistical measures of this structure can be used to both verify the growth mechanism ( see e.g. , bernardeau 2002 and references therein ) and quantify the initial pertubations . a different kind of growth of structure is expected when we consider the neutral hydrogen density field during the epoch of reionization ( see e.g. , the review by loeb & barkana 2001 ) . in this case , bubbles of ionized material form first around bright sources and grow as the ionization fronts overlap until the universe is fully ionized .
statistical measures applied to this `` radiation induced structure '' ( hereafter ris ) can be used in a similar way to the gravitational instability picture above , but this time to verify the process of reionization and the nature of the sources of radiation .
the effect of ris is likely to be qualitatively different from that of gravity , and as a result the statistical signatures will be different . our aim in this paper is to explore the differences , using ray traced simulations of reionization .
we aim to both find out how reionization is different from gravity in the way it forms structure and how to use these differences to categorize reionization scenarios .
the statistical properties of ris have been explored in many other works , e.g. , using the power spectrum of hi fluctuations by furlanetto ( 2004 ) , zaldarriaga ( 2004 ) , morales & hewitt ( 2004 ) , and using minkowski functionals by gleser ( 2006 ) . in the present paper
we will focus on the autocorrelation function .
theoretical studies of reionization include both analytic work , e.g. , miralda - escud ( 2000 ) , wyithe & loeb ( 2003 ) , cen ( 2003 ) , liu ( 2004 ) , furlanetto ( 2004 ) , and numerical simulations , e.g. , razoumov & scott ( 1999 ) , abel ( 1999 ) , gnedin ( 2000 ) , ciardi ( 2001 ) , sokasian ( 2001 ) , razoumov ( 2002 ) , sokasian ( 2004 ) .
recently , n - body simulations with radiative transfer post processing have been performed by iliev ( 2006 ) , and zahn ( 2007 ) with box sizes as large as @xmath7 and able to resolve halos down to masses of @xmath8 .
kohler ( 2007 ) have performed simulations with an extremely large box size ( up to @xmath9 ) in which hydrodynamics and radiative transfer are coupled self consistently .
these simulation rely on higher resolution simulations to calibrate the sub grid ( @xmath10 ) physics .
other recent advances in simulating reionization include the work of trac & cen ( 2006 ) , a hybrid n - body dark matter and rt approach , a code which was used to study the growth of bubbles during reionization by shin ( 2007 ) .
analytic work has been carried out using perturbation theory to predict inhomogeneities in the density of neutral hydrogen and photons ( zhang 2007 ) , as well as press - schechter based analyses ( e.g. , furlanetto 2004 . )
most relevant to the work here , the autocorrelation function of 21 cm emission has been examined in analytic models by wyithe & morales ( 2007 ) and barkana ( 2007 ) . in the present paper
we do not aim to simulate particular observational probes of this epoch , such as the doppler scattering of cmb photons on relativistic electrons , or the 21 cm emission from neutral hydrogen .
instead , we concentrate on the differences between @xmath11 for the density field and the density field modulated by ris . in principle
, this will be directly observable in the future , through the various observational probes ( e.g. , carilli 2004 , peterson 2005 , valds 2006 ) we leave exploration of this to future work .
the format of this paper is as follows . in 2 we describe the simulations , including the n - body outputs and the radiative transfer code .
we also describe our different models for the sources of ionizing radiation and different physical conditions we simulate . in
3 we show how global properties such as the mean ionized fraction evolve as a function of redshift in the different runs , and then in 4 examine the morphology of the neutral and ionized hydrogen density field . in
5 we measure the autocorrelation function in our different models and explore how it can differentiate between them .
we summarise our results and discuss them in
we work in the context of the standard cosmological constant dominated universe , with parameters @xmath12 , @xmath13 @xmath14 , and a hubble constant @xmath15 .
the initial linear power spectrum is cluster - normalised with a linearly extrapolated amplitude of @xmath16 at @xmath17 .
the radiative transfer is carried out as post - processing on n - body simulation outputs .
we run our n - body simulations with the cosmological hydrodynamic code gadget ( springel 2001 ) .
our fiducial simulation is run in a @xmath18 cubical volume .
iliev ( 2006 ) have shown in tests using subvolumes of a larger ( 100 @xmath19 ) simulation that @xmath20 is the smallest box side length for which the scatter in reionization histories in different volumes is reasonably small .
we use @xmath21 dark matter and @xmath21 gas particles .
the mass resolution is therefore @xmath22 for gas particles and @xmath23 for dark matter .
we also run higher resolution models in tests , as detailed below .
we do not include radiative cooling or star formation when computing the gas dynamics so that our simulations are similar to those carried out by sokasian ( 2001 ) .
we run the rt as postprocessing , so that there is no coupling between the hydrodynamics and radiation . in this respect ,
the gas serves to trace the dark matter distribution closely ( there is little difference at these high redshifts . ) we chose our sources of radiation to be associated with dark matter haloes ( see below ) .
in addition to the fiducial run , we run another model with identical box size and particle number but with different random initial phases in order to roughly indicate the effect of simulation cosmic variance .
we also run a model with a different amplitude of mass fluctuations ( @xmath24 . ) for resolution tests , we run a simulation with @xmath25 gas and dark matter particles in a box of size @xmath26 ( the same mass resolution as the fiducial run ) and a simulation with @xmath21 gas and dark matter particles in a box of size @xmath26 ( eight times better mass resolution than the fiducial run ) .
all models were started at @xmath27 and run until @xmath28 .
we output snapshots of the density field every 25 myr , so that there are approximately 40 snapshot files per run . after choosing models for the sources of radiation ( see below )
we carry out raytracing simulations of radiative transfer ( rt ) to study the evolution of the neutral hydrogen density .
the code we use to do this carries out monte carlo rt to follow photon packets through the distribution of matter .
the code is based on that used by croft ( 2004 ) to study the fluctuating radiation background field at lower redshift but incorporates time dependent rt in a monte carlo manner similar to the crash code of maselli ( 2003 ) ( we actually only treat hydrogen here , so we are in fact closer to the earlier work of ciardi ( 2001 ) .
the code in the present paper traces directly through the sph particle kernels ( see also kessel - deynet & burkert , 2000 , susa , 2006 , yoshida 2007 , and daqle 2007 , for non grid based radiative transfer , and semelin 2007 , and oxley & woolfson 2003 for sph codes that trace through a barnes - hut tree ) and so requires no regridding of the density field between outputs . the spatial resolution of the rt is therefore in principle higher in dense regions than would be possible with a uniform grid .
this approach is described in detail in altay ( 2007 ) , where test problems are carried out .
we also outline some of the relevant features of the code briefly below . in the present paper we model
only the hydrogen component of the universe ( assumed to comprise 0.76 of the baryonic mass ) .
we also do not explicitly follow the temperature evolution of the gas , beyond taking temperatures of particles to be @xmath29 k when they are ionized and @xmath30 when they are not .
we follow collisional ionization , photoionization and recombinations using the rates given in cen ( 1992 ) .
we randomly sample source photons from a power law distribution , @xmath31 , of photon energies ( more details on the sources are given in section 2.3 , below . ) in the present work , photon packets are monochromatic and are emitted isotropically from sources , again using a random number generator to pick directions .
every time a ray is traced through a particle , the number of recombination photons which have been produced in that particle since the last time it was visited are added to a stack .
when the stack size reaches one packet , a recombination photon packet is emitted from the particle where this occurs .
the frequency of the recombination radiation is given by the milne relation ( osterbrock , 1989 . )
we have one numerical code parameter , @xmath32 which sets the size of the photon packet , and therefore the time resolution of the code .
this parameter @xmath32 is the number of fully neutral simulation gas particles which could be ionized by one photon packet if the energy in that packet was split up into 13.6 ev photons .
for example , if a packet consists of @xmath33 13.6 ev photons and a gas particle contains @xmath34 neutral hydrogen atoms then @xmath35 . by trying runs with different values of @xmath32
we change how well the code can resolve the recombination timescale by governing the average interval between rays visiting particles and updating their ionization states .
packets of recombination radiation as well as source radiation are governed by @xmath32 so that shot noise arising from the discreteness of recombination modelling can be controlled .
we choose a suitable value of @xmath32 by carrying out convergence tests ( see e.g. ,
5.1 ) . in practice
we find that @xmath36 is adequate for modelling clustering of hi and is the value we use in our fiducial simulations .
this results in from @xmath37 photon packets being used in each of the runs . the paper altay ( 2007 )
is mainly concerned with presenting a closely related code , the publically available code sphray , which has many additional features to the one used in this paper , including the ability to model temperature evolution and helium species .
eventually , observations of the ris at the epoch of reionization will be useful as probes of the sources of radiation as well as cosmology .
the statistical measures of this structure are likely to be correlated with the properties of the sources , their luminosities , lifetimes , and clustering .
one of the goals of this paper is to study the ris produced by various extremely simple models for the ionizing source population , in order to see how they can be differentiated ( principally through the autocorrelation function , which we focus on ) and which features appear to be generic to the models considered .
the range of possible sources for reionization is extremely wide , including decaying dark matter ( e.g. , mapelli 2006 ) , primordial black holes ( ping & fang 2002 , ricotti 2007 ) , high redshift miniquasars ( e.g. , madau 2004 ) , population iii stars ( e.g. , sokasian 2004 ) , population ii stars ( e.g. , sokasian 2003 ) , some more or less likely than others .
rather than attempting to simulate particular models in detail , we restrict ourselves to simply parametrized models which relate the ionizing radiation intensity directly to the dark matter distribution .
this is on the understanding that in most reasonable models of reionization there would be some relationship between the two ( either through galaxies and stars associated with dark matter overdensities , or directly through dark matter clumps decaying to ionizing photons ) .
in particular , we associate sources of radiation to dark matter halos .
this approach has been used also by mellema ( 2006 ) , who use a constant mass to light ratio to assign ionizing radiation to each halo , as well by zahn ( 2007 ) , who populate each halo with a single ionizing source whose luminosity is proportional to host halo mass .
mcquinn ( 2006 ) have also simulated 17 different variations of this type of model with various ionizing photon efficiencies , prescriptions for feedback and and minihalos .
mcquinn ( 2006 ) focus on the morphology of hii regions .
all the runs we use for the main studies in this paper have a simulation box length of @xmath18 and particle number of @xmath38 , although as explained in 2.1 , for resolution studies we have some runs with different mass resolutions and box sizes .
we find halos using a standard friends - of - friends routine , with a linking length of 0.2 times the mean interparticle separation .
the minimum halo mass we use as a source in our fiducial run is @xmath39 , containing only 8 ( gas and dark matter ) particles .
this is approaching the scale of mini halos expected to host numerous weak ionizing sources and provide small scale clumpiness to the igm , but as with the calculations of zahn ( 2007 ) ( who do have a smaller particle mass ) it is still approximately an order of magnitude too large .
although we only resolve such halos with a small number of particles , we have checked using simulations with 8 times better mass resolution ( see 5.1 ) that this does not affect our computation of the autocorrelation function of neutral hydrogen , the statistic we focus on . in general , the limited mass resolution of simulations will affect results both through the absence of sinks ( minihalos ) and sources hosted by small halos .
recent simulations ( e.g. , santos 2007 ) are beginning to address this directly through increases in simulation particle number .
we return to this point in 5.1 and 6.2 .
we use the same density field for 10 of the 12 main runs , carrying out the rt as postprocessing using different source prescriptions .
the other two are a low fluctuation amplitude model ( @xmath24 ) run with the same random phases and another model with the same amplitude as the fiducial case but with different phases .
the 12 runs are differentiated by their different halo mass to light ratios , treatment of the relationship between halo mass and ionizing radiation , the recombination rate , and spectrum of radiation .
an overview is given in table [ runs ] , and they are descibed in detail below .
we label the different simulations by short descriptive names rather than numbers or letters in order to avoid the necessity of the reader referring back to a table when examining the results . [
cols="^,^,^,^,^,^,^ " , ] @xmath40 the * fiducial * run . here
the instantaneous luminosity of sources is proportional to the dark matter halo mass , with @xmath41 in our fiducial run , we take @xmath42erg / s/@xmath43 , and a source spectrum with @xmath44 , appropriate for population ii stellar sources ( e.g. , sokasian 2003 ) .
this simple model is also used by zahn ( 2007 ) , who assume a number of photons proportional to halo mass , ( with a conversion factor @xmath45 photons / sec/@xmath43 . )
this corresponds to our fiducial model having a luminosity per halo mass 4 times larger than that of zahn . using the computation of zahn , the source output of our fiducial model can be considered to be very roughly equivalent to one with population ii stars forming with an efficiency of @xmath46 from a salpeter imf , with a stellar lifetime of @xmath47 yrs and an escape fraction of @xmath48 .
the recombination rate in the fiducial simulation is taken to be that computed directly from the gas density , and diffuse recombination photons are treated .
@xmath40 runs * l/2 , l/4 * and * l/8*. these are the same as the fiducial run in every respect except that the ionizing luminosity has been reduced by an overall factor of 2,4 or 8 .
these models can be considered to be equivalent to for example reducing the efficiency of star formation , and/or the ionizing escape fraction with respect to the fiducial run .
the l/4 model therefore corresponds to the model simulated by zahn ( 2007 ) .
* l indep m * we use the same halo list as the fiducial run , but instead of a luminosity proportional to halo mass , we assign the same luminosity to all halos . the total integrated luminosity to @xmath49 is set to be equal to that in the fiducial run . @xmath40 * m @xmath50 10@xmath51 @xmath43 , m @xmath52 10@xmath51 @xmath43 * for these two runs , the same halo source list is used as in the fiducial run , but with either an upper or lower cutoff applied in the mass of a halo which can host a source . as in the previous model above , the total integrated luminosity to @xmath49 is set to be equal to the fiducial run .
these runs can be considered to roughly model the effects of feedback which might cause disruption of galaxy sources in small halos .
@xmath40 * @xmath24 * this run is the same as the fiducial run , except using a simulation which has a significantly lower amplitude of mass fluctuations , leading to later halo formation times .
the source luminosity was proportional to the halo mass in the same fashion as for the fiducial run , also with @xmath42erg / s/@xmath43 . the total integrated luminosity to @xmath49 is therefore less than in the fiducial run .
@xmath40 * no recomb , 2@xmath53 recomb * in these runs , the recombination rate of ionzed hydrogen was either set to zero or doubled , and the number of recombination photon packets adjusted accordingly . otherwise , the runs are the same as the fiducial run .
these models can be thought of as parametrizing the effects of changing the clumping factor of unresolved gas .
@xmath40 * @xmath54 spectrum * a harder spectrum than the @xmath55 spectrum used for the fiducial run was used here .
the total number of ionizing photons integrated to @xmath49 was set to be the same as the fiducial run , so that @xmath56erg / s/@xmath43 .
this run gives a rough indication of the effects of a harder spectrum than that of population ii stars , although it was not chosen to reproduce the spectrum of any particular source .
for example , at the zams , pop iii stars are expected to have a spectrum with @xmath57 , with @xmath58 close to the hydrogen ionizing edge ( tumlinson 2003 ) . for representative composite spectra of agn ( e.g. , telfer 2002 )
a value of @xmath59 is possible .
the effect of the harder spectrum will be to allow the photons to penetrate further .
the `` preheating '' effect of hard photons is not treated here as we do not follow the temperature evolution of the gas explicitly .
@xmath40 * other random phases * as a rough indicator of the effects of cosmic variance , this run has the same parameters as the fiducial run , but uses a different underlying density field realization .
we evolve all the simulations to redshift @xmath60 , irrespective of whether they have achieved full reionization by then .
because we have many different simulation runs with different treatments of recombination and models for the sources , we need to decide at which redshifts to compare our measures of clustering .
the different runs reach @xmath61 mean ionized fraction by mass , @xmath62 at redshifts ranging from @xmath63 , so that comparing them at the same redshift will mean comparing different stages of reionization .
we will therefore concentrate in the paper on snapshots taken at the same mean ionized fractions in the different runs , rather than at the same redshifts . in figure [ nvsz ]
we show how the mean mass weighted neutral fraction ( linear scale ) and mean mass weighted ionized fraction ( log scale ) vary with redshift for the different runs . as expected , the run without recombinations reionizes first , reaching a @xmath64 ionized fraction by redshift @xmath65 and being @xmath64 neutral at @xmath66 . as reionization proceeds ,
the relationship between mean ionized fraction @xmath62 and redshift is approximately exponential , @xmath67 , ( here @xmath68 is the redshift when the model is fully ionized ) for this and the other models .
this is roughly true for all models , except for the model where only large galaxies ( halo masses @xmath69 ) are sources , and the model with @xmath24 , both of which have a substantially more rapid change in ionization with redshift , @xmath70 .
this can be explained by the fact that galaxies massive enough to be sources only form relatively late in these two models .
the models where the luminosity per dark matter halo atom was varied , fiducial , @xmath71 , @xmath72 , @xmath73 reach the @xmath74 point later by @xmath75 for each halving of the luminosity .
the slope of @xmath76 vs @xmath77 is very similar for @xmath78 for these models but then changes as reionization proceeds .
we note that the @xmath72 model reaches @xmath74 at @xmath79 , similar to the model of zahn ( 2007 ) , which has a similar spectrum and source luminosity ( see 2.3 above . ) more physical insight can be gained by looking at two other quantities as a function of redshift , the number of recombinations per atom and the photon mean free path , which are plotted in figure [ recpath ] .
to compute the former , we divide the cumulative number of recombination photons by the total initial number of hydrogen atoms in the simulation volume .
we plot a symbol on the curves at the point when the ionized fraction by mass reaches @xmath74 and another at @xmath80 ( not all models reach this stage ) .
most of the models have below 0.5 recombination photons per atom by the time they have fully reionized , indicating that recombinations do not play the major role in the process of reionization .
the curves for the different models tend to flatten off slightly for the last half of the reionization process .
one striking feature of these curves is the importance of the amplitude of mass fluctuations , @xmath81 .
as we would expect , the low amplitude model ( @xmath24 ) , having less clumping has a lower recombination rate .
when compared at the point when @xmath74 , this model has experienced @xmath82 times fewer recombinations than the fiducial model . this is greater than the ratio of @xmath83 for the two models ( 1.65 ) , due to nonlinearity of clustering and the greater cosmic time in the low amplitude model to get to this point .
the photon mean free path as a function of redshift is also shown in figure [ recpath ] .
the mean path length between absorptions has been calculated by averaging over the most recent @xmath84 photon packets , a relatively small number . as a result ,
sawtooth features in the curves can be seen , which correspond to the times at which the new density field snapshots were inputted ( every 25 myr ) .
we have tested with more widely spaced outputs ( 50 myr ) that this does not affect the underlying curves .
another artifact of the measurement is the behaviour at large mean free path .
when averaging the path length , we only include photon packets that have not wrapped more than once round the box . as a result ,
when the volume is close to fully ionized , the average is taken over a biased sample , those photons which have run into an absorber in less than a box length .
the mean free path we plot therefore falls .
the overall behaviour of the mfp is as we would expect , with a rapid rise for most models at early times , with a mfp corresponding roughly to the size of the hii regions around sources .
for example at redshift @xmath85 , when the ionized fraction is @xmath86 , the mfp in the fiducial model is @xmath87 . in the @xmath54 spectrum case
, the low cross section of neutral hydrogen for the hard photons means that the mfp starts off nearly an order of magnitude larger , @xmath88 .
the hardest photons will travel much farther than this mean , and there is a low level of ionization ( ionized fraction @xmath89 ) present throughout the whole volume , even at these very early redshifts . in order to understand how reionization proceeds , it is instructive to look at the relation between neutral fraction and density
. this is plotted for the fiducial run in figure [ scatterfid ] at 4 different stages in the evolution of the model . in each panel
, we also show a histogram of particle density values and neutral fractions as well as a scatter plot of one versus the other .
if we go from panel to panel , we can see that the high density regions are reionized first ( the inside - out scenario also found in iliev 2006 ) .
for example , when @xmath90 , the cloud of partially ionized particles to the right of the panel is centered around @xmath91 , whereas at @xmath92 it is centered around @xmath93 . within this cloud of points
, the higher rate of recombinations in high density regions means that within the partially ionized volume there is a trend for higher neutral fractions at high densities .
this trend is additional to the opposite trend for high density regions to be reach this partially ionized state first .
as the universe comes close to being fully ionized , @xmath94 , particles close to the mean density are the only ones still fully neutral and the particles with the highest densities have neutral fractions @xmath95 . in the next section
, we examine the morphology of the neutral and ionized regions in order to investigate this in more detail . for now , we can compare the scatter plots of neutral fraction and density in different models .
we do this at @xmath74 in figure [ scatter_m ] .
the absence of a cloud of partially ionized points in the `` no recomb '' simulation is expected , as once particles are ionized , they drop off the bottom of the plot . also as expected , the @xmath54 spectrum simulation and the `` @xmath96 recomb '' run have a greater density of particles in this region than the fiducial model .
just as looking at structure in galaxy redshift surveys ( e.g. , schectman 1996 ) revealed filaments , voids and clusters , the ris is expected to give rise to a complex morphology . in the case of structure in the density field , reproducing the visual characteristics of observational data was one of the drivers in searching for the correct theory of structure formation , and cosmological n - body simulations are expected to give rise to a `` cosmic web '' with the same appearance ( e.g. , bond 1996 ) . deciding how to view the morphology of ris is complicated by the question of whether to plot the neutral fraction , neutral density , ionized density or ionized fraction , each of which can lead to a potentially different impression . also , unlike pure density fluctuations , which evolve relatively slowly , the exact time the ris is plotted as reionization proceeds can yield very different results .
again , we will choose to compare different models at the same value of @xmath62 . in figure [ slice_05
] , we show a thin ( 1 @xmath19 ) slice through the fiducial simulation volume , at the time when @xmath74 .
we use a two dimensional color scale to show both the density and the neutral fraction of hydrogen , as well as overplotting the positions of the individual sources of radiation , associated with dark matter halos .
it is apparent from the plot that low density regions where there are only a few isolated sources have not yet created noticeable bubbles , but that the highly clustered regions , associated with filaments and proto clusters have appreciable stromgren spheres around them .
iliev ( 2006 ) have measured the size of the bubbles in their simulations by fitting spheres into the ionized regions , finding a median bubble radius of @xmath97 ( see their figure 13 ) at this late stage of reionization .
many other studies , e.g. , shin ( 2007 ) , zahn ( 2007 ) have been carried out on the size of bubbles in simulations . visually , our simulation appears to be broadly consistent with the sizes found by iliev , and we leave to future work detailed statistical characterization of the size and shape of voids in the neutral hydrogen . for
now , we will comment on the obvious differences apparent between the morphology of ris in the simulation and that which can be seen in the underlying density field .
the ris has a much higher contrast level , with the neutral fraction in hii regions being @xmath98 times less than in the rest of the volume .
the edges of hii regions are consequently much sharper than those of voids in the matter distribution , lending themselves to easier detection by void - finding techniques ( e.g. , colberg 2005 ) . the matter density field over the redshift range relevant to reioniziation evolves little compared to the neutral density field .
reionization ( say the change from @xmath99 to @xmath100 ) in these models takes place over an change in scale factor @xmath101 of @xmath102 , and as @xmath103 to good approximation at these redshifts , linear growth of matter fluctuations occurs by the same factor . because of this , most of the change in the morphology and structure of neutral density field occurs in the ris .
plotting a slice through the neutral density field at different epochs allows us to see this well . in figure
[ slice_fbs ] , we show this for 3 different models ( fiducial , @xmath104 , @xmath105 ) , at times when @xmath62 varied from 0.1 to 0.9 in steps of 0.2 .
mcquinn ( 2007 ) have shown that the large scale morphology of ionized hydrogen bubbles depends most strongly on @xmath62 and the properties of the ionizing sources , and is relatively less affected by the specific subgrid model used to determine small scale source suppression and clumping factors . in figure [ slice_fbs ] , we can see that models in the same column ( same value of @xmath62 ) are indeed fairly similar .
however , because we use the mass - weighted ionization fraction @xmath62 rather than the volume weighted fraction @xmath106 , our conclusions about the similarity of the morphologies is somewhat different than those of mcquinn ( 2007 ) .
for example , the panel with @xmath74 ( middle panel ) for the fiducial model ( top row ) seems to be most similar to the @xmath107 ( 2nd panel ) for the @xmath104 model ( middle row ) .
this is because in the @xmath104 model , the ionized material is all concentrated in large bubbles , whereas in the fiducial model there are many smaller hii regions around the fainter sources which are hard to see but which nevertheless account for much of the mass in ionized material .
a good way to see that this is the case is to refer to figure [ slice_fbsi ] , which shows the same slices through the same models , but this time plotting the ionized density rather than neutral density .
the hii regions around the fainter sources are clearly seen in the top and bottom rows .
the early hii regions appear sharper and perhaps more spherical when seen directly in terms of their ionized density , rather than as shadows in the neutral hydrogen plot ( figure [ slice_fbs ] ) .
this is understandable because of the fact that the edges of the bubbles and the totality of bubbles that are smaller than the slice thickness ( 1 @xmath19 ) in size will be somewhat obscured in figure [ slice_fbs ] by neutral hydrogen that lies in front or behind the bubble but is still in the slice . in figures [ slice_fbsi ] and [ slice_fbs ]
we see little difference between the @xmath108 model and the fiducial model , indicating that the absence of the most massive halos does not greatly affect the morphology , as long as we compare at the same value of @xmath62 .
if we look at the panels at the farthest right of these plots , the end stages of reionization ( @xmath109 ) , we can see that the models with small halos allowed do have hi remnants with more ragged edges and more small scale structure apparent in them than the @xmath69 model .
in future work , it would be interesting to investigate the mass function of the disconnected hi remnants present at these times as they may have constraining power ( as well as being likely detectable in 21 cm emission .
the @xmath62 value at which the ionized regions percolate seems likely to depend on the source model also , as for example the @xmath74 panel for the @xmath69 model consists less of disconnected hii regions than the other two panels .
this makes sense , as the ionized material is more clustered , being closer to the more massive haloes .
this @xmath74 epoch is therefore a good one at which to compare the morphology of the other models also . in figure
[ slice_m ] , we show the same thin slice through the neutral density field for the 12 different models .
the panel most different from the others is not suprisingly that for the `` other random phases '' model , for which the slice intersects a fairly spherical bubble ( at the bottom of the plot ) and a large region which has little sign of reionization ( in the middle ) .
comparing this panel to the fiducial model , it might seem as though the latter is closer to the percolation stage , although they both have the same @xmath74 . in the later stages of reionization ,
when the ionized structures are a substantial fraction of the box size , we must naturally be careful in our interpretation of the morphology , due to the effects of cosmic variance and finite volume . as stressed by iliev ( 2006 ) , and others , large boxes are necessary to capture these processes , particularly at the later epochs .
looking systematically at the different panels of figure [ slice_m ] , we can see that in the fiducial , l/2 , l/4 and l/8 models as the luminosity per halo atom is reduced , the visual extent of the structures becomes smaller and smaller . from figure [ recpath ]
we can see that in these models , the number of recombinations becomes progressively higher as we reduce the luminosity , and although one would expect that this would be compensated for in the morphology by the fact that we compare all models at the same @xmath62 , there is in fact still a large difference .
we can therefore state that changing the luminosity of sources at fixed @xmath62 does affect the visual morphology quite strongly . for the l/8 model ,
much of the ionized hydrogen is again hidden from view in the strongly neutral regions , as happened in the comparison of the fiducial and @xmath110 models . from a rough visual comparison of the different columns of figure
[ slice_fbs ] and models l/2-l/8 in figure [ slice_m ] it seems as though scaling the luminosity of sources by a factor of @xmath111 changes the visual morphology of the hi density field in a similar fashion to changing @xmath62 by @xmath112 . of course
this will not hold for the morphology of the ionized regions , and we will investigate this statistically when we look at the correlation function in the next section .
quantitative comparisons of the structure in the various models can be carried out by looking at the autocorrelation function , @xmath113 .
we will focus on @xmath11 measured for the neutral density distribution , although we will briefly examine @xmath11 for the ionized density field .
we note that the fourier transform of @xmath11 , the power spectrum @xmath114 of fluctuations has been studied in reionization models by many authors .
we compute @xmath11 for the gas density directly from the particle positions in the simulation , and for @xmath0 we weight each particle by its neutral fraction : @xmath115 where @xmath116 is the number of pairs of particles in a bin centered on @xmath117 , @xmath118 is the expected number for a random distribution , @xmath119 and @xmath120 are the neutral fractions of particles in a pair , and @xmath121 is the mean neutral fraction by mass . as with our examination of morphology of the neutral density , we will compute @xmath11 for different simulation outputs chosen by their mass weighted ionized fraction , @xmath62 . in order to test the range of validity of our @xmath11 results ,
we carry out several resolution and boxsize tests . because the bubble like structures which overlap during the end stages of reionization occupy a large volume
, one would expect that the simulation box size may have a strong effect on our results .
we therefore compute @xmath11 for the neutral gas and the total gas density for two simulations with different box sizes ( we vary the box side length by a factor of 2 ) but the same mass and spatial resolution ( this is kept the same as our fidcucial model ) .
the results are show in figure [ xi_boxsize ] , where the 3 panels show @xmath122 and @xmath123 for at different stages of reionization ( parametrized by @xmath62 ) in the fiducial model .
we can see that the @xmath122 curves are extremely similar over the range @xmath124 for the 3 values of @xmath62 .
in particular , a power law fit to the curves over this range gives virtually identical parameters .
this is a very good thing , because it shows that no non - linear gravitational mode coupling has taken place with large scale density modes of order the box size .
this is one advantage of working at these high redshifts ( @xmath125 for @xmath126 in this case ) , where we can see that even a @xmath26 box is large enough to study gravitational clustering over this range of length scales .
the correlation function of the neutral gas density has a broadly similar behaviour .
the position of the break in the powerlaw form of @xmath11 is due to the finite size of the simulation volume and so we will restrict our analysis of @xmath11 to smaller scales .
interestingly , the panel with results closest to the end of reionization ( @xmath62 ) does not show any greater disagreement for large @xmath117 for @xmath123 than for @xmath122 .
this means that at least over this limited range of scales , our @xmath113 measurements will also be reliable for the neutral density . on the smaller scales ( @xmath127
, there appears to be a rapid drop off in @xmath113 for the neutral density for both boxsizes .
we shall see below when we consider mass and spatial resolution that our results will not be useful below these scales in any case . for our next test
, we keep the box size fixed at @xmath26 but vary the particle number ( and hence mass resolution ) by a factor of 8 .
we also vary the spatial ( force ) resolution by a factor of 2 .
the coarser mass / spatial resolution is the one we use in our fiducial case ( only with a @xmath18 box ) .
the dark matter halos which we use to place our sources of radiation have the same lower mass cutoff @xmath128 @xmath43 in both runs , which corresponds to 8 times fewer particles in the low resolution case .
the results for @xmath122 and @xmath123 are shown in figure [ xi_res ] , again for different values of @xmath62 .
the @xmath122 curves show good agreement on large scales ( the two simulations were run with the same initial phases for the density field ) . on scales
@xmath129 , the two curves diverge , indicating the effects of mass and spatial resolution on the gravitational evolution of the gas density field .
we note that on scales comparable to this the clustering in the gas density will be influenced by cooling and star formation , which we do not include here in any case .
the @xmath123 correlation functions also agree well down to @xmath129 , when they diverge even more sharply .
we notice that as will the boxsize test , the worst disagreement on large scales occurs rather unexpectedly for the early stages of reionization ( @xmath130 ) .
the final simulation parameter which we vary is @xmath32 , the maximum number of simulation particles which can be ionized by a single photon packet .
this parameter ( see section 2.2 for more details ) is inversely proportional to the total number of photon packets used to carry out the monte carlo rt . for larger values of @xmath32 ,
the radiation field will not be as smooth , and there will be more shot noise in the neutral density field .
our fiducial value of @xmath131 , and in figure [ xi_cf ] we show what happens when this is varied from @xmath132 to @xmath133 , with all other simulation parameters the same as our fiducial model .
the @xmath122 curves are almost identical , as might be expected , with the small differences due to the fact that reionization occurs at slightly different times in the different @xmath32 runs .
the curves for @xmath123 are also very similar , and there is no apparent systematic effect even when @xmath32 is made 10 times larger than our fiducial value .
at least as far as the correlation function is concerned , we have therefore converged to stable results with our fiducial number of photon packets .
the tests in this section have therefore revealed that our results for @xmath134 should be reliable over at least the scales @xmath135 .
we will concentrate on this range in our analysis , for example looking at the power law nature of @xmath134 . for looking at larger scales , approaching the scale of bubbles at the time of percolation ,
larger simulation volumes should be run in the future .
we note that because @xmath134 for a @xmath26 volume converged with a larger box on scales below @xmath136 it is probably safe to assume that we can draw information from our fiducial volume ( @xmath18 box ) on scales up to @xmath137 .
we show @xmath11 for the fiducial model in figure [ xifid ] , for values of the ionized fractions @xmath62 ranging from 0.1 to 0.99 , which corresponds to a redshift range of @xmath138 to @xmath139 . before reionization starts , when @xmath140 , @xmath134 and @xmath0 are identical , by definition .
however , once @xmath62 has reached 0.1 , one can see that @xmath0 is somewhat lower than @xmath134 , by a factor of @xmath141 , but has the same shape , a power law : @xmath142 with slope @xmath143 on scales @xmath144 and a gentle break above it ( at a scale dictated by the finite box volume ) . we will explore power law fits to @xmath11 in section 5.4 below . for
now , we explore qualitatively the behaviour of @xmath134 as reionization proceeds .
we note that this behaviour ( a decrease and then an increase in the amplitude of clustering ) has been seen in analytic calculations of the 21 cm brightness correlations ( e.g. , wyithe & morales , 2007 ) as @xmath62 increases , @xmath134 exhibits the usual linear growth , with the amplitude increasing as expected under gravitational instability , and the shape remaning constant .
as can be seen from the solid lines in figure [ xifid ] , this gravitational amplification of structure is very small ( a factor @xmath145 ) over the time of reionization .
@xmath0 on the other hand shows dramatic growth , by a factor 100 or more over the same interval , as we would predict for example by examining the morphology of structure in the hi density field in figure [ slice_fbs ] .
the initial stage of reionization affects primarily the high density regions around sources of radiation .
removing their contribution from the clustering of hi leads to an antibias on all scales in @xmath0 with respect to @xmath134 .
after this , the effect of removing highly clustered neutral regions competes with the amplifying effect of ris , with the latter winning after @xmath146 , raising @xmath0 above @xmath134 at this point .
as the ris grows in scale , the shape of @xmath0 begins to change , with the slope of the powerlaw region becoming flatter , reaching @xmath147 when @xmath80 .
the relationship between @xmath0 and @xmath134 can be examined by plotting the bias as a function of scale , defined by @xmath148^{1/2 } \label{beq}\ ] ] .
this quantity is plotted in figure [ bfid ] for the fiducial model , where it can be seen that @xmath149 is approximately constant with scale for @xmath150 for @xmath151 , with a value less than 1 for low values of @xmath62 .
after this , @xmath149 takes on a positive power law slope .
if @xmath0 and @xmath134 are fit by powerlaws , with slopes @xmath152 and @xmath153 , then we expect the slope of a power law fit to @xmath154 , @xmath155 to give @xmath156 .
if we examine @xmath154 for @xmath80 , we find a slope @xmath157 , as expected .
we can see that this slope is reached quite rapidly after @xmath62 reaches @xmath158 and does not appear to evolve much after this .
we shall explore this further below .
the behaviour of @xmath154 in this strongly ionized regime , is not constant with scale , indicating that during the late stages of reionization , 21 cm tomography measurements ( if they were possible on these small scales , which is observationally difficult ) would not result in a straightforward measurement of the underlying gas or dark matter clustering . during the earlier stages , however , the ( anti)bias is linear , although its amplitude may be related to reionization process in a complex way .
if instead we compute @xmath159 , the correlation function of the ionized hydrogren ( e.g. , as plotted in figure [ ionxifid ] ) , we find that the situation is simpler .
this is shown for the fiducial model for different @xmath62 values in figure [ ionxifid ] . at the highest values of @xmath62
, we obviously expect @xmath160 , which has the characteristic power law behaviour for @xmath161 . at earlier times
, @xmath162 has the same slope , but a larger amplitude , indictating that the ionized regions trace a biased subset of the density distribution , around peaks in the density field ( see e.g. , kaiser , 1984 . ) when @xmath90 ( at @xmath138 ) , the value of @xmath163 is @xmath164 , ( compare to @xmath165 for @xmath166 galaxies at the present day , zehavi 2005 ) .
we can see from figure [ ionxifid ] that @xmath167 breaks from a powerlaw at around @xmath168 , which , from looking at figure [ slicefbs ] corresponds approximately to the size of ionized bubbles at this early epoch .
we have seen that @xmath0 in the fiducial model has a similar shape to that which arises under gravitational instability .
we now investigate @xmath0 in the other reionization models in order to see if there is a dependence on the properties of the sources of radiation or the physics of reionization .
we again compute @xmath0 at output times corresponding to different specific values of @xmath62 .
the results for our 12 models are shown in figure [ xi_m ] , along with power - law fits ( described more fully in section 5.4 below ) to the region @xmath169 .
it is apparent that all models display behaviour broadly similar to the fiducial case , with a form reasonably approximate to a power law over @xmath170 decades in scale .
as indicated by our tests above , the break on scales @xmath171 is largely caused by the finite box volume and is expected to be similar in all cases .
all models have a certain level of antibias between @xmath0 and @xmath134 at first , with the most extreme antibias being reached in the `` l/8 '' model .
the `` @xmath172 '' model has only a small level of antibias , indicating that the large bubbles formed rapidly by the very luminous sources quickly modulate the hi field .
the correlation function becomes shallower and its amplitude increases dramatically for all models towards the end of reioniziation .
when @xmath80 , the models all have a very similar amplitude and a shallow power law slope @xmath147 .
it should be noted that model `` l/8 '' and `` l indep m '' do not have outputs for the very end stages of reionization .
the latter has very similar behaviour to the fiducial case , and the former is rather different , as we shall see when we consider the evolution of the power fit parameters .
overall one can see from the panels in figure [ xi_m ] that the ris leads in all models to a strong increase in @xmath0 over the course of reionization . unlike linear growth of fluctuations under gi , the slope the correlations changes , reaching a similar shallow value in all models .
no obvious features with any particular physical are created in @xmath0 in any of our variations of the reionization scenario . despite the relatively wide variety of models employed , the differences in @xmath0 are subtle , and will be explored in @xmath173 space below .
the galaxy - galaxy correlation function has only small deviations from power law behaviour over @xmath174 decades in length scale ( e.g. , peebles , 1980 , and zehavi 2002 ) , despite the undoubted complexities of galaxy formation .
the dark matter correlation function in simulations also exhibits this behaviour ( e.g. , jenkins 1998 , kravtsov 2004 . )
it is therefore not unreasonable to expect that this type of scale invariance might also be created by ris .
we have seen from figure [ xi_m ] that this is indeed the case . in this subsection
we examine the power law behaviour of @xmath0 more quantitatively .
we fit power laws to the region @xmath169 , based on the resolution and boxsize tests of section 5.1 .
the power - law fits were carried out assuming poisson errors on the @xmath0 points , so that the error is @xmath175 , where @xmath116 is the number of pairs of particles in a bin .
we have tried changing @xmath116 to include only pairs of particles above a threshold neutral fraction ( e.g. , @xmath176 ) , but this has a negible effect on the fit parameters .
also , changing the fitting region to @xmath177 does not change any of our conclusions below . in the top 2 panels of figure [ fitxi ]
, we show the dependence of the power law fit parameters @xmath163 and @xmath178 on @xmath62 for our different models .
looking at @xmath163 first , we can see that there is a strong dependence on @xmath62 .
the models all follow a trend roughly equivalent ( within a factor of 2 in @xmath163 for all but one model ) to @xmath179 , with the @xmath180 model being the most extreme outlier .
the fact that the lines for nearly all the different models are clustered together so tightly is rather surprising , and appears to be an indication that @xmath62 plays a dominant role setting @xmath163 . at least in this fashion
, the neutral fraction governs the structure in the neutral hydrgogen density field .
the visual morphology of different models with the same @xmath62 values ( e.g. , figure [ slice_m ] ) did appear to be noticeably different , rather more than one would expect give the tight locus of @xmath181 curves .
given the visual impression of the hi slices , however , it is understandable that the model with perhaps the greatest difference to the others ( the @xmath182 model ) is also discrepant in terms of @xmath163 . for any value of @xmath62 , this model has a larger @xmath163 than the others .
the @xmath163 differences between models are small , but measurable , as are the purely gravitational instability based differences ( note that the models reach different values of @xmath62 at different redshifts , so that the underlying @xmath183 curves will be different . )
we do find a wider variation in the values of the slope of @xmath1 , for fixed values of @xmath62 .
this can be seen from the middle panel of figure [ fitxi ] , where for example @xmath178 varies from 1.6 to 0.6 when @xmath74 .
the fiducial model has @xmath184 when @xmath74 .
if we look horizontally to find out at what @xmath62 value the other models have the same slope , we find a range from @xmath185 ( for the l/4 model ) to @xmath186 ( for the @xmath180 model ) .
all curves ( except for the l/8 model which does not get more than half ionized ) do follow the same pattern , with the slope getting asymptotically flatter as reionization proceeds , all ending up with @xmath147 .
the rapidity with which this asymptotic value is reached does vary with the different models , with the @xmath180 model doing this most quickly , as the large - scale modulation effects of the bright sources tilt the correlation function upwards on large scales .
the slope of @xmath1 stays approximately constant as @xmath163 increases rapidly during the end stages of reionization , as can be seen clearly from the bottom panel of figure [ fitxi ] , where @xmath173 curves are plotted . in the region to the left of the plot , clustering of hi in all models is dominated by the clustering in the underlying density field .
the ris then takes over , and again all models follow a rather similar locus of @xmath1 parameters .
we have seen above that the variation between the @xmath173 lines is mostly due to the variation of @xmath178 with @xmath62 in the different models .
overall , the change in slope for the different models can be qualitively explained by the different length scales of the ris features that occur in each , even at the same stages of reionization ( @xmath62 values ) .
there is a competition between small and large scale features that sets the slope of the correlation function .
the behaviour of @xmath163 is more puzzling , and we shall return to these questions in section 6.2 we have seen in previous sections that once reionization is visually progressing , the amplitude of fluctuations in the hi field grows extremely rapidly .
it is interesting to compare this quantitatively to the growth expected of perturbations under gravitational instability . in the latter case
, the amplitude of linear density perturbations will grow at the same rate independent of scale , so that @xmath134 will retain the same shape , but increase in amplitude by an overall factor @xmath187^{2}$ ] which can be computed from first order perturbation theory ( e.g. , peebles 1980 ) . at the high redshifts relevant here ( @xmath188 )
, the lcdm universe behaves in a manner very close to an einstein de sitter model so that the linear growth factor @xmath189 . from figure [ xi_m ] ,
we have seen that @xmath163 for the matter correlation function is between 0.1 and 0.5 @xmath19 for the models over the range of redshifts when reionization occurs , so that linear theory should be accurate over most of the range of scales indicated by our resolution and boxsize tests ( section 5.1 ) .
we can also see from figure [ xi_m ] that @xmath134 does keep the same shape , and increases in amplitude slowly , as expected .
this can also be inferred from figure [ gfid ] , where we plot the square root of the ratio of the correlation function @xmath134 at redshift @xmath77 to that at redshift @xmath190 , as a function of @xmath77 .
this is proportional to the growth factor of pertubations between redshifts .
there are three lines on the plot , corresponding to @xmath191 and @xmath192 .
the lines are close together , indicating that the shape of @xmath134 does not change dramatically . in figure [ gfid ]
, we plot the same quantity for @xmath0 , for the same @xmath117 values , for the fiducial model of reionization . in this case , we see both that growth of pertubations is much more rapid , due to the ris , than the linear density growth .
we draw a smooth curve corresponding to @xmath193 alongside the simulation growth factor for @xmath194 in order to show roughly how extreme the growth in fluctuations as a function of redshift is during the epoch of reionization . the growth is faster than exponential over the short interval between @xmath195 and @xmath196 , it is roughly an exponential function of a power law .
we also see that the different scales exhibit different growth rates , with the large - scale fluctuations changing most rapidly . on @xmath197 scales ,
the fluctuations are first suppressed , with the square root of @xmath0 decreasing by a factor @xmath198 , before rapidly increasing after @xmath199 .
this stronger behaviour relative to the smaller scales results in the flattening of the correlation function .
the roughly parallel nature of the curves after @xmath200 indicates that at the late stages of reionization the flatter power law form of @xmath0 has been reached , and the amplitude grows similarily on all scales . if we now consider the growth of perturbations in the different models , a better way to compare them is to look at the growth as a function of @xmath62 . in figure
[ gm ] we show this for the @xmath201 scale . as the models become more ionized , from left to right , the amplitude of fluctuations first dips and then rises steeply . the model which dips the least on these scales is the @xmath69 model , as we expect from looking at figure [ xi_m ] .
the main growth phase of fluctuations starts between @xmath90 @xmath202 , depending on the model , with an exponential relationship between @xmath203 and @xmath62 .
the curves appear to converge towards the end of reionization , so that all models exhibit roughly the same amount of growth , within @xmath204 from the start of reionization until @xmath94 .
the models which started off with less growth in @xmath1 therefore have steeper dependence on @xmath62 .
we find that @xmath205 approximately holds over the range @xmath206 for the fiducial model .
the steepest curve has @xmath207 ( the l/4 model ) and the shallowest has @xmath208 ( for the l @xmath209 model ) .
if we look at the growth factor from the point of view of the parameters varied in each model , we can see that decreasing the luminosity ( from the fiducial model through @xmath71 , @xmath72 and @xmath73 ) monotonically changes how abrupt the growth of hi fluctuations is during the bulk of the reionization process .
the model with least luminous sources ( @xmath73 ) exhibits fastest growth with respect to @xmath210 , although of course with respect to @xmath77 , this is not necessarily the case .
we have examined the growth factor vs @xmath77 for the different models ( not plotted ) and find that the growth of fluctuations for all models has ( at least for the @xmath211 scale ) a form roughly consistent wit the @xmath193 curve drawn on figure [ gfid ] . to zeroth order , the growth of hi fluctuations as a function of @xmath77 does not seem to be dependent on the source physics . we have seen that a simple power law fit to the hi correlation function works well on scales for which the simulation has sufficient boxsize and spatial resolution . as reionization proceeds ,
the power law becomes shallower , moving away from the slope of the underlying matter correlation function . while this paper s focus is on numerical modelling and phenomenology of hi clustering rather than analytic modelling ,
may neverthless be instructive to consider a different fitting function , for @xmath0 , derived from very simple model .
many analytic models of bubble growth during reionization , of varying complexity ( e.g. , using the excursion set formalism , furlanetto 2004 , and perturbation theory , zhang 2007 . ) have been proposed , and most recently , tests of the model of furlanetto , zhan , and hernquist have been carried out by zahn ( 2007 ) using rt simulations . here
we consider the model of babul & white ( 1991 ) originally proposed by those authors as a simple description of the way galaxy clustering could be modulated by `` spheres of avoidance '' around quasars .
the model makes the simplifying assumption that the sources are poisson distributed and that they give rise to spheres of avoidance of a fixed physical size .
the two point correlation function of material distribution uniformly in the interbubble regions is then ( babul & white 1991 ) : here @xmath213 is the nominal filling factor of spherical bubbles of radius @xmath214 and mean number density @xmath215 . if we furthermore assume that the distribution of sources is independent of the density distribution , modulation of the hi by bubbles leads to an hi correlation function given by : @xmath216[1+\xi_{\rho } ] \label{modu}\ ] ] of course , the distribution of sources _ is _ closely related to the density distribution , something that leads to the antibiasing seen in the early stages of reionization ( see figure [ bfid ] ) . at later times , however , this correlation may become less important in governing @xmath1 , and the simple interbubble model may be useful . rather than computing values of the @xmath217 and @xmath218 parameters from theory , we use equation [ interbubble ] to fit @xmath219 measured from the simulations , allowing the values of these parameters to vary . before fitting , we first compute @xmath219 using the measured value of @xmath134 in the simulation and equation [ modu ] .
@xmath219 for the fiducial model for various values of @xmath62 is shown in figure [ babul1 ] . for @xmath220 ,
@xmath219 is negative , an unphysical result which occurs because of the correlation between sources and @xmath221 .
as reionization proceeds , @xmath219 becomes dramatically larger in amplitude , and the right hand cut off moves to larger scales .
the correlation function is close to flat on smaller scales than the break .
we can interpret this behaviour approximately in the context of the babul and white model fit , which is also shown in figure [ babul1 ] .
we have fit the curves in the same manner as was carried out for the power law fits in section 5.4 ( assuming poisson errors ) .
it is evident that the simple model can reproduce the rough shape of @xmath219 , with agreement becoming best towards the end stages of reionization .
it is the modulation of the density correlation function by this form for the ris @xmath222 results in the flattening of @xmath0
. it would be interesting if the best fit value of @xmath218 for a given @xmath219 could be used as a measure of the bubble size .
there are two problems with this , however .
the first is that in the early stages of reionization when bubbles are better defined , they are strongly correlated with sources , so that the fit does nt work .
the second problem is that at late times , the simulation box size we have used becomes comparable to the bubble size and the results will presumably not be very reliable .
there is a range of reionization stages for which this might be useful though . for the fits in figure [ babul1 ]
the @xmath218 values did vary monotonically with @xmath210 , with @xmath223 for @xmath202 , @xmath224 for @xmath74 and @xmath225 for @xmath92 . above @xmath126 , @xmath218 saturated at a value of @xmath226 , possibly indicating the limitations of finite boxsize .
the other parameter , the filling factor @xmath217 can also be examined , and could also be a useful diagnostic . in this case
the value steadily increases , with @xmath227 for @xmath228 .
we note from the definition of @xmath217 above that it is a nominal filling factor , and as a result @xmath217 will become greated than unity as bubbles overlap in the late stages of reionization .
the difference between @xmath219 for the different models ( all with @xmath229 ) is shown in figure [ babul2 ] .
we can see that the peak amplitude of the curves varies by a factor of @xmath230 between models , but that they all display the same shallow curves behaviour with a cutoff .
as we would expect , the model with the largest obvious bubbles in figure [ slice_m ] , @xmath231 has the largest amplitude and largest cutoff scale .
a few of the models ( for example l/4 and l/8 ) have a negative @xmath219 and are not plotted . for clarity , we do not show the model fits , but we note that the @xmath232 has the largest value of @xmath217 , 0.27 , and the l/2 model has the smallest value , 0.06 . these values for the bubble filling factor appear by eye to be at least indicative of the trends seen in figure [ slice_m ] .
in this paper we have explored the radiation - induced structures ( ris ) produced during the epoch of reionization and compared them to the results of gravitational instability .
our conclusions can be summarized as follows : + ( 1 ) for most of the models we have tested , we find that the mean ionized fraction of hydrogen increases exponentially , @xmath67 ( where @xmath68 is the redshift of full ionization ) as reionization proceeds . for models adjusted to have emitted the same number of source ionizing photons by @xmath49 ,
there is still quite a wide spread in the redshift of reionization , with models reaching @xmath74 from @xmath233 to @xmath234 ( the last one to reionize is the low @xmath81 model . ) + ( 2 ) at a fixed @xmath62 , the morphology of the ris is most strongly affected by the lower cutoff in source luminosity , which changes the size of bubbles .
the mass to light ratio of sources also has a substantial effect , but the recombination rate and the amplitude of mass fluctuations , @xmath235 only minimally change the appearance of the hi density field .
+ ( 3 ) the hi correlation function , @xmath1 exhibits a generic behaviour for all models tested .
@xmath1 initially becomes linearly antibiased with respect to the matter @xmath183 , as the high density hi around sources is ionized .
the linear bias factor reaches a minimum of @xmath236 when @xmath237 .
the amplitude of @xmath1 then increases rapidly , and @xmath1 keeps a scale invariant shape , but the power law index flattens to an asymptotic value of @xmath238 .
+ ( 4 ) we find that @xmath163 , the correlation length of @xmath1 has the essentially the same functional relationship with @xmath62 in all but one of the models we have tested . how the power - law index @xmath178 varies with @xmath62 on the other hand depends much more widely on the different source and physics prescriptions adopted .
+ ( 5 ) the growth factor of hi perturbations is seen to change much more rapidly than that of gravitionally evolving matter perturbations over a redshift range @xmath6 during which the bulk of reionization occurs .
we find pertubations on a scale of @xmath211 to be evolving @xmath239 compared to @xmath240 for gravitational growth .
this is valid for all models tested , so that the source physics does not appear to affect relation .
+ ( 6 ) during the late stages of reionization , the shape evolution of @xmath1 can be approximately reproduced by a simple model due to babul & white ( 1991 ) in which ionizing sources are uncorrelated with the density field and produce spherical bubbles .
fitting the parameters of this model to @xmath1 therefore form a method for inferring simple morphological characteristics from measurements of @xmath1 .
+ in this paper , we have largely avoided discussing directly observational probes of the ris and the reionization epoch .
this said , many of our findings can be related closely to the possible results of a survey of 21 cm brightness , as a function of angular position and wavelength .
we have concentrated mainly on the correlation function of the hi , and so in the most likely scenario in which there have been enough early x - ray sources to heat up the primordial gas just before reionization occurs , its temperature is higher than the cmb and we are in the emission regime . in this case , the 21 cm brightness temperature is independent of the spin temperature .
the brightness temperature @xmath241 , where @xmath242 is the overdensity of hydrogen and @xmath243 the neutral fraction ( see e.g. , madau 1997 , di matteo 2004 , barkana 2007 . ) looking at the autocorrelation function of the hi , the power law behaviour we see is striking . for structure growing through gravitational instablity ,
the initial fluctuations are scale - invariant over a large range and they evolve in a scale free manner ( peebles 1974 , davis & peebles 1977 . ) in the case of ris , the onset of significant clustering is extremely rapid , and it might be expected that this would lead to features in the correlation function related to for example the stromgren radius of sources dominant at that time . on the other hand , cosmic variance would lead to a substantial scatter in the scale length of features from place to place .
our fiducial simulation volume , at @xmath244 comoving @xmath19 side length also limits our ability to capture the late stages of reionization .
wyithe and loeb ( 2004 ) have predicted a comoving radius of bubbles @xmath245 at the end of the overlap stage .
we have seen in 5.6 that the simple babul and white model fit does capture the shape of @xmath11 reasonably well even though there is a distinct bubble scale in the model .
scale invariance therefore appears relatively easy to achieve and detection of departures from it probably requires a wide range of scales to be available .
furlanetto ( 2006 ) have used the analytic model of furlanetto ( 2004 ) to describe how the hii bubble size is related to both the bias of galaxies and the underlying matter power spectrum .
the success of such analytic models in comparisons to simulations ( e.g. , zahn 2007 ) has opened up the way for their use in analyzing future 21 cm observations . our related work , attempting to directly simulate a relatively wide range of models has found that the autocorrelation function of 21 cm emission can be used to infer the broad signatures of ris compared to gravitational structure ( e.g. , non - monotonic growth , flatter asymptotic slope @xmath238 ) which will help in analysis of the first observations . among the other analytic models which have been developed
, the pertubation theory approach of zhang ( 2007 ) is different from many in that it does not make a step function bubble approximation to the hi distribution . as our simulation approach includes recombinations and the contribution of residual hi in the ionized regions , future comparisons will be beneficial . for example , zhang ( 2007 ) compute the rapid rise in the bias of hi clustering as a function of redshift , finding some qualitatively similar results to ours , although they find a scale - independent bias on large scales .
one aspect which we have not covered is the effect of redshift distortions on the ris .
we expect the autocorrelation function to be affected by the usual line of sight amplification ( kaiser 1987 ) on large scales , and small scale suppression from the velocity dispersion ( e.g. peebles 1980 ) on small scales .
the latter effect in particular is likely to be strongly modified by the fact that most of the hi around bright sources and hence in dense regions is ionized early on .
these effects will likely be important in the use of 21 cm emission maps to carry out tests of cosmic geometry ( see e.g. , nusser 2005 . )
it would be simple enough in future work to look at the simulations in redshift space , such as would be seen with observational data . in carrying out our numerical experiments , we have simulated a range of models , which we expect to have many of the features likely in most scenarios for the reionization of the universe .
it was not possible to be completely general , however , and it is certainly possible to imagine other interesting models and tests of the physics that could be included .
for example , mcquinn ( 2007 ) ran as one of their many models one in which all cells in the simulation were set to the mean density .
this had the effect of changing the amount of structure in the ionization fronts , and of course drastically reducing the recombination rate .
with our postprocessing rt carried out on the hydrogen distribution only , by keeping track of the ionization state but not predicting the temperature we necessarily simplified much of the physics involved in reionization .
as our intention was to capture the broad differences between ris and gravitational structure , we do not regard this as important .
however in future work extending our simulation approach so that it is directly relevant to upcoming observational data , we plan to capture more detail .
for example , the code sphray ( altay 20007 ) which represents an extension of the present method includes different ionized species and explicit temperature evolution .
other work such as ciardi ( 2006 ) and mcquinn ( 2007 ) specifically include the effect of minihalos , sources below the resolution scale of the simulation .
the latter group finds that they have a strong effect on the growth of large bubbles in the late stages of reionization . in the present work
, we have seen from our resolution studies that the autocorrelation function at least is not sensitive to increases in mass resolution and small scale structure , at least to the accuracy and range of models that we have considered . in principle
, the high spatial resolution of our gridless ray tracing approach could allow the effects of lyman - limit systems to be modelled , which become dominant in limiting the photon mean free path when the universe is mostly ionized ( e.g. , miralda - escud 2000 . ) future work on realistic models should also include radiative cooling in the formation of sources and modelling of specific sources and spectra .
quasars and miniquasars have been investigated by including the formation and growth of black holes together with a model for feedback directly into cosmological sph simulations ( di matteo 2007 , sijacki 2007 , pelupessy 2007 ) .
using such models as sources in rt calculations would enable us to investigate how the ris caused by harder sources is different to softer sources , in ways which are not constrained by the simple association of source and halo mass as has been carried out here .
this work was supported by the nasa astrophysics theory program , contract nng 06-gh88 g and nsf grant ast-0507665 .
peterson , j. b. , pen , u. & wu , x. , 2005 , from clark lake to the long wavelength array : bill erickson s radio science asp conference series , vol .
345 , proceedings of the conference held 8 - 11 september , 2004 in santa fe , new mexico , usa . edited by n. kassim , m. perez , m. junor , and p. henning , p.441 i. , blanton m. r. , frieman j. a. , weinberg d. h. , mo h. j. , strauss m. a. , anderson s. f. , annis j. , bahcall n. a. , bernardi m. , briggs j. w. , brinkmann j. , burles s. , carey l. , castander f. j. , connolly a. j. , csabai i. , dalcanton j. j. , dodelson s. , doi m. , eisenstein d. , evans m. l. , finkbeiner d. p. , friedman s. , fukugita m. , gunn j. e. , hennessy g. s. , hindsley r. b. , ivezic z. , kent s. , knapp g. r. , kron r. , kunszt p. , lamb d. q. , leger r. f. , long d. c. , loveday j. , lupton r. h. , mckay t. , meiksin a. , merrelli a. , munn j. a. , narayanan v. , newcomb m. , nichol r. c. , owen r. , peoples j. , pope a. , rockosi c. m. , schlegel d. , schneider d. p. , scoccimarro r. , sheth r. k. , siegmund w. , smee s. , snir y. , stebbins a. , stoughton c. , subbarao m. , szalay a. s. , szapudi i. , tegmark m. , tucker d. l. , uomoto a. , vanden berk d. , vogeley m. s. , waddell p. , yanny b. , york d. g. , 2002 , apj , 571 , 172 , i. , zheng , z. , weinberg , d. h. , frieman , j. a. , berlind , a. a. , blanton , m. r. , scoccimarro , r. , sheth , r. k. , strauss , m. a. , kayo , i. , suto , y. , fukugita , m. , nakamura , o. , bahcall , n. a. , brinkmann , j. , gunn , j. e. , hennessy , g. s. , ivezi , . , knapp , g. r. , loveday , j. , meiksin , a. , schlegel , d. j. , schneider , d. p. , szapudi , i. , tegmark , m. , vogeley , m. s. and york , d. g. , 2005 , apj , 630,1
|
the structures produced during the epoch of reionization by the action of radiation on neutral hydrogen are in principle different to those that arise through gravitational growth of initially small perturbations .
we explore the difference between the two mechanisms using high resolution cosmological radiative transfer .
our computations use a monte carlo code which raytraces directly through sph kernels without a grid , preserving the high spatial resolution of the underlying hydrodynamic simulation .
because the properties of the first sources of radiation are uncertain , we simulate a range of models with different source properties and recombination physics .
we examine the morphology of the neutral hydrogren distribution and the reionization history in these models .
we find that at fixed mean neutral fraction , structures are visually most affected by the existence of a lower limit in source luminosity , then by galaxy mass to light ratio , and are minimally affected by changes in the recombination rate and amplitude of mass fluctuations .
we concentrate on the autocorrelation function of the neutral hydrogen , @xmath0 as a basic quantitive measure of radiation induced structure ( ris . )
all the models we test exhibit a characteristic behaviour , with @xmath1 becoming initially linearly antibiased with respect to the matter correlation function , reaching a minimum bias factor @xmath2 when the universe is @xmath3 ionized . after this
@xmath1 increases rapidly in amplitude , overtaking the matter correlation function .
it keeps a power law shape , but flattens considerably , reaching an asymptotic logarithmic slope of @xmath4 .
the growth rate of @xmath5 fluctuations is exponentially more rapid than gravitational growth over a brief interval of redshift @xmath6 .
[ firstpage ] cosmology : observations large - scale structure of universe
|
the new generation of high - angular resolution cosmic microwave background ( cmb ) ground - based experiments represented by the the atacama cosmology telescope ( act ) @xcite and the south pole telescope ( spt ) @xcite are currently targeting their observations in a common area in the southern sky that will ultimately cover several hundreds to thousands of square degrees . these experiments will provide a blind survey of the oldest light in the universe at wavelengths of @xmath1 mm and angular scales beyond the resolution limits of the wmap and planck satellites . at these arcminute angular scales , temperature fluctuations in the cmb are dominated by secondary effects arising from the formation of large - scale structure in the universe .
one of the strongest effects is the imprint left by galaxy clusters though the sunyaev zeldovich effect ( sze ) @xcite in which cmb photons suffer inverse compton scattering by the hot intracluster gas .
act and spt are designed to detect the sze , through its frequency - dependence : these experiments will measure temperature shifts of the cmb radiation corresponding to a decrement below and an increment above the `` null '' frequency around 220 ghz .
much can be learned about the universe from these surveys .
first , accurate systematics - free maps will allow measurement of the primary power spectrum of temperature fluctuations at all scales on which they are the dominant contribution .
second , these data sets will result in a complete census of massive clusters to arbitrarily large distances , limited only by a minimum cluster mass set largely by the instrumental sensitivity and expected to be several @xmath2 @xcite .
thanks to the relatively clean selection function as well as the redshift independence of the sze , the cluster sample , especially the evolution of the number density of clusters with redshift , will be quite sensitive to the growth of structure in the universe offering a potentially powerful probe of dark energy @xcite .
moreover , the sze data in combination with optical , uv and x - ray observations can teach us a great deal about the detailed physics of cluster atmospheres and galaxy evolution in these dense environments . significant observing time and effort
has been devoted to the development of techniques and the detection of galaxy clusters using large - area optical catalogs and x - ray observations .
several projects have taken advantage of large - area ccd imaging and have developed automated cluster detection schemes to produce large catalogs of clusters of galaxies ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for example ) which target the relative over - abundance of galaxies over a range of redshifts .
similarly , x - ray surveys such as the _ rosat _ all sky survey @xcite produced catalogs with hundreds of galaxy clusters , while pointed x - ray observations have discovered systems up to @xmath3 @xcite . with this article
we inaugurate the southern cosmology survey ( scs ) .
this project , funded by the national science foundation under the partnership in research and education ( pire ) program , is a multiwavelength ( radio , mm - band , optical , uv , and x - ray ) large area survey specifically coordinated with act observations of the southern sky .
the goal of the scs is to maximize the scientific return from the new ground - based cmb experiments and therefore focuses on specific observational studies relevant to this science , such as the selection function of galaxy clusters across wavebands , cluster mass determination , and the establishment a `` gold '' sample of clusters for cosmology and galaxy evolution studies .
here we present results from an @xmath4 deg@xmath5 optical imaging survey of the southern sky that overlaps the common sz survey region .
the purpose of this paper is twofold : ( 1 ) to present the details of our data reduction pipeline and analysis software and ( 2 ) identify new galaxy clusters , constrain their redshifts and masses , and predict their sz signals .
photometric redshifts come from the 4-band imaging data , while our mass estimates are inferred from the optical luminosity ( @xmath6 ) and richness ( 200 ) of the clusters , using relations calibrated by the sloan digital sky survey ( sdss ) . for the eight massive clusters , out of 38 identified ( 37 are new sources ) in the survey area
, we present positions , richness estimates , masses , and predictions for the integrated compton @xmath7-distortion of the sze using empirical power - law relations based on @xmath8-body simulations . throughout this paper
we assume a flat cosmology with @xmath9 km s@xmath10 mpc@xmath10 , @xmath11 and matter density @xmath12 .
our study is based on the optical multi - band analysis of public data from the blanco cosmology survey ( bcs ) .
this is a noao large survey project that was awarded 45 nights over three years on the blanco 4-m telescope at the cerro tololo interamerican observatory ( ctio ) .
the survey aims to image two 50 square - degree patches of the southern sky in four optical bands ( _ griz _ ) using the @xmath13 pixel ( 0.36 deg@xmath5 ) mosaic ii camera in order to attain a sensitivity about an order of magnitude deeper than the sdss imaging .
the targeted areas are centered near declinations of @xmath1455@xmath15 and @xmath1452@xmath15 at right ascensions of 23 hr and 5 hr respectively ; each of these patches lies within a larger common region of the southern sky that both act and spt plan to survey .
the bcs began in 2005 and has completed three years of data taking .
for this paper we have processed and analyzed public data from the first year of the survey using an independent software pipeline developed by us at rutgers university .
the data we present was obtained on 15 nights of observing near the end of november and the beginning of december 2005 and cover an area of @xmath4 deg@xmath5 in the 23 hr region . in table
[ tab : obs ] we show the observing dates , photometric conditions , lunar illumination and observed bands for the 19 tiles that make up the full extent of the observations analyzed here . in the following
we describe the steps followed and tasks performed by the pipeline .
ccccccc 18 nov 2005 & yes & 89.4% & 1.0 & 1.0 & 3.0 & 3.0 + 19 nov 2005 & yes & 0.0% & 4.5 & 4.5 & 1.0 & 1.0 + 20 nov 2005 & no & 0.0% & 8.0 & 8.0 & 0.0 & 0.0 + 22 nov 2005 & no & 0.0% & 0.0 & 0.0 & 0.0 & 0.0 + 24 nov 2005 & yes & 0.0% & 2.0 & 2.0 & 0.0 & 0.0 + 26 nov 2005 & yes & 0.0% & 3.0 & 3.0 & 1.0 & 1.0 + 28 nov 2005 & yes & 0.0% & 0.0 & 0.0 & 2.7 & 2.7 + 30 nov 2005 & yes & 0.0% & 0.0 & 0.0 & 0.3 & 0.3 + 02 dec 2005 & yes & 0.0% & 0.5 & 0.5 & 0.0 & 0.0 + 04 dec 2005 & no & 15.3% & 0.0 & 0.0 & 0.0 & 0.0 + 05 dec 2005 & no & 24.5% & 0.0 & 0.0 & 1.3 & 1.7 + 06 dec 2005 & yes & 35.1% & 0.0 & 0.0 & 1.3 & 1.0 + 08
dec 2005 & yes & 57.8% & 0.0 & 0.0 & 2.3 & 2.7 + 10 dec 2005 & yes & 78.4% & 0.0 & 0.0 & 3.0 & 3.0 + 11 dec 2005 & yes & 86.6% & 0.0 & 0.0 & 3.0 & 3.3 + [ tab : obs ] the rutgers southern cosmology image analysis pipeline is written in python with a scalable object - oriented design based on existing public astronomical software that is aimed at processing a large dataset in a repeatable , stable and semi - automated fashion .
the initial standard image processing steps for each observing night are handled by the iraf / mscred @xcite procedures via the stsci / pyraf interface .
these include : overscan trim , bias correction , ccd cross - talk coefficients corrections as well as dome flat field correction .
the pipeline also executes secondary ccd calibration steps on the science images which include the creation of super sky - flats , fringe patterns for @xmath16 and @xmath17bands and their corresponding correction and removal .
additionally , procedures affecting the cosmetic appearance of the images , such as cosmic ray rejection , removal of saturated star bleed - trails , and generation of bad pixel masks , are automatically performed at this stage .
astrometric re - calibration and wcs plate solution are also handled automatically at this stage on the pre - stack science images using iraf s mscred / mscmatch task by matching several hundred sources within each tile with stars from the us naval observatory catalog .
we achieve good accurate astrometric solutions ( the residual error in matched source positions was typically @xmath18 ) as tested using the overlapping regions between neighboring pointings .
photometric standard star fields were processed together with the normal science images and photometric zero points for each observing night were obtained using a few hundred standards from the southern hemisphere standards stars catalog @xcite . like these authors we use the ab magnitude system .
the survey strategy followed a predetermined observing pattern , which typically consisted of exposures of @xmath19 , @xmath20 , @xmath21 and @xmath22 in the @xmath23 and @xmath17bands respectively with offsets of @xmath24 arcmin ( within each filter ) intended to provide significant overlap between neighboring mosaic ii tiles and fill in the gaps between ccds chips .
we used the overlapping regions between tiles to adjust the photometric zero points of non - photometric nights using matched sources from adjacent photometric tiles .
this ensured a homogeneous photometric calibration across the full survey region with typical variations below @xmath25 mags .
image alignment , stacking and combination as well as catalog generation are performed at a secondary stage by the pipeline using association files , which describes a logical group of exposures and filters , created for each tile .
science images were mosaiced , aligned and median combined using swarp @xcite to a plate scale of @xmath26/pixel .
source detection and photometry measurements for the science catalogs were performed using sextractor @xcite in dual - image mode in which sources were identified on the @xmath16 band images using a @xmath27 detection threshold , while magnitudes were extracted at matching locations from all 4 bands .
-band magnitude limit for 90% completeness . ] as our data set is composed of observations taken over several weeks under varying conditions during the 2005 campaign , we determined , for each tile individually , the @xmath28-band magnitude limit at which the galaxy detection limit was complete to @xmath29 .
to compute this limit we use the fact that the galaxy number counts follow a power - law function , which we fitted in the magnitude range @xmath30 in each tile and extrapolated to obtain the magnitude at which the galaxy number counts dropped by @xmath31 .
we took this as the @xmath29 completeness limit for the tile .
we found variations of roughly 1 magnitude on the limits among the 19 tiles and , in an attempt to understand this , we investigated a possible correlation with lunar illumination at the time of observation .
figure [ fig : illum ] shows a clear trend between the @xmath28-band magnitude limit and the lunar illumination .
we report a mean limit @xmath32 and we set a conservative magnitude limit of @xmath33 for our full catalog .
+ > from the multi - band photometry the pipeline computes photometric redshifts and redshift probability distributions @xmath34 for each object using the @xmath35 isophotal magnitudes , as defined by the i - band detection , and the bpz code @xcite .
we use a magnitude - based empirical prior ( benitez , private communication ) taken from the sdss and hdf - n spectroscopic redshift distributions , which accounts for the tendency of fainter galaxies to be more likely found at higher redshifts ( see fig 4 . from * ? ? ?
* as an example ) . because the area covered by the available noao imaging does not include any publicly available spectroscopic redshift information for @xmath36 ( nasa / ipac extragalactic database , ned ) , we investigated the accuracy of our photometric redshift estimates using ancillary data .
as the ability of bpz to estimate photometric redshifts at fainter magnitudes ( @xmath37 ) from multi - band photometry has been consistently established in the past using filter sets similar to ours here ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for example ) , we focused on the redshift accuracy at @xmath38 . to this end
we extracted @xmath35 photometry from the dr6 sdss for 5000 randomly - selected bright galaxies ( @xmath39 mag ) with reliable spectroscopic redshifts , matching the depth and signal - to - noise ratio of our galaxy sample .
we computed photometric redshifts for the sdss spectroscopic sample in the same way as just described and compared the resulting values to the spectroscopic redshifts .
we found , not surprisingly , that simply employing the maximum - likelihood ( ml ) condition is an ill - suited approach for redshifts below @xmath40 in the absence of a bandpass bluer than 3000 as it largely over - estimates redshifts and produces an unacceptable number of catastrophic outliers .
recently , @xcite have demonstrated how the addition of bluer bands using galex uv imaging greatly improves ml estimates and reduces the need for priors . on the other hand ,
bayesian estimates give results with typical rms errors of @xmath41 and with almost no catastrophic outliers . in fig .
[ fig : bpz ] we show the results of our comparison between ml and bayesian photometric versus spectroscopic redshifts color - coded according to the spectral energy distribution ( sed ) determined by the bpz code .
it is clear from the figure that at higher redshifts the sdss population is dominated by early types as these tend to be the most luminous objects .
we also note ( see fig .
[ fig : bpz ] , lower panel ) that on average our photometric redshifts tend to overpredict the true redshifts .
the mean bias level @xmath42 for galaxies with e / s0s seds is largest around @xmath43 where it is on the order of @xmath44 . in table
[ tab : bpz ] we show the mean bias and standard deviation for three redshift intervals for @xmath45 as well as the standard @xmath46 defined as @xmath47 . in summary we are able to determine the redshifts for early - type galaxies to an accuracy better than 0.1 across the redshift range of the survey .
this is encouraging since early - type galaxies are the predominant population in clusters of galaxies and good photometric redshift determination is essential for successful cluster finding , as we discuss in the next section . in figure
[ fig : dndz ] we show the photometric redshift distribution for all galaxies within our flux completeness limit , @xmath48 , as well as the filter responses for the survey .
our distribution peaks around @xmath49 , which sets a conservative upper limit to the redshift at which we are able to detect optical clusters . mag ) . ]
ccccc @xmath50 & -0.017 & 0.042 & -0.015 & 0.038 + @xmath51 & -0.027 & 0.059 & -0.020 & 0.047 + @xmath52 & -0.002 & 0.070 & -0.001 & 0.048 + [ tab : bpz ] one of the main goals of the current sze experiments is to define a mass - selected sample of galaxy clusters out to large redshifts . at long last
this is beginning to happen @xcite , after a number of successful individual detections of the sze in well - known optical or x - ray clusters ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
if we want to understand the systematics of sze surveys it is crucial to compare with cluster identifications using independent methods . in this section , we describe our effort to select clusters of galaxies from multi - wavelength optical imaging .
there are several methodologies and a plethora of papers describing these techniques ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for example ) but they all rely on the same well - known properties of galaxy clusters : a ) early - type galaxies are the dominant population , b ) cluster galaxies have very similar colors , and display tight color - magnitude relationships across several orders of magnitude in luminosity , and c ) the surface number density of cluster galaxies falls off with distance from the center roughly as a power law @xmath53 .
we search for clusters using a matched filter approach similar to the one described in @xcite and then define membership and estimate richness of the clusters using the maxbcg prescription @xcite .
our cluster finder method folds in the contributions from a cluster spatial profile filter function @xmath54 , a luminosity weight @xmath55 and the bpz redshift probability distribution @xmath34 from each source to generate likelihood density maps ( at pixel positions denoted by @xmath56 ) or a `` filtered '' galaxy catalog @xmath57 over the area covered by the survey as a function of redshift , namely , @xmath58)~l(m_k ) \int_{z-\delta z}^{z+\delta z } p_{\rm bpz}(z_k)dz.\ ] ] specifically we use a profile with the form @xmath59 which is normalized as @xmath60 and where @xmath61 is the typical cluster core radius and @xmath62 is the cutoff limit for the function . in our analysis
we chose @xmath63 kpc and @xmath64 .
we also use a luminosity weight @xmath55 given by @xmath65 where @xmath66 is the apparent magnitude corresponding to @xmath67 .
this function is normalized as @xmath68 where @xmath69 is the flux limit of the sample ( @xmath33 ) , @xmath70 is the number of background galaxies and @xmath71 is the @xcite galaxy luminosity function .
we use the parameters computed from @xcite for the evolving luminosity function of red galaxies , with a faint - end slope @xmath72 and @xmath73 between @xmath74 . for our estimation of
@xmath70 we use the number counts from @xcite .
we generate likelihood density maps with a constant pixel scale of @xmath75 arcmin at @xmath76 intervals between @xmath77 over the surveyed regions . in figure
[ fig : dense ] we show an example of a likelihood density map centered at @xmath78 on which we superpose outlines of the 19 tiles that define the region studied here . and width @xmath76 over the 23hr field .
bright regions in the image represent denser areas .
the red lines represent the area covered by each of the 19 tiles that comprise the area studied . ]
cluster candidates are selected from the peaks of the likelihood density maps . in order to define peaks
consistently we constructed noise maps by randomizing the positions of the input catalog to produce likelihood density maps following exactly the same procedure as described above .
these maps represent the noise floor level above which we desire to detect clusters .
we define our initial list of cluster candidates from @xmath79 peaks in the likelihood density maps , where @xmath80 is defined as the median value in the noise maps .
cluster candidates were checked for multiple detection in different likelihood maps . to avoid duplication we considered a system unique if detected in two adjacent redshift maps and with the same center ( i.e. , within 3@xmath81 ) .
we perfomed simulations to investigate our cluster selection function by estimating the contamination and recovery rates of our cluster finding technique .
the lack of distance information in imaging surveys is the principal source of contamination as fluctuations in the projected two - dimensional galaxy distribution as well as random alignments of poor groups may result in false apparent overdensities .
we explore this issue following the same methodology as used by @xcite and @xcite which rely on generating monte carlo representations of the galaxy sample with an angular two - point correlation function similar to that observed @xcite .
as described in @xcite we implemented the rayleigh - lvy ( rl ) random walk process using mandelbrot s ( 1975 ) elegant fractal prescription to simulate galaxy positions on the sky , such that galaxy pairs are placed in a randomly chosen direction at distance @xmath82 drawn from the distribution : @xmath83 where we chose @xmath84 and @xmath85 to match the observed galaxy distribution of our sample . in practice
we generate simulated distributions by starting from a randomly selected location within the survey boundaries and generate positions following the rl random walk allowing up to seven galaxies to be drawn around this location .
we then select a new center randomly and the process is repeated until we generate the same number of galaxies as in the observed sample .
we then process the rl distribution to generate likelihood maps using the same procedure and parameters as for the real data and use these to investigate the rate of false detections as a function of estimated redshift .
since the rl distributions by construction do not explicitly include clusters , we assess the false positive cluster detection fraction by taking the ratio of detections in the simulations per area unit to the observed number of candidates in the real data .
we find that at low redshift the false positive fraction is zero ( there is virtually no contamination ) , while at redshifts of @xmath86 and @xmath87 the false positive fraction grows to values of 1% and 19% , respectively .
we conclude that false positives are not an important source of spurious detections .
we investigate the selection function for our galaxy sample by simulating galaxy clusters of various richness and shapes at different redshifts and examining their recovery fractions .
specifically we generated clusters with random ellipticities uniformly between @xmath88 using an @xmath89 profile for the galaxy distribution @xcite with @xmath90 and @xmath91 mpc and using the luminosity function for red galaxies from @xcite .
these clusters are inserted 20 at a time in the observed catalogs at random positions and redshifts , but avoiding the locations where clusters were detected . in total
we generate 10,000 simulated clusters with richness values , @xmath92 , of 15 , 20 , 25 , 35 , 50 , 80 and 120 galaxies uniformly distributed in redshift .
values we give for the detected clusters . ]
we process each realization using the same parameters as for the observed data and record the number of clusters recovered as a function of redshift and galaxy richness . in figure
[ fig : sims ] we show the results of this exercise where we plot the recovery fraction as a function of redshift for the seven cluster richnesses simulated .
we conclude that for the rich clusters @xmath93 we are always nearly complete ( @xmath94 ) for @xmath95 while for poorer clusters we only detect at best @xmath96 around @xmath97 .
and @xmath98 galaxies and a profile with slope @xmath90 and core radius @xmath99 mpc.,width=288 ]
one of our main drivers in searching for clusters at optical wavelengths is to correlate them with sz detections in the new blind sz surveys .
the signal to be detected in the mm - band experiments ( i.e. , the @xmath7-distortion due to inverse compton scattering ) is related to the number of hot electrons in the intracluster medium , and simulations have shown ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for example ) that the sze signal is closely linked to cluster mass . our analysis of the optical survey has provided positions , redshifts , and fluxes of galaxies , from which we infer the underlying cluster mass using scaling relations established by the sdss , where cluster masses were determined from weak lensing .
the current state - of - the - art mass tracers for clusters of galaxies using optically observed parameters @xcite have been extracted from a sample of around @xmath100 optically - selected clusters from the sdss maxbcg catalog @xcite . in this paper , we apply these scaling relations to our cluster sample and obtain mass estimates from which we additionally predict sz distortions .
to be fully consistent with previous analyses , we define membership and all other relevant cluster observables following the same method as @xcite and @xcite .
we begin by examining each candidate - cluster peak in the density maps and select the brightest elliptical galaxy in the cluster ( bcg ) , which is taken to be the initial center and redshift @xmath101 of the system .
we then use galaxies defined as e or e / s0s ( i.e. , bpz sed types 0 and 1 only ) within a projected radius of @xmath102 mpc and redshift interval @xmath103 to obtain a local color - magnitude relation ( cmr ) for each color combination , @xmath104 , @xmath105 and @xmath106 , and the cluster mean redshift , @xmath107 , for all cluster members , using a @xmath108 median sigma - clipping algorithm .
we use these to determine @xmath109 , the number of galaxies within @xmath110mpc of the cluster center .
formally , we compute @xmath111 by including those galaxies within a projected 1@xmath112mpc from the cluster center that satisfy three conditions : ( a ) the galaxy must have the sed of an e , e / s0 according to bpz , ( b ) it must have the appropriate color to be a cluster member ( i.e. , colors within @xmath108 of the local cmr for all color combinations ) and ( c ) it must have the right luminosity ( i.e. , dimmer than the bcg and brighter than @xmath113 , where we use the corresponding absolute magnitude @xmath114 from @xcite redshifted with the elliptical sed template from bpz ) .
we designated cluster members according to the estimated cluster size @xmath115 , defined as the radius at which the cluster galaxy density is @xmath116 times the mean space density of galaxies in the present universe .
we estimated the scaled radius @xmath115 using the empirical relation from @xcite , @xmath117mpc which is derived from the sdss and we assume it holds beyond @xmath118 for our higher redshift clusters .
in our analysis we use 200 , 200 , and to scale cluster optical parameters with mass , following @xcite .
the cluster richness , 200 , is the number of e / s0 galaxies within @xmath115 with colors and luminosities that satisfy conditions ( b ) and ( c ) above .
similarly , @xmath6 is the total rest - frame integrated @xmath119-band luminosity of all member galaxies included in 200 in units of @xmath120 and @xmath121 is the rest - frame @xmath119-band luminosity of the bcg . in order to have reliable estimates of @xmath122
it is necessary to determine the galaxy background contamination and implement an appropriate background subtraction method .
the lack of spectroscopic redshifts in our sample only allowed a statistical removal of unrelated field galaxies with similar colors and redshifts that were projected along the line of sight to each cluster .
we assumed that the presence of a cluster at some redshift is independent of the field population seen in projection .
therefore we estimate the surface number density of ellipticals in an annulus surrounding the cluster ( within @xmath123 ) with @xmath124 and the same colors as the cluster members .
we measure this background contribution around the outskirts of each cluster and obtain a corrected value @xmath125 which is used to compute @xmath115 and then corresponding values of 200 and @xmath6 .
the magnitude of the correction ranges between @xmath126% .
we will refer to the corrected values hereafter .
moreover as our analysis is based on a magnitude - limited sample it is worth considering the fraction of lower luminosity galaxies that will fall below our magnitude limit ( @xmath33 ) at higher redshifts . as in menanteau & hughes ( 2009 ) , if we make the assumption that the cluster population is like that of the five clusters at @xmath127 in our sample ( see table [ tab : data2 ] ) and @xmath67 evolves passively , then we can compute the fraction of @xmath6 missed for clusters at higher redshifts .
we estimate that we are missing 4% , 13% , 29% and 38% of the cluster luminosity at @xmath128 and @xmath129 respectively . given the uncertainty in this correction factor , we do not include it in our quoted luminosity values for the higher redshift clusters .
this means our cluster masses , @xmath130 ) , are underestimated by roughly these factors .
the area covered by the bcs in the 23hr region is a virtual desert in terms of known clusters and spectroscopic redshifts for galaxies with @xmath36 .
we found one catalogued x - ray selected cluster from the 160 deg@xmath5 _ rosat _ survey @xcite : rxj2325.6@xmath145443 at @xmath131 with an x - ray flux of @xmath132 erg @xmath133 s@xmath10 in the 0.52 kev band .
our cluster finding algorithm easily recovered this cluster and produced a photometric redshift estimate of @xmath134 .
figure [ fig : c0 ] shows the @xmath135 color composite optical image of the cluster as well as the color magnitude diagrams for cluster members . using the techniques described below we estimate the mass of rxj2325.6@xmath145443 to be @xmath136 which is just below the detectability limit of act and therefore will not be included in our sze predictions . from the @xmath137-@xmath138 @xcite and @xmath139-@xmath138 @xcite relations we estimate a temperature of @xmath140 kev and an x - ray flux of @xmath141 erg @xmath133 s@xmath10 ( 0.52 kev band ) which is in rough agreement ( factor of 2 ) with the published value . considering the entire sample of 38 clusters ,
rxj2325.6@xmath145443 is the closest and , based on our estimated masses from the optical properties and the @xmath137-@xmath138 and @xmath139-@xmath138 relations , has the highest predicted x - ray flux of the sample .
still , even this value is below the x - ray detection threshold of the _ rosat _ all sky survey ( rass ) and so , as expected , we find that none of our new optical clusters are significant rass x - ray sources
. observed in the @xmath4 deg@xmath5 region analyzed here .
the thin and thick lines show the predicted number of clusters in the region for cluster masses greater than @xmath142 and @xmath143 respectively , using the wmap 5-year best - fit cosmological model and the @xcite halo mass function . ]
rrrrrrrrrrr @xmath144 & 1.54 & 7.77 & 0.41 & 0.67 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + @xmath151 & 1.64 & 7.92 & 0.43 & 0.66 & @xmath152 & @xmath153 & @xmath154 & @xmath155 & @xmath156 & @xmath157 + [ tab : pars ] in this section we use the mass - richness relations based on 1mpc to weigh our new optical clusters .
both @xcite and @xcite found that the luminosity - mass and richness - mass relations were well described by power - law functions and they measured the normalizations and slopes in these relations using @xmath158 minimization .
their values are in broad agreement , but we will use the fits provided by @xcite since they restrict their fits to clusters with @xmath159 and give results for two redshift bins : @xmath160 and @xmath161 .
it is important to note that our clusters go well beyond @xmath97 and that we extrapolate the relation in the last redshift bin for clusters with @xmath162 .
we investigated the two fitting functions based on @xmath6 and 200 , ( see section 5.2.1 from @xcite for full details ) , which are described as : @xmath163 @xmath164 where @xmath137 is the mass observational equivalent of @xmath165 is the halo mass enclosed within @xmath115 , defined as a radius of spherical volume within which the mean density is 200 times the critical density . ] in units of @xmath166 , @xmath6 is in units of @xmath120 and the dependence is normalized by its mean value .
this is also described by a power - law function for a given value of 200 and 200 : @xmath167 @xmath168 the published best - fitting parameters for @xmath169 , @xmath170 and @xmath171 in eqs .
( [ eq : m1 ] ) and ( [ eq : m2 ] ) as well as the values of @xmath172 , @xmath173 for eqs .
( [ eq : l1 ] ) and ( [ eq : l2 ] ) are shown in table [ tab : pars ] . the combination of equations ( [ eq : m1 ] ) , ( [ eq : m2 ] ) and ( [ eq : l1 ] ) , ( [ eq : l2 ] ) for 200 and 200 respectively enable us to obtain mass estimates for any cluster with @xmath174 . the left panel of fig .
[ fig:3 m ] compares the masses obtained using @xmath175 and @xmath6 for our 38 optical clusters .
the solid line denotes equality between the estimates while the two dashed lines show a factor of 2 range .
most clusters fall within this range , establishing a lower bound on our mass error .
the two most significant outliers correspond to nearby clusters ( see the right panel of fig .
[ fig:3 m ] ) for which @xmath175 is evidently overpredicting the mass compared to @xmath6 . in one of these cases ( which is rxj2325.6@xmath145443 )
, the mass inferred by @xmath175 grossly overpredicts ( by more than an order of magnitude ) the estimated x - ray flux .
the natural concentration of luminosities in clusters ( i.e. , that more luminous galaxies dominate the central regions of clusters ) suggests that the mass estimate derived from @xmath6 should be more robust than that from @xmath175 as a function of redshift . for these reasons we use the cluster mass estimate derived from @xmath6 in predicting the sze signal .
hereafter we refer to this mass as @xmath176 .
table [ tab : data ] displays the 8 clusters with @xmath177 in the 8 square - degree sky area covered by our analysis .
these clusters also have @xmath178 above the same mass limit . in figure
[ fig : nclusters ] we show the cumulative number distribution of these clusters as a function of redshift , compared with expectations from simulations ( h. trac , private communication ) , using the mass function of dark matter haloes from @xcite and wmap5 cosmology @xcite .
the heavy solid curve shows expectations for a mass limit of @xmath179 , while the light solid curve indicates a mass limit of @xmath180 .
this figure demonstrates that the number of clusters we observe is consistent with a mass limit in the range of @xmath181 . in table
[ tab : data2 ] we display the properties for the remaining clusters with @xmath182 .
scaling relations between the integrated thermal sze signal and cluster mass have emerged from current n - body plus hydrodynamic simulations of galaxy clusters . in this section
we use these relations and our optically - derived mass estimates , @xmath183 , to predict the sze signal to be observed by act and spt when these experiments survey this sky region .
the sze signal consists of small distortions to the cmb spectrum originating from inverse compton scattering by electrons in the hot plasma of clusters of galaxies @xcite .
here we consider the thermal sz flux @xmath184 , defined as the integrated compton @xmath185parameter , @xmath186 where @xmath187 and @xmath188 are the number density and temperature of hot electrons in the cluster , @xmath189 is electron rest mass , @xmath190 is the speed of light , @xmath191 is the thompson scattering cross section and @xmath192 is the angular diameter distance .
the projected area , @xmath193 , and solid angle , @xmath194 , are related by the angular diameter distance as @xmath195 .
self - similar scaling relations @xcite predict that the virialized mass @xmath137 in clusters scales with the gas temperature as @xmath196 where @xmath197 for a flat cosmology . if clusters were isothermal , we would expect their sze signal to scale like @xmath198 and therefore the self - similar sz flux - mass scaling relation should have the shape @xmath199 where @xmath200 is the cluster mass fraction .
however , clusters are not always isothermal or in hydrostatic equilibrium and physical processes like star formation and feedback will also contribute to deviations from self - similarity . prompted by the upcoming sz surveys ,
several studies have characterized in detail the @xmath201 scaling relation using simple power - law fits to cosmological n - body simulations . here
we will use the recent @xmath201 fits from @xcite , who included gas simulations employing small - scale cluster physics such as star formation and feedback into a large cosmological n - body simulation .
> from their catalog of @xmath202 simulated clusters with @xmath203 , @xcite fit the relation @xmath204 where @xmath205 and @xmath206 are the projected sz compton @xmath7-parameter and mass respectively in a disk of radius @xmath115 .
we use the best fit values @xmath207 and @xmath208 for all clusters regardless of redshift , as the redshift dependence of the fits is very weak ( see table 2 from * ? ? ?
the power - law index that they report is slightly steeper than ones quoted by some previous hydrodynamic simulations ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* for example ) , which @xcite attribute to their more realistic feedback prescription ( i.e. , including the effects of agn and supernovae ) .
we use the power - law model from equation [ eq : y200 ] and the optical @xmath176 mass estimates for our 8 new massive clusters to predict the integrated @xmath205 signal to be observed by act and spt .
table [ tab : data ] gives the results in physical units of mpc@xmath209 and observable ones of arcmin@xmath5 .
rrrrrrrrr scso j233430.2543647.5 & 0.35 & @xmath210 & @xmath211 & @xmath212 & @xmath213 & @xmath214 & @xmath215 & @xmath216 + scso j233556.8560602.3 & 0.52 & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 & @xmath222 & @xmath223 + scso j233425.6542718.0 & 0.53 & @xmath224 & @xmath225 & @xmath226 & @xmath227 & @xmath227 & @xmath228 & @xmath229 + scso j232211.0561847.4 & 0.61 & @xmath230 & @xmath231 & @xmath232 & @xmath220 & @xmath233 & @xmath234 & @xmath235 + scso j233731.7560427.9 & 0.61 & @xmath236 & @xmath237 & @xmath238 & @xmath239 & @xmath239 & @xmath240 & @xmath241 + scso j234012.6541907.2 & 0.62 & @xmath242 & @xmath243 & @xmath244 & @xmath245 & @xmath246 & @xmath247 & @xmath248 + scso j234004.9544444.8 & 0.66 & @xmath249 & @xmath250 & @xmath251 & @xmath252 & @xmath253 & @xmath254 & @xmath255 + scso j232829.7544255.4 & 0.68 & @xmath256 & @xmath257 & @xmath258 & @xmath259 & @xmath260 & @xmath261 & @xmath262 + [ tab : data ] rrrrrrr scso j232540.2544430.9 & 0.10 & @xmath263 & @xmath264 & @xmath265 & @xmath266 & @xmath267 + scso j232230.9541608.3 & 0.12 & @xmath268 & @xmath269 & @xmath270 & @xmath271 & @xmath272 + scso j233000.4543707.7 & 0.14 & @xmath273 & @xmath274 & @xmath275 & @xmath252 & @xmath276 + scso j232419.6552548.9 & 0.18 & @xmath277 & @xmath278 & @xmath279 & @xmath280 & @xmath276 + scso j233106.9555119.5 & 0.19 & @xmath281 & @xmath282 & @xmath283 & @xmath284 & @xmath285 + scso j233252.9561454.1 & 0.20 & @xmath286 & @xmath287 & @xmath288 & @xmath289 & @xmath276 + scso j233215.5544211.6 & 0.20 & @xmath290 & @xmath291 & @xmath292 & @xmath293 & @xmath294 + scso j233037.1554338.8 & 0.20 & @xmath295 & @xmath296 & @xmath297 & @xmath298 & @xmath299 + scso j233550.6552820.4 & 0.22 & @xmath300 & @xmath301 & @xmath302 & @xmath303 & @xmath304 + scso j232200.4544459.7 & 0.27 & @xmath305 & @xmath306 & @xmath307 & @xmath308 & @xmath294 + scso j233522.6553237.0 & 0.29 & @xmath217 & @xmath309 & @xmath310 & @xmath311 & @xmath289 + scso j233807.5560304.9 & 0.30 & @xmath312 & @xmath313 & @xmath314 & @xmath253 & @xmath315 + scso j232956.0560808.3 & 0.32 & @xmath316 & @xmath317 & @xmath318 & @xmath308 & @xmath319 + scso j232839.5551353.8 & 0.32 & @xmath320 & @xmath321 & @xmath322 & @xmath294 & @xmath323 + scso j232633.6550111.5 & 0.32 & @xmath324 & @xmath325 & @xmath326 & @xmath311 & @xmath327 + scso
j233753.8561147.6 & 0.33 & @xmath328 & @xmath329 & @xmath330 & @xmath253 & @xmath331 + scso j232156.4541428.8 & 0.33 & @xmath332 & @xmath333 & @xmath334 & @xmath267 & @xmath276 + scso j233003.6541426.7 & 0.33 & @xmath335 & @xmath336 & @xmath337 & @xmath331 & @xmath338 + scso j233231.4540135.8 & 0.33 & @xmath339 & @xmath340 & @xmath341 & @xmath342 & @xmath294 + scso j233110.6555213.5 & 0.39 & @xmath343 & @xmath344 & @xmath345 & @xmath289 & @xmath323 + scso j233618.3555440.3 & 0.49 & @xmath346 & @xmath347 & @xmath348 & @xmath349 & @xmath350 + scso j233706.3541903.8 & 0.51 & @xmath351 & @xmath352 & @xmath353 & @xmath315 & @xmath272 + scso j233816.9555331.1 & 0.52 & @xmath354 & @xmath354 & @xmath355 & @xmath349 & @xmath356 + scso j232619.8552308.8 & 0.52 & @xmath357 & @xmath358 & @xmath359 & @xmath267 & @xmath276 + scso j232215.9555045.6 & 0.56 & @xmath360 & @xmath361 & @xmath362 & @xmath284 & @xmath363 + scso j232247.6541110.1 & 0.57 & @xmath364 & @xmath365 & @xmath366 & @xmath276 & @xmath276 + scso j232342.3551915.1 & 0.67 & @xmath367 & @xmath368 & @xmath369 & @xmath319 & @xmath298 + scso j233403.7555250.7 & 0.71 & @xmath370 & @xmath371 & @xmath372 & @xmath323 & @xmath338 + scso j233951.1551331.3 & 0.73 & @xmath373 & @xmath374 & @xmath375 & @xmath284 & @xmath356 + scso j233720.2562115.1 & 0.75 & @xmath376 & @xmath377 & @xmath378 & @xmath379 & @xmath380 + [ tab : data2 ]
in this article we have laid out the techniques and methods for our analysis of a large multi - band optical survey with the blanco telescope and mosaic - ii instrument under the aegis of the southern cosmology survey .
we have obtained sub - arcsecond astrometric precision and sufficient photometric accuracy for the estimation of redshifts to @xmath381 .
forty - two optical cluster candidates were identified from an area of the sky covering @xmath3828 deg@xmath5 ; the richness and integrated galaxy luminosity of the clusters were measured . based on correlations between these optical observables and cluster mass as established by sdss cluster surveys @xcite we provide mass estimates ( in addition to positions and redshifts ) for 8 new clusters whose inferred masses lie above @xmath383 .
these clusters are all likely to be detected by act and spt if these experiments reach their expected sensitivity levels .
although the uncertainties on the estimated mass and inferred sze signal are large ( factors of 2 or so ) , the accuracy of our cluster positions and redshifts are quite good and typical for 4-band imaging survey data .
moreover , it is worth noting that the @xmath201 relation varies quite a bit from simulation to simulation ( differences of a factor of 2 in @xmath137 at fixed @xmath184 ) , so our predicted masses are well within the uncertainty range of the latest theoretical predictions . in an effort to reduce the mass errors we have begun to estimate weak lensing masses from the current blanco data @xcite .
we also note that significant additional areas of the bcs are now publicly available and will be presented in future publications .
the strength of the scs is its multi - wavelength aspect and large sky area coverage .
the 23hr region of the sky analyzed here has now been surveyed in the uv by galex and in the x - ray band by _ xmm - newton _ ; over the next several months as these data become available , we will incorporate them into the scs .
correlation analyses of these multi - wavelength data should allow us to reduce the large errors on inferred cluster masses and study the cluster selection biases across wavebands .
secure confirmation of all candidates , as well as additional mass constraints , rest on follow - up optical spectroscopy , which we are pursuing at the southern african large telescope and elsewhere .
finally , once act and spt data become available , the rich optical data set analyzed here will be a valuable source for understanding and quantifying the impact of large scale structure on secondary anisotropies in the cmb .
fm would like to thank txitxo benitez for discussions and the use of bpz priors and templates .
jph would like to thank kent patterson for help during the early phases of this project and the aresty center for undergraduate research at rutgers for financial support .
ak has been supported by nsf grant ast-0546035 .
li thanks centro de astrofisica fondap .
we thank armin rest and frank valdes for helpful discussion on the image analysis and we also acknowledge david spergel for several important insights and suggestions .
we are thankful to wayne barkhouse , mark brodwin , will high , yen - ting lin and the entire bcs team for the careful planning and execution of the noao observations on which this paper is based .
financial support was provided by the nasa ltsa program ( grant number nag5 - 11714 ) and the national science foundation under the pire program ( award number oise-0530095 ) .
|
we present first results from the southern cosmology survey , a new multiwavelength survey of the southern sky coordinated with the atacama cosmology telescope ( act ) , a recently commissioned ground - based mm - band cosmic microwave background experiment .
this article presents a full analysis of archival optical multi - band imaging data covering an 8 square degree region near right ascension 23 hours and declination -55 degrees , obtained by the blanco 4-m telescope and mosaic - ii camera in late 2005 .
we describe the pipeline we have developed to process this large data volume , obtain accurate photometric redshifts , and detect optical clusters . our cluster finding process uses the combination of a matched spatial filter , photometric redshift probability distributions and richness estimation .
we present photometric redshifts , richness estimates , luminosities , and masses for 8 new optically - selected clusters with mass greater than @xmath0 at redshifts out to 0.7 .
we also present estimates for the expected sunyaev - zeldovich effect ( sze ) signal from these clusters as specific predictions for upcoming observations by act , the south pole telescope and atacama pathfinder experiment .
|
up to now the standard model ( sm ) managed to pass all the experimental challenges unscathed , providing an overall good description of particle physics up to the energy scales probed in experiments so far , namely hundreds of gev .
the higgs boson , the last missing building block of the sm , is being searched for at the lhc experiments atlas and cms , which will be able to find or exclude it . in spite of the phenomenological success , however , the sm is not satisfactory for several theoretical reasons , including the instability of the fundamental scale of weak interactions , the fermi scale , against radiative corrections , the absence of a dark matter candidate , an amount of cp violation too small to account for the matter - antimatter asymmetry in the universe , the lack of an explanation for the origin of flavor and cp violation and the non - unified description of the interactions , with gravity completely missing . for these reasons ,
the sm is regarded as a low - energy theory bound to fail at some energy scale larger than the fermi scale , where new physics ( np ) effects become important .
the search for these effects beyond the sm is the main goal of particle physics in the next decades , both at present and future experimental facilities .
the most straightforward way to search for np is producing and observing new particles in colliders . to this end
, the key ingredient is the available center - of - mass energy : the higher the energy , the heavier the particles one can produce , thus probing higher np scales . pushing forward the energy frontier then means building colliders with higher and higher center - of - mass energy .
the problem with direct searches is that their success depends on the unknown production thresholds of the new particles . some of the sm problems , however , have solutions , such as the natural stabilization of the fermi scale or the wimps ( weakly interacting massive particles ) as dark matter candidates , which point to np at the tev scale giving to the lhc good chances to discover it . a complementary way to reveal np is measuring the effect of virtual heavy particle exchange in processes involving only sm particles as external states . for this kind of searches , called indirect ,
the production threshold is not an issue .
since quantum effects become typically smaller as the mass of the virtual particles increases , higher np scales are probed by increasing the precision of the measurements , controlling at same time the sm contributions with an accuracy sufficient to disentangle genuine np effects from sm uncertainties .
progress at the intensity frontier entails building experimental facilities which can deliver high - intensity beams at energies where one can make a set of precision measurements not limited by systematic and theoretical uncertainties .
flavor physics is the best candidate as a tool for indirect np searches for several reasons .
flavor changing neutral currents ( fcnc ) , neutral meson - antimeson mixing and cp violation occur at the loop level in the sm and therefore are potentially subject to large virtual corrections from new heavy particles .
in addition , flavor violation in the sm is governed by weak interactions and , in the case of quarks , suppressed by small mixing angles .
these features are not necessarily shared by np which could then produce very large effects . on quite general grounds
, the indirect np search in flavor processes explores a parameter space including the np scale and the np flavor- and cp - violating couplings . in specific models ,
these are related to fundamental parameters such as masses and couplings of new particles .
in particular , an observable np effect could be generated by small np scales and/or large couplings .
indeed the inclusion in the sm of generic np flavor - violating terms with @xmath0 couplings is known to violate present experimental constraints unless the np scale is pushed to large values , from @xmath1 to @xmath2 tev depending on the flavor sector , with the larger scales required by kaon physics @xcite the difference between the np scale emerging from flavor physics when flavor couplings are not suppressed and the tev scale suggested by the solution of the sm problems mentioned above is usually referred to as the np flavor problem : np models with new particles at the tev scale need a mechanism to suppress np contributions to fcnc and cp - violating processes . on the other hand , this clearly indicates that flavor physics has either the potential to push the explored np scales well beyond the tev region or , if the np scale is indeed close to @xmath3 tev , that the np flavor structure is non - trivial , calling for the experimental determination of new flavor - violating couplings in order to elucidate the suppression mechanism at work . in spite of its potential , so far flavor physics has not provided a clear np signal , although deviations with low significance are present here and there .
this failure , however , should not be considered discouraging .
broadly speaking , flavor physics probed and excluded only the region of small masses / large np flavor couplings . on the other hand
, we should not forget that flavor physics already proved itself effective in predicting the existence and in some cases the mass of most of the heavy particles which are nowadays part of the sm .
the list of these successes includes the prediction of the existence of the charm quark from the suppression of the rate @xmath4 in the early seventies @xcite , the prediction of the existence of the third generation from the measurement of cp violation in @xmath5@xmath6 mixing , again in the seventies @xcite and the indication of a heavy mass for the top quark from the measurements of the semileptonic @xmath7 decay rates and the @xmath7@xmath8 mass difference in the late eighties @xcite .
the lesson to be learned from this short historical excursus is that flavor physics can indeed provide very valuable information on unknown heavy particles and it is therefore worth pushing experiments and theory to higher precision in order to probe heavier particles and/or smaller flavor couplings . in this paper
, we give an overview of the physics opportunities opened by the next - generation flavor experiments , mainly the two approved super @xmath7-factories superb in italy and belle - ii / superkekb in japan .
these are actually super flavor - factories , expected to produce precision measurements in @xmath7 , @xmath9 and @xmath10 physics , increasing by two orders of magnitude the luminosity of their predecessors pep - ii and kekb .
the two super - flavor factories are planned to start to take data in the second part of this decade and to collect more than 50 ab@xmath11 integrated luminosity by around 2020 .
it has to be noted that by @xmath12 the lhcb experiment at lhc will collect the design integrated luminosity of 10 fb@xmath11 and an upgrade ( after 2020 ) is actually under study to considerably increase the collected data sample up to 50 fb@xmath11 .
we begin with a brief primer on flavor and cp violation in the sm , introducing also the some common flavor observables , in section [ sec : primer ] .
we then summarize in section [ sec : status ] the present knowledge of the sm flavor sector and point out some tensions which could be the harbingers of physics beyond the sm .
the most interesting measurements in the physics programme of the next - generation flavor factories are reviewed in section [ sec : opportunities ] , giving estimates of the expected statistical and systematic errors .
the selection of these measurements is done taking into account the physics program which should have been already accomplished by lhcb . in section [ sec : bsm ] , we discuss the impact of these measurements on a selection of np models in different scenarios , giving few examples of the potential implications of precision flavor physics .
finally , our conclusions are presented in section [ sec : conclusions ] .
we are particularly indebted with the authors of refs .
@xcite , as many arguments and results presented in this review are based on their studies .
in this section we briefly review the sm flavor structure , focusing mainly on the quark sector and collecting some basic formulae used throughout the review .
the sm matter content is organized as follows : @xmath13 where the transformation properties under the gauge groups @xmath14 are indicated .
the index @xmath15 spans the flavor space . in the sm with three generations
, one has @xmath16 the subscripts @xmath17 and @xmath18 denote the left - handed and right - handed components of the fields .
the sm lagrangian can be written as @xmath19 in the absence of @xmath20 , the sm lagrangian is invariant under the global symmetry @xmath21 corresponding to the separate unitary rotation of each matter multiplet in flavor space .
@xmath20 contains the yukawa interactions between the higgs doublet and the matter fields : @xmath22 where @xmath23 is the higgs doublet transforming as @xmath24 and @xmath25 .
the yukawa couplings @xmath26 are @xmath27 complex matrices in flavor space ( the sum over the flavor indices @xmath28 is understood ) .
the presence of non - vanishing yukawa couplings breaks this large group down to few @xmath29 factors , @xmath30 { } u(1)_b \otimes u(1)_e \otimes u(1)_\mu \otimes u(1)_\tau\,,\ ] ] which account for the conservation of the barion and the three lepton numbers .
they can be diagonalized using the singular value decomposition , giving @xmath31 where @xmath32 are four unitary matrices and @xmath33 are diagonal matrices with real non - negative matrix elements .
therefore it is always possible to choose a flavor basis for the quark fields where , for example , @xmath34 @xmath35 is the cabibbo - kobayashi - maskawa matrix @xcite which contains all flavor non - diagonal couplings and the only cp - violating phase of the sm lagrangian ( notice however that the flavor symmetry is broken also by @xmath33 as long as the diagonal entries are different ) .
clearly @xmath36 originates from the misalignment in flavor space of the components of the left - handed quark doublet .
after the electroweak symmetry breaking ( ewsb ) , the yukawa interactions generate the fermion masses . replacing @xmath37 in eq .
( [ eq : y ] ) , one obtains @xmath38 one can go to the mass eigenstate basis , where @xmath39 and @xmath40 are both diagonal , applying unitary flavor transformations to the quark fields , namely making the replacement @xmath41 where @xmath32 are the four unitary matrices defined in eq .
( [ eq : svd ] ) .
the only leftover of these transformations in the sm lagrangian is the ckm matrix appearing in the charged current weak interactions @xmath42 where @xmath43 is the @xmath44 coupling constant .
let us have a closer look at the ckm matrix @xmath36 .
a generic @xmath27 unitary matrix can be represented using @xmath45 euler angles and @xmath46 phases , but @xmath47 of the latter can be removed by re - phasing the quark fields .
therefore @xmath36 can be parameterized by @xmath45 angles @xmath48 $ ] and @xmath3 phase @xmath49 .
notice that the choice of the paramters is not unique . in the parameterization used by the pdg @xcite
, the ckm matrix reads @xmath50 where @xmath51 and @xmath52 .
the unitarity of @xmath36 implies @xmath61 conditions involving products of rows ( or columns ) of the form @xmath62 .
the off - diagonal relations are called triangular as they define triangles in the complex plane .
remarkably , the area of all these triangles is a constant equal to @xmath63 and thus a measure of cp violation in the sm .
one of them , of particular phenomenological interest , is referred to as the unitarity triangle ( ut ) : @xmath64 the ut can be rewritten in the normalized form @xmath65 with @xmath66 being two sides and two angles as sketched in fig .
[ fig : ut ] .
the third side is the unity vector , while the third angle is @xmath67 .
similarly , it is useful to define the angle @xmath68 from the triangle @xmath69 , as the phase of the @xmath70@xmath71 mixing amplitude is @xmath72 in the phase convention of eq .
( [ eq : v ] ) .
the ut sides and angles are observables and we discuss their present determination from @xmath5- and @xmath7-meson physics in section [ sec : status ] . given the definition in eq .
( [ eq : ut ] ) , all the information related to the ut is encoded in one complex number @xmath73 corresponding to the coordinates @xmath74 in the complex plane of the only non - trivial apex of the ut .
it is worth mentioning another popular ckm parameterization introduced by wolfenstein @xcite which allows to write an expansion of the ckm matrix in terms of a small parameter @xmath75 : @xmath76 this parameterization makes explicit the hierarchy of ckm matrix elements , showing that quark flavor - changing transitions are suppressed in the sm .
the exact relations between the pdg and the wolfenstein parameterizations are @xmath77 notice that @xmath78 ( the sine of the cabibbo angle @xmath79 ) is indeed a good expansion parameter .
the relation between the ut apex coordinates and the wolfenstein parameters is @xmath80 showing that @xmath81 and @xmath82 at the lowest order in @xmath75 . for massless neutrinos ,
the sm lepton sector is flavor diagonal and cp conserving , as @xmath83 can be chosen real and diagonal .
once the sm is trivially extended to include right - handed neutrinos and account for neutrino dirac masses , introducing an additional yukawa matrix @xmath84 , the formalism for lepton flavor and cp violation is similar to the quark one .
the lepton mixing matrix , called the pontecorvo - maki - nakagawa - sakata matrix , @xcite , is parameterized much as the ckm matrix .
the only difference is that the possibility of having majorana mass terms for neutrinos allows for two additional cp - violating phases . in spite of these formal similarities ,
the flavor phenomenology of the lepton sector , in particular of neutrinos , is rather different from the quark one and its discussion goes beyond the scope of this primer .
we just remind that , given the smallness of neutrino masses , charged lepton flavor violation is negligible in the sm .
most of flavor physics deals with flavor - changing transitions involving mesons .
as far as decays are concerned , cp violation appears as a difference between the rates of a decay and its cp - conjugate , accounted for by the direct cp asymmetry @xmath85 where @xmath86 is the decaying meson , @xmath87 is the final state , @xmath88 and @xmath89 are the cp - conjugate states and @xmath90 and @xmath91 are the two decay amplitudes .
charged mesons can only violate cp in the decay .
on the other hand , neutral mesons , with the exception of the pion , are subject to the mixing phenomenon , _
i.e. _ the mass eigenstates are a superposition of the flavor ones . in this case
, cp violation can also occur in the mixing itself and in the interference between mixing and decay , giving additional opportunities to observe it .
we briefly summarize in the following the main formulae related to meson mixing and cp violation .
the time evolution of a system of unstable neutral meson - antimeson states @xmath86@xmath88 can be described by a @xmath92 non - hermitean matrix hamiltonian @xmath93 @xmath94 @xmath93 can be decomposed using the two hermitean matrices @xmath95 and @xmath96 representing its dispersive and absorbitive part respectively . in particular , assuming cpt invariance , one can write @xmath97 the eigenstates of @xmath93 , denoted as @xmath98 , can be written as @xmath99 where @xmath17 ( @xmath23 ) corresponds to the upper ( lower ) sign .
the corresponding eigenvalues are @xmath100 giving the mass and width differences between the two eigenstates @xmath101 these observables , or the corresponding dimensionless variables @xmath102 and @xmath103 ( @xmath104 is the average lifetime ) , characterize the mixing .
in addition , the absolute value @xmath105 is also observable .
the deviation of @xmath105 from one is a measure of cp violation in the mixing , as shown by the expression of the semileptonic cp asymmetry . for @xmath106 mesons ,
the leptons in the final state are cp conjugated . ]
@xmath107 depending on the meson , these formulae can be approximated in different ways .
for @xmath7 mesons , where @xmath108 , one finds @xmath109 for the @xmath9 meson such simplification is not possible and the full expressions have to be used . besides , eqs .
( [ eq : dmdg ] ) only give the short - distance part of the corresponding observables , which contain large long - distance contributions . for kaons
, the eigenstates are usually parameterized in terms of @xmath110 and different approximations hold , see for example ref .
@xcite for details on kaon mixing .
finally , the third manifestation of cp violation in meson decays is through the interference between mixing and decay . the key observable in this case
is the time - dependent cp asymmetry of a meson state @xmath111 decaying into a final state @xmath87 . taking a cp - eigenstate final state and up to @xmath112 corrections , the asymmetry is given by @xmath113 where @xmath114 with @xmath115 @xmath116 is the @xmath86 mixing parameter and @xmath90 ( @xmath91 ) is the amplitude for @xmath117 ( @xmath118 ) . the coefficient @xmath119 , the direct cp asymmetry defined in eq .
( [ eq : acpdir ] ) .
the coefficient @xmath120 , instead , is a new measurement of @xmath121 violation generated by the interference of mixing and decay amplitudes .
time - dependent cp asymmetries for @xmath7 decays ( where @xmath122 ) give access to the ut angles . for example , as the amplitude for @xmath123 is dominated by a single term with a definite weak phase , one finds @xmath124 up to doubly cabibbo - suppressed corrections . in other cases ,
the decay amplitude is not dominated by a single term and the extraction of the ut angles is less straightforward , as hadronic amplitudes no longer cancel in @xmath125 .
previous formulae can also be generalized to non cp - eigenstate final states @xcite .
in this section , we briefly review the present status of flavor physics in the sm , discussing the determination of the ckm parameters and possible deviations found in present data .
the ckm matrix elements @xmath126 and @xmath127 can be measured from super - allowed @xmath128 decays @xcite and semileptonic / leptonic kaon decays @xcite respectively , determining accurately the sine of the cabibbo angle .
the other ckm parameters are determined through a fit to the ut in eq .
( [ eq : ut ] ) , as shown in fig .
[ fig : ckm_fit_today ] , using the latest determinations of the theoretical and experimental parameters @xcite . the basic constraints are @xmath129 from semileptonic @xmath7 decays , @xmath130 and @xmath131 from @xmath132 oscillations , @xmath133 from @xmath134@xmath135 mixing , @xmath136 from charmless hadronic @xmath7 decays , @xmath137 and @xmath138+@xmath137 from charm hadronic @xmath7 decays , @xmath139 from @xmath140 decays and the br@xmath141 @xcite . and @xmath142 selected from constraints on @xmath129 , @xmath130 , @xmath131 , @xmath133 , @xmath136 , @xmath128 , @xmath137 , @xmath138+@xmath137 and br@xmath141 . @xmath143 and
@xmath144 total probability contours are shown , together with @xmath144 probability regions from the individual constraints .
, scaledwidth=53.0% ] the consistency between the different constraints clearly establishes the ckm matrix as the dominant source of flavor mixing and cp - violation , described by a single parameter @xmath142 .
the consistency of the ckm picture can be quantitatively tested comparing the direct measurement of a flavor observable entering the ut fit with the sm prediction obtained from the ut fit without using the constraint being tested .
table [ tab : pull ] contains a set of flavor observables : for each of them , the sm prediction , the measurement , and the pull ( the difference of prediction and measurement in unit of @xmath145 ) are shown . at present , the main tensions in the ut fit come from br@xmath146 , which is found to deviate from the measurement by @xmath147 , and @xmath148 , which is larger than the experimental value by @xmath149 .
it is interesting to note that the former prefers large values of @xmath129 , while the latter wants @xmath129 to be small .
any change of the measured value of @xmath150 and @xmath151 will not improve the situation . in the @xmath70 sector
, the possibility of a large deviation in @xmath152 is still open , although new tevatron data on @xmath153 , still to be included in the average @xcite , go in the sm direction @xcite .
moreover , the dimuon asymmetry @xmath154 measured by d @xcite also displays a large deviation from the sm , pointing to a large @xmath152 and possibly to np contributions in @xmath155 .
.summary of different measurements and corresponding sm predictions . in the last column
the pull is explicitly indicated.[tab : pull ] [ cols="<,^,^,^",options="header " , ] lht effects in @xmath9 physics have been studied in ref .
a large cp violation , a clear np signal , can be present both in the mixing amplitude ( @xmath156 ) and in the time - dependent cp asymmetry of @xmath157 ( @xmath158 ) .
the possibility to measure @xmath9 time - dependent cp asymmetries is under study at superb .
randall - sundrum models of extra dimensions exploits the geometry of the fifth dimension to naturally stabilize the fermi scale @xcite .
in addition , by putting fermions in the bulk , one can address the sm flavor problem : the hierarchies in quark masses and mixings are obtained from the localization of fermions along the fifth dimension @xcite .
new flavor effects in these models arise from the exchange of kaluza - klein excitations .
they depend crucially on the specific realization of the model , although in general they are severely constrained by cp violation in @xmath5@xmath6 mixing . in the minimal scenario with only the sm gauge group in the bulk , one finds that large effects are possible in both @xmath7 and @xmath5 decays @xcite , see for example the correlation between @xmath159 and @xmath160 in fig .
[ fig : minrs ] .
the former @xmath161 will be measured soon at lhcb , while the transition @xmath162 can be measured in the exclusive decays @xmath163 with full statistics at superb . and @xmath164 in the minimal rs model from ref .
@xcite.[fig : minrs ] ] other realizations , such as the custodially - extended bulk gauge symmetry , suppress the np contributions to @xmath7 fcncs , although large effects in kaons and in @xmath70@xmath71 mixing are still possible @xcite .
in this review we have considered a large subset of flavor measurements , briefly discussing their present status and the expected sensitivities at next - generation experiments , including possible systematic or theoretical limitations .
we have then presented some examples of flavor effects generated in a selection of np models which could be measurable in the future . in particular , we selected those correlations among observables which could provide a distinctive mark of the np model , showing how flavor physics could help telling it apart from the others
. the next decade will be crucial for the future of particle physics in general and of collider physics in particular .
the lhc already started looking for the higgs boson and for new particles at the tev scale . in this context
, flavor physics could have a prominent role in characterizing a tev np and looking for signals from the multi - tev region .
it is even possible that the first evidence for np come from flavor , with lhcb measuring np - sensitive channels like @xmath165 and @xmath153 . in any case , reconstructing the np lagrangian from the data is a collective effort requiring the interplay of both direct and indirect searches , together with an accurate theoretical control of the sm contributions . improved experimental results at the energy and intensity frontiers and the most advanced theoretical techniques are therefore required ingredients to go steadily beyond the sm .
precision flavor physics can certainly give a major contribution to this effort .
next - generation experiments will offer new opportunities by pushing existing flavor measurements to an unprecedented accuracy and by opening the possibility to measure new np - sensitive observables in all flavor sectors .
99 bona m , et al .
[ utfit collaboration ] _ jhep _ 0803 , 049 ( 2008 ) ; isidori g , nir y and perez g arxiv:1002.0900 [ hep - ph ] glashow sl , iliopoulos j and maiani l , _ phys .
d _ 2 , 1285 ( 1970 ) kobayashi m and maskawa t _ prog .
_ 49 , 652 ( 1973 ) bigi iiy _ slac - pub-4521 _ ; lusignoli m , maiani l , martinelli g and reina l _ nucl .
b _ 369 , 139 ( 1992 ) bona m et al .
[ superb collaboration ] arxiv:0709.0451 [ hep - ex ] browder t et al .
_ jhep _ 0802 , 110 ( 2008 ) hitlin dg et al . arxiv:0810.1312 [ hep - ph ] aushev t et al . arxiv:1002.5012 [ hep - ex ] oleary b et al .
[ superb collaboration ] [ arxiv:1008.1541 [ hep - ex ] ] jarlskog c _ phys .
_ 55 , 1039 ( 1985 )
wolfenstein l _ phys .
_ 51 , 1945 ( 1983 ) pontecorvo b _ sov .
jetp _ 6 , 429 ( 1957 ) ; maki z , nakagawa m and sakata s _ prog . theor .
_ 28 , 870 ( 1962 ) dunietz i and rosner jl _ phys .
d _ 34 , 1404 ( 1986 ) towner is and hardy jc _ rept .
_ 73 , 046301 ( 2010 ) antonelli m et al . _
j. c _ 69 , 399 ( 2010 ) bevan aj et al .
[ utfit collaboration ] arxiv:1010.5089 [ hep - ph ] and online update at ` http://www.utfit.org ` abazov vm et al .
[ d0 collaboration ] _ phys .
d _ 82 , 032001 ( 2010 ) ciuchini m , pierini m , silvestrini l _ phys . rev .
_ 95 , 221804 ( 2005 ) ; faller s , jung m , fleischer r and mannel t _ phys .
d _ 79 , 014030 ( 2009 ) faller s , fleischer r and mannel t _ phys .
d _ 79 , 014005 ( 2009 ) beneke m _ phys .
b620 , 143 - 150 ( 2005 ) ; cheng h - y , chua c - k , soni a _ phys .
d72 , 014006 ( 2005 ) ; cheng h - y , chua c - k , soni a _ phys .
d72 ( 2005 ) 094003 m. ciuchini , e. franco , g. martinelli , m. pierini , l. silvestrini , phys .
lett . * b674 * , 197 - 203 ( 2009 ) .
[ arxiv:0811.0341 [ hep - ph ] ] .
atwood d , eilam g , gronau m , soni a _ phys .
b341 , 372 - 378 ( 1995 ) ; atwood d , dunietz i , soni a _
_ 78 , 3257 - 3260 ( 1997 ) giri a , grossman y , soffer a , zupan j _ phys .
d68 , 054018 ( 2003 ) buchalla g , hiller g and isidori g _ phys .
d _ 63 , 014015 ( 2000 ) altmannshofer w , buras aj , straub dm and wick m _ jhep _ 0904 , 022 ( 2009 ) bona
[ utfit collaboration ] _ phys .
b687 , 61 - 69 ( 2010 ) the cdf collaboration _ cdf public note _
9892 fullana e , sanchis - lozano ma _ phys .
b653 , 67 - 74 ( 2007 ) ; domingo f , ellwanger u _ jhep _ 0712 , 090 ( 2007 ) ; ellwanger u , hugonie c , teixeira am _ phys .
_ 496 , 1 - 77 ( 2010 ) mcelrath b _ phys .
d _ 72 , 103508 ( 2005 ) abe k et al .
[ sld collaboration ] _ phys .
_ 73 , 25 ( 1994 ) [ aleph and delphi and l3 and opal and sld and lep electroweak working group and sld electroweak group and sld heavy flavour group collaborations ] _ phys .
_ 427 , 257 - 454 ( 2006 ) van oers wth et al .
[ qweak ] experimental proposal for jlab e02 - 020 laplace s , ligeti z , nir y and perez g _ phys .
d _ 65 , 094040 ( 2002 ) ; bona m et al .
[ utfit collaboration ] _ jhep _ 0603 , 080 ( 2006 ) ; bona m et al .
[ utfit collaboration ] _ phys .
_ 97 , 151803 ( 2006 ) brod j , gorbahn m and stamou e _ phys .
d _ 83 , 034030 ( 2011 ) martin sp arxiv : hep - ph/9709356 haber he _ nucl .
_ 62 , 469 ( 1998 ) hall lj , kostelecky va and raby s _ nucl .
b _ 267 , 415 ( 1986 ) altmannshofer w et al .
phys . b _ 830 , 17 ( 2010 ) girrbach j et al .
arxiv:1101.6047 [ hep - ph ] chivukula rs and georgi h _
lett . b _ 188 , 99 ( 1987 ) ; hall lj and randall l _ phys .
_ 65 , 2939 ( 1990 ) ; gabrielli e and giudice gf _ nucl .
b _ 433 , 3 ( 1995 ) [ erratum - ibid . _ b _ 507 , 549 ( 1997 ) ] ; ciuchini m , degrassi g , gambino p and giudice gf _ nucl .
b _ 534 , 3 ( 1998 ) dambrosio g , giudice gf , isidori g , strumia a _ nucl . phys . _
b645 , 155 - 187 ( 2002 ) buras aj _ acta phys .
polon . _
b34 , 5615 - 5668 ( 2003 ) bobeth c et al .
b _ 726 , 252 ( 2005 ) deschamps o et al . _ phys .
d _ 82 , 073012 ( 2010 ) ; bona m et al.[utfit collaboration ] _ phys .
b _ 687 , 61 ( 2010 ) altmannshofer w , buras aj and paradisi p _ phys .
b _ 669 , 239 ( 2008 ) barbieri r , lodone p and straub dm arxiv:1102.0726 [ hep - ph ] hou ws , nagashima m and soddu a _ phys .
d _ 72 , 115007 ( 2005 ) ; hou ws , nagashima m and soddu a _ phys .
d _ 76 , 016004 ( 2007 ) ; herrera ja , benavides rh and ponce wa _ phys .
d _ 78 , 073008 ( 2008 ) soni a et al .
d _ 82 , 033009 ( 2010 ) ; buras aj et al .
_ jhep _ 1009 , 106 ( 2010 ) arkani - hamed n , cohen ag and georgi h _ phys .
86 , 4757 ( 2001 ) ; arkani - hamed n , cohen ag and georgi h _
b _ 513 , 232 ( 2001 ) ; schmaltz m and tucker - smith d _ ann . rev
. nucl . part .
_ 55 , 229 ( 2005 ) blanke m et al .
_ jhep _ 0612 , 003 ( 2006 ) blanke m et al .
_ acta phys .
b _ 41 , 657 ( 2010 ) ellis jr , hisano j , raidal m and shimizu y _ phys .
d _ 66 , 115013 ( 2002 ) paradisi p _ jhep _ 0602 , 050 ( 2006 ) bigi ii , blanke m , buras aj and recksiegel s _ jhep _ 0907 , 097 ( 2009 ) randall l and sundrum r _ phys .
_ 83 , 3370 ( 1999 ) gherghetta t and pomarol a _ nucl .
b _ 586 , 141 ( 2000 ) agashe k , perez g and soni a _
_ 93 , 201804 ( 2004 ) blanke m et al .
_ jhep _ 0903 , 001 ( 2009 ) bauer m , casagrande s , haisch u and neubert m _ jhep _ 1009 , 017 ( 2010 )
|
starting with next - generation experiments , flavor physics fully enters the era of precision measurements .
the focus shifts from testing the standard model to finding and characterizing new physics contributions .
we review the opportunities offered by future flavor experiments , discussing the expected sensitivities of the most important measurements .
we also present some examples of measurable deviations from the standard model in the flavor sector generated in a selection of new physics models , demonstrating the major contribution that precision flavor physics could give to the effort of going beyond the standard model .
* lal 11 - 245 * + october 2011 * physics opportunities at the next generation + of precision flavor physics * + + + +
|
the process of forming collimated jets / outflows due to disk accretion onto central compact objects is thought to depend on how magnetic fields behave when they are swirled around by the accretion disk .
the progress of understanding this process has , however , been hindered by the significant lack of knowledge on the global magnetic field configuration in / near the accretion disk ( see okamoto 1999 for detailed critiques on many mhd models ; see also blandford 2000 for a recent review ) .
an ordered magnetic field is widely thought to have an essential role in jet formation from a rotating accretion disk .
two main regimes have been considered in theoretical models ( see lovelace , ustyugova & koldoba 1999 for a review ) , the _ hydromagnetic regime _ where the energy and angular momentum is carried by both the electromagnetic field and the kinetic flux of matter , and the _ poynting flux regime _ where the energy and angular momentum outflow from the disk is carried predominantly by the electromagnetic field .
major progress has been made recently in the hydromagnetic regime of jet formation , originally proposed by blandford & payne ( 1982 ) .
dynamic mhd simulations of the near jet region have been carried out by several groups ( bell 1994 ; ustyugova et al .
1995 , 1999 ; koldoba et al .
1995 ; romanova et al .
1997 , 1998 ; meier et al . 1997 ; ouyed & pudritz 1997 ; krasnopolsky , li , & blandford 1999 ) .
the simulation study of ustyugova et al .
( 1999 ) in the hydromagnetic wind regime indicates that the outflows are accelerated to super - fast magnetosonic , super - escape speeds close to their region of origin ( @xmath3 the inner radius of the disk ) , whereas the collimation occurs at much larger distances . poynting flux models for the origin of jets were proposed by lovelace ( 1976 ) and blandford ( 1976 ) and developed further by lovelace , wang , and sulkanen ( 1987 ) , lynden - bell ( 1996 ) , and colgate & li ( 1999 ) . in these models the rotation of a keplerian accretion disk twists a poloidal field threading the disk , and this results in outflows from the disk which carry angular momentum ( in the twist of the field ) and energy ( in the poynting flux ) away from the disk , thereby facilitating the accretion of matter .
most recent computer simulations using the full axisymmetric mhd equations have been in the hydromagnetic regime
. however , recent simulation studies have found jet outflows in the poynting flux regime ( romanova et al . 1998 ; ustyugova et al .
2000 ) .
one class of models deals with a simplified limit , where magnetostatic and force - free conditions are assumed . this kind of problem is very clearly stated in the introduction of lynden - bell & boily ( 1994 , hereafter lb94 ; see also lynden - bell 1996 ) : a field rooted in a heavy conductor on @xmath4 ( i.e. , the rotating disk ) pervades a perfectly conducting force - free medium in the region @xmath5 . on the disk surface ( @xmath4 ) , the field passes upwards from the region @xmath6 and returns downwards in the region @xmath7 .
the disk is now rotated about its axis according to the keplerian motion , @xmath8 .
the problem is to determine the magnetostatic field configuration when the disk has gone through a specific number of turns .
they found that fields , instead of collimating along the rotation axis , expand along an angle of @xmath9 away from the rotation axis .
the behavior of twisted magnetic field lines that are anchored in a perfectly conducting medium has originally been considered in the solar corona context ( see , e.g. , aly 1984 , 1991 , 1995 ; mikic et al .
1988 ; finn & chen 1990 ; sturrock et al . 1995 ) .
the direct application of those studies to accretion disks has been fairly recent ( e.g. , appl & camenzind 1993 ; knigl & ruden 1993 ; lb94 ; lynden - bell 1996 ; bardou & heyvaerts 1996 ; goodson et al .
1999 ; uzdensky et al .
. the important role of an external plasma pressure was first emphasized by lynden - bell ( 1996 ) where he argued that field lines can not simply expand to infinity as shown by lb94 because the required @xmath10 work becomes too large .
he then constructed a simplified cylindrical model where a magnetic helix / jet is bounded at large radius by external plasma pressure .
but the question left open is whether the plasma pressure can indeed play a confining role in a self - consistent _ global _ treatment of the twisting and expansion of magnetic fields driven by the disk shear rotation . in this study
, we obtain self - consistent global solutions of axisymmetric magnetostatic configurations by strictly enforcing the keplerian shear condition on the flux surfaces that are anchored in the disk .
the assumptions we use are explained in detail in [ sec : assum ] .
we formulate our problem in [ sec : eqs ] , along with the relevant equations and methods to solve them .
results are given
[ sec : result ] and final conclusions and discussions are in [ sec : diss ] .
the equation of motion in the non - relativistic ideal magnetohydrodynamics ( mhd ) limit is simply @xmath11 where @xmath12 is the plasma density , @xmath13 the flow velocity , @xmath14 the current density , @xmath15 the gravitational acceleration , and @xmath16 is the plasma pressure .
we restrict our attention to axisymmetric magnetostatic configurations .
the underlying assumption is that the speed of the field line twisting by the disk is slow so that the system quickly reaches an equilibrium .
we can then treat the evolution of field configurations as sequences of magnetostatic equilibria .
we make the further assumption that the magnetic field plays the dominant role with plasma flow @xmath13 and gravity @xmath17 much less important .
then , the steady state equation is @xmath18 the main astrophysical motivation for keeping this pressure term is that at sufficiently large distances away from the disk , the plasma pressure will be comparable to the magnetic pressure , thus becoming dynamically important .
when @xmath19 , we have the so - called force - free ( ff ) limit .
we now discuss the physical conditions where these assumptions apply .
there are at least two relevant speeds in this problem which are related to two physical aspects .
the first one is the velocity of keplerian disk rotation , @xmath20 , which describes the rate of field line footpoint movement .
the second is the relaxation speed , @xmath21 , for magnetic fields to reach equilibrium , which is usually achieved by mhd waves going back and forth in the system .
this is essentially the alfvn speed @xmath22 and is ultimately limited by the speed of light @xmath23 . together with these speeds ,
there are two relevant timescales : the disk rotation period @xmath24 , and the relaxation time @xmath25 , where @xmath26 is the inner disk radius , and @xmath27 the system dimension , respectively . in order to ensure the steady state condition on the timescale of @xmath28 ,
one requires that @xmath29 . in order to simulate a system that is much larger than @xmath26 ,
these considerations indicate that in a black hole accretion disk system , field lines close to the black holes are likely twisted too rapidly to be treated by the magnetostatic equations .
in other words , the dynamic pressure from the inertial term @xmath30 in equation ( [ eq : motion ] ) is likely to be very important during the expansion , thus the conclusions from this study might not apply in this limit . on the other hand , for field lines further away from the black hole and/or for accretion disks around systems like young stars , the rotation speed is small enough that the magnetostatic limit could still apply .
another relevant process is the footpoint drift across the disk . magnetic field threading
the disk tends to be advected inward with the accretion flow , but at the same time it may diffuse through the disk owing to a finite magnetic diffusivity @xmath31 of the disk .
the outward drift of the magnetic field in the disk occurs at speed @xmath32 , where @xmath33 is the half - thickness of the disk , @xmath34 , and a smaller , second order diffusion term ( @xmath35 ) has been omitted ( lovelace _ et al .
_ 1997 ) . for cases where the diffusivity is of the order of the viscosity and where the viscosity is given by the shakura and sunyaev ( 1973 ) prescription @xmath36 ( with @xmath37 and @xmath38 the midplane sound speed ) , the diffusive drift speed is @xmath39 .
this is larger than the radial accretion speed @xmath40 due to viscosity alone .
but what is important here is that the field drift and the accretion speeds are much less than the keplerian velocity of the disk for @xmath41 .
for this reason the fields can be treated as frozen into the disk .
this point was made by ustyugova et al .
( 2000 ) and also discussed by uzdensky et al .
we have formulated the problem discussed in [ sec : intro ] in both cylindrical @xmath42 and spherical @xmath43 coordinates .
axisymmetry is assumed in both cases .
the magnetic field can be written as : @xmath44 where the poloidal component @xmath45 is in the @xmath46 or @xmath47 plane , @xmath48 is the total poloidal flux through the disk , and contours of @xmath49 label the poloidal field lines .
the poloidal magnetic fields are @xmath50 for cylindrical coordinates and @xmath51 for spherical coordinates .
the current density @xmath52 where the operator @xmath53 is , for cylindrical coordinates , @xmath54 and for spherical coordinates , @xmath55 equation ( [ eq : ff1 ] ) implies that @xmath56 in other words , the gas pressure is constant along field lines . in cylindrical coordinates , equation ( [ eq : ff1 ] )
can be written as the well - known grad - shafranov equation , @xmath57 from which we define @xmath58 and we get @xmath59/d\psi + 4\pi r^2 dp / d\psi = 0~.\ ] ] similarly , in spherical coordinates , we define @xmath60 and @xmath61/d\psi + 4\pi ( r\sin\theta)^2
dp / d\psi = 0~.\ ] ] in our spherical calculations , we actually use @xmath62 to concentrate the uniform grid in @xmath63 for better resolution at small @xmath64 . equation ( [ eq : ff_s1 ] ) then becomes @xmath65/d\psi + 4\pi ( r^4\sin\theta^2 ) dp / d\psi = 0~,\ ] ] where @xmath66 the quantity @xmath67 is @xmath68 times the total current flowing through a circular area of radius @xmath69 ( with normal @xmath70 ) labeled by @xmath71= const .
the development of the toroidal field component from an initial purely poloidal field comes from the differential rotation of the disk onto which footpoints of the same field lines are anchored .
since there is no direct observational information on how magnetic fields are distributed on the surface of an accretion disk , we have made the following assumptions .
if the magnetic fields on the disk is initially generated via an accretion disk dynamo that makes a quadrupole or dipole - like field , magnetic fields will emerge out of the disk at smaller radii and go back to the disk at large radii .
we use @xmath72 to represent the o - point of the field in the disk , i.e. , the radius where @xmath73 on the disk .
furthermore , in a fully self - consistent treatment including the back - reaction on the disk , @xmath72 also marks a separation location inside which angular momentum is lost and transmitted to the outer part of the disk by field line tension .
we study the expansion of this cylindrical magnetic arcade anchored on the disk .
we further assume that the overall strength of @xmath74 decreases as a function of radius , except near the o - point where @xmath75 .
this is roughly consistent with the fact that the thermal pressure of the disk ( which anchors the fields ) decreases radially as well .
specifically , we have chosen a computational domain that is a box with @xmath76 and @xmath77 in the cylindrical case and/or a sphere with @xmath78 in the spherical case .
the outer boundary is assumed to be perfectly conducting ( @xmath79 ) . at @xmath4 ( along the disk )
, we assume that the disk is perfectly conducting as well . an initial poloidal field distribution is specified @xmath80 [ or @xmath81 as @xmath82 where we have joined these parts smoothly .
figure [ fig : init_psibz ] shows the distribution of @xmath49 and @xmath74 on the disk surface .
the poloidal flux is distributed between @xmath83 and @xmath84 .
a core radius @xmath85 is used , inside of which the @xmath86 component approaches a constant .
the index @xmath87 is chosen to be @xmath88 , so that @xmath86 is decreasing as @xmath89 , reversing its sign at the o - point @xmath90 , and continuing decreasing until @xmath91 . both @xmath80 and @xmath74 remain zero between @xmath91 and @xmath92 .
these choices for @xmath72 and @xmath91 ensure that the initial magnetic field is distributed far inside the outer boundary @xmath93 .
we have tried many different initial field configurations and found that our main conclusions do not depend on these particular choices of parameters , as long as magnetic fields stay away from the outer boundary @xmath93 or @xmath94 .
the dependence on @xmath95 near the o - point is very weak since those field lines are hardly twisted at all .
the physical picture we have in mind for accounting for the pressure of an external plasma is that magnetic field close to the disk is force - free with little influence from the plasma pressure . at large distances , however , the magnetic pressure decreases sufficiently so that plasma pressure becomes important . in other words ,
the magnetic fields are bounded above and on the sides by an ambient medium which is perfectly conducting and mostly devoid of any magnetic flux .
this forms a conducting , in general , `` deformed box '' inside of which the magnetic field dominates and the gas pressure is very small . to model the external plasma pressure , we adopt the dependence @xmath96~,\ ] ] so that @xmath97~,\ ] ] where @xmath98 and @xmath99 are the input parameters . for most of the results presented here ,
we have chosen @xmath100 and @xmath101 .
the physical meaning of @xmath98 is that its magnitude gives an estimate of the overall importance of plasma pressure .
using the example given in equation ( [ eq : psi_disk ] ) , the maximum @xmath86 near @xmath26 is @xmath102 , ( see figure [ fig : init_psibz ] ) , so the ratio of the maximum magnetic pressure to the maximum gas pressure in the computational domain is @xmath103 .
the quantity @xmath99 gives the fraction of poloidal flux that is affected by the plasma pressure .
the prescribed plasma pressure ( equation [ eq : press ] ) is kept fixed at all times throughout the sequence of equilibria .
this is because the prescribed pressure here is meant to mimic a _
constant _ external pressure boundary ( such as pressure from interstellar or intergalactic medium ) that is reacting to the `` push '' by magnetic fields .
this interpretation is precise in the limit of @xmath104 .
we have used a very small but finite @xmath99 for the convenience of modeling both the lorentz force and pressure gradient terms simultaneously , rather than having to specify the pressure gradient term through a boundary condition , especially when the shape of this boundary is determined dynamically by the push from the magnetic fields and is not known _ a priori_. the pressure in the flux tubes with @xmath105 , however , might change due to the flux volume expansion .
if the entropy of each flux tube is conserved , then pressure has to decrease with expanding volume . on the other hand , if there is enough heat flux between the disk and the flux tubes , then the entropy of a flux tube is not conserved ( finn & chen 1990 ) .
either way , since the pressure in the flux tubes with @xmath105 is already set to be exponentially small ( equation [ eq : press ] ) initially , any further decrease in pressure will not alter our conclusions .
we make two more observations about the plasma pressure term .
first , equation ( [ eq : press ] ) gives that , when @xmath106 ( i.e. , at the boundary ) , the pressure @xmath107 and its gradient @xmath108 .
both features are physically plausible .
second , it is interesting to note that the pressure effect enters equations ( [ eq : ff_c ] ) and ( [ eq : ff_s ] ) with a geometric factor @xmath109 or @xmath110 .
this factor actually comes from the @xmath111 term .
it is only under the equilibrium condition can it be `` moved '' to be a multiplier of the pressure term .
consequently , the pressure tends to prevent expansion away from the rotation axis but has relatively little constraining effect along the rotation axis . from the distribution of @xmath80 and the keplerian rotation @xmath8
, we can define the required twist on each field line @xmath112 .
the twist of a given field line going from an inner footpoint at @xmath113 to an outer footpoint at @xmath114 is proportional to the differential rotation of the disk . using a cylindrical coordinate , for a given field line
, we have @xmath115 where @xmath116 is the poloidal arc length along the field line , and @xmath117
. then the total twist of a field line is @xmath118 we denote the quantity @xmath119 .
the field line twist after a time @xmath120 is @xmath121~ , \nonumber \\ & = & ( \omega_{o}~t ) ~{\cal f}(\psi)\end{aligned}\ ] ] where @xmath122 is the keplerian angular frequency at @xmath72 of an object of mass @xmath123 , and @xmath124 is dimensionless .
the rotation profile is assumed to begin deviating from keplerian and approach a constant when @xmath125 ( the bottom panel of figure [ fig : init_psibz ] ) .
there exists a region , however , between @xmath126 and @xmath127 where the twist decreases as @xmath49 decreases ( i.e. , getting closer to the @xmath128-axis and the surrounding wall ) .
these field lines are not explicitly followed in our calculations .
but since @xmath129 as @xmath106 , the twist @xmath112 approaches zero as well [ equation ( [ eq : path ] ) ] .
we solve equations ( [ eq : ff_c ] ) and ( [ eq : ff_s ] ) following the approach given in finn & chen ( 1990 ) , which employs several levels of iterations .
for the very first step , we will use a trial @xmath130 , then we solve equation ( [ eq : ff_c ] ) or ( [ eq : ff_s ] ) using the successive over - relaxation ( sor ) method ( see for example , potter 1973 ) for @xmath131 . from @xmath132 , we integrate along the field lines to obtain @xmath133 .
we typically trace @xmath134 field lines with @xmath135 for calculating the poloidal current profile @xmath130 . using equation ( [ eq : path ] )
, we update @xmath130 using the input @xmath112 .
this procedure is then repeated .
the advantage of the simple sor solver is that the nonlinear source terms @xmath136 and @xmath137 can be easily included in the iterations so that convergence can be achieved fairly quickly ( with a typical relative residue less than @xmath138 ) in summary , we compute global magnetostatic solutions of @xmath139 using the grad - shafranov equation in axisymmetry by requiring that the twist on each field line follows a specified function @xmath140 , derived based on the keplerian disk rotation .
we have made many runs with different choices of the initial field configuration on the disk @xmath80 , the twist profile @xmath140 , and the pressure profile @xmath141 .
runs are made in both cylindrical and spherical coordinates .
we present most of our results using the following parameters : a @xmath142spherical coordinate system is used , with @xmath143 .
the radial and @xmath144angle grids are @xmath145 .
the initial field configuration is the same as shown in figure [ fig : init_psibz ] , where @xmath146 and @xmath147 .
the smallest @xmath49 value of the field lines we track is @xmath148 .
the plasma pressure parameters are @xmath149 and @xmath150 .
the maximum @xmath74 on the disk is @xmath151 , which implies a ratio of @xmath152 between the maximum magnetic pressure and the plasma pressure .
( even larger pressure ratios can be achieved with relative ease . ) to indicate the progressively increasing twist added to the field lines , we use the notation time `` @xmath120 '' [ equation ( [ eq : twist ] ) ] in units of revolutions of the inner most flux line ( @xmath153 ) around the @xmath128-axis , i.e. , @xmath154 .
figure [ fig : model ] describes our overall physical picture with two different physical regimes . for @xmath155 ,
the plasma pressure is dominant .
this region is labeled as `` plasma - pressure '' ( pp ) . for @xmath156 , the magnetic pressure dominates ,
the region is force - free ( labeled as `` ff '' ) .
so , we have effectively made two ideal mhd fluids , one ( pp ) has a plasma @xmath157 and the other ( ff ) has @xmath158 .
the key question we are addressing is : `` how does the shape of @xmath159 boundary change while field lines are being twisted by the keplerian disk rotation ? '' in other words , the boundary between the pp and ff regions evolves according to both the expansion and `` pushing '' from the ff region and the `` hindrance '' of the plasma pressure . although some details may differ , we generally find that magnetic fields evolve in the following manner : ( 1 ) in the ff region , field lines expand primarily towards the large radius along an angle ( from the @xmath128-axis ) of @xmath160 ; ( 2 ) in the pp region , field lines expand predominantly vertically ( along the @xmath128-axis ) with some slight radial expansion ; ( 3 ) more twist causes further expansion along the @xmath128-axis but subject to a break - down of the magnetostatic , equilibrium assumption .
we now discuss these features in detail .
figure [ fig : seq ] shows the evolution of magnetic fields in the poloidal plane as twist is added , with @xmath161 and @xmath162 turns , respectively .
the field lines shown are evenly spaced in @xmath163 , with the outer - most line having @xmath164 and the inner - most line having @xmath165 . here
, we use the term `` outer '' to refer to field lines in the pp region , which originate from the smallest radii and return back to disk at the largest radii .
the term `` inner '' refers to field lines around the o - point in the ff region .
consequently , because the field lines are twisted according to the keplerian rotation ( largest rotation at the smallest radius ) , the outer field lines show substantial `` movement '' due to the added twist , whereas field lines around the o - point show little change . the corresponding results for the poloidal current @xmath130 and the resultant twist @xmath166 are plotted in figure [ fig : htp ] .
it is interesting to note that @xmath130 can be approximated by two power - laws of @xmath49 , where @xmath167 when @xmath168 and flattens somewhat for @xmath105 .
the twist shown in the lower panel is derived by integrating along each field line .
they indeed follow closely the input profile due to keplerian rotation .
the striking feature of the sequence of equilibria in figure [ fig : seq ] is the way the outer field lines expand .
to best understand this behavior , we separate the field lines into two groups according to the ratio of @xmath169 .
these two groups have fundamentally different expansion behavior , which is illustrated in figure [ fig : expan ] , where the expansion of three different field lines ( @xmath170 ) is shown for @xmath171 ( solid ) , @xmath172 ( dotted ) , and @xmath173 ( dashed ) , respectively .
field line expansion in the ff region has been studied in detail in many previous studies ( see introduction ) .
the lorentz force is the only force available ( i.e. , the magnetic pressure gradient is balanced by the tension force ) . as found for example by lb94 ,
the field lines expand to large radii along an angle of @xmath9 from the @xmath174axis .
this is consistent with the left panel of figure [ fig : expan ] ( also the inner field lines in figure [ fig : seq ] ) .
the field line expansion in the pp region , however , is fundamentally different from the ff regime .
this is because the magnetic pressure becomes very small at large radii and eventually at some distance it becomes comparable to the surrounding plasma pressure . at this location ,
further expansion of the magnetic fields is greatly hindered by the plasma pressure .
at the same time , increasing the twist acts to increase the ratio of @xmath175 .
consequently , these field lines are increasingly `` pinched '' around the @xmath128-axis , and they eventually expand along the @xmath128-axis in a `` collimated '' fashion , as shown by the middle and right panels of figure [ fig : expan ] and the late time in figure [ fig : seq ] .
the middle panel of figure [ fig : expan ] is particularly important .
it shows how the critical boundary at @xmath176 ( see figure [ fig : model ] ) evolves with increasing twist .
note that its initial shape is quasi - spherical , but a clear collimation along the @xmath174axis has developed by @xmath173 .
a small amount of poloidal flux ( with @xmath177 ) is in the pp region and is twisted as well .
one can ask whether these outer field lines have contributed importantly to forming the new shape of the @xmath159 boundary . to answer this question ,
we have performed runs with a very small @xmath140 for @xmath177 , instead of the near constant @xmath140 presented in figure [ fig : htp ] .
we find that as long as @xmath178 remains the same , its shape does not depend on the twist profile for @xmath179 .
this means that the small amount of magnetic flux beyond the @xmath159 boundary does not change our results noticeably .
we have also performed runs with various magnitudes of the plasma pressure . for the purely force - free case
, we confirm the results by lb94 that the field lines expand to large radii along an angle of @xmath9 from the @xmath174axis and only a small total twist ( less than half a turn ) can be added before the fields are packed against the outer boundary . with a finite pressure , however ,
the collimation effect is always observed as long as the simulation region is large enough or the pressure is large enough so that magnetic fields are not directly packed against the outer boundary .
thus , the external plasma pressure , no matter how small it may be , plays a fundamental role in causing the field lines to collimate along the rotation axis .
it essentially stops ( or greatly slows down ) the radial expansion of the field lines , and in the meantime allows the build - up of the toroidal component .
this interpretation is also consistent with the fact that field lines are closely packed at large radius ( stopped by plasma pressure ) but are loosely spaced along the @xmath128-axis ( see the `` @xmath173 '' panel in figure [ fig : seq ] ) .
the results shown in figures [ fig : seq ] and [ fig : expan ] indicate that the solution suggested by lynden - bell ( 1996 ) is indeed possible ( at least in the early expansion stage of the helix formation ) , even though details are different . here , we have shown global self - consistent solutions where the field lines are twisted according to the disk keplerian rotation .
figure [ fig : twist ] shows how its twist ( @xmath180 ) is distributed along this field line as a function of @xmath128 at @xmath173 .
we have plotted another field line with a larger @xmath181 at @xmath173 as well .
there is an important difference between these two field lines in the distribution of the twist @xmath140 . for @xmath182 ,
most of its twist ( @xmath183 ) is distributed on the ascending portion of the field line and near the @xmath128-axis ; the rest is distributed while the field line spirals down back to the disk . for @xmath184 , however , the twist is distributed uniformly between ascending and descending portions . to understand this difference , we can write @xmath185 note that @xmath130 is constant along a field line .
imagine a small flux tube ( @xmath186 ) originating from the inner region of the disk .
since the field line remains at small @xmath69 for the ascending section in the presence of plasma pressure , @xmath187 .
this is the reason why @xmath188 following a flux tube is roughly constant for the ascending part of the flux tube . for the @xmath182 flux tube ,
however , its expansion at large radius ( the descending portion ) is strongly constrained by the plasma pressure , so that the cross section area for the descending flux tube is much smaller than it would have been without the plasma pressure .
consequently , its @xmath86 component on the descending portion of the @xmath189 surface varies more slowly than @xmath190 , so that the rate @xmath191 is smaller .
overall , the twist is then non - uniformly distributed on a flux line .
( see ch .9 of parker 1979 ) .
thus , the presence of plasma pressure leads to a strong collimation and a concentration of the twist to the collimated region .
the total injected toroidal flux @xmath192 by the disk rotation into the system can be evaluated as @xmath193 we find that the ratio of @xmath192 to the initial total poloidal flux , @xmath194 , increases linearly with time as expected by equation ( [ eq : twist ] ) , reaching a maximum of @xmath195 when @xmath173 .
so , the total injected toroidal flux is still a small fraction of the total initial poloidal flux .
this information is useful for non - axisymmetric stability considerations for future studies .
figure [ fig : b2 ] shows the spatial distribution of magnetic energy density and how it evolves with added twist .
the quantity @xmath196 ( i.e. , a scaled magnetic energy density ) is shown for @xmath171 ( top left ) and @xmath162 ( top right ) , where @xmath197 and @xmath100 .
it is clear that essentially all the magnetic energy is enclosed by the plasma pressure .
the expansion results in a clear collimation of magnetic energy along the @xmath174axis .
the two bottom panels show , at @xmath173 , the poloidal @xmath198 ( lower left ) and the toroidal @xmath199 ( lower right ) components , respectively .
note the dramatic increase of magnetic energy ( mostly poloidal component ) along the @xmath174axis .
we find that it is not possible to increase the twist beyond what is shown in figure [ fig : seq ] .
when more twist is added , the system evolves from a configuration with all field lines being tied to the disk to a new topology with some poloidal field lines close on themselves instead of connecting to the disk ( i.e. , forming an isolated `` island '' of poloidal fluxes or `` plasmoid '' ) . at this point ,
our method used to solve equation ( [ eq : ff_s ] ) becomes unstable .
the plasmoid formation is associated with the field lines that are in the ff regime ( @xmath105 ) .
the transition to the plasmoid formation is sudden , giving a sense of eruptive behavior of the solution .
however , as discussed below , this numerical behavior is associated with instability of the method but not necessarily related to instability of the mhd equations .
such eruptive behavior has been the subject of intense research in the solar flare community ( aly 1984 , 1991 , 1995 ; sturrock et al .
1995 ) . the formation of a plasmoid from a single arcade ( inhester et al .
1992 ) is a loss - of - equilibrium bifurcation related to tearing instability of the current sheet which forms at the center of the arcade when it is sheared strongly ( finn et al .
in addition , it has been shown that plasmoid formation can occur directly as a consequence of linear instability when multiple arcades exist ( mikic et al .
1988 ; biskamp et a. , 1989 ; finn et al . 1992 ) .
this behavior of the solutions with increasing twist can be traced to a mathematical nature of the elliptical equation we are solving ( e.g. , equation ( [ eq : ff_c ] ) and ( [ eq : ff_s ] ) ) .
we can write this equation in a pseudo - time - dependent form , e.g. , @xmath200~,\ ] ] where a steady state is achieved progressively when @xmath201 . here , @xmath202 is a positive constant .
the computational domain can then be divided into three parts depending on the ratio of @xmath169 .
region i : large @xmath69 .
the pressure term `` @xmath203 '' dominates .
since it is negative , it drives @xmath49 to @xmath204 exponentially for small @xmath49 ( i.e. , @xmath205 , with @xmath206 ) .
the solution in this region is quite stable .
region ii : small @xmath69 and small @xmath128 ( close to the disk ) , i.e. , the ff region with a negligible pressure term .
region iii : small @xmath69 but large @xmath128 ( along the rotation axis ) , where all three terms contribute .
the eruptive behavior we found could occur in both regions ii and iii , whenever the poloidal current term @xmath207 exceeds a certain critical value . with a large current
, magnetic fields tend to expand enormously and fill up the whole computational domain .
an important question is whether this eruptive behavior actually occurs in a system described with the full set of dynamical equations .
recent axisymmetric simulations using the full set of mhd equations by ustyugova et al .
( 2000 ) and goodson et al .
( 1999 ) indicate that field lines can reconnect and become open .
clearly , the rate of reconnection is enhanced by the artificially large resistivity in such codes ; the exact role of resistivity in allowing reconnection when it might not otherwise occur is not completely clear . in ustyugova
et al . ( 2000 ) , two regimes have been found after @xmath208 tens of rotation periods of the inner disk : a hydromagnetic outflow from the outer part of the disk , and a poynting outflow , which has negligible mass flux but is dominated by the electromagnetic field along the rotation axis .
we have shown that a static plasma pressure is fundamental in shaping the overall magnetic equilibrium , despite the fact that the plasma pressure is exceedingly small ( the maximum magnetic pressure over the plasma pressure is @xmath209 in this study ) .
this is essentially different from the pure force - free models ( e.g. , lb94 ; see however lynden - bell 1996 ) .
this difference comes from the fact that the field lines that are being twisted the most expand the furthest so that they are most affected by the plasma pressure .
this is opposite to the behavior in the solar flare models where most of the shear is concentrated around the o - point .
the least amount of shear is applied near the o - point in the accretion disk case . in regions where the plasma pressure is negligible ,
the physics of expansion is dominated by the force - free condition ( @xmath210 ) .
their expansion is predominantly towards larger radii along an angle of @xmath211 from the @xmath174axis ( lb94 ) .
but for field lines affected by the plasma pressure , their radial expansion ( away from the @xmath174axis ) is slowed or stopped by this pressure .
the twisting of those field lines causes them to expand much more strongly along the @xmath174axis due to the build - up of the toroidal field component @xmath1 .
the build - up of @xmath1 is non - uniform along those field lines , with more twist distributed on the ascending portion of the field lines .
this is , again , a direct result of external plasma pressure .
we have obtained magnetostatic equilibria up to a maximum of @xmath162 turns of the inner most field line . for larger twist the solutions exhibit a change of topology and our method breaks down .
the larger values of twist should be treatable by including the inertia term @xmath212 in the full set of dynamic mhd equation .
an important question is the stability / instability of these axisymmetric equilibria .
instabilities in 3d involving flux conversion between toroidal and poloidal components is probably unavoidable and this effect will be important astrophysically : this is because the initial poloidal flux on the disk is probably too small to be responsible for the observed total magnetic flux in many astrophysical systems ( colgate & li 2000 ) .
the flux multiplication by the disk rotation / twisting , and subsequent flux conversion , could however generate enough flux .
|
we study the evolution of a magnetic arcade that is anchored to an accretion disk and is sheared by the differential rotation of a keplerian disk . by including an extremely low external plasma pressure at large distances , we obtain a sequence of axisymmetric magnetostatic equilibria and
show that there is a fundamental difference between field lines that are affected by the plasma pressure and those are not ( i.e. , force - free ) .
force - free fields , while being twisted by the differential rotation of the disk , expand outward at an angle of @xmath0 away from the rotation axis , consistent with the previous studies .
these force - free field lines , however , are enclosed by the outer field lines which originate from small disk radii and come back to the disk at large radii .
these outer fields experience most of the twist , and they are also affected most by the external plasma pressure . at large cylindrical radial distances ,
magnetic pressure and plasma pressure are comparable so that any further radial expansion of magnetic fields is prevented or slowed down greatly by this pressure .
this hindrance to cylindrical radial expansion causes most of the added twist to be distributed on the ascending portion of the field lines , close to the rotation axis .
since these field lines are twisted most , the increasing ratio of the toroidal @xmath1 component to the poloidal component @xmath2 eventually results in the collimation of magnetic energy and flux around the rotation axis .
we discuss the difficulty with adding a large number of twists within the limitations of the magnetostatic approximation .
|
as it is known , the dark matter problem arose from analysis of astronomical data when considerable discrepancy between orbital velocities of stellar objects at periphery of large galaxies compared to velocities calculated for usual gravitational fields has been found ( see , for instance , reviews in [ 8 , 14 ] ) .
this phenomenon was explained by the dark matter action that resulted in increased mass of such stellar objects . discovered effect of gravitational lensing considerably supported such assumption .
a particular obstacle was to provide experimental verification for existence of the dark matter particles .
many candidates for such particles have been proposed as well as several theories for their interaction with ordinary matter were considered [ 1 , 15 ] .
still , particles of dark matter have not been found .
another problem was caused by relatively low density of dark matter inside galaxies , where the theories [ 14 , 15 ] predicted formation of dark matter clouds .
present paper proposes a model for resonance amplification of interaction between two astronomical stellar objects stimulated by additional interaction with dark matter particles ( hereinafter dark particles ) .
in other words , it is considered that the dark particles not just increase gravitational mass of stellar objects , but , more importantly , do noticeably change effective interaction between the two heavy bodies .
such changes became distinguishable at specific large distances between the heavy particles only .
we refer to the two heavy stellar objects as to a system of one massive body located in a center of a galaxy and another stellar body located at the galaxy@xmath0s periphery orbiting the center of the galaxy .
resonant amplification of the effective interaction between the two heavy bodies imitates increase in their mass while their true gravitational mass does not change at that .
such increased interaction results in more pronounced gravitational lensing of the bypassing light .
so , the model can describe main effects and phenomena of the dark matter problem . for simplicity
, we consider the task in the non - relativistic case .
our considerations below are presented within the quantum - mechanical theory of the three - body scattering .
scattering of a dark particle at the system of two heavy bodies is considered .
the heavy bodies in our considerations are massive stellar objects as described above and a dark particles has a negligibly small mass , for example , in compared with the electron mass . we propose a simple potential model for interaction of a dark particle with a massive body ; such approach allows solving the three - body problem analytically .
these solutions distinctly demonstrate the resonance dependence of the effective interaction between two massive bodies on distance between these bodies .
so , at relatively small distances the catalytic amplification is negligibly small and effective interaction here coincides with direct ( gravitational ) interaction between the heavy bodies . at higher distances ,
catalytic action of the dark particles increases , and effective interaction of the heavy bodies also increases . at even larger distances , catalytic action can reverse and , instead of attraction between the heavy bodies , cause their repulsion .
this to certain degree can imitate dark energy action . in reality
, the nature of interactions between dark particles and a heavy body can be more complex and the catalytic action may happen to be far from simple .
let us assume that the dark particles do not interact with each other .
this simplifies the problem . of course
, we must take into account that it is necessary to consider the flux of particles per unit time , the total action of which is accepted here as the action of a single particle , taken with a certain energy and momentum .
we would also not consider the three - body type forces .
solutions for scattering amplitudes in the three - body problem are based on solutions for scattering amplitudes in two - body subsystems [ 5,6 ] . for simplicity ,
let us set the pair interaction of a dark particle with a heavy body by a potential in separable form .
the task therefore is reduced to determination of interaction in a system of three bodies : two heavy ones interacting gravitationally and one dark particle interacting with each of the heavy bodies .
another simplification will be related to utilization of the born - oppenheimer approximation which leads to the three - body problem solution in analytical form [ 16 ] .
such approximation can be used here since the mass ratio of the dark particle and the heavy body is extremely small .
let us take the potential of two heavy body interaction in non - relativistic limit and take its dependence on distance same as of coulomb interaction , for simplicity . in non - relativistic quantum mechanics the hamiltonian @xmath1 of heavy body with mass @xmath2 in newtonian gravitational field
is given by : @xmath3 where @xmath4 is newtonian gravitational potential at distance @xmath5 .
with no magnetic fields presented , this equation is the same as schr@xmath6dinger equation and the gravitational potential @xmath4 is the same as for the classical field [ 7 ] .
the equation ( 1 ) can be generalized by inclusion of additional non - newtonian forces ( see , for example in [ 3 , 7 ] ) . in the present paper
we choose the simplest interaction to clearly reveal the resonant amplification of the effective interaction between the heavy bodies .
the faddeev equations [ 5 , 6 ] are the key to solve the three - body scattering problem .
when a system of two heavy bodies and one dark particle is considered , these equations in combination with the born - oppenheimer approximation lead to simplification of the initial equations . at that , for a set of model pair potentials the solutions can be obtained analytically what makes the math better comprehended .
additional amplification of interaction between the heavy bodies due to their interaction with the dark particle is determined .
this third particle acts as a catalyst particle ( some comments in [ 10 ] ) . unlike regular pair interaction between heavy bodies ,
the additional effective interaction can resonantly increase at certain distances between the heavy particles at certain impulse of the dark particle .
let us call such certain distances the resonant ones .
the amplification or additional forces decrease abruptly at distances considerably less or , on the contrary , larger than the resonant distance . as a result ,
regular spatially monotonous pair interaction forces become dominant again [ 16 , 17 ] .
let us use a system of units where @xmath7 , @xmath8 .
we also assume that energy of a dark particle is small , i.e. we deal with non - relativistic particles .
interaction of the dark particle and the heavy body in separable form can be described as : @xmath9 here @xmath10 is the coupling constant , the @xmath11 subscript denotes interaction of the @xmath12 particle and the @xmath13 body .
then , based on the lippmann - schwinger equation @xmath14 where @xmath15 - is the free green@xmath0s function of system ; one can get the solution for the @xmath16-matrix also in separable form : @xmath17 @xmath18 @xmath19 @xmath20 , @xmath21 , @xmath22 - denotes reduced mass of the pair .
hereinafter we use the following normalization : @xmath23 , but @xmath24 . for simplicity ,
let us consider the @xmath25-wave and the potential form - factor in the form : @xmath26 where @xmath27 , @xmath28 ; we get ( @xmath29 , @xmath30 - potential parameter ) @xmath31 we have to normalize @xmath32 to make the coupling constant to be dimensionless
. then , @xmath33 would correspond to the attraction forces . for @xmath34
, the amplitude pole corresponds to the virtual level , and for @xmath35 it corresponds to the bound state since for @xmath36 , @xmath37 . here and below @xmath10 is real by definition .
the value @xmath38 is frequently referred to as an amplification coefficient to highlight the similarity between the potential and the @xmath16-matrix forms as well as the dependence of @xmath38 on the initial energy of the pair . in the case of @xmath39
we would have no bound states in the system of a dark particle and a heavy body .
note , in the case of @xmath40 the solutions for the two - body amplitude can be presented in breit - wigner form .
our goal is to determine the behavior of the effective interaction between the heavy massive bodies as a function of the distance between them .
mathematically rigorous solution of the three - body problems was provided by l.d .
faddeev [ 5 ] .
faddeev equations for components of the three - particle @xmath41-matrix can be written as [ 6 ] : @xmath42 where @xmath43 , @xmath44 are pair interaction potentials and @xmath45 - their corresponding pair @xmath41 -matrixes , @xmath46 - kronecker delta .
usually for conciseness , the interacting pair of particles is denoted by the number of the third particle .
the complete @xmath41-matrix would correspond to the sum : @xmath47 .
the index @xmath48 in @xmath49 stands for the number of the pair which interacts last at the left asymptotic , i.e. corresponds to the number of a particle which first escapes the interaction region .
similarly , the index @xmath50 denotes the number of the pair which interacts last at the right asymptotic .
let us first solve the problem taking into account interaction of a dark particle with the heavy body : at this stage we assume that there is no pair interaction between the two heavy bodies .
the pair of heavy particles is denoted with @xmath51 .
two other subsystems with one dark and one heavy body are denoted by @xmath52 , i.e. @xmath53 .
their potentials are taken in the separable form ( 2 ) .
related components @xmath49 with non - connected and connected parts can be written as [ 16 ] : @xmath54 and transferred into the matrix equations for @xmath55 @xmath56 where @xmath57 , @xmath58 .
it is important that the diagonal elements of this matrix are identically equal to zero , i.e. @xmath59 .
this peculiarity of the faddeev equations assures compactness of the core of the integral equations [ 5,6,9 ] .
let us consider a limiting case when @xmath60 , where @xmath61 - mass of a dark particle and @xmath2 - mass of a heavy body . in this limit
the total energy of the system is @xmath62 , where @xmath63 - initial momentum of the dark particle .
pair interactions between a dark particle and heavy bodies would not depend on momentums of the heavy particles @xmath64 , @xmath65 , where @xmath66 .
correspondingly , @xmath67 would also be the functions of initial momentum of the dark particle .
the value @xmath68 takes the form : @xmath69 here @xmath70 .
we denote with the prime the variables of heavy particles at the reaction outlet ; those without a prime @xmath71 at the reaction entrance .
@xmath68 can be written in the form : @xmath72 where @xmath73 similar transformation for the amplitude gives us : @xmath74\ ] ] since @xmath75-functions call off the integration in the right part of ( 15 ) , the equation for @xmath76 takes a simple form : @xmath77 moreover , @xmath75-functions in ( 15 ) give the equalities : @xmath78 if @xmath79 , and @xmath80 .
it means that @xmath81 and @xmath82 are counted from the center of symmetry between the positions of the heavy bodies . in this case for @xmath83 and @xmath84 the indeces @xmath48 and
@xmath50 can be omitted .
the solutions can be written in a convenient form [ 16 ] : @xmath85 where @xmath86^{-1}_{il } j_{lj}(\vec{r } ) \ ; \ ; , \ ; \ ; j\neq i \ ; \ ; , \ ] ] @xmath87^{-1}_{il } k_{li } \ ; \ ; , \ ] ] and @xmath88_{ij } = k_{ij}(\vec{r } ) = \sum_{l } j_{il}(\vec{r})\eta_l(p_0)j_{lj}(-\vec{r } ) \ ; \ ; .\ ] ] here we note that in our three - body problem we deal with two variables : @xmath89 initial momentum ( wavenumber ) of the dark particle and @xmath90 ( @xmath91 distance between the heavy bodies ) .
in this section we denote with index @xmath92 the solutions and interactions in the three body system , but with the @xmath93 potential excluded .
we also introduce notations @xmath94 , where @xmath95 , and @xmath96 - green@xmath0s function , and @xmath97 - wave function for this system .
the solutions are determined in ( 17 ) - ( 20 ) .
now one can move forward to solve a problem with inherent pair interaction between heavy particles taking into account the potential @xmath93
. the pair potential @xmath93 acts directly between the two heavy bodies of the system . the lippmann - schwinger equation for scattering at two potentials
can be written as follows [ 9 , 10 ] : @xmath98 where @xmath99 - total wave function for a system of three particles .
this symbolic equation is a short form of the system of equations in matrix form similar to the faddeev equations .
the first term in ( 21 ) at right discribes the resonant behaviour , but without direct interactions between two heavy bodies .
since @xmath100 where @xmath101 is the matrix with the components as in ( 10 ) , we can rewrite ( 21 ) in a symbolic form @xmath102 where @xmath103 this effective potential corresponds to the distorted potential @xmath93 @xmath104v_1 \ ; \ ; .\ ] ] it then follows that the effective potential as long as the amplitude @xmath105 would depend resonantly on both distance between the heavy particles and energy of the dark particle . in order to compare the interaction potentials @xmath93 and @xmath106
, acting between the two heavy bodies , it is necessary to record the first potential in the three particle representation . for this purpose
, @xmath93 can be supplemented with born interaction of the dark particle .
a similar addition should be made for the second potential - @xmath106 .
then the ratio of these quantities , written in the form @xmath107 will lead us to the result of the interaction between heavy bodies due to participation of an additional dark particle .
accordingly , the enhancement factor will be equal to : @xmath108 this important outcome demonstrates that the interactions with the third particle can substantially change the character of forces acting between the heavy bodies .
figures 1 and 2 show the behavior of real and imaginary parts of the enhancement factor @xmath109 for @xmath110 . ;
@xmath111 ; the resonance at @xmath112 . ] , while the dotted line shows the imaginary part of ( @xmath113 ) : a ) as function of @xmath114 at fixed value of @xmath115 ; b ) as function @xmath116 for a fixed value of @xmath117 ; the resonance at @xmath112 ; @xmath111 . ] following the model in section 4 above , one can get for @xmath118 : @xmath119 where dimensionless variables @xmath120 and @xmath121 are introduced .
it is necessary to note that location of the amplitude pole @xmath122 on the @xmath123 plane would change depending on the coupling constant @xmath10 ( fig .
the key factor here is the resonant dependence of the amplitudes @xmath122 on distance between the heavy particles .
it is important that the resonances take place not only when distance between the heavy bodies varies , but when the dark particle wavenumber changes @xmath124 .
it means that even at resonant distances the resonance takes place only at certain wavenumber values of the dark particles .
this is a distinguishing feature of resonances in quantum - mechanical few - body systems as regards the continuum .
such phenomenal behavior of the effective interaction between the heavy particles is due to three - particle quantum - mechanical effects . here
one should note that the described above model can potentially help explaning phenomenological interactions at galactic scales . ; @xmath125 ; resonance regions at @xmath126 and @xmath127 . ]
let us assess the characteristic distances .
for example in our model , resonance takes place at the following parameters : @xmath110 , @xmath128 . assuming that the distance between the massive bodies is about @xmath129 , one can get @xmath130 and @xmath131 . then , taking into account the resonant behavior of the effective interaction at such distances ( see fig .
2 ) , one can get from ( 27 ) the following multiplication of the interaction potential for the factor : @xmath132 at @xmath133 .
so we can say that exchange dark particle flux between heavy gravitating bodies acts like the dark matter .
the orbital velocity of the peripheral body becomes higher than that at normal gravity owing to the enhancement factor .
moreover this flux would glue up these two heavy bodies and , with respect to other particles and fields ( gamma quanta , for instance ) such a system would appear as a single object . the system would have the effective mass much higher than its own mass .
it is quite possible that such mechanism can contribute to gravitational lensing of electromagnetic radiation .
it is remarkable that the enhancement factors can be negative at certain values in the case of two - body attractive potentials ( see figs 1@xmath1344 ) .
for example , in the case of @xmath111 we have @xmath135 at @xmath136 .
then the resulting potential between two heavy bodies looks like a repulsive force . here
the question arises - to what extent can this potential also simulate the action of dark energy ?
4 shows the dependence of @xmath137 on @xmath138 at @xmath111 .
@xmath137 is the overall result of @xmath139 on the interval @xmath140 .
the opposite example is given in fig .
5 that demonstrates @xmath109 values in the case of repulsive potential between the dark particle and heavy body ( @xmath141 ) .
in the proposed paper we consider a region of the continuous spectrum ( continuum ) . coupling constant is taken as @xmath142 ,
so the bound states in the pair of the dark particle and the heavy body does not arise .
moreover , we consider the reverse limit when the interaction radius of @xmath143forces is taken large , but its ratio to the scattering length remains unchanged @xmath144 . the direct gravitational interaction between two heavy bodies in the model is complemented by interaction with the third dark particle ( or a flow of dark particles in order to be more exact ) .
this additional interaction has a specific resonant behavior as a function of the distance between the heavy bodies .
the resonant behavior of the effective potential implies its significant increase at certain distances between these bodies . ; @xmath110 . ]
the figures 1@xmath1344 show that the strengthening of effective interaction occurs at certain distances between the heavy particles while at other distances we observe sharp decrease or even repulsion .
this particular fact reveals a certain similarity in the action of such effective potential and dark matter .
as it follows from the analysis of astronomical observations , dark matter in dwarf galaxies appears to be quite small [ 12 , 13 ] , but its effect is huge in big galaxies [ 11 , 14 ] .
i.e. dark matter is stronger between massive bodies located in the center and on the periphery of a galaxy in large enough galaxies . here
, we assume that the ensemble of heavy particles in a distant star should be considered as a single massive body , as well as an ensemble of heavy particles in another massive body located in the center of the galaxy .
these massive bodies will interact with each other not only via gravity but also through exchange flows of dark particles .
there is another important fact .
ultra - cold dark particles , apparently can not be registered in terrestrial environments .
indeed , the corresponding wavelengths are large or even of galactic dimensions .
these dark particles do not form bound states with other particles and massive objects .
their impact will be insignificant within the earth and the solar system [ 13 ]
. however , the effect of these particles within large galaxies may already be quite noticeable .
strengthening of effective attractive forces will alter the orbits of massive objects in such galaxies . ;
@xmath141 . ] at even greater distances than the galactic ones , this model provides effective repulsive forces .
it requires further investigation to understand how this effect simulates the dark energy action .
of course , this model of pair interaction is very simple and can be improved .
for example , one can also include in a consideration @xmath145-wave and high waves , and several kinds of dark particles .
accordingly , the sum interactions will be more complex .
the effect obtained within the proposed model imitates the dark matter amplification of interaction forces at very large but definite distances .
the important feature of the resonant amplification is its selectiveness .
the effect is negligible at distances much less or , vice versa , much greater than the resonance distances .
this resonant amplification is based on well - known quantum - mechanical considerations and , in particularly , on the quantum considerations of few - body interactions .
|
a non - relativistic quantum mechanical model for interaction of dark particles with matter is proposed ; the model describes resonant amplification of effective interaction between two massive bodies at large distances between them .
the phenomenon is explained by catalytic action by a third dark particle scattered at a system of two heavy bodies . it is shown that effective interaction between the heavy bodies is changed at larger distances and can transform into repulsion contributing in that case to the dark energy action . * n. takibayev * institute of experimental and theoretical physics al - farabi kazakh national university al - farabi 71 , almaty 050040 , kazakhstan * subject classification : * 03.65nk ; 21.45.+v ; 95.35.+d + * keywords : * dark particles , three - body system , resonance distances
|
cantilevers of micrometer size are nowadays present in many applications , ranging from chemical and biological sensors@xcite to scanning probe microscopy@xcite . in many cases , these tiny mechanical systems operate in a fluid environment ( air or water for instance ) which has a great influence on their dynamical behavior : the viscosity of the medium will broaden the structural resonances while the added mass due to the fluid moving along with the cantilever will shift their frequencies .
a few theoretical models have been proposed to account for these effects @xcite , and validated experimentally @xcite . among the prediction of these approaches ,
the power spectrum of thermal noise induced fluctuations is of particular interest for its applications in atomic force microscopy ( afm ) or micro - electromechanical systems ( mems ) .
we will focus here on the approach by sader et al .
@xcite , and derive a generic formula for the thermal noise using the fluctuation dissipation theorem for each mode , extending the work of paul and cross @xcite in a simpler framework than dorignac et al.@xcite .
this careful analysis is indeed incorrect in the original work of sader et al .
, as noted by paul and cross @xcite , dorignac et al .
@xcite and sader et al . themselves
( see ref .
44 of @xcite )
. we will then compare the prediction of the model to a noise measurement on a commercial afm cantilever in air : an excellent agreement is obtained for the full frequency behavior of the two first resonances , as well as at low frequencies and in the intermediate region between the two pics .
, width @xmath0 and thickness @xmath1 .
deformation are supposed to be only along axis @xmath2 and to depend only of time @xmath3 and spacial coordinate @xmath4 , they are described by the deflection @xmath5 . ] we will first recall the general theory to describe the motion of a cantilever immersed in a viscous fluid and subject to an arbitrary external forcing , using the same notations and approach as j. sader @xcite , and then derive the expression for the thermal noise of the mechanical system .
the cantilever is sketched on fig.[fig : cantilever ] .
its length @xmath6 is supposed to be much larger than its width @xmath0 , which itself is much larger than its thickness @xmath1 .
we will limit ourself in this study to the flexural modes of the cantilever : the deformations are supposed to be only perpendicular to its length ( along axis @xmath2 of fig.[fig : cantilever ] ) and uniform across its width .
theses deformation can thus be described by the deflection @xmath5 , @xmath4 being the spacial coordinate along the beam normalized by @xmath6 , and @xmath3 the time .
the equation of describing the dynamics of @xmath7 is @xcite : @xmath8 where @xmath9 is the young modulus , @xmath10 the second moment of inertia of the cantilever , @xmath11 the mass per unit length and @xmath12 the external force per unit length .
the boundary conditions for eq .
[ eq : beam ] correspond to the classical clamped ( @xmath13 ) and free end ( @xmath14 ) conditions : @xmath15 in the fourier space , we can rewrite eq . [
eq : beam ] as @xmath16 with @xmath17 the pulsation in the fourier space , @xmath18 the mass and @xmath19 the spring constant of the cantilever . solving this equation in absence of external forces leads to the well known normal modes of oscillation of a clamped beam in vacuum with spatial profiles @xmath20 where @xmath21 is the @xmath22 solution of equation @xmath23 which leads to @xmath24 , @xmath25 , , @xmath26 .
the normals modes @xmath27 for @xmath28 form an orthonormal basis of the functions of @xmath4 in @xmath29 $ ] @xcite : @xmath30 the pulsations of these resonant modes are given by the dispersion equations @xmath31
when the cantilever is moving inside a fluid , the force acting on the cantilever can be decomposed into 2 components : the force corresponding to the hydrodynamic load @xmath32 due to the motion of the beam in the fluid , and the actual external driving @xmath33 .
following sader s approach @xcite , the hydrodynamic load can be approximated by @xmath34 where @xmath35 is the hydrodynamic function corresponding to the rectangular cantilever beam and @xmath36 the density of the fluid .
the real part @xmath37 of @xmath38 corresponds to the added mass due the fluid moving along with the cantilever during its motion ( normalized to @xmath39 , the mass of a cylinder of fluid of diameter @xmath0 and length @xmath6 ) , and the imaginary part @xmath40 to the viscous drag . to solve the equation of motion of the cantilever in the fluid ,
let us decompose the deflection @xmath41 and external driving @xmath33 on the orthonormal basis of the normal modes @xmath42 : @xmath43 we note here that this decomposition is also valid in the real time space , with amplitude @xmath44 and external driving @xmath45 of each mode simply being the inverse fourier transform of @xmath46 and @xmath47 .
these 2 variables are coupled by the hamiltonian @xmath48 of the system .
indeed , let us compute the infinitesimal work @xmath49 of @xmath50 when the deflection changes by @xmath51 : @xmath52 for a reversible transformation , we can write @xmath53 , hence @xmath54 this last equation shows that the amplitude @xmath55 and external driving @xmath56 of each mode are coupled variables by the hamiltonian of the system .
we can thus apply the fluctuation dissipation theorem @xcite : the power spectrum density @xmath57 of fluctuation of amplitude for mode @xmath58 is given by @xmath59 where @xmath60 is the boltzmann constant , @xmath61 the temperature of the thermostat and @xmath62 is the imaginary part of its argument .
we will now explicit the right part of eq .
[ eq : s_beta_n - fdt ] . using the orthonormalization of the base @xmath42
, we can easily show that eq .
[ eq : beam fourier ] leads to the following equation for each mode : @xmath63 the amplitude @xmath55 of each mode is thus governed by the equation of motion of an oscillator with stiffness @xmath64 , mass @xmath65 and damping coefficient @xmath66 , forced by external driving @xmath56 . if @xmath35 is a slow varying function of the frequency , close to its resonance each mode behaves like an harmonic oscillator . using the above expression , eq .
[ eq : s_beta_n - fdt ] can be rewritten as @xmath67 with @xmath68 . in general , dissipation couples the modes calculated from the non - dissipative equations , but within the approximation of position independent dissipative forces ( eq [ eq : f_hydro ] ) , the noises are uncorrelated
. we can thus write a generic expression of the power spectrum density of the deflection using eq .
[ eq : w = sum beta_n phi_n ] : @xmath69 this last equation is in agreement with the one derived by dorignac et al .
@xcite in a more complex framework .
to test the validity of the above formalism , we measure the thermal noise of a commercial afm cantilever ( budget sensor bs - cont ) with the following nominal geometry : @xmath70 , @xmath71 , @xmath72 .
the cantilever is made of silicium and is coating - less , thus for theoretical expressions we used the tabulated values of si for the young modulus ( @xmath73 for si@xmath74 , the usual crystalline orientation along the length of afm cantilever ) and density ( @xmath75 ) .
the length and width were checked with an optical microscope , and the thickness is inferred from the resonance frequency of the cantilever ( from eq .
[ eq : omega_n ] ) , which eventually leads to @xmath70 , @xmath76 , @xmath77 .
the measurement is performed with a home made interferometric deflection sensor @xcite , inspired by the original design of schonenberger @xcite with a quadrature phase detection technique @xcite : the interferences between the reference laser beam reflecting on the base of the cantilever and the sensing laser beam on the free end of the cantilever directly gives a calibrated measurement of the deflection @xmath78 , with very high accuracy .
we illustrate in fig.[fig : measurement ] the performance of our detection with the power spectrum density ( psd ) of a rigid mirror ( in black ) : the background noise of our system is as low as @xmath79 , just @xmath80 higher than the shot noise limit of our detection system .
typical noise of a measurement in a commercial afm would be at least two orders of magnitude higher . to @xmath81 range .
we clearly see the two first flexural modes of the mechanical beam , well above the background noise of the apparatus acquired on a rigid mirror .
the difference between the measurement and the background noise spectrums gives a full view of the behavior of the thermal fluctuations of the cantilever in this frequency window .
for comparison , the typical noise floor of a well tuned and calibrated commercial afm is plotted : only the resonances could be studied , the remaining part of the spectrum would be useless . ] and eq .
[ eq : psdw ] of the present work .
insets present zooms around the resonances . ]
we plot in fig.[fig : measurement ] the psd of thermal noise driven deflection at the free end of the cantilever ( in green ) . we clearly see in this spectrum the two first modes of oscillation of the mechanical beam , around @xmath82 and @xmath83 , but we also have a lot of information out of these resonances as the measurement is always above the background noise of the system .
if we subtract this background noise spectrum from the measurement ( blue curve ) , we have an estimation of the thermal noise of the system on the whole @xmath84 to @xmath81 frequency interval , which we can compare to the theoretical expectation of the model presented in the previous section
. the same measurement in a commercial afm would yield much less information , as it would be limited to the resonances , and we would not have any information between them or at low frequency . in fig.[fig :
model ] , we compare the measured psd of fluctuations ( in blue ) with the prediction of 3 different models ( in red ) for the thermal noise of the cantilever , all based on sader s approach @xcite . the hydrodynamic function @xmath35 used for these predictions
was taken from eq .
20 , 21a and 21b of ref .
@xcite , as it correspond to the first two resonances and the extension to higher order modes of ref .
@xcite is not necessary .
the first model ( fig.[fig : model](a ) ) corresponds to eq .
29a of the original article of sader @xcite , where the frequency dependence of the thermal noise forcing as been omitted .
both resonances are really well described with this model , but the off - resonance behavior is not : the low frequency noise measured is not flat as expected by the model , which also slightly underestimates the noise the intermediate region .
the second model ( fig.[fig : model](b ) ) corresponds to eq .
17a of ref .
@xcite , a corrected expression of the thermal noise proposed by green and sader ( see ref .
44 of @xcite ) to account for frequency dependance of the dissipation following the approach of paul and cross @xcite .
let us first note that this expression is dimensionally incorrect , as it is homogeneous to @xmath85 and not to @xmath86 , so we had to rescale the formula by @xmath87 to match the first resonance .
however , this model perfectly describes the frequency dependance of the first mode , catching the slow variation of the noise in the low frequencies .
the behavior of the second mode could also be well described , but it would need a higher rescaling factor .
we came to the conclusion that the scaling factor is mode dependent , and conducted the analysis of the previous part to derive the correct expression .
indeed , it can easily be shown that the frequency behavior of each mode ( eq . [ eq : s_beta_n ] ) is exactly the same as in ref .
@xcite , but the scaling factors are different .
we plot in ( fig.[fig : model](c ) ) the prediction of eq .
[ eq : s_beta_n ] and eq .
[ eq : psdw ] , and see that we match the whole power spectrum density of fluctuation of deflection : both resonances , low frequency and intermediate region are very well described by the model .
we stress here that the only parameter we have adjusted in the model is the thickness of the cantilever @xmath1 , to match the actual frequencies of resonance . the sader approach to describe the effect of the embedding fluid on the behavior of
the cantilever turns out to be very good on the whole frequency range probed here .
in this paper , we have presented a simple theoretical framework to describe the thermal noise of a mechanical beam in a viscous fluid . under the assumption of an infinitely thin and long cantilever
, we used the sader approach to describe the effect of the surrounding fluid ( added mass and viscous drag ) . using the fluctuation dissipation theorem for each flexural modes of the system
, we derived a general expression for the power spectrum density of fluctuations as the sum of the contribution of the different modes .
this prediction has been compared with an experimental measurement on a commercial afm cantilever in a frequency range covering the 2 first resonances .
a very good agreement has been found on the whole spectrum , with no adjustable parameters but the thickness of the cantilever .
this analysis can be very useful in the afm area , where the thermal motion of the cantilever is of great practical importance ( as a lower bound to measurable forces or as a mechanical driving on its own ) : it is not limited to a fit of resonances only , but gives an _ a priori _ knowledge of the cantilever noise on the whole frequency spectrum .
i thank f. vittoz and f. ropars for technical support , and p. paolino , s. ciliberto , a. petrosyan , j.p .
aim and f. bertin for stimulating discussions .
this work has been partially supported by contract anr-05-blan-0105 - 01 of the agence nationale de la recherche in france .
|
we present a simple theoretical framework to describe the thermal noise of a microscopic mechanical beam in a viscous fluid : we use the sader approach to describe the effect of the surrounding fluid ( added mass and viscous drag ) , and the fluctuation dissipation theorem for each flexural modes of the system to derive a general expression for the power spectrum density of fluctuations .
this prediction is compared with an experimental measurement on a commercial atomic force microscopy cantilever in a frequency range covering the two first resonances .
a very good agreement is found on the whole spectrum , with no adjustable parameters but the thickness of the cantilever .
|
for a self - ignited tokamak plasma , confinement of alpha particles is a crucial issue .
the alpha particles must effectively heat the bulk plasma without a spontaneous loss .
alpha particles heat electrons first and then ions through electron - ion thermal equilibriation .
recent gyrokinetic simulation studies have pointed out that the microinstabilities strongly affect the high energy particles transport@xcite .
another gyrokinetic study of ion temperature gradient turbulence suggested that the diffusion of the high energy particles is small@xcite .
note that the ion temperature gradient turbulence is generated by selfconsistent particle - in - cell method by the thermal ions , while the high energy particles effect on the turbulence is absent . in this work ,
we study transport of alpha particles in the presence of driftwave turbulence .
we employ the guiding center orbit calculation with the toroidal driftwave model@xcite .
the model provides us with good tools to understand the basic mechanism of the alpha particle transport . by our guiding center calculation , we are able to compare analytical aspects and numerical work more closely . furthermore , the model is flexible so that we can apply the simulation to large tokamak size , for example iter .
our recent work@xcite clarified the onset condition of guiding center stochasticity for the passing particles ( guiding center islands widths are inverse proportional to the square root of parallel velocities ) . on the other hand ( while the resonance in the phase velocity is possible ) the radial excursion of trapped particles is limited in finite radial domains .
in this work , we conduct numerical experiments on collisional effects on the high energy particles .
the small collisional effects can free the trapped particles from the limited radial domain .
this paper is organized as follows . in sec.2
, we describe the computational model and present our orbit following calculation results in sec.3 . in sec.4
, we discuss the collisional effects .
we summarize this work in sec.5 .
for the guiding center equation , we employ littlejohn s guiding center lagrangian @xmath1@xcite in the mks unit @xmath2 the guiding center position is given by @xmath3 , @xmath4 is magnetic moment , @xmath5 is the gyro - phase , and @xmath6 is the parallel velocity .
the values @xmath7 and @xmath8 are particle s mass and charge , respectively . here , @xmath9 is the effective vector potential , while @xmath10 is the effective magnetic field .
the unit vector along magnetic field is given by @xmath11 . here
, @xmath12 represents the electrostatic potential , @xmath13 is the magnetic field strength . from eq.([eq :
lagrangian - hamiltonian ] ) , euler - lagrange equation gives rise to the equation of motion along with the definition of parallel velocity , magnetic moment , and the gyro - phase .
the guiding center equation is given by @xmath14 and the acceleration along the magnetic field by @xmath15 where @xmath16 is the effective electric field and @xmath17 .
the dot operator is for the time derivative .
hereafter , @xmath18 represents time .
the ballooning type electrostatic perturbation is employed in the orbit - following calculation@xcite .
the ballooning mode posses translational invariance@xcite : fourier modes with different @xmath19 ( poloidal mode number ) but the same @xmath20 ( toroidal mode number ) are correlated . while modes with different poloidal / toroidal numbers are orthogonal in a cylindrical geometry , in a toroidal geometry , due to the toroidal curvature @xmath21 ( @xmath22 is the major radius , @xmath23 is @xmath22 at the magnetic axis , @xmath24 and
@xmath25 are minor radius and poloidal angle respectively , @xmath26 is the inverse aspect ratio ) , the amplitude of neighboring poloidal modes `` @xmath19 '' ( with the same toroidal mode number `` @xmath20 '' , @xmath27 ) , are correlated with each other and grow at the same rate@xcite .
we describe the turbulence potential @xmath12 by@xcite @xmath28 \nonumber \\ & \times&\cos\left ( m \theta - n \zeta \pm { \omega}_{\star n } t + \delta_n \right ) \end{aligned}\ ] ] where @xmath29 is the toroidal angle and @xmath30 is the radial location of the mode rational surfaces .
the amplitudes of @xmath31 ( at the order of @xmath32 of the electron temperature ) , the diamagnetic frequencies @xmath33 and the phase of the mode @xmath34 are given by a random number generator within physically reasonable range . here , @xmath35 is the gaussian localization parameter@xcite .
we parameterize @xmath19 and @xmath20 so that the shortest wavelength of turbulence must be larger than that gyro - radius of particle . a contour plot of @xmath36 by eq.(4 )
is given in fig.1 . and
minor radius at the wall is given by @xmath38.,width=245 ]
we now discuss the behavior of passing and trapped particles in the presence of background fluctuation . in the calculation
we use an iter like profile , @xmath39 , minor radius @xmath40 , and @xmath41 at the magnetic axis .
figure.2 is a poincar plot of guiding center of passing particles at a toroidal angle @xmath42 for 1.5kev passing particles .
the pitch angle is taken as @xmath43 ( @xmath44 is the total velocity ) . in fig.2 , we turned the curvature term off ( mirror force and the gradient - b drift terms are automatically zero when @xmath45 and thus @xmath46 ) for the demonstration purpose , but the particles still respond to the background fluctuation through the @xmath0 term . in fig.2 ,
perturbation modes of @xmath47 with a single @xmath48 mode are incorporated [ see eq.(4 ) ] . in the presence of multiple @xmath20 fluctuation
, the passing particles will be stochastic due to dense formation of guiding center islands and the overlapping of the islands ( see fig.2 at @xmath49)@xcite .
for the passing particles , the transport is much larger for thermal ions than the alpha particles , because the island width is proportional to @xmath50@xcite .
this is because the net @xmath0 drift for resonating particles is larger for the slower particles which spend longer time at the same phase of the driftwave .
note that the latter effect is different from the idea of _ orbit averaging_@xcite which is a crude averaging idea only applied for toroidal ripple toroidal modes `` @xmath51 '' ( no resonance condition considered)@xcite .
the finite @xmath52 terms in eq.(4 ) gives rise to the radial shift of the resonant location [ positive ( negative ) @xmath52 for upshift ( downshift ) in @xmath24 ] which match with the analytical estimation@xcite . in this paper , we set @xmath54 , however . for @xmath55 deuterium ions
the pitch angle is taken as @xmath43 and the initial position of the particles are at @xmath56 . to demonstrate the guiding center formation the mirror force and the magnetic curvature terms are off but the @xmath0 drift effects is kept .
the number of island chains corresponds to the poloidal mode number @xmath19 while having the same toroidal mode number @xmath20 .
the safety factor profile is given by @xmath57 .
note that the widths of the guiding center islands are inverse proportional to the square root of parallel velocities.,width=245 ] in fig.3 , poincar plots of trapped @xmath55 deuterium ions are shown ( at @xmath58 ) by keeping all the drift terms and the mirror force .
toroidal modes of @xmath59 are incorporated .
near the banana tips , @xmath60 is close to zero so that particle is sensitive to the @xmath0 drift .
the driftwave effects , as in the passing particle case , are larger for the lower energy particles .
note that the @xmath0 drift itself is not particle s energy dependent but the energy dependence comes in as a combination of the parallel and perpendicular motion . if the parallel velocity is small , they will stay for a long time at the same phase of the perturbation , and drift largely in the radial direction .
the trajectories of trapped particles are not resonant and do not form any islands .
the trajectories are distorted ( invariant tori ) without a topological change . in fig.4 ,
poincar plots of trapped @xmath61 alpha particles are shown .
figure 4 demonstrates that the second adiabatic invariant ( `` @xmath62 '' ) for trapped particles is not easy to break by the background fluctuation ( except for those due to the resonance in the velocity space@xcite ) . in fig.3 , lower energy trapped particles
are seemingly diffusive but by if one follows the particle orbits for a longer time , they can end up in a long term periodic motion .
the kolmogorov entropy@xcite for the trapped particles in fig.3 are estimated which does not necessarily show exponential divergence ( even for the @xmath55 ions ) .
the radial excursion of trapped particles is limited in finite radial domains . in the next section ,
we investigate collisional effects which can free the trapped particles from the limited radial domain . for @xmath55 deuterium ions .
the pitch angle is taken as @xmath63 .
all the drift terms are kept .
the plot absent region in the center ( @xmath64 ) corresponds to the strong magnetic field side of the tokamak .
particles are set initially at @xmath65 and equi - distantly along the @xmath24 axis.,width=245 ] for @xmath61 alpha particles .
the rest of the settings is exactly the same with fig.3.,width=245 ]
we employ a monte carlo collision operator in the orbit following code .
the density and the temperature profile for the bulk ions and electrons are assumed to be @xmath66 and @xmath67 where @xmath68 and @xmath69 .
the collision effects are given by the energy scattering@xcite and the pitch angle scattering@xcite which obeys the lorentz collision operator@xcite .
the collision frequency for the energy scattering by the electrons is given by the inverse of the energy relaxation time@xcite while the collision frequency for the pitch angle scattering ( test ion particles against bulk ions ) is given by the inverse of the perpendicular momentum relaxation time@xcite .
the latter collision frequency is changed as a function of the test particle energy .
the diffusion coefficients @xmath70 in mks units are obtained from estimating the second order cumulant@xcite .
figure 5 shows the the diffusion coefficients versus test ion energy , for ( black ) trapped particles without collisions but with the turbulent background , and ( red ) trapped particles with collisions and the turbulence . in estimating the diffusion , @xmath71 ( @xmath72 is the initial radial location and @xmath73 is the sampling time ) , we have set @xmath74 test particles separated equi - distantly in @xmath75 at @xmath76 ( where the background bulk temperature is @xmath77 ) and @xmath78 with an initial pitch angle @xmath79 , which are all trapped particles . to subtract the finite banana width effects ( which is large for high energy particles ) from the diffusion estimation , radial displacement
is measured when the banana motions cross @xmath80 after each bounce motion . in the guiding center calculation ,
all the drift terms are incorporated .
the toroidal modes of @xmath59 are employed .
the maximum amplitude of the electrostatic fluctuation is given by @xmath81 where @xmath82 is the unit charge .
as shown in fig.5 , the diffusion level decreases as the energy increase for both the curves ( the same trend is found for the passing particles@xcite ) .
the collisional effects are visible toward the lower energy side , @xmath83 .
note that both the abscissa and the ordinate are in the logarithmic scale .
although the collision frequencies are very small in the high temperature collisionless plasmas , it is suggested that the collisions ( pitch angle scattering per se ) can possibly remove the trapped particles from a long term periodic motion . versus test ion energy @xmath84 ; ( black curve )
trapped particles without collisions but with the turbulent background , and ( red curve ) trapped particles with collisions and the turbulence .
initial positions of the test particles are given by @xmath76 , @xmath78 , and @xmath75 .
pitch angle of @xmath79 taken for all the cases.,width=283 ]
in this work , we have studied high energy particle transport in the presence of toroidal driftwave turbulence . to model more realistic
trapped particle transport , energy and pitch angle scattering are incorporated .
it is suggested that small collisional effects can change the transport process of the trapped particles .
the authors would like to thank discussions with dr .
s. satake of national institute for fusion science .
one of the authors yn would like to thank discussions with dr .
this work is supported by national cheng kung university top university project and by national science council of taiwan , nsc 100 - 2112-m-006 - 021-my3 .
9 c. estrada - mila , j. candy , and r. e. waltz , phys .
plasmas * 13 * , 112303 ( 2006 ) .
w. zhang , z. lin , and l. chen , phys .
* 101 * , 095001 ( 2008 ) .
r. b .white and h. mynick , phys .
fluids b * 1 * , 980 ( 1989 ) .
j. w. connor , r. j. hastie , and j. b. taylor , proc .
london a * 365 * , 1 ( 1979 ) . c. z. cheng and l.chen , phys .
fluids * 23 * , 1770 ( 1980 ) .
y. nishimura and m. azumi , phys .
plasmas * 4 * , 2365 ( 1997 ) . b. huang , _ fusion born alpha particle simulation with background turbulence by gpu computing _ , master s thesis , national cheng king university , taiwan ( 2013 ) .
r. g. littlejohn , j. plasma phys .
* 29 * , 111 ( 1983 ) . l. e. zakharov , proc .
innsbruck , vol * 1 * , iaea , vienna , 689 ( 1978 ) . m. azumi and m. wakatani , j. plasma and fusion res . * 66 * , 494 ( 1991 ) .
b. huang , y. nishimura , and c. z. cheng , _ alpha particle transport in the presence of toroidal driftwave turbulence _ , submitted ( 2013 ) ; http://www-jt60.naka.jaea.go.jp/next/next19/[email protected] .
y. nishimura , computer physics communications * 182 * , 158 ( 2011 ) .
y. nishimura , y.xiao , and z.lin , contributions to plasma physics * 48 * , 224 ( 2008 ) .
a. h. boozer and g. kuo - petravic , phys .
fluids * 24 * , 859 ( 1981 ) .
k. miyamoto , _ plasma physics for nuclear fusion _ ,
( iwanami , tokyo , 1987 ) , p.83 and p.85 . y. nishimura , ieee transactions on plasma science * 38 * , 3442 ( 2010 ) .
|
high energy particles behavior including fusion born alpha particles in an iter like tokamak in the presence of background driftwave turbulence is investigated by an orbit following calculation .
the background turbulence is given by the toroidal driftwave eigenmode combined with a random number generator .
the transport level is reduced as the particle energy increase ; the widths of the guiding center islands produced by the passing particles are inverse proportional to the square root of parallel velocities . on the other hand ,
the trapped particles are sensitive to @xmath0 drift at the banana tips whose radial displacement is larger for lower energy particles .
coulomb collisional effects are incorporated which modifies the transport process of the trapped high energy particles whose radial excursion resides in limited radial domains without collisions .
|
since the discovery of the higgs boson @xcite special effort has been made to determine its properties .
the motivation for such an effort resides on understanding the mechanism for electroweak symmetry breaking . at present ,
several aspects of the higgs boson are to some extent well known , in particular those related with some of its expected `` standard '' decay modes , namely : @xmath7 , @xmath8 , @xmath9 , @xmath10 and @xmath11 .
currently , measurements of these decay modes have shown compatibility with the standard model ( sm ) expectations , although with large associated uncertainties @xcite . indeed , it is due to these large uncertainties that there is still room for nonstandard decay properties , something that has encouraged such searches at the lhc as well .
searches for invisible higgs decays have been published in @xcite , while direct searches for lepton flavor violating higgs decays ( @xmath12 ) have been recently reported by the cms collaboration in @xcite . in this letter
we focus on the latter , for which the cms collaboration , using the 2012 dataset taken at @xmath13 tev with an integrated luminosity of 19.7 fb@xmath14 , has found a 2.5@xmath1 excess in the @xmath15 channel , which translates into br@xmath16%
. indirect bounds on higgs lepton flavor violating decay modes arise from low - energy data .
muon and tau rare decays ( e.g. @xmath17 , @xmath18 , @xmath19 and @xmath20)induced by higgs lepton flavor breaking couplings
place upper bounds on the higgs flavor violating modes : @xmath12 , @xmath21 , @xmath22 .
since muon decays have the most tight limits , it is for the @xmath23 mode for which consistency with low - energy data demands a branching fraction well below the lhc reach ( @xmath24 ) .
constraints on tau rare processes , being less stringent , allow larger @xmath25 and @xmath26 branching ratios @xcite , hence stimulating these searches at the lhc .
although these bounds follow from a fairly model - independent analysis ( see @xcite and references therein ) , one may also wonder what type of frameworks are capable of producing sizeable lepton flavor violating higgs decays .
efforts in such direction have been done in different contexts , with pioneer works in refs .
@xcite . more recenty , ref .
@xcite studied the problem in the mssm , while @xcite in the r - parity violating mssm .
flavor violating decays have been considered as well in the inverse seesaw model in @xcite .
possible effects due to vectorlike leptons have been investigated in @xcite .
extended scalar sectors involving several higgs doublets and flavor symmetries ( yukawa textures ) have been examined too @xcite . finally ,
the type - iii two higgs doublet model ( 2hdm ) has been considered in refs .
basically , the bottom line of all these analyses is that unless one deals with extra higgs doublets , lepton flavor violating higgs decays are below the lhc reach . in this paper
, we study the viability of producing the cms excess signal in the type - iii 2hdm . for simplicity , we assume flavor violation only in the lepton sector and adopt a pure phenomenological approach , that is to say all the parameters of the model are treated as free parameters subject only to phenomenological constraints .
the phenomenological restrictions we consider are the following .
first of all , the new degrees of freedom are constrained by electroweak precision data .
thus , in our analysis we calculate the contributions to the @xmath27 parameter ( contributions to the @xmath28 and @xmath29 parameters in the 2hdm are small @xcite ) . since the couplings that induce @xmath30 induce as well tau rare processes , we consider the restrictions arising from @xmath31 ( for which we include barr - zee diagrams contributions @xcite as done in @xcite ) , @xmath32 and @xmath33 .
we also study whether our scenario can explain the @xmath34 discrepancy .
in contrast to previous studies in the 2hdm , we also verify that the higgs standard properties deviate only within the allowed measured ranges @xcite . these include the following higgs decay channels : @xmath11 , @xmath10 , @xmath4 and @xmath5 .
we check as well for theoretical constraints , namely that the scalar potential is bounded from below and contains perturbative parameters .
the rest of the paper is organized as follows . in sec .
[ sec:2hdm - description ] we discuss basic aspects of the type - iii 2hdm and introduce the formulas employed in our numerical calculation . in sec .
[ sec : pheno ] we describe the strategy followed in our numerical analysis and present our results . finally , in sec .
[ sec : conclusions ] we summarize and present our conclusions .
we consider a 2hdm @xcite of type - iii .
contrary to other versions of the 2hdm , the type - iii 2hdm does not include any discrete symmetry that serves to distinguish between higgs doublets .
therefore , both higgs doublets are allowed to couple to all fermion species .
it is common to present the 2hdm in an arbitrary basis in higgs space . instead ,
following @xcite , we prefer to introduce the model in the so - called higgs basis . in this basis , only one higgs doublet acquires a vacuum expectation value ( vev ) and the scalar potential of the model ( assuming cp conservation ) is given by @xmath35 \ , , \end{aligned}\ ] ] where @xmath36 are the higgs doublets in the higgs basis , such that @xmath37 and @xmath38 . here
@xmath39 and @xmath40 are cp - even neutral higgs fields , @xmath41 is a cp - odd neutral higgs field , @xmath42 is a charged higgs field and @xmath43 and @xmath44 are goldstone bosons . since we assume cp conservation , @xmath41 is the physical pseudoscalar higgs and does not mix with @xmath39 and @xmath40 .
the relation between @xmath39 and @xmath40 and the scalar mass eigenstates @xmath45 and @xmath46 ( with @xmath47 ) is @xmath48 here we have introduced @xmath49 , the physical mixing angle that relates the higgs basis and the mass basis for the cp - even scalar states .
the potential parameters are related to the physical higgs masses as @xmath50 ^ 2 + 4 \ , \lambda_6 ^ 2 v^4 \ , , \\
\sin \left [ 2(\beta - \alpha ) \right ] & = & - \frac{2 \ , \lambda_6 v^2}{m_h^2 - m_h^2 } \ , . \label{eqlast}\end{aligned}\ ] ] we now turn to the yukawa interactions of the model . in the higgs basis for the higgs doublets and the mass basis for the fermions , the yukawa lagrangian of the model can be written as @xmath51 here we denote @xmath52 , the fermions @xmath53 are mass eigenstates , @xmath54 is the ckm matrix and @xmath55 are generation indices .
@xmath56 , @xmath57 and @xmath58 are the up - type quark , down - type quark and charged lepton masses , respectively , and @xmath59 , @xmath60 and @xmath61 are general @xmath62 complex matrices in flavor space . for simplicity ,
in the following we will assume that the @xmath63 matrices are hermitian .
now , using eqs . , and we can rewrite the leptonic part of @xmath64 as @xmath65 \
, , \label{eqyuk}\end{aligned}\ ] ] where @xmath29 is the pmns matrix , @xmath66 is the usual right - handed chirality projector , @xmath67 and @xmath68 . from eq .
we can extract the higgs couplings to fermions @xmath69 where @xmath70 for down - type quarks and charged leptons and @xmath71 for up - type quarks .
finally , we will consider a specific structure for the @xmath63 matrices inspired in the _ cheng - sher ansatz _
first , we normalize @xmath61 as @xmath72 where @xmath73 and @xmath74 is the physical mixing angle defined by the ratio since one can always apply rotations in higgs space , acting also on the higgs vevs , @xmath75 and @xmath76 .
therefore , a specific ( physical ) definition is required @xcite .
since we are interested in tau flavor violation , we choose to define @xmath77 as the relative size of the tau yukawa coupling and @xmath78 , in analogy to the usual definition of @xmath79 in the type - ii 2hdm .
[ foot : tanb ] ] @xmath80 note that , by definition , @xmath81 .
however , the other @xmath82 s , and in particular , @xmath83 , are free parameters . for the quark @xmath63 matrices
we assume the following type - ii values @xmath84 this parameterization of the @xmath63 matrices is not the most general one but , as we will see , it leads to results in good agreement with the experimental constraints .
furthermore , this ansatz can be understood as a minimal correction beyond the type - ii 2hdm , with the only departure in the @xmath0 coupling @xcite .
finally , the higgs couplings to gauge bosons are fully dictated by the gauge symmetry .
one has @xmath85 , @xmath86 and @xmath87 .
the couplings to a pair of @xmath88-bosons follow the same proportionality .
we now proceed to describe our phenomenological analysis .
our results are based on a random scan of the parameter space , with the following ranges for the relevant input parameters : @xmath89 \ , \text{gev}\ , \qquad m_a \subset \left [ 400 , 1000 \right ] \ , \text{gev}\ , \\ m_{h^\pm } & = m_a + \delta m \ , \qquad \text{with}\qquad \delta m \subset \left [ -5 , 5 \right ] \ , \text{gev}\ ,
\\ \sin ( \beta - \alpha ) & \subset [ 0.7 , 1.0]\ , \qquad \tan \beta_\tau \subset [ 0.1 , 40]\ , \\ |\kappa_{\tau \mu}| & \subset [ 0.1 , 3]\ .
\label{eq : parameter - ranges - last}\end{aligned}\ ] ] in addition , we assume @xmath90 gev . since we are driven by phenomenological considerations , we use as input the scalar masses ( @xmath91 , @xmath92 , @xmath93 and @xmath94 ) rather than the parameters in the scalar potential .
our choice of a small mass difference between the pseudoscalar and charged higgses is motivated by the reduction of the @xmath27 parameter , see @xcite for details . ] , whereas the lower limits on @xmath95 and @xmath94 ( as induced by the lower limit on @xmath93 ) are motivated by experimental constraints : smaller values of @xmath95 would be excluded by current measurements of the higgs couplings to fermions and gauge bosons while lower @xmath94 would lead to certain tension with flavor physics bounds ( mainly b physics , see @xcite for a review ) .
we emphasize that the mass ranges selected for our numerical scan are quite conservative since lower values for @xmath92 , @xmath93 and @xmath94 are allowed by lhc data , see e.g. @xcite . after choosing input parameters three checks
are performed : * we determine @xmath96 , @xmath97 , @xmath98 and @xmath99 by means of eqs . - and only allow those parameter points where the resulting values are below @xmath100 .
the exact value of this perturbativity constraint is somehow arbitrary ( see for example the discussion in @xcite ) .
however , we have checked that it has no impact on our numerical results .
furthermore , we also ensure boundedness from below @xcite by properly choosing the remaining @xmath101 parameters .
* we explicitly compute the @xmath27 parameter and discard parameter points outside the current 1 @xmath1 region , given by @xmath102 $ ] @xcite . * finally , we check whether the higgs couplings agree with the current cms measurements @xcite
. more precisely , we determine the signal strengths for @xmath103 , defined as @xmath104 , and compare them to the cms 1 @xmath1 ranges @xcite .
we assume that higgs production is given by gluon fusion ( neglecting other production mechanisms , a fairly good approximation ) and compute the signal strengths as @xmath105 here @xmath106 and @xmath107 are the higgs boson couplings to a pair of top quarks and @xmath108 , respectively .
the superscript sm indicates standard model values .
these constraints impose severe restrictions on the allowed parameter space of the model ( most parameter points are actually excluded ) and their consideration is required to properly address the type - iii 2hdm as responsible for the cms @xmath109 signal excess . after these checks are passed ,
the observables are computed using the analytical expressions of @xcite . in particular , we compute the branching ratio for @xmath109 as @xmath110 where @xmath111 is the higgs boson total decay width turns out to be close to the sm value , @xmath112 mev . ] . for the analytical expressions required for the computation of br@xmath113 we refer the reader to appendix [ ap : taumugamma ] .
[ fig : htmu ] shows our results for br@xmath114 as a function of br@xmath113 .
the horizontal lines define the 1 @xmath1 interval around the observed br@xmath114 at cms @xcite .
the vertical lines show the current experimental limit set by babar , br@xmath115 @xcite , and an expected sensitivity at belle - ii of about @xmath116 @xcite .
as can be clearly seen , there is a correlation between br@xmath114 and br@xmath117 .
however , cancellations among the different contributions to br@xmath113 preclude any definitive prediction for this observable .
we find that the dominant contribution is typically given by 2-loop barr - zee diagrams @xcite with internal @xmath118 bosons , although the other contributions considered in our numerical evaluation . ]
may have similar sizes ( and even become dominant in some cases ) .
we conclude that it is possible to explain the cms excess in @xmath109 , while being compatible with the current bound on br@xmath113 as well as the abovementioned constraints on the scalar potential , the higgs couplings and the @xmath27 parameter .
this is the main result of this paper .
regarding the required values for the model parameters , a sizable @xmath119 is necessary .
this is the key parameter in the determination of br@xmath114 .
in contrast , we find little dependence with the other parameters ( when taken individually ) , although specific ranges are favored by the experimental constraints . in what concerns @xmath77
, a value @xmath120 is necessary to be compatible with the constraints on the higgs boson couplings .
the @xmath49 angle should departure from @xmath121 ( which would correspond to the decoupling limit ) and have values in the @xmath122 ballpark .
finally , we find a preference for large @xmath101 parameters , required to generate a hierarchy in the scalar spectrum . as a function of br@xmath123 for the relevant parameters fixed according to eqs . - .
the horizontal lines show the 1 @xmath1 interval around the best - fit value for br@xmath114 , as observed by cms @xcite .
the right vertical line corresponds to the current experimental limit br@xmath124 @xcite , whereas the left vertical line represents an expected sensitivity at belle - ii of about @xmath125 @xcite .
[ fig : htmu ] ] in fig .
[ fig : signal ] we present our results for the allowed signal strengths in the @xmath126 , @xmath127 and @xmath128 channels . as usual , they are normalized to their sm values and thus the vertical line at @xmath129 represents the sm prediction .
the purple bars cover the complete 1 @xmath1 ranges compatible with measurements by cms @xcite , with the black dots at the best - fit values .
the orange bars represent the allowed signal strengths in the type - iii 2hdm as required to obtain br@xmath130% .
we do not show the results for the @xmath131 channel because they do not provide any information , as the purple and orange bars extend over the same range .
the most interesting results are those for the @xmath127 and @xmath128 channels , @xmath132 $ ] . for both , a measurement implying @xmath133 would rule out our scenario .
in contrast , for the @xmath126 channel both @xmath134 and @xmath135 are compatible with the cms measurement of br@xmath136 , although the allowed bar shrinks to a narrower range , @xmath137 $ ] .
nevertheless , we note that the results for this channel are related to the ansatz assumed in eq . .
finally , we have also considered the lfv processes @xmath138 and @xmath139 .
however , we have found that the constraints on the model derived from their experimental bounds are weaker than those obtained from the radiative @xmath140 .
furthermore , our numerical results show that the @xmath34 anomaly can not be explained within our scenario , in agreement with @xcite . in agreement with
the signal found at cms ( within 1 @xmath1 ) while at the time satisfying electroweak precision data ( @xmath27 parameter constraints ) as well as low - energy data .
the purple ( thin ) bars cover the 1 @xmath1 uncertainties obtained by cms ( with the black dots at the best - fit values ) , whereas the orange ( thick ) bars represent the allowed signal strengths in the type - iii 2hdm as required to obtain br@xmath141% .
[ fig : signal ] ]
the excess of signal events in the @xmath26 higgs boson decay channel reported by the cms collaboration poses a challenge for beyond sm physics models .
basically , apart from models involving extra higgs doublets , no models capable of producing higgs lepton flavor violating measurable at lhc have been put forward .
certainly , the 2hdm is the simplest of such extensions , and in its type - iii `` incarnation '' combines the elements that in principle can yield an explanation to cms data . driven by this motivation , in this paper
, we have done a detailed study of the type - iii 2hdm . in our analysis
, we have payed special attention to all relevant constraints , which include : ( i ) boundedness from below and perturbativity of the scalar potential ; ( ii ) electroweak precision data ; ( iii ) low - energy @xmath142 lepton flavor violating decay constraints ; ( iv ) standard higgs decay channels .
our findings show that generating a large @xmath12 branching fraction , as required by cms data , turns out to be possible while satisfying criteria ( i)-(iv ) .
conditions in ( i ) do not lead to significant contraints due to large parameter freedom .
conditions in ( ii ) and ( iii ) do not represent either any unavoidable restriction .
we have found that the most stringent constraints arise from requiring the higgs boson to obey current limits on its standard decay channels : @xmath10 , @xmath143,@xmath4 , @xmath5 , although never threatening the production of the cms signal . in summary ,
in this paper we have demonstrated that the type - iii 2hdm with a _ cheng - sher ansatz_-inspired flavor structure naturally accounts for the cms excess signal . of course , if this excess persist pinning down its origin will require an extensive experimental effort
. however , we stress that it might be the first indication of a more complex scalar structure .
moreover , since _ cheng - sher _ flavor structures emerge from flavor symmetries , these data might even reveal the presence of a fundamental flavor symmetry @xcite .
das and av want to thank martin hirsch for drawing their attention to the cms results and sacha davidson for the illuminating and inspiring conversations that led to this work , for her suggestions , encouragement and comments on the manuscript .
av is grateful to alejandro celis for fruitful discussions on the subject of higgs lepton flavor violation in the 2hdm .
we would like to thank as well ben oleary and igor ivanov for comments .
das would like to acknowledge financial support from the belgian fnrs agency through a `` charg de recherche '' contract .
av acknowledges partial support from the expl / fis - nuc/0460/2013 project financed by the portuguese fct .
in this appendix we give the required analytical expressions for the computation of the @xmath140 rate .
this radiative process is induced by the dipole operator @xmath144 where @xmath145 are the flavors of the external leptons , @xmath146 is the electromagnetic strength tensor and @xmath147 $ ] .
the coefficients @xmath148 can be related to the form factors @xmath149 and @xmath150 , @xmath151 which appear in the @xmath140 branching ratio as @xmath152 in the model under consideration @xmath153 , which leads to @xmath154 . as in @xcite
, we will consider three contributions to the form factor @xmath41 : 1-loop diagrams with neutral higgs bosons and charged leptons in the loop , 2-loop barr - zee diagrams with an internal photon and a third generation quark , and 2-loop barr - zee diagrams with an internal photon and a @xmath118-boson @xcite .
therefore , we write @xmath155 the different contributions are @xcite @xmath156 \ , .
\label{eq : a2b}\end{aligned}\ ] ] here @xmath157 , @xmath158 = @xmath159 , the coupling @xmath160 of the internal loop fermion to the @xmath161 scalar is given in eqs .
- and the couplings @xmath162 are given in the last paragraph of sec .
[ sec:2hdm - description ] . in eqs .
- lepton masses have been neglected whenever possible . finally
, the loop functions in the previous expressions are @xcite @xmath163 \ , .\end{aligned}\ ] ] s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) [ arxiv:1207.7235 [ hep - ex ] ] .
g. aad _ et al . _
[ atlas collaboration ] , phys .
b * 716 * , 1 ( 2012 ) [ arxiv:1207.7214 [ hep - ex ] ] .
cms collaboration , `` precise determination of the mass of the higgs boson and studies of the compatibility of its couplings with the standard model '' , _ cms physics analysis summary _ cms - pas - hig-14 - 009 ( 2014 ) g. aad _ et al . _
[ atlas collaboration ] , phys .
lett . * 112 * , 201802 ( 2014 ) [ arxiv:1402.3244 [ hep - ex ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , eur .
j. c * 74 * , 2980 ( 2014 ) [ arxiv:1404.1344 [ hep - ex ] ] .
cms collaboration , `` search for lepton flavor violating decays of the higgs boson '' , _ cms physics analysis summary _ cms - pas - hig-14 - 005 ( 2014 ) .
g. blankenburg , j. ellis and g. isidori , phys .
b * 712 * , 386 ( 2012 ) [ arxiv:1202.5704 [ hep - ph ] ] .
r. harnik , j. kopp and j. zupan , jhep * 1303 * , 026 ( 2013 ) [ arxiv:1209.1397 [ hep - ph ] ] .
a. goudelis , o. lebedev and j. h. park , phys .
b * 707 * ( 2012 ) 369 [ arxiv:1111.1715 [ hep - ph ] ] .
a. celis , v. cirigliano and e. passemar , phys .
d * 89 * ( 2014 ) 013008 [ arxiv:1309.3564 [ hep - ph ] ] .
a. pilaftsis , phys .
b * 285 * ( 1992 ) 68 .
j. l. diaz - cruz and j. j. toscano , phys .
d * 62 * ( 2000 ) 116005 [ hep - ph/9910233 ] .
m. arana - catania , e. arganda and m. j. herrero , jhep * 1309 * , 160 ( 2013 ) [ arxiv:1304.3371 [ hep - ph ] ] .
a. arhrib , y. cheng and o. c. w. kong , europhys .
* 101 * , 31003 ( 2013 ) [ arxiv:1208.4669 [ hep - ph ] ] .
e. arganda , m. j. herrero , x. marcano and c. weiland , arxiv:1405.4300 [ hep - ph ] .
a. falkowski , d. m. straub and a. vicente , jhep * 1405 * , 092 ( 2014 ) [ arxiv:1312.5329 [ hep - ph ] ] .
g. bhattacharyya , p. leser and h. ps , phys .
d * 83 * , 011701 ( 2011 ) [ arxiv:1006.5597 [ hep - ph ] ] .
g. bhattacharyya , p. leser and h. ps , phys .
d * 86 * , 036009 ( 2012 ) [ arxiv:1206.4202 [ hep - ph ] ] .
m. arroyo , j. l. diaz - cruz , e. diaz and j. a. orduz - ducuara , arxiv:1306.2343 [ hep - ph ] .
m. d. campos , a. e. c. hernndez , h. ps and e. schumacher , arxiv:1408.1652 [ hep - ph ] .
g. dambrosio , g. f. giudice , g. isidori and a. strumia , nucl .
b * 645 * , 155 ( 2002 ) [ hep - ph/0207036 ] . c. d. froggatt and h. b. nielsen , nucl .
b * 147 * , 277 ( 1979 ) .
a. dery , a. efrati , y. nir , y. soreq and v. susi , arxiv:1408.1371 [ hep - ph ] .
s. davidson and g. j. grenier , phys .
d * 81 * , 095016 ( 2010 ) [ arxiv:1001.0434 [ hep - ph ] ] .
j. kopp and m. nardecchia , arxiv:1406.5303 [ hep - ph ] .
d. atwood , l. reina and a. soni , phys .
d * 55 * , 3156 ( 1997 ) [ hep - ph/9609279 ] .
a. crivellin , a. kokulu and c. greub , phys .
d * 87 * ( 2013 ) 9 , 094031 [ arxiv:1303.5877 [ hep - ph ] ] .
d. oneil , arxiv:0908.1363 [ hep - ph ] . s. m. barr and a. zee , phys
. rev .
* 65 * ( 1990 ) 21 [ erratum - ibid .
* 65 * ( 1990 ) 2920 ] .
t. d. lee , phys .
d * 8 * ( 1973 ) 1226 .
g. c. branco , phys .
d * 22 * ( 1980 ) 2901 .
s. davidson and h. e. haber , phys .
d * 72 * ( 2005 ) 035004 [ erratum - ibid .
d * 72 * ( 2005 ) 099902 ] [ hep - ph/0504050 ] .
t. p. cheng and m. sher , phys .
d * 35 * ( 1987 ) 3484 .
s. kanemura , t. ota and k. tsumura , phys .
d * 73 * ( 2006 ) 016006 [ hep - ph/0505191 ] .
a. celis , v. ilisie and a. pich , jhep * 1312 * ( 2013 ) 095 [ arxiv:1310.7941 [ hep - ph ] ] .
gerard and m. herquet , phys .
lett . * 98 * , 251802 ( 2007 ) [ hep - ph/0703051 [ hep - ph ] ] .
p. m. ferreira and d. r. t. jones , jhep * 0908 * ( 2009 ) 069 [ arxiv:0903.2856 [ hep - ph ] ] .
d. eriksson , j. rathsman and o. stal , comput .
commun .
* 181 * ( 2010 ) 189 [ arxiv:0902.0851 [ hep - ph ] ] .
a. broggio , e. j. chun , m. passera , k. m. patel and s. k. vempati , arxiv:1409.3199 [ hep - ph ] .
i. f. ginzburg and i. p. ivanov , phys .
d * 72 * ( 2005 ) 115010 [ hep - ph/0508020 ] .
i. p. ivanov , phys .
d * 75 * ( 2007 ) 035001 [ erratum - ibid .
d * 76 * ( 2007 ) 039902 ] [ hep - ph/0609018 ] .
i. p. ivanov , phys .
d * 77 * ( 2008 ) 015017 [ arxiv:0710.3490 [ hep - ph ] ] .
m. baak , m. goebel , j. haller , a. hoecker , d. kennedy , r. kogler , k. moenig and m. schott _ et al .
_ , eur .
j. c * 72 * ( 2012 ) 2205 [ arxiv:1209.2716 [ hep - ph ] ] .
b. aubert _ et al . _
[ babar collaboration ] , phys .
* 104 * ( 2010 ) 021802 [ arxiv:0908.2381 [ hep - ex ] ] . t. aushev , w. bartel , a. bondar , j. brodzicka , t. e. browder , p. chang , y. chao and k. f. chen _ et al .
_ , arxiv:1002.5012 [ hep - ex ] .
j. d. bjorken , k. d. lane and s. weinberg , phys .
d * 16 * ( 1977 ) 1474 .
d. chang , w. s. hou and w. y. keung , phys .
d * 48 * ( 1993 ) 217 [ hep - ph/9302267 ] .
j. bernon , b. dumont and s. kraml , arxiv:1409.1588 [ hep - ph ] .
a. antaramian , l. j. hall and a. rasin , phys .
* 69 * ( 1992 ) 1871 [ hep - ph/9206205 ] .
|
direct searches for lepton flavor violating higgs boson decays in the @xmath0 channel have been recently reported by the cms collaboration .
the results display a slight excess of signal events with a significance of 2.5@xmath1 , which translates into a branching ratio of about @xmath2% . by interpreting these findings as a hint for beyond the standard model physics ,
we show that the type - iii 2hdm is capable of reproducing such signal while at the same time satisfying boundedness from below of the scalar potential , perturbativity , electroweak precision data , measured higgs standard decay modes and low - energy lepton flavor violating constraints .
we have found that the allowed signal strength ranges for the @xmath3 , @xmath4 and @xmath5 standard channels shrink as soon as br@xmath6% is enforced .
thus , we point out that if the excess persists , improved measurements of these channels may be used to test our type - iii 2hdm scenario .
|
di - boson production processes are important channels to test the standard model ( sm ) at the lhc .
they have been studied intensively in the past years both from the theoretical and the experimental side . as a signal , they are sensitive to triple gauge boson couplings , and therefore , provide a unique avenue to quantify deviations from the sm predictions .
furthermore , they are a background to many sm and beyond standard model analyses . due to the large size of the next - to - leading order(nlo ) corrections and the expected percent precision measurement at the lhc ,
the theoretical community has pursued in the last years the task to provide next - to - next - to - leading order ( nnlo ) qcd results .
this task has been almost completed in the last years and exact results at nnlo are known for most of the processes , not only for total cross sections @xcite , but also for differential distributions @xcite . at the same time , due to the large collection of results known at nlo for processes with different jet multiplicities , a field by its own has emerged with the aim to merge in a consistent way processes with different jet multiplicities at nlo . in this letter , following the loopsim approach @xcite , we will merge @xmath1 and @xmath1+jet samples and review results at approximate nnlo accuracy for several di - boson production processes .
furthermore , we will show preliminary results of anomalous couplings effects in @xmath0 production .
using the loopsim approach , we merge samples at nlo accuracy , provided by the vbfnlo monte carlo program @xcite , for several @xmath1 and @xmath1+jet production processes .
the merged sample is simultaneously accurate at nlo for both the @xmath1 and @xmath1+jet sample and provides results at nnlo accuracy for the @xmath1 production process in certain regions of the phase space since it includes consistently the double - real and virtual - real contributions to the @xmath1 nnlo contributions , simulating in a unitarity approach , the missing two loop corrections , such that by construction the merged sample is infrared finite .
thus the sample consistently includes all the new phase space regions opening up first at nnlo , including the double soft and collinear emission of the weak bosons , which leads to numerically large logarithms of the form @xmath2 and therefore to potentially large nnlo corrections .
furthermore , it includes consistently the new partonic sub - processes opening up at nnlo , in this case , @xmath3 and @xmath4 initiated processes . thus in regions of phase space where the lo kinematics are not dominant and therefore the missing finite piece of the two loop corrections is suppressed , like in inclusive anomalous coupling searches , the merged sample should provide most of the nnlo contributions . ,
defined in eq .
( [ eq : ht ] ) , for the lhc at @xmath5 .
the bands correspond to varying @xmath6 by factors 1/2 and 2 around the central value .
the solid bands give the uncertainty related to the @xmath7 parameter varied between 0.5 and 1.5 .
the distribution is a sum of contributions from same - flavor decay channels ( @xmath8 and @xmath9 ) and the different - flavor channel ( @xmath10 ) in @xmath11 production .
, title="fig : " ] , defined in eq .
( [ eq : ht ] ) , for the lhc at @xmath5 .
the bands correspond to varying @xmath6 by factors 1/2 and 2 around the central value .
the solid bands give the uncertainty related to the @xmath7 parameter varied between 0.5 and 1.5 .
the distribution is a sum of contributions from same - flavor decay channels ( @xmath8 and @xmath9 ) and the different - flavor channel ( @xmath10 ) in @xmath11 production .
, title="fig : " ] of the hardest lepton at @xmath5 without ( left ) and with jet veto ( right ) .
bands are defined as in fig .
[ fig : ht - standard ] .
we include the channels @xmath12 , @xmath13 , @xmath14 and @xmath15 .
the contribution from the gluon - fusion box and higgs diagrams is included in the nlo and @xmath16curves .
the left panels correspond to the inclusive sample , while the results shown in the right panel were obtained with vetoing events containing jets which fulfill the criteria @xmath17 and @xmath18 .
, title="fig : " ] of the hardest lepton at @xmath5 without ( left ) and with jet veto ( right ) .
bands are defined as in fig .
[ fig : ht - standard ] .
we include the channels @xmath12 , @xmath13 , @xmath14 and @xmath15 .
the contribution from the gluon - fusion box and higgs diagrams is included in the nlo and @xmath16curves .
the left panels correspond to the inclusive sample , while the results shown in the right panel were obtained with vetoing events containing jets which fulfill the criteria @xmath17 and @xmath18 .
, title="fig : " ]
in the following , results for @xmath11 and @xmath19 production are given at @xmath16 to refer to our approximated results . ] .
they were studied in ref .
@xcite and ref .
@xcite , respectively . the input parameters and a detailed description of the analysis can be found there .
we take into account the leptonic decay of the weak bosons including all off - shell and spin correlation effects .
however , we refer to the processes by the on - shell production mode for simplicity . independently of the order of a prediction
, we used the nnlo mstw2008
@xcite pdf set with @xmath20 . as central value for the factorization and renormalization scales we chose @xmath21=
@xmath22 @xmath23 @xmath24 @xmath25 where @xmath26 and @xmath27 are the transverse momenta and invariant masses of the decaying vector bosons , respectively
. the scale uncertainty is obtained by varying simultaneously the factorization and renormalization scale by a factor two around the central scale .
additionally , to assess the uncertainties associated with the recombination method used by loopsim , we show the uncertainty bands associated with variations of the clustering radius , @xmath28 , of @xmath29 around the central value @xmath30 .
@xmath28 is used in loopsim to establish the sequence of emissions , which is used later on to identify the born type particles ( @xmath31 or @xmath32 ) of the event .
[ fig : ht - standard ] , for @xmath11 production , shows the differential distribution for the effective mass , @xmath33 , defined as a scalar sum of transverse momenta of leptons and jets @xmath34 the set of cuts closely follows the atlas @xcite analysis for inclusive searches and is described in detail in ref .
@xcite . in the left panel ,
we require that the invariant masses of the reconstructed @xmath35 bosons satisfy the cut @xmath36 and label the pair closer ( further ) to the on - shell value @xmath37 as @xmath38 .
one can observe the large @xmath16contributions , which clearly exceed the scale uncertainties , and the small loopsim uncertainty provided by the @xmath39 variation .
the origin of the size of the corrections is well understood and is due to the sensitivity of this observable to additional jet radiation , leading to enhanced logarithms of the form @xmath40 . on the right
, one observes smaller corrections once we reduce the size of the appearing logs by imposing @xmath41 .
the plots also show results at @xmath16for the gluon loop - induced contributions , which formally contribute at nnnlo and use the amplitudes of ref .
@xcite . in fig .
[ fig : ptww - standard ] , we show the differential distribution for the hardest lepton for @xmath19 production with ( left ) and without ( right ) applying a fixed jet veto . to a large extent ,
our cuts match the ones by the cms experiment in ref .
@xcite . the loop - squared gluon - fusion box and higgs contributions
are separately shown .
on the left , one can see that the @xmath16corrections are of the order of @xmath42 and beyond the nlo scale uncertainties . on the right ,
we show results for the vetoed contributions to point out that exclusive samples are subject to potentially large negative sudakov corrections and to show that the entire @xmath43-vetoed result has larger scale uncertainties than the nlo - vetoed curves .
this reveals , partially , accidental cancellation happening at nlo .
however , as discussed in ref .
@xcite , jet - vetoed exclusive samples are potentially subject to further corrections from the constant term of 2-loop diagrams which are not accounted for by the @xmath44 uncertainty band . of the hardest lepton for the lhc at @xmath45 without ( left ) and with a dynamical jet veto ( right ) for different values of the anomalous coupling parameter @xmath46 ( in @xmath47 ) .
the light - grey ( green ) and grey ( blue ) bands correspond , respectively , to the sm nlo and sm @xmath16contributions varying @xmath6 by factors 1/2 and 2 around the central value .
the dark - grey ( orange ) band correspond to the sm @xmath16 uncertainty related to the @xmath7 parameter varied between 0.5 and 1.5 .
dashed and solid lines refer to nlo and @xmath16 , respectively .
, title="fig : " ] of the hardest lepton for the lhc at @xmath45 without ( left ) and with a dynamical jet veto ( right ) for different values of the anomalous coupling parameter @xmath46 ( in @xmath47 ) .
the light - grey ( green ) and grey ( blue ) bands correspond , respectively , to the sm nlo and sm @xmath16contributions varying @xmath6 by factors 1/2 and 2 around the central value .
the dark - grey ( orange ) band correspond to the sm @xmath16 uncertainty related to the @xmath7 parameter varied between 0.5 and 1.5 . dashed and solid lines refer to nlo and @xmath16 , respectively .
, title="fig : " ]
in the following , we show how higher order corrections can fake anomalous couplings ( ac ) effects for @xmath0 production .
we closely follow the setup defined in ref .
@xcite and use the amplitudes from ref . @xcite . on the left plot of fig .
[ fig : ac ] , we present the sm predictions for lepton @xmath48 distributions with a finite anomalous coupling parameter , @xmath49 , corresponding to the dimension 6 operator @xmath50 .
the coupling values used are within the range allowed by the global fit to present data in ref .
we use a dipole form factor to preserve tree level unitarity , with a form factor scale derived from unitarity constraints .
one can clearly see in the left plot that higher order qcd contributions can fake ac effects , if nlo predictions are taken . on the right , to increase the sensitivity to ac ,
we apply a dynamical veto , @xmath51 , as described in ref .
@xcite and given by @xmath52 .
part of this work was performed on the computational resource bwunicluster funded by the ministry of science , research and arts and the universities of the state of baden - wrttemberg , germany , within the framework program bwhpc .
we acknowledge support from the bmbf ( `` verbundprojekt hep - theorie '' , 05h12vkg ) and the dfg ( grk 1694 ) .
fc thanks support to the spanish government and erdf funds from the european commission ( fpa2014 - 53631-c2 - 1-p and fpa2014 - 54459-p ) .
99 m. grazzini , s. kallweit , d. rathlev and a. torre , phys .
b * 731 * ( 2014 ) 204 ; f. cascioli _ et al .
_ , phys .
b * 735 * ( 2014 ) 311 ; t. gehrmann , m. grazzini , s. kallweit , p. maierhfer , a. von manteuffel , s. pozzorini , d. rathlev and l. tancredi , phys .
* 113 * ( 2014 ) 21 , 212001 .
s. catani , l. cieri , d. de florian , g. ferrera and m. grazzini , phys .
lett . * 108 * ( 2012 ) 072001 ; m. grazzini , s. kallweit and d. rathlev , jhep * 1507 * ( 2015 ) 085 ; phys .
b * 750 * ( 2015 ) 407 .
m. rubin , g. p. salam and s. sapeta , jhep * 1009 * ( 2010 ) 084 .
k. arnold , m. bahr , g. bozzi , f. campanario , c. englert _ et al .
_ , comput .
commun . *
180 * , 1661 ( 2009 ) ; arxiv:1107.4038 [ hep - ph ] ; j. baglio _ et al .
_ , arxiv:1404.3940 [ hep - ph ] .
f. campanario , m. rauch and s. sapeta , jhep * 1508 * ( 2015 ) 070 .
f. campanario , m. rauch and s. sapeta , nucl .
b * 879 * ( 2014 ) 65 .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , eur .
j. c * 63 * ( 2009 ) 189 .
g. aad _ et al . _ [ atlas collaboration ] , jhep * 1303 * ( 2013 ) 128 f. campanario , q. li , m. rauch and m. spira , jhep * 1306 * ( 2013 ) 069 .
f. campanario and s. sapeta , phys .
b * 718 * ( 2012 ) 100 .
f. campanario , c. englert and m. spannowsky , phys .
d * 82 * ( 2010 ) 054015 .
t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , phys .
d * 87 * ( 2013 ) 015022 .
f. campanario , r. roth and d. zeppenfeld , phys .
d * 91 * ( 2015 ) 054039 .
|
in these proceedings , we review results for several di - boson production processes beyond nlo qcd at high transverse momenta using the vbfnlo monte - carlo program together with the loopsim method . additionally , we show for the @xmath0 production process how higher order qcd corrections can resemble anomalous coupling effects .
|
complex oxide heterostructures have been the subject of many recent papers , both experimental and theoretical , as their interesting interface properties promise to pave the way towards novel electronic devices . with the current available experimental techniques oxide thin films
can be produced with a high degree of cristallinity and the electronic structure of their surfaces and interfaces can be precisely determined .
the most studied system of this type has been laalo@xmath1/srtio@xmath1 ( lao / sto ) , that consists of two perovskite structures stacked along the ( 001 ) direction@xcite , in which they present alternate layers of lao , alo2 , tio2 and sro with a very small lattice mismatch between them . a similar but
less studied interface is that of lao with tio@xmath2-anatase , that presents an even smaller lattice mismatch , namely less than 0.1% , along the ( 001 ) direction @xcite .
tio2 is a key material for most applications , including catalytic and optical devices , sensors , optoelectronics and spintronics .
anatase thin films are frequently grown over lao by pulsed laser deposition .
it seems therefore important to perform a careful characterization , both from theory and experiments , of the film / substrate interface .
just to mention one example where this study may be relevant , room temperature ferromagnetism has been obtained from doped and undoped anatase films grown over lao and the results strongly depend on the growth conditions @xcite .
there are two possible interfaces for the system lao / tio@xmath2-anatase , namely lao facing tio@xmath2 and alo@xmath2 facing tio@xmath2@xcite . due to the ionicity of the component oxides , and in particular , to the fact that the layers of lao ( alo@xmath2 and lao ) have alternating formal charges ( -1 and + 1 , respectively ) , there is an interfacial formal excess charge that should be compensated either by the presence of terraces with different stacking , or by oxygen vacancies or by atomic interdiffusion . .
[ fig:1 ] experimental work @xcite has shown that , in the case of lao / tio@xmath2-anatase , terraces with both types of interfaces appear and approximately in the same proportion . for the theoretical modeling of this system the fact that the tio@xmath2-anatase layers along the ( 001 ) direction are not strictly planar must be taken into account . as a consequence ,
the anatase interface can be either oxygen - terminated and thus neutral ( with formal valence + 4 for the interfacial ti ion ) , or ti - terminated .
the latter situation is equivalent to considering surface oxygen vacancies and gives rise to a change in the valence of the interfacial ti ion . in this paper
we study the electronic and magnetic properties of interfaces for different stackings of lao / tio@xmath2 using density functional theory@xcite in the local density approximation ( lsda ) and also considering electronic correlations in the framework of the lsda+u approximation .
to simulate the interfaces , we use a periodically repeated slab - geometry with 5 layers of each of the component materials and enough empty space between slabs so that they do not interact with each other ( see fig.[fig:1 ] ) .
the cell parameters in the plane of the interface are kept fixed to those of the experimental bulk lao ( 3.79 ) @xcite and the positions of all atoms are allowed to relax in the out of plane direction until the forces are smaller than 0.05 ev/ . as in our previous work@xcite , we use an odd number of layers of lao to avoid the formation of a large dipole moment , but we are aware that with this procedure it is difficult to assure whether the interface will be conducting or not .
the reason lies in the fact that an lsda calculation for a positively ( negatively ) charged system will locate the fermi level in the conduction ( valence ) band .
we use the wien2k code , that is an implementation of the full potential linear augmented plane waves method ( fp - lapw)@xcite .
the calculations are scalar relativistic and the parameters used are listed on ref . .
it is well known that in the case of oxides there is a band gap underestimation when using local exchange correlation functionals within density functional methods .
for this reason , we have performed calculations with both the lsda and the lsda+u approximations@xcite , using u=0.4 ry for the ti @xmath4 orbitals@xcite , as in previous works .
the comparison between the two procedures will evidence whether there is a qualitative difference in the interfacial properties due to electron correlations , as for example , a change in the valence of the ti ions .
the total energies of different possible structures ( at 0k and with collinear spins ) have been evaluated in previous works@xcite , and in this paper we present the results for the lowest energy ones .
for the interface in which lao faces tio@xmath2
( fig.[fig:1](a ) ) , the oxygen - termination for anatase is preferred .
however , when alo@xmath2 faces tio@xmath2 , the lower energy is for the ti - termination , thus with oxygen vacancies@xcite .
two types of ti - terminated interfaces are studied : one where the interfacial ti atom faces the hollow site of the al atoms in the alo@xmath2 layer ( fig.[fig:1](b ) ) and another one where the ti atom faces a bridge position of the al sites in the alo@xmath2 plane ( fig.[fig:1](c ) ) . in all the cases considered
there is a total formal charge of + 1 arising in ( a ) from the lao slab while in cases ( b ) and ( c ) it is due to the contribution of both the lao slab and the oxygen vacancy in the anatase slab .
we point out that there are more possible structures for each type of stacking , and that in this work we consider three of the lowest energy ones , those that require more energy to separate lao from anatase .
lao / tio@xmath2 has the lowest energy , followed by alo@xmath2/tio where oxygen vacancies are present@xcite ( i.e. o2 is removed in fig.[fig:1](b ) and ( c ) ) . in the hollow structure
one oxygen atom from the alo@xmath2 layer moves towards the interfacial ti thus increasing its number of neighbors .
this relaxation decreases the total energy of the system with respect to the bridge structure by 0.2 ev in lsda and 0.4 ev in lsda+u .
fig.[fig:2 ] shows the densities of states ( dos ) for the three structures depicted in fig.[fig:1 ] .
we present the results for the lsda+u approximation as they are similar to the lsda ones and if there are any differences , they are mentioned in the text . on the left side of fig.[fig:2 ] , we present the total dos and the partial contributions from the la atoms . on the right side ,
the dos projected on the ti atoms are shown in a zoomed - in region close to the fermi energy ( e@xmath5 ) .
as in the case of pure anatase , the valence band is mostly from the oxygen atoms and the conduction band from the ti atoms , with the @xmath4 orbitals clearly separated by symmetry : the @xmath6 levels closer to e@xmath5 and the e@xmath7 ones at higher energies .
there are no al states close to e@xmath5 , but oxygen atoms from lao contribute significantly to the higher energy part of the valence band .
@xmath8{fig2a_1.ps } & \includegraphics[width=0.36\textwidth , keepaspectratio , angle=-90]{fig2a_2.ps}\\ \includegraphics[width=0.36\textwidth , keepaspectratio , angle=-90]{fig2b_1.ps } & \includegraphics[width=0.36\textwidth , keepaspectratio , angle=-90]{fig2b_2.ps}\\ \includegraphics[width=0.36\textwidth , keepaspectratio , angle=-90]{fig2c_1.ps } & \includegraphics[width=0.36\textwidth , keepaspectratio , angle=-90]{fig2c_2.ps } \end{array}$ ] @xmath9{fig3a.eps } & \includegraphics[width=0.3\textwidth , keepaspectratio]{fig3b.eps } & \includegraphics[width=0.3\textwidth , keepaspectratio]{fig3c.eps } \end{array}$ ] * lao / tio@xmath2 interface * in this interface ( fig.[fig:1](a ) ) ,
e@xmath5 is in the conduction band , as expected due to the formal charge in the lao slab .
the largest contribution to the occupied states close to e@xmath5 comes from the interfacial ti ion and the rest of the ti atoms contribute less as the distance to the interface increases , as shown in fig.[fig:2](top - right ) .
this is true for both lsda and lsda+u approximations .
it is interesting to note that similar layer projected dos were obtained for lao / sto @xcite even though in that structure , the tio@xmath2 planes alternate with sro layers .
the authors also find that the charge distribution is not affected by the inclusion of correlation .
there is certainly no contribution from the interfacial la ions to the occupied states , as that peak is shifted towards the right when compared with the la atom at the free surface of the slab ( see fig.[fig:2](a ) ) .
the charge density at the interface is plotted in fig.[fig:3](a ) and shows clearly the shape of the ti @xmath4 orbitals .
the main occupation comes from d@xmath10 and d@xmath11 .
a small total magnetic moment is obtained for the unit cell , distributed among all the ti atoms ( see fig.[fig:4 ] ) .
* alo@xmath2/tio - hollow interface * this interface ( fig.[fig:1](b ) ) , presents a difference between lsda and lsda+u approximations . in lsda+u , the majority spin of the interfacial ti d@xmath12 orbital shifts towards lower energies , so that the system becomes a semiconductor with a small gap of 0.1 ev ( see fig.[fig:2](b ) ) this shift is smaller in the lsda , so that the system is metallic . in fig.[fig:3](b ) we show the charge density at the anatase side of the interface where the @xmath13 symmetry is evident .
both the charge and the magnetic moment at the ti ions are large at the interface and decrease abruptly inside the anatase slab ( see fig.[fig:4 ] ) .
* alo@xmath2/tio - bridge interface * in this interface ( fig.[fig:1](c ) ) , e@xmath5 lies in the conduction band , and the la peaks are further away from it , overlapping the @xmath14 orbitals of the ti atoms .
the same result is obtained with lsda .
the interfacial ti orbitals have different symmetry from those of the hollow case , being of @xmath15 character : @xmath16 and @xmath17 , as can be seen in fig.[fig:2 ] and fig.[fig:3 ] . in this case , the @xmath15 levels split into two regions : one occupied band close to e@xmath5 and another one at energies above the @xmath14 orbitals , in the conduction band .
the extra charge and the magnetic moment are both localized at the interface , near the oxygen vacancy , and their magnitude decreases abruptly inside the anatase slab as can be seen in fig.[fig:4 ] . in this case , the interfacial ti formal valence and the magnetic moment are smaller than in the hollow structure .
it is interesting to note that in alo@xmath2/tio interfaces , there is no contribution of the ti atom at the free surface in the occupied region of the conduction band , showing that a slab with 5 layers is a good enough approximation to account for interface properties .
there has been some controversy concerning the appearance of interfacial ferromagnetism in the lao / sto interface @xcite , which may also be the case here . for this reason we performed calculations with a double size unit cell ( 54 atoms ) , for the alo@xmath2/tio interfaces which present localized magnetism at the interface .
we studied the magnetic interaction between the interfacial ti atoms within the lsda+u method , setting them both in parallel and antiparallel configurations .
the two solutions exist but the antiparallel one is lower in energy , indicating that a long - range ferromagnetic interaction will not arise from the considered concentration and distribution of oxygen vacancies in alo@xmath2/tio interfaces .
we do not exclude the possibility of ferromagnetism in lao / tio@xmath2 , but it would be minor and due to fact that all the ti atoms have local magnetic moments , many magnetic configurations should be considered for this study . to explore the consequences of a lower vacancy concentration , and also of a formally neutral system , we considered the double size unit cell in the hollow configuration with only one interfacial oxygen vacancy . in this calculation none of the ti ions resulted ti+3 and no magnetic moments appeared .
thus , a charge imbalance is needed to obtain ti+3 ions and is also a necessary ( but not sufficient ) condition for the appearance of magnetism .
the principal conclusion of this work is that whatever the origin of the charge at the interfaces , its main effect is to change the valence of the ti atoms , either if it is due to the layered - structure of lao or to the presence of oxygen vacancies in anatase . in alo@xmath2/tio
, there are interfacial oxygen vacancies and the interfacial ti atom close to the vacancy position , acquires an extra charge . in lao / tio@xmath2 , there are no oxygen vacancies as they are not electrostatically favored and the extra charge is distributed among all ti atoms , with decreasing value as the distance from the interface increases .
our results indicate that the lao / tio@xmath2 interface spans through several anatase layers while the alo@xmath2/tio one is more localized and presents large local magnetic moments . in experimental samples ,
co - existence of the studied interfaces and possibly others as well , is expected . as a consequence ,
small patches with different magnetic order might appear @xcite , thus giving rise to sample - dependent results .
we thank r. weht for a careful reading of the manuscript . m.w . and
are members of conicet - argentina and acknowledge support by grant pip - conicet00038 .
p. blaha , k. schwarz , g.k.h .
madsen , d. kvasnicka , and j. luitz , wien2k , an augmented plane wave + local orbitals program for calculating crystal properties ( karlheinz schwarz , techn .
universitt wien , austria ) , 2001 .
isbn 3 - 9501031 - 1 - 2 . for the fp - lapw calculations we use the following parameters : atomic sphere radii of 1.7 bohr for ti and al , 1.4 bohr for o and 2.5 for la .
the convergence control parameter rk@xmath18 is chosen to be 6 , that corresponds to an energy cutoff of 250 ev .
the employed vacuum space between slabs is 20 . in this work
the number of vacancies considered is large , but the simulation of smaller vacancy concentrations would imply , due to the anatase structure , to start from corrugated interfaces that may be the the subject of future work .
|
ab initio calculations using the local spin density approximation and also including the hubbard @xmath0 have been performed for three low energy configurations of the interface between laalo@xmath1 and tio@xmath2-anatase .
two types of interfaces have been considered : lao / tio@xmath2 and alo@xmath2/tio , the latter with ti - termination and therefore a missing oxygen .
a slab - geometry calculation was carried out and all the atoms were allowed to relax in the direction normal to the interface . in all the cases considered ,
the interfacial ti atom acquires a local magnetic moment and its formal valence is less than @xmath3 .
when there are oxygen vacancies , this valence decreases abruptly inside the anatase slab while in the lao / tio@xmath2 interface the changes are more gradual .
+ 3 mm
|
peg solitaire is one of the earliest known puzzles , with a 300 year history .
the puzzle consists of a game board together with a number of pegs , or more commonly marbles .
the board contains a grid of holes in which these pegs or marbles are placed .
figure [ fig1 ] shows the two most common shapes for a peg solitaire board , the 33-hole cross - shaped board , and the 15-hole triangular board .
the game is played by jumping one peg over another into an empty hole , removing the peg that was jumped over . on the cross - shaped board of figure [ fig1]a ,
these jumps must be made along columns or rows , whereas on the triangular board of figure [ fig1]c , jumps are allowed along any of the six directions parallel to the sides of the board . the goal is to finish with one peg , a more advanced variation is to finish with one peg at a specified hole .
the basic game begins from a full board with one peg removed , as in figure [ fig1]a or c. starting from figure [ fig1]a and ending with one peg in the center of the board is known as the central game " @xcite . in figure [ fig1]b or d
we show starting configurations to be referred to later . only relatively recently in the history of the game
have computers been used as an interface to play the game , as well as solve the game .
there are now dozens of versions of the game available for playing on your computer or even your cell phone .
a computer version of the game is in many ways less satisfying than a physical game .
however , there are some definite advantages to playing the game on a computer
. the board can be reset instantly , and you wo nt be chasing marbles that fall off the board ! you can take back a move , all the way back to the beginning if desired .
this tends to make the game easier as you can more easily backtrack from dead ends .
the sequence of moves leading to a solution can be recorded and played back .
the computer can also be programmed to tell the user if the jump they are considering leads to a dead end or not . adding this ability to a computer version of the game is tricky , and most versions do not have this ability .
it is the goal of this paper to describe efficient techniques to enable a computer to point out all good and bad jumps from the current board position .
we label the holes in the 33-hole board using cartesian coordinates ( figure [ fig2]a ) , but with @xmath0 _ increasing _ downward . for the triangular boards , we use skew - coordinates " as shown in figure [ fig2]b . by adding 1 to each coordinate , and converting the first to a letter
, we obtain the standard board labellings used by beasley @xcite and bell @xcite ( for example the central hole @xmath1 in figure [ fig2]a becomes d4 " ) .
the 33-hole board has square symmetry .
there are eight symmetry transformations of the board , given by the identity , rotations of @xmath2 , @xmath3 and @xmath4 , and a reflection of the board followed by these 4 rotations @xcite ( the dihedral group @xmath5 ) .
the triangular boards have 6-fold symmetry , with 3 possible rotations of @xmath6 , @xmath7 or @xmath8 , plus a reflection followed by a rotation ( the dihedral group @xmath9 ) . to store a particular board position on a computer , we convert it to an integer by taking one bit per hole
. the most obvious way to do this is to take the board , row by row , top to bottom , as in figure [ fig2]c .
we will use @xmath10 to denote the total number of holes on the board , so each board position is represented by an @xmath10 bit integer . if @xmath11 is a board position we ll denote this integer representation by @xmath12 .
most computer languages use a 32-bit integer , so we have one bit too many for storing the 33-hole board .
beasley @xcite gives a technique for storing a board position on the 33-hole board using 4 fewer bits .
the additional complexity is often not worth it , and computer memory is cheap these days . for boards with more than 32 holes ,
we usually split @xmath12 into several 4-byte integers .
the * complement * of a board position @xmath11 is obtained by replacing every peg by a hole ( i.e. removing it ) , and replacing every hole by a peg .
the complement of @xmath11 will be denoted as @xmath13 .
we note that @xmath14 , where @xmath15 is the board position where every hole contains a peg , @xmath16 .
the starting position for the central game " in figure [ fig1]a therefore has code @xmath17 .
two board positions are symmetry equivalent if one can be converted to the other by a symmetry transformation .
this equivalence relation introduces a set of _ equivalence classes _ of board positions , which we will call * symmetry classes*. the symmetry class of a board position does not change after it is rotated or reflected .
one way to choose a representative from each symmetry class is to take the one with the smallest code .
we use the notation @xmath18 to denote this operation .
for example , the board position @xmath11 in figure [ fig1]d has code @xmath19 , and the other 5 codes obtained by symmetry transformation are : @xmath20 , @xmath21 , @xmath22 , @xmath23 and @xmath24 , so @xmath25 .
we also have @xmath26 .
a peg solitaire problem which begins with one peg missing , with the goal to finish with one peg , will be called a * single vacancy to single survivor * problem , abbreviated svss . when the starting hole @xmath27 and finishing hole @xmath28 are the same , the svss problem is called a * complement problem * , because the starting and ending board positions are complements of each other .
a simple parity argument gives a necessary condition for solvability of a svss problem ( * ? ? ?
* chapter 4 ) . on a square lattice ( like the standard 33-hole board ) , the requirement is that @xmath29 and @xmath30 must differ by a multiple of 3 , or that @xmath31 , and @xmath32 . starting and ending board positions satisfying the above conditions
are said to be in the same * position class*. the main result is that _ the position class does not change as the game is played_. on a triangular lattice , the requirement is weaker : @xmath33 . ] .
we will not go into the theory of position classes , the reader should see beasley ( * ? ? ?
* chapter 4 ) or bell @xcite for triangular peg solitaire .
it is interesting to see what happens to the position class after the board is rotated or reflected .
for the central game on the 33-hole board ( figure [ fig1]a ) , the position class is not changed by rotations or reflections of the board .
if we begin with @xmath1 vacant we can finish at @xmath1 , or the rotationally equivalent holes @xmath34 , @xmath35 .
any board position which begins from any of these five holes is in the position class of one peg in the center .
moreover , if we reflect and/or rotate the board at any time during the game , it remains in the same position class .
thus , this set of holes forms a set of svss problems of the same * type * whose general solutions are all interconnected , shown in figure [ fig3]a .
more commonly , the position class does change after the board is rotated or reflected .
problems of this type are shown in figure [ fig3]b . in this case when we rotate the board , the position class changes , but only among the 4 with single peg representatives in the holes shown in figure [ fig3]b .
another way to look at figure [ fig3]b is that we can begin with @xmath27 at any 3 " or 4 " , and finish at any 3 " or 4 " , if we allow peg solitaire jumps plus rotations and reflections of the board . the third type of problem on this board is given in figure [ fig3]c .
the three problem types are in a sense completely separate it is never possible to move from a svss problem of one type to another , even if you are allowed to rotate or reflect the board .
wiegleb s board ( figure [ fig4 ] ) is an extension of the standard 33-hole board and has 45 holes @xcite .
figure [ fig4 ] shows that this board also has three types of svss problems , but there are more of them ( 36 in all , see @xcite ) .
figure [ fig5 ] shows that the @xmath36 square board also has three problem types . in triangular peg
solitaire the situation is somewhat different . on the triangular board of side @xmath37 ,
if @xmath38 , the board is not null - class and no complement problem can be solved ( see bell @xcite ) .
the only svss problem on the 10-hole triangular board that is solvable is of the form : vacate @xmath39 finish at @xmath40 .
this gives only one type of problem that is solvable ( see figure [ fig6]a ) , and any problem starting from an unmarked hole in figure [ fig6]a can not be solved to one peg ( even if you are allowed to rotate and flip the board ) .
the 15-hole triangular board is null - class ( figure [ fig6]b ) .
here there are 4 different starting locations , the standard starting holes " for each of the 4 are shaded .
the fact that there is only one type indicates that all problems on this board are all interrelated .
the 21-hole triangular board is also null - class ( figure [ fig6]c ) , and contains 5 different starting locations . on this board
it is possible to start with any peg missing , and finish at any hole using peg solitaire jumps , _ plus _ rotating and flipping the board . here
are several peg solitaire problems we are interested in solving : the * complement problem * : from a starting vacancy @xmath27 , execute an arbitrary number of jumps , then determine if the board position can be reduced to one peg at @xmath27 .
the * general ( svss ) problem * : from a starting vacancy @xmath27 , execute an arbitrary number of jumps , then determine if the board position can be reduced to one peg ( anywhere on the board ) .
the * general problem * : from a given board position , determine if it can be reduced to one peg ( anywhere on the board ) .
we will consider primarily the first two problems in this paper .
we also want to solve these problems quickly ideally within a web browser . in determining if a jump leads to a dead end or
not , a delay of one second is unacceptable .
it is important to realize that problems # 2 and # 3 are different .
for example , on the 33-hole board , a popular problem to solve is cross " ( figure [ fig1]b ) .
this board position can never appear during the solution to any svss problem .
how do we know this ? because the complement of this board position can not be reduced to a single peg .
see bell @xcite to clarify why a board position during a svss problem must have this property .
a fair question is , why not go for the most general and difficult problem # 3 ?
the reason is that the complement problem # 1 and general svss problem # 2 are significantly easier , because we can take advantage of special properties of their solutions .
the simplest technique for solving a peg solitaire problem on a computer is to store the sequence of jumps , together with the current board position .
one then performs a depth - first search by jump ( extending the jump sequence and backtracking when no further move is possible ) .
the 15-hole triangular board can be easily solved using this technique , but it is much slower on the 33-hole board .
the reason is that there are a large number of jump sequences that result in the same board position , so there is a tremendous amount of duplicated work .
this difficulty can be removed by storing board positions seen previously in a hash table or binary tree .
a better technique is to stop recording the jump sequence entirely , and look at the whole problem as a sequence of board positions .
given a set of board positions @xmath41 , we denote by @xmath42 the set of board positions that can be obtained by performing every possible jump to every element of @xmath41 .
we call @xmath42 the * descendants * of @xmath41 . as a programming task calculating @xmath42
is straightforward .
for example the set @xmath41 can be stored on the disk as a sequence of integer codes , we read each code and convert it to a board position .
from this board position we execute every possible jump , resulting in a large number of board positions which are stored in a binary tree ( or hash table ) to remove duplicates .
this binary tree is then dumped to a file as a sequence of codes , the set @xmath42 .
the problem that eventually occurs is that the binary tree becomes too large to fit into memory . at this point
the problem is easily split into @xmath43 smaller pieces that are calculated separately .
it is best if @xmath43 is a prime number , chosen so as to reduce the problem size so that it fits in memory .
we now go through the board positions in @xmath41 as before , but instead of storing each board position @xmath11 in a binary tree , we convert it to a code and write that code into a temporary file numbered @xmath44 . here
% " operator represents the remainder upon division by @xmath43 ( as in c ) .
after all board positions in @xmath41 have been considered , we now go back through each of the @xmath43 temporary files , filling a binary tree for each @xmath43 to remove duplicates and finally writing the unique board positions to the disk ( in @xmath43 separate files ) .
in most cases we do not want to store two board positions that are in the same symmetry class . for example , for the central game ( figure [ fig1]a , there are four first jumps , but these result in identical board positions which are rotations of one another .
a nice way to select a single representative from the symmetry class is to use the one with the smallest code . in the above algorithms
, we use @xmath18 in place of @xmath12 . for a set of board positions @xmath41 , we denote by @xmath45 the number of elements in the set .
@xmath46 is the set of complemented board positions . in other words
@xmath47 if and only if @xmath48 .
we note that @xmath46 is not a set complement in the traditional sense ( i.e. @xmath47 if and only if @xmath49 ) , but rather _ a set of complemented board positions_. let @xmath50 be the initial board position with one peg missing .
let @xmath51 be the final board position with one peg .
let @xmath52 and @xmath53 .
we then define the set of board positions that can be reached from @xmath50 by : @xmath54 note that every element of @xmath55 has exactly @xmath37 pegs .
this produces a playing forward " sequence of sets @xmath56 where the number of elements in each set increases exponentially , at least initially .
we are calculating the nodes in the game tree " , but have lost all information about the links connecting them ( however , this link information is easily recovered ) .
the problem has a solution if and only if @xmath57 .
a sequence of sets can also be obtained from the finishing board position @xmath51 by playing backwards " , which in our notation we write as : @xmath58 again , every element of @xmath59 has exactly @xmath37 pegs , and the problem has a solution if and only if @xmath60 .
it is worth noting that the sets @xmath59 contain _ every _ board position which can be reduced to @xmath51 .
thus , the sets @xmath59 can be used to solve any problem # 3 ( p. 6 ) which finishes at @xmath51 , or symmetric equivalents .
if we calculate @xmath59 over all possible one peg finishes , we can solve any problem # 3 .
the set of winning board positions " with @xmath37 pegs is defined as @xmath61 if we have _ any solution _ , and play this solution until reaching board @xmath11 with @xmath37 pegs , then it must be that @xmath62 .
the sets @xmath63 are usually much smaller than the @xmath55 and @xmath59 , these are the nuggets of gold that we seek , because they enable us to quickly recognize any winning board position . as a practical matter , to find @xmath63 it is not necessary to calculate every @xmath55 and @xmath59 for each @xmath37 between @xmath64 and @xmath65 and perform their intersection ( intersecting two sets with potentially billions of elements is not a trivial computation ) .
suppose we can calculate the forward sets to @xmath66 , and the backward sets to @xmath67 for some @xmath68 between @xmath65 and @xmath64 . if the problem has a solution , then @xmath69 is not empty .
we then compute @xmath70 recursively using @xmath71 and @xmath72 using @xmath73 to calculate using equation ( [ eq : wdown ] ) , we take each element of @xmath74 , complement it , calculate all descendants and complement each .
this yields the set @xmath75 , and we now save each element which is in common with @xmath55 , giving us @xmath63 . the recursive calculations ( [ eq : wup ] ) and ( [ eq : wdown ] ) are much easier than calculating all @xmath55 and @xmath59 because the sets @xmath63 tend to be orders of magnitude smaller .
the determination of @xmath63 using ( [ eq : wup ] ) and ( [ eq : wdown ] ) is considerably faster than the initial task of calculating the sets @xmath66 and @xmath67 .
finally , we note that the sets @xmath55 , @xmath59 and @xmath63 can be defined in two subtly different ways .
first , they can simply be sets of board positions .
if @xmath76 is the starting position for the central game ( figure [ fig1]a ) , then @xmath77 , and @xmath78 has 4 elements which are rotations of one another
. we will sometimes refer to these sets as
raw @xmath55 " .
in most cases , however , we will consider @xmath55 , @xmath59 and @xmath63 as sets of symmetry classes . now
the set @xmath78 only has a single element , which can be taken as any representative of this symmetry class , and generally we choose the one with the smallest @xmath79 .
these sets are called symmetry reduced " @xmath55 .
if we refer to an unqualified @xmath55 or @xmath63 it can be assumed to be symmetry reduced .
for a complement problem , we have @xmath80 , which implies @xmath81 @xmath82 @xmath83 equation ( [ eq : w ] ) states that _ the sets of winning board positions are complements of one another_. this is a remarkable result , and tells us that for a complement problem , we only need to store half the winning board positions . in order to calculate @xmath63
the work is halved as well , for we need only calculate the forward sets @xmath55 down to @xmath84 . after performing the intersection ( [ eq : wcalc ] ) , the remaining @xmath63 are then calculated using equation ( [ eq : wup ] ) or equivalently ( [ eq : wdown ] ) . is storing winning board positions the most efficient technique ? during the start of a game , it does not seem so , because all board positions that can be reached are winning .
perhaps it is better to store losing board positions " , or positions from which a one peg finish at the starting hole can not be reached ?
we could define the set of losing board positions " with @xmath37 pegs as those elements of @xmath55 which are not in @xmath63 .
a more efficient technique is to store _ only _ those losing board positions which are one jump away from a winning board position .
thus , we define @xmath85 table [ table1 ] shows the sizes of @xmath55 , @xmath63 and @xmath86 for the 15- and 21-hole triangular boards . all winning board positions for any corner complement problem can be identified by storing just @xmath87 board positions ( 15-hole board ) or @xmath88 board positions ( 21-hole board ) .
if we store losing board positions as defined by equation ( [ ldef ] ) , we need to store more than four times as many board positions .
.size of @xmath55 , @xmath63 and @xmath86 for the corner complement problem on the 15 and 21-hole triangular boards .
@xmath89 only half of the @xmath63 need to be stored , due to equation ( [ eq : w ] ) .
[ cols="^ , > , > , > , > , > , > , > " , ] for example , since @xmath90 in binary , then @xmath91 contains all @xmath37-peg positions that are common to problems 1 , 2 and 3 .
we note the since problem 4 is unsolvable as a complement problem @xcite , @xmath92 and all sets @xmath93 with @xmath94 between @xmath95 and @xmath96 are also empty .
a trickier question is how to resolve the degeneracy at the @xmath97 starting location .
the sets @xmath98 can not be symmetry reduced , yet @xmath99 and @xmath100 are symmetry reduced , and @xmath92 .
we see from figure [ fig7 ] that @xmath98 contains code @xmath101 , while @xmath99 contains mincode @xmath102 , a board position in the same symmetry class . the solution is to use an algorithm which keeps all codes in @xmath98 but removes all symmetry equivalents in the intersecting sets .
this is the reason that the degenerate starting locations are indexed first .
table [ table5 ] summarizes the number of board positions by @xmath94 and number of pegs @xmath37 .
the total number of board positions over all sets is @xmath103 , which is only @xmath104 more than are needed to solve the general svss problem by itself .
we can infer that there are @xmath104 board positions among the @xmath103 that are duplicated
these must be @xmath104 pairs of board positions in @xmath98 that are in the same symmetry class .
figure [ fig8 ] shows representative 5-peg board positions in @xmath105 for values of the _ index _ 0 to 7 .
let us interpret two of the board positions in this diagram .
the board position with code @xmath106 is in _ index _ 0 , meaning that this board position can not appear in any complement problem .
this board position has a mincode of @xmath107 .
we can finish with one peg from this board position at @xmath108 or @xmath97 , but we can not start from either of these holes and reach this board position .
but it must be possible to reach this board position from some start , and it turns out this start is @xmath109 . the board position with _ index
_ 6 is @xmath110 , which is the mincode .
we can play from this board position to finish at @xmath108 , @xmath97 or @xmath109 , and we can reach this board position from @xmath108 or @xmath109 . therefore , this board position can be reached during the solution to the @xmath108 or @xmath109 complements , so is in _ index _ 6 .
we now present pseudocode for identification of winning board positions for either the complement problem ( # 1 ) or the general svss problem ( # 2 ) : .... w[5][1 ] = { 16,64,1,8 } ! w^1 , index=0,1 , ... 15 , for the 15-hole triangular board end[5][1 ] = { 1,2,3,3,4,4,4,4,4,4,4,4,4,4,4,4 }
! ends of each index 0 - 15 ... see triangle??winning.txt ... !
board is the current board position !
side is the triangular board side ( 4,5 , or 6 ) !
i is the number of this svss problem !
ksym is the symmetry code of the starting board position !
comp is true for complement problems , otherwise any finish is assumed boardissolvable(board , side , i , ksym , comp ) { int code[6 ] totholes = side*(side+1)/2 if ( side==4 ) { ! 10-hole triangular board topindex = 1 degen = 0 } if ( side==5 ) { ! 15-hole triangular board topindex = 2 ^ 4 ! power set of 4 problems degen = 1 !
number of degenerate problems } if ( side==6 ) { ! 21-hole triangular board topindex = 2 ^ 5 ! power set of 5 problems degen = 2 ! number of degenerate problems } n = countpegs(board ) if ( n > tot/2 ) then { code[0 ] = 2^tot - 1 - code(board ) n = tot - countpegs(board ) } else code[0 ] = code(board ) ! get the 6 symmetry codes , code[0 ] to code[5 ] code[1 ] = rotatecode(code[0 ] ) code[2 ] = rotatecode(code[1 ] ) code[3 ] = reflectcode(code[2 ] ) code[4 ] = rotatecode(code[3 ] ) code[5 ] = rotatecode(code[4 ] ) if ( comp ) { ! complement problem kstart = 0 kend = 6 if ( i<=degen ) { kstart = ksym kend = ksym + 1 } for ( index=1 ; index < topindex ; index++ ) { if ( ( 1<<i ) & index ) { !
true if the i'th bit of index is set for ( j = end[side][n][index-1 ] ; j < end[side][n][index ] ; j++ ) { for ( k = kstart ; k < kend ; k++ ) if ( code[k]==w[side][n][j ] ) return true } } } } else { ! comp = false , finish anywhere for ( j=0 ; j < end[side][n][topindex ] ; j++ ) { for ( k=0 ; k<6 ; k++ ) if ( code[k]==w[side][n][j ] ) return true } } return false } ....
i have created a javascript game @xcite for playing peg solitaire on the 10 , 15 and 21-hole triangular boards .
this game can begin from any starting vacancy , and the program will point out all jumps which lead to winning or losing board positions .
the game can be specified as either a complement problem ( the user must finish with one peg at the location of the original hole ) , or the general problem with a one peg finish anywhere on the board . the algorithm used to identify
winning and losing board positions is the algorithm ` boardissolvable ( ) ` in this paper converted to javascript .
* ` triangle10winning.txt ` a text file of the board positions @xmath111 , given in table [ table4 ] .
note : each file ` triangle??winning.txt ` contains two versions of the solution sets @xmath111 .
first , a version for identifying winning board positions in the finish anywhere " problem ( # 2 ) using the algorithm ` problemissolvable ( ) ` .
second , an _ indexed _ version as described in section 6 , which is slightly larger due to the symmetry degeneracy .
these sets can be used to identify winning board positions for any complement problem ( # 1 ) as well as the finish anywhere " problem ( # 2 ) using the more complex algorithm ` boardissolvable ( ) ` .
each set is sorted by index ( if present ) , then by code . on this 10 hole board
the two versions are identical so the same sets appear twice . * ` triangle15winning.txt ` a text file of the board positions @xmath111 for the 15-hole triangular board .
this is given as a single set subdivided by index .
the array ` end[index ] ` ( summarized by table [ table5 ] ) shows where the end of the codes for each index occurs .
the function ` boardissolvable ( ) ` shows how this information is used .
+ the sets for the corner complement problem ( problem @xmath112 ) : + @xmath1131@xmath114 , @xmath11510@xmath114 , @xmath11628 , 112@xmath114 , @xmath11723 , 58 , 85 , 120 , 1108 , 1616 , 2076 , 2210 , 2272@xmath114 , @xmath11831 , 93 , 115 , 601 , 1054 , 1138 , 1140 , 1562 , 1648 , 2183 , 2218 , 2245 , 2280 , 2348 , 2472 , 2616 , 2728 , 2819@xmath114 , @xmath119125 , 633 , 1086 , 1111 , 1594 , 1621 , 2191 , 2253 , 2275 , 2289 , 2343 , 2467 , 2589 , 2723 , 2785 , 2841 , 2889 , 3126 , 3250 , 3298 , 3428 , 3634 , 3845 , 4220 , 4270 , 4282 , 4691 , 4728 , 4817@xmath114 , @xmath1201567 , 1651 , 2235 , 2365 , 2413 , 2537 , 2731 , 2793 , 3159 , 3196 , 3320 , 3374 , 3388 , 3607 , 3642 , 3667 , 3669 , 3704 , 3859 , 3921 , 4215 , 4339 , 4341 , 4469 , 4701 , 4849 , 5302 , 5350 , 5746 , 5810 , 6881 , 6985 , 10053 , 10065 , 12065@xmath114 * ` triangle21winning.txt ` a text file of the board positions @xmath111 for the 21-hole triangular board . * ` english33winning_n.bin ` 16 binary files of @xmath37-peg winning board positions @xmath98 for the central game on the 33-hole cross - shaped board .
although 33 bit integers are needed to store a general board position , all ( symmetry reduced ) board positions in @xmath98 can be stored as 32 bit ( 4 byte ) integers .
the data format is _ unsigned _ 4 byte integer .
we give the beginning and end of each set below as a check that they are being interpreted correctly .
table [ table2 ] gives the sizes of each of these sets .
+ @xmath121 , @xmath122,@xmath123,@xmath124153 , 1680 , 16688 , 17928 , 66432 , 82976 , 147984 , 352256@xmath114,@xmath125158 , 692 , 793 , @xmath126 , 4554760 , 6684688 , 8601616@xmath114,@xmath127691 , 729 , 798 , @xmath126 , 137626128 , 138674576 , 138690824@xmath114,@xmath128734 , 1213 , 1459 , @xmath126 , 184696868 , 234962962 , 270942217@xmath114,@xmath129957 , 1786 , 1853 , @xmath126 , 287589385 , 305203236 , 305528850@xmath114,@xmath1301789 , 2941 , 3038 , @xmath126 , 409747465 , 439631908 , 439746596@xmath114,@xmath1316909 , 18174 , 19326 , @xmath126 , 1208836146 , 1208836241 , 1510760466@xmath114,@xmath13221502 , 22398 , 23294 , @xmath126 , 1552443410 , 2416730291 , 2418878546@xmath114,@xmath13324062 , 24555 , 27615 , @xmath126 , 2452121690 , 2452123930 , 2484693018@xmath114,@xmath13454271 , 55167 , 56063 , @xmath126 , 2586714211 , 2588438678 , 2592630883@xmath114,@xmath13553247 , 56831 , 57279 , @xmath126 , 2651879594 , 2655805539 , 3098292302@xmath114,@xmath136127999 , 128895 , 129791 , @xmath126 , 3793449102 , 3793531059 , 3793629859@xmath114,@xmath137126975 , 130559 , 229359 , @xmath126 , 3864553651 , 3928764638 , 3929805043@xmath114 .
* ` triang.zip` a folder of javascript programs which can identify all winning positions on the 10 , 15 , and 21-hole triangular boards ( source files for @xcite ) .
|
we consider the one - person game of peg solitaire played on a computer .
two popular board shapes are the 33-hole cross - shaped board , and the 15-hole triangular board we use them as examples throughout .
the basic game begins from a full board with one peg missing and finishes at a board position with one peg .
first , we discuss ways to solve the basic game on a computer .
then we consider the problem of quickly distinguishing boards positions that can be reduced to one peg ( winning " board positions ) from those that can not be solved to one peg ( losing " board positions ) .
this enables a computer to alert the player if a jump under consideration leads to a dead end .
on the 15-hole triangular board , it is possible to identify all winning board positions ( from any single vacancy start ) by storing a key set of 437 board positions . for the central game " on the 33-hole cross - shaped board
, we can identify all winning board positions by storing 839,536 board positions .
+ [email protected] + .2 in
|
in the last decade , multiphysics astrophysical simulations have increased exponentially in both sophistication and size @xcite ; however , the software tools to mine those simulations have not kept pace .
typically , methods for examining data suffer from a lack of agility , discouraging exploratory investigation . to accommodate this
, massively parallel visualization tools such as visit and paraview @xcite have been repurposed as domain - specific astrophysical tools .
this repurposing , while effective , does not benefit from domain - specific analysis or data structures .
the lack of domain - specific quantitative analysis tools designed for astrophysical data leads to the development of specialized tools by individual researchers or research groups , most of which are never shared outside the research group .
this can substantially inhibit collaboration between different groups even those using the same simulation code .
this fractionation of the astrophysical community demonstrates a clear need for a flexible and cross - code software package for quantitative data analysis and visualization . in this paper
we present ` yt ` @xcite , a data analysis and visualization package that works with several astrophysical simulation codes . `
yt ` is developed openly and is freely available at ` http://yt-project.org/ ` .
it has been designed to be a common platform for simulation analysis , so that scripts can be shared across groups and analysis can be repeated by independent scientists . by making this tool available , we hope not only to encourage cross - group collaboration and validation of results , but to remove or at least greatly lower the barrier to entry for exploratory simulation analysis . ` yt ` provides mechanisms for conducting complete analysis pipelines resulting in publication quality figures and data tables , as well as the necessary components for constructing new methods for examining data .
the concepts for data handling and representation in ` yt ` are certainly not new , but their application to astrophysical data enables complex , detailed analysis pipelines to be shared between individuals studying disparate phenomena using disparate methods .
this enables and even encourages reproducibility and independent verification of results .
we have built this analysis and visualization code in python , using numpy @xcite for fast mathematical operations , ` mpi4py ` for mpi - parallelism @xcite , and optionally matplotlib for 2d visualization @xcite .
additionally , several core library routines in ` yt ` such as the amr volume rendering , multi - dimensional binning , and file access routines , are written in cython .
in addition to utilizing community - developed python modules , ` yt ` is itself a python module suitable for direct scripting or access as a library .
a community of users and developers has grown around the project , with over 20 committers in the history of the project , and it has been used in numerous published papers and posters .
( see , for example , @xcite . ) in order to accomodate the diverse computing environments on which astrophysical simulations are run , ` yt ` was designed to use primarily off - screen rendering and scripting interfaces , although several smaller tools are provided for specific , interactive visualization tasks .
the former method is well - suited to remote visualization and can be run via a job execution queue on a batch - compute cluster , such as those on which the underlying simulation are run . ` yt ` is subdivided into several sub - packages for data handling , data analysis , and plotting .
this modularity encourages the creation of reusable components for multi - step analysis operations . while work continues on the exploratory , post - processing methods of data analysis and visualization that ` yt ` was originally designed for , current development has focused on analysis during the course of a simulation , or so - called _ in situ _ analysis .
this allows for high - cadence analysis to be conducted without writing data to disk .
future simulations , such as those to be conducted on petascale machines , will require analysis and visualization during the simulation rather than exclusively as a post - processing technique . in this paper
, we will describe the mechanisms that ` yt ` provides for accessing data ( [ sec : mechanisms ] ) , methods of interacting with ` yt ` ( [ sec : interacting ] ) , the visualization techniques offered by ` yt ` ( [ sec : visualization ] ) , the parallelism strategy for data analysis and generation of visualizations ( [ sec : parallelism ] ) , and end with a discussion of the process of embedding ` yt ` in running simulation codes and how this will be inverted in the future ( [ sec : embedding ] ) .
` yt ` primarily operates on adaptive mesh refinement ( amr ) data , where the simulation domain is divided into spatially - organized zones along a regular mesh .
regions requiring higher degrees of refinement to capture shocks , collapse , instabilities and so on are replaced with higher - resolution meshes ; in this manner , a stable and accurate solution is ensured at all locations .
( for more information , see one of several astrophysics papers describing amr implementations , such as @xcite . )
while ` yt ` is able to analyze collisionless particles ( such as dark matter or star particles ) , it is currently ill - suited to analysis of smoothed particle hydrodynamics ( sph ) simulations ; future versions will improve support .
the vast majority of amr calculations in the astrophysical literature are computed on a rectilinear grid ; while this affords a number of computational efficiencies and conveniences , astrophysical phenomena as a whole are not rectangular prisms and thus are poorly suited for analysis as rectangular prisms .
this presents a fundamental disconnect between the data structures utilized by simulations and the geoemetries found in nature .
furthermore , the task of selecting geometric regions in space requires substantial overhead : masking of overlapping simulation regions , selective io , data selection , and so on . ` yt ` provides a number of convenience functions and mechanisms for addressing data within astrophysical simulations that make the process of handling and manipulating data straightforward .
the ` yt ` codebase has been organized along several conceptual lines , each corresponding to a set of tasks or classes in python .
the primary mechanisms for handling data are contained in the python module ` yt.data_objects ` , while all code and data structures specific to a particular simulation code resides within a submodule of ` yt.frontends ` ( such as ` yt.frontends.enzo ` , ` yt.frontends.orion ` , etc ) . in the current version of ` yt ` full support is provided for accessing and reading enzo @xcite , flash @xcite , orion @xcite and nyx codes , with preliminary support for ramses , art @xcite and several others . to open a dataset , the user creates an instance of a simulation code - specific subclass of ` staticoutput ` , a lightweight class that scans a parameter file and obtains the necessary information to orient the dataset : the current time in the simulation , the domain information , the mechanisms for converting units , and the necessary file locations on disk . a convenience function ( ` load ` ) for automatically creating such an instance
is provided , such that it only requires a path on disk to the dataset of interest .
however , geometric information about the manner in which data is laid out on disk or in the simulation domain is compartmentalized to a ` amrhierarchy ` object .
these objects are comparatively expensive to construct , as they contain a hierarchy of ` gridpatch ` objects , all of which posses spatial and parentage information .
these objects are not instantiated or constructed until requested .
all data access is mediated by ` amrhierarchy ` objects , as noted below . by relegating data handling to individual instances of classes ,
we compartmentalize datasets ; because each dataset is merely a variable , the number that can be opened and simultaneously cross - compared is only limited by the available memory and processing power of the host computer .
furthermore , datasets from different simulation codes can be opened and compared simultaneously in memory . when handling astrophysical data , it is appropriate to speak of geometric regions that outline the rough boundaries of physical objects : dark matter halos as ellipsoids , protostars as spheres , spiral galaxies as cylinders , and so on .
the central conceit behind ` yt ` is the presentation to the user of a series of physical objects with the underlying simulation largely abstracted . for amr data
, this means hiding the language of grid patches , files on disk and their interrelationships , and instead describing only geometric or physical systems ; these intermediate steps are handled exclusively by ` yt ` , without requiring any intervention on the part of the user .
for instance , to select a spherical region , the user specifies a center and a radius and the underlying ` yt ` machinery will identify grid patches that intersect that spherical region , identify which grid patches are the most highly - refined at all regions within the sphere , locate the appropriate data on disk , read it and return this data to the user . by abstracting the selection of and access to data in this manner , operations that can be decomposed spatially or that are embarrassingly parallel " can be transparently parallelized , without requiring the user s intervention .
the data containers implemented in ` yt ` include spheres , rectangular prisms , cylinders ( disks ) , arbitrary regions based on logical operations , topologically - connected sets of cells , axis - orthogonal and arbitrary - angle rays , and both axis - orthogonal and arbitrary - angle slices .
data containers provide several methods for data access .
the data can be accessed directly , as in the above code listing , or through abstractions such as _ object quantities _
, where bulk operations are conducted such as calculating the angular momentum vector or the total mass .
the abstraction of data into data containers leads to the creation of systems of components : data containers become `` sources '' for both analysis procedures as well as visualization tasks .
these analysis procedures then become reusable and the basis for chains of more complicated analysis tasks . using such chains
, a user can volume render a set of halos based on their angular momentum vectors , color particles by merger history , and even calculate disk inclination angles and mass fluxes .
once a region of the simulation is selected for analysis , ` yt ` must process the raw data fields themselves . its model for handling this data and processing fundamental data fields into new fields describing _ derived values _ is built on top of an object model with which we can build automatically recursive field generators that depend on other fields .
all fields , including derived fields , are allowed to be defined by either a component of a data file , or a function that transforms one or more other fields .
this indirection allows multiple layers of definition to exist , encouraging the user to extend the existing field set as needed , using python functions as transformation and mathematical operators .
the primary interface to ` yt ` is through a programmatic api .
scripts are written and then executed , either in serial or in parallel through a batch queue .
interactive helper functions , implemented using ipython , are also provided for tab - completion , figure handling and so forth .
we provide a command - line utility with many common functions : plotting , statistics , volume rendering , halo finding , pastebinning , image uploading , bug reporting , and even uploading a script to the yt hub ( ` http://hub.yt-project.org/ ` ) to share with other users .
recently , the ability to spawn a google maps - like interface has been added , to allow interactive panning and zooming of multi - resolution datasets from a web - browser .
the most recent version of yt ( 2.2 ) features a new gui , entitled `` reason , '' as seen in figure [ fig : reason ] . at many supercomputing centers , toolkits for constructing graphical user interfaces are not available or are extremely difficult to build and install .
this greatly reduces the utility of creating a traditional gui . to circumvent this limitation ` yt ` provides a fully - integrated gui written in html and javascript ( extjs ) and served by a webserver ( bottlee and rocket ) running inside ` yt ` itself . rather than a large , bulky framework within which operations could be constructed and executed
, this gui will presents a simple interactive interpreter that can be accessed through a web browser .
this hosted interpreter dynamically creates user interface widgets as well as enabling inline display of newly - created images .
the primary user interface is a single - cell which can be submitted to execute on the server ; additional convenience features such as embedded irc , pastebin support , image upload and bug reporting are also included .
as this gui requires no client - side libraries or widgets , and as python itself provides all of the necessary tools on top of which this type of gui could be built , we believe this will be more maintainable and straightforward than a traditional gui .
a user creates a new server on - demand on a supercomputing center login node and connects to it through an ssh tunnel from a local machine such as a laptop .
remote analysis and visualization are then guided and driven through the locally - rendered web page , with results and images passed back asynchronously and displayed inline in the same web page .
future versions will allow for parallel execution in batch queues and detachment and reattachment operations .
` yt ` provides methods for creating 2d and 3d visualizations of simulation data .
the mechanisms for creating 2d visualizations have two primary components : the data - handling portion and the figure creation or `` pixelization '' step .
the former is composed of a set of objects which provide uniform access to 2d data objects , while the latter is a simple method for making plots quickly , which can be wrapped into other convenience functions ( both created by ` yt ` and external to ` yt ` . ) the figure creation in ` yt ` is motivated by a desire for simplicity : rather than attempting to accommodate the myriad use cases and user preferences , ` yt ` seeks to provide a set of routines that can be extended easily .
users requiring complex figures for specific publications can take the 2d image pixel buffers provided by ` yt ` and feed them to any plotting package , though ` yt ` integrates most naturally with the matplotlib python module @xcite . here , we first describe each of the 2d pixalization mechanisms , and then the 3d volume rendering algorithms . futher information on the simple , built - in figure generation can be found in the ` yt ` documentation .
the simplest means of examining data is plotting grid - axis aligned slices through the dataset .
this has several benefits - it is easy to calculate which grids and which cells are required to be read off disk ( and most data formats allow for easy striding of data off disk , which reduces this operation s io overhead ) and the process of stepping through a given dataset is relatively easy to automate . when handling astrophysical simulation data , one often wishes to examine either the sum of values along a given sight - line or a weighted - average along a given sight - line , in a _
projection_. ` yt ` provides an algorithm for generating line integrals in an adaptive fashion , such that every returned @xmath0 point does not contain data from any points where @xmath1 or @xmath2 ; the alternative to this is a simple 2d image array of fixed resolution perpendicular to the line of sight whose values are filled in by all of the cells of the source object with overlapping domains .
but , by providing this list of _ all _ finest - resolution data points in a projected domain , images of any field of view can be constructed essentially instantaneously ; conversely , however , the initial projection process takes longer , for reasons described below .
we term the outputs of this process _
adaptive projections_. for the santa fe light cone dataset @xcite , to project the entire domain at the highest resolution would normally require an image with @xmath3 values .
utilizing this adaptive projection method , we require less than @xmath4 of this amount of image storage .
direct ray casting through a volume enables the generation of new types of visualizations and images describing a simulation . `
yt ` has the facility to generate volume renderings by a direct ray casting method .
currently the implementation is implemented to run exclusively on the cpu , rather than faster hardware - based rendering mechanisms , but this also allows for clearer descriptions of the algorithms used for compositing , calculation of the transfer function , and future advances in parallelization .
furthermore , it eases the task of informing volume renderings with other analysis results : for instance , halo location , angular momentum , spectral energy distributions and other derived or calculated information . in ` yt ` , volume rendering is exposed through a `` camera '' interface that allows for camera paths , zooms , stereoscopic rendering and easier access to the underlying vector plane .
transfer functions that can automatically sample colormaps as well as one that provides off - axis line integrals are supplied , as well as a transfer function whose colors correspond to johnson filter - convolved planck emission with approximate scattering terms , as in @xcite . utilizing the healpix algorithm for equal latitudinal decomposition of a sphere @xcite ` yt ` also provides the ability to render @xmath5 images , suitable both for creating outward - facing sky maps and planetarium images . .
isocontours were placed at @xmath6 and @xmath7 @xcite.,scaledwidth=46.0% ] by allowing for detailed control over the specification of the transfer function , viewing angle and generation of images , volume renderings that contain a scientific narrative are easier to create .
for instance , in figure [ fig : vr_binary ] we have constructed a volume rendering of the population iii star formation simulation described in @xcite , where a collapsing metal - free halo has been found to fragment into two distinct clumps .
this volume rendering has been aligned such that the normal vector to the image plane is aligned with the angular momentum vector of the two - clump system .
furthermore , the isocontours visible in the image have been selected such that they coincide with transitions between chemical states in the cloud .
additional volume renderings based on derived fields describing chemical and kinetic quantities could be constructed , as well .
as the capabilities of supercomputers grow , the size of datasets grows as well .
most standalone codes are not parallelized ; the process is time - consuming , complicated , and error - prone . therefore ,
the disconnect between simulation time and data analysis time has grown ever larger . in order to meet these changing needs , ` yt ` has been modified to run in parallel on multiple independent processing units on a single dataset .
specifically , utilizing the message passing interface @xcite via the ` mpi4py ` python module @xcite , a lightweight , numpy - native wrapper that enables natural access to the c - based routines for interprocess communication , the code has been able to subdivide datasets into multiple decomposed regions that can then be analyzed independently and joined to provide a final result .
a primary goal of this process has been to preserve at all times the api , such that the user can submit an unchanged serial script to a batch processing queue , and the toolkit will recognize it is being run in parallel and distribute tasks appropriately .
the tasks in ` yt ` that require parallel analysis can be divided into two broad categories : those tasks that act on data in an unordered , uncorrelated fashion ( such as weighted histograms , summations , and some bulk property calculation ) , and those tasks that act on a decomposed domain ( such as halo finding and projection ) .
all objects and tasks that utilize parallel analysis exist as subclasses of ` parallelanalysisinterface ` , which provides a number of functions for load balancing , inter - process communication , domain decomposition and parallel debugging . furthermore , ` yt ` itself provides a very simple parallel debugger based on the python built - in ` pdb ` module .
to parallelize unordered analysis tasks , a set of convenience functions have been implemented utilizing an initialize / finalize formalism ; this abstracts the entirety of the analysis task as a transaction .
signaling the beginning and end of the analysis transaction initiates several procedures , defined by the analysis task itself , that handle the initialization of data objects and variables and that combine information across processors .
these are abstracted by an underlying parallelism library , which implements several different methods useful for parallel analysis . by this means , the intrusion of parallel methods and algorithms into previously serial tasks is kept to a minimum ; invasive changes are typically not necessary to parallelize a task .
this transaction follows four steps .
first , the list of grids to process is obtained .
this is followed by initialization of the parallelism on the data object .
each grid is then processed , and a finalize process is conducted on the data object .
this is implemented through the python iterator protocol ; the initialization of the iterator encompasses the first two steps and the finalization of the iterator encompasses the final step . inside the grid selection routine , ` yt ` decomposes the relevant set of grids into chunks based on the organization of the datasets on disk .
implementation of the parallel analysis interface mandates that objects implement two gatekeeper functions for both initialization and finalization of the parallel process . at the end of the finalization step
, the object is expected to be identical on all processors .
this enables scripts to be run identically in parallel and in serial . for unordered analysis
, this process results in close - to - ideal scaling with the number of processors . in order to decompose a task across processors , a means of assigning grids to processors
is required . for spatially oriented - tasks ( such as projections )
this is simple and accomplished through the decomposition of some spatial domain . for unordered analysis tasks ,
the clear means by which grids can be selected is through a minimization of file input overhead .
the process of reading a single set of grid data from disk requires the opening of a file , seeking to the position of the dataset in that file , the actual reading of the data , and the file close operation . for those data formats where
multiple grids are written to a single file , this process can be consolidated substantially by performing multiple reads inside a single file once it has been opened .
if we know the means by which the grids and fields are ordered on disk , we can simplify the seeking requirements and instead read in large sweeps across the disk . by
futher pre - allocating all necessary memory , this becomes a single operation that can be accomplished in one `` sweep '' across each file . by allocating as many grids from a single `` grid output '' file on a single processor , this procedure can be used to minimize file overhead on each processor .
each of these techniques are implemented where possible . in figure
[ fig : scaling ] ( left panel ) we show the results of a strong - scaling study of conducting profiles of the final dataset from the santa fe light cone @xcite project .
this dataset consists of @xmath8 computational elements .
the dashed black corresponds to profiling in one dimension , and the solid line corresponds to profiling in two dimensions .
overplotted in thin solid lines are the ideal scaling relationships , as calibrated to the time taken by 16 processors .
we see nearly ideal strong scaling up to 128 processors , at which point overhead dominates ; we are essentially starving the processors of work at this scale .
the overall time taken to conduct a profile is quite low , on one of the largest amr datasets in the published literature .
we note also that the substantial speed difference between the two mechanisms of profiling , which is counter - intuitive , is a result of a difference in implementation of the histogramming method ; 1d profiles use a pure - python solution to histogramming , whereas 2d profiles use a hand - coded c routine for histogramming . future versions of ` yt ` will eliminate this bottleneck for 1d profiling and we expect to regain parity between the two methods .
several tasks in ` yt ` are inherently spatial in nature , and thus must be decomposed in a spatially - aware fashion .
mpi provides a means of decomposing an arbitrary region across a given number of processors . through this method
, the ` parallelanalysisinterface ` provides mechanisms by which the domain can be divided into an arbitrary number of subdomains , which are then realized as individual data containers and independently processed .
for instance , because of the inherently spatial nature of the adaptive projection algorithm implemented in ` yt ` , parallelization requires decomposition with respect to the image plane ( however , future revisions of the algorithm will allow for unordered grid projection . ) to project in parallel , the computational domain is divided such that the image plane is distributed equally among the processors ; each component of the image plane is decomposed into rectangular prisms ( ` amrregion ` instances ) along the entire line of sight .
each processor is allocated a rectangular prism of dimensions @xmath9 where the axes have been rotated such that the line of sight of the projection is the third dimension , @xmath10 is constant across processors , and @xmath11 is the entire computational domain along the axis of projection .
following the projection algorithm , each processor will then have a final image plane set of points , as per usual : @xmath12 but subject to the constraints that all points are contained within the rectangular prism as prescribed by the image plane decomposition . at the end of the projection step
all processors join their image arrays , which are guaranteed to contain only unique points . in figure
[ fig : scaling ] ( right panel ) we show the results of a strong - scaling study of adaptively projecting the same dataset as above .
the dashed line represents a projection of the density field , whereas the solid line represents projection in the absence of disk io .
clearly the algorithmic overhead dominates the cost of disk io , but we also see strong scaling between 4 and 64 processors ; at 128 processors we see deviation from this . the relatively early termination of strong scaling ( 64 processors for this dataset , but we expect this to be higher for larger datasets ) as a result of algorithmic overhead is one of the motivations behind future improvements to the projection algorithm .
however , from a pragmatic perspective , because ` yt ` creates adaptive projections , the time taken to project is a one - time investment and thus not a rate - determining step for post - processed analysis . for non - adaptive projections
, the process of handling all of the data , conducting parallel reductions and outputting images must be undertaken for every chosen field of view .
k @xmath13 k , with densities less than 50 times the cosmic mean density .
the left panels show the results of this analysis performed using 252 simulation datasets written to disk , while the right panels show the identical analysis performed at every single integration timestep with an in - situ instantiation of ` yt ` . of note
are three additional input and output phases identified in the in - situ analysis that were missed with the on - disk analysis.,scaledwidth=48.0% ] an outstanding problem in the analysis of large scale data is that of interfacing with disk storage ; while data can be written to disk , read back , and then analyzed in an arbitrary fashion , this process is not only slow but requires substantial intermediate disk space for a substantial quantity of data that will undergo severely reductionist analysis @xcite . to address this problem ,
the typical solution is to insert analysis code , generation of derived quantities , images , and so forth , into the simulation code .
however , the usual means of doing this is through either a substantial hand - written framework that attempts to account for every analysis task , or a limited framework that only handles very limited analysis tasks . ` yt ` provides an explicit embedding api to enable in - line analysis . by enabling in - line analysis ,
the relative quantity of analysis output is substantially greater than that enabled by disk - mediated analysis ; the cadence of analysis tasks can be increased , leading to greater time - domain resolution . removing numerous large files dumped to disk as a prerequisite for conducting analysis and generating visualization allows for a much more favorable ratio of data to analyzed data .
for example , in a typical population iii star formation simulation , such as in @xcite , the size of the data dumps can be as much as 10 gigabytes per timestep ; however , the relative amount of information that can be gleaned from these outputs is significantly smaller @xcite .
using smaller data output mechanisms as well as more clever streaming methods can improve this ratio ; however , by enabling in - line analysis , images of the evolution of a collapsing population iii halo can be output at every single update of the hydrodynamical time , allowing for true `` movies '' of star formation to be produced . by allowing for the creation and exporting of radial profiles and other analytical methods ,
this technique opens up vast avenues for analysis while simulations are being conducted , rather than afterward . at the opposite end of the spectrum from simulations of population iii star formation , which involve deep and complex adaptive - mesh hierarchies , cosmological simulations of the evolution of large scale structure , e.g. @xcite , are characterized by static meshes that are much simpler but vastly greater in size .
the largest simulations of @xcite , which had 1024@xmath14 grid cells , required roughly 135 gb of disk space for each dataset , of which 252 evenly spaced in time were written .
this allowed for the study of time - dependent phenomenon with a resolution of approximately 55 million years .
one of the primary goals of @xcite was to understand the movement of matter from one `` phase '' to another over cosmic time , where a phase is determined simply by the density and temperature .
the flux of material from one phase to another was calculated by comparing two consecutive datasets written to disk and identifying grid cells in each dataset in which the phase had changed .
an example of this is shown in the left panels of figure [ fig : phase_flux ] , in which all of the inputs ( top ) and outputs ( bottom ) into a single phase , the diffuse warm whim , are plotted .
more recently , even larger simulations of this nature , with 1536@xmath14 grid cells , were run with the identical analysis performed .
however , instead of using consecutively written datadumps , the analysis was performed during the simulation at every integration timestep using the in - situ capabilities of ` yt ` with grid data stored in memory .
this allowed the phase flux analysis to be performed 5000 times instead of just 252 , for a factor of 20 higher time resolution , resulting in far greater insight into the phase evolution of matter , as is illustrated in the right panels of figure [ fig : phase_flux ] .
the use of in - situ analysis was all the more necessary in this simulation as each dataset requires nearly 0.5 tb of disk space . at this scale ,
performing this analysis on datasets written to disk with the equivalent time resolution would have require nearly 3 pb of storage space .
the python / c api allows for passage of data in - memory to an instance of the python interpreter . `
yt ` has been instrumented such that it can be accessed by an embedded python interpreter inside a simulation code , such that one interpreter instance exists for every mpi task . ` yt ` provides a clear api for passing the necessary geometric information from the simulation code to the analysis package . by utilizing thin wrappers around the memory
in which field values and simulation data exist , the contents of the running simulation are exposed to ` yt ` and analysis can be conducted on them . while this currently works for many relatively simple tasks ,
it is not currently able to decompose data spatially ; as we are constrained by the parallel nature of most domain decomposition algorithms , we attempt to avoid passing data between mpi tasks .
this means if a grid resides within mpi task 1 , it will not be passed to mpi task 2 during the analysis stage .
currently this mechanism for inline analysis has been exposed to enzo simulations , and we hope to extend this in the future to additional simulation codes .
inline analysis will only become more important as simulations increase in size and scope , and future development in ` yt ` will make it easier , more robust , and more memory efficient . the primary mechanism by which ` yt ` will be embedded will change ; future iterations of the inline analysis interface will rely on communication between separate mpi jobs for simulation and analysis , rather than an analysis task that shares memory space with the running simulation code .
this mechanism will allow asynchronous analysis tasks to be run , enabling the simulation to proceed while the user controls the data that is examined .
additionally , the method for interfacing ` yt ` and simulation codes will be provided as a single c++ library that can be linked against , allowing it to be embedded by other developers .
development on ` yt ` is driven by the pragmatic needs of working astrophysics researchers .
several clear goals need to be met in the future , particularly as the size and scope of simulation data increases .
we also hope to work with other research groups to add support for the visualization and analysis of output from other popular astrophysics simulation codes such as art , gadget , pluto @xcite , and zeus - mp @xcite .
the most relevant improvement for very large simulation datasets is to improve load balancing for parallel operations .
as noted above , for some operations this can be accomplished by avoiding image - plane decomposition .
several efforts are underway to this end .
both the volume rendering and projection algorithms load balance through decomposition of the image plane , which often leads to poor work distribution .
these shortcomings are being addressed algorithmically : adaptive projections will utilize a quad tree , enabling better load balancing , and volume rendering will utilize a kd - tree approach combined with intermediate image composition .
however , experimentation in quad tree projection algorithms have indicated that the most rate - determining step shifts , from io resulting from poor load balancing to a time - consuming merger step , wherein quad trees from different processors are merged .
however , an underlying problem with the parallelization as it stands is the global state ; each instance of a python interpreter running ` yt ` currently runs in either `` parallel '' or `` serial '' mode .
future versions of the ` yt ` parallel analysis interface will allow this to be toggled based on the task under consideration , as well as more convenience functions for distributing work tasks between processors for instance , scatter / gather operations on halos .
we intend to implement this on top of mpi , utilizing non - blocking probes to function as a queueing and task distribution system .
improvements to the communication mechanisms for parallel analysis in ` yt ` will be necessary as _ in situ _ analysis grows more pervasive in large calculations . _ in situ _ analysis is challenging as it must necessarily proceed asynchronously with the simulation ; this will require careful data transport between ` yt ` and the simulation code .
abstracting and isolating the engine that drives this communication will be necessary to enable ` yt ` to be embedded in simulation codes other than enzo .
to this end , we have implemented a
stream " frontend , suitable for supplying arbitrary data to ` yt ` .
this can function either as a remote endpoint for mpi intercommunicators , or as input from paraview , translating vtk objects obtained through paraview s co - processing functionality into ` yt ` objects .
the process of instrumenting simulation codes for inline analysis provides additional avenues for deeper control of the simulation code .
typically , the process of execution of a simulation involves an initialization step , a main loop where modules that update the physical state of the simulation are executed in sequence and the current time of the simulation is updated , and then a finalization step , where memory is de - allocated , final outputs written and the simulation is terminated .
the mechanism for calling physics modules is either extremely specific to a given code or relatively cumbersome .
however , once the simulation code s internal structures have been exposed to the broader runtime environment , the conversion of this loop to a higher level language becomes more practical and useful .
this would enable rapid testing of components such as load balancing schemes and physics modules .
the process of modifying and debugging a code would be greatly simplified , and an interactive iteration through the main loop would ease the process of inspecting and debugging a simulation .
new users would be able to interactively step through the physics modules , manually inspecting the updates to physical quantities and learning how the simulation code behaves , rather than tediously examining and recompiling .
additionally , this provides the opportunity to interface more readily with co - scheduled visualization tools through mpi inter - communicators . by abstracting the interface to underlying physics modules ,
individual physics modules become trivially portable . in the future
we intend to position ` yt ` as the outermost control structure for simulations of tera- and peta - scale problems
. this will require substantial effort ; the first steps in this will be to identify an _ de facto _ , rather than _ de jure _ common api for physics modules typically used in astrophysical simulation codes such as enzo .
we are developing wrappers for fundamental physics modules using cython .
file system latency and read times become problematic even on the highest performance lustre file systems at processor counts in excess of @xmath15 , which requires the usage of a static - linking of utilized python libraries as well as the usage of zipfile - based module importing .
in addition to this , in order to sidestep the issue of abi incompatibilities with mismatched c++ compilers , we have eschewed the usage of matplotlib for this purpose and instead have developed a simplified png file writer that wraps both libpng and freetype for rapid inspection of colorbars and plots . through this process
, we intend to push forward in efforts to unify simulation and analysis through high - level `` glue code '' such as ` yt ` for the largest scale simulations of star formation , galaxy formation , and the evolution of galaxy clusters and the intergalactic medium .
the ` yt ` project is fully free and open source software , released under the gnu general public license , with no dependencies on external code that is not also free and open source software .
the development process occurs completely in the open at ` http://yt-project.org/ ` , with publicly - accessible source control systems , bug tracking , mailing lists , and regression tests . building
a community of users has been a priority of the ` yt ` development team , both to encourage collaboration and to solicit contributions from new developers ; both the user and developer communities are highly distributed around the world . ` yt ` is developed using mercurial , a distributed version control system that enables local versioned development and encourages users to make and contribute changes upstream .
many of the operations conducted in ` yt ` : fluid analysis , phase diagrams , volume rendering , parallelism , and in situ analysis could feasibly be applied to domains other than astrophysics .
we intend to generalize the underlying code base such that it can be applied to many other data formats in astrophysics , and ultimately we hope to provide these tools and techniques to domains other than astrophysics .
our first steps toward this , providing a generic and arbitrary data loader , have shown that it is feasible .
future versions of ` yt ` will generalize fields and particle handling , and should make this process much easier
. the creation of a freely available , publicly inspectable platform for simulation analysis allows the community to disentangle the coding process from the scientific process .
simultaneously , by making this platform public , inspectable and freely available , it can be openly improved and verified . the availability and relatively approachable nature of ` yt ` , in addition to the inclusion of many simple analysis tasks , reduces the barrier to entry for young scientists . rather than worrying about the differences between enzo and flash hierarchy formats , or row versus column ordering , or hdf4 versus hdf5 versus unformatted fortran data formats
, researchers can focus on understanding and exploring their data . more generally , however , by orienting the development of an analysis framework as a community project , the fragmentation of methods and mechanisms for astrophysical data analysis is greatly inhibited .
future generations of simulations and simulation codes will not only benefit from this collaboration , but they will require it .
m.j.t was supported in this work by nsf ci tracs fellowship award oci-1048505 .
b.d.s acknowledges support by nasa grants atfp nnx09-ad80 g and nnz07-ag77 g and nsf grants ast-0707474 and ast-0908199 .
both authors would like to thank the users and developers of ` yt ` , in particular jeffrey s. oishi , samuel w. skillman , stephen skory , cameron hummels , and john h. wise .
we also thank tom abel , greg l. bryan , berk geveci , charles law , michael l. norman , brian w. oshea , jorge poco , and george zagaris for thoughtful discussions . `
yt ` has been supported directly or indirectly over the years by a number of different funding agencies , including nsf , doe and academic institutions , and we are grateful for their support .
v. springel , s. d. m. white , a. jenkins , c. s. frenk , n. yoshida , l. gao , j. navarro , r. thacker , d. croton , j. helly , j. a. peacock , s. cole , p. thomas , h. couchman , a. evrard , j. colberg , f. pearce , simulations of the formation , evolution and clustering of galaxies and quasars , 435 ( 2005 ) 629636 . http://arxiv.org/abs/arxiv:astro-ph/0504097 [ ] , http://dx.doi.org/10.1038/nature03597 [ ] .
m. kuhlen , j. diemand , p. madau , m. zemp , the via lactea incite simulation : galactic dark matter substructure at high resolution , journal of physics conference series 125 ( 1 ) ( 2008 ) 012008+ .
http://arxiv.org/abs/0810.3614 [ ] , http://dx.doi.org/10.1088/1742-6596/125/1/012008 [ ] .
d. r. reynolds , j. c. hayes , p. paschos , m. l. norman , self - consistent solution of cosmological radiation - hydrodynamics and chemical ionization , journal of computational physics 228 ( 2009 ) 68336854 . http://arxiv.org/abs/0901.1110 [ ] , http://dx.doi.org/10.1016/j.jcp.2009.06.006 [ ] . p. ocvirk , c. pichon , r. teyssier , bimodal gas accretion in the horizon - marenostrum galaxy formation simulation , 390 ( 2008 ) 13261338 . http://arxiv.org/abs/0803.4506 [ ] , http://dx.doi.org/10.1111/j.1365-2966.2008.13763.x [ ] . a. almgren , j. bell , d. kasen , m. lijewski , a. nonaka , p. nugent , c. rendleman , r. thomas , m. zingale , maestro , castro , and sedona petascale codes for astrophysical applications , arxiv e - printshttp://arxiv.org / abs/1008.2801 [ ] . b. fryxell , k. olson , p. ricker , f. x. timmes , m. zingale , d. q. lamb , p. macneice , r. rosner , j. w. truran , h. tufo , flash : an adaptive mesh hydrodynamics code for modeling astrophysical thermonuclear flashes , 131 ( 2000 ) 273334 . http://dx.doi.org/10.1086/317361 [ ] .
g. h. weber , s. ahern , e. w. bethel , s. borovikov , h. r. childs , e. deines , c. garth , h. hagen , b. hamann , k. i. joy , d. martin , j. meredith , prabhat , d. pugmire , o. rbel , b. van straalen , k. wu , recent advances in visit : amr streamlines and query - driven visualization , in : numerical modeling of space plasma flows : astronum-2009 ( astronomical society of the pacific conference series ) , 2010 , lbnl-3185e . to appear .
m. j. turk , b. d. smith , j. s. oishi , s. skory , s. w. skillman , t. abel , m. l. norman , yt : a multi - code analysis toolkit for astrophysical simulation data , 192 ( 2011 ) 9+ .
[ ] , http://dx.doi.org/10.1088/0067-0049/192/1/9 [ ] .
j. d. hunter , http://dx.doi.org/10.1109/mcse.2007.55[matplotlib : a 2d graphics environment ] , computing in science & engineering 9 ( 3 ) ( 2007 ) 9095 . http://dx.doi.org/10.1109/mcse.2007.55 [ ] .
d. w. silvia , b. d. smith , j. m. shull , numerical simulations of supernova dust destruction .
i. cloud - crushing and post - processed grain sputtering , 715 ( 2010 ) 15751590 . http://arxiv.org/abs/1001.4793 [ ] , http://dx.doi.org/10.1088/0004-637x/715/2/1575 [ ] . j. o. burns , s. w. skillman , b. w. oshea , galaxy clusters at the edge : temperature , entropy , and gas dynamics near the virial radius , 721 ( 2010 ) 11051112 .
[ ] , http://dx.doi.org/10.1088/0004-637x/721/2/1105 [ ] .
m. j. turk , t. abel , b. oshea , the formation of population iii binaries from cosmological initial conditions , science 325 ( 2009 ) 601. http://arxiv.org/abs/0907.2919 [ ] , http://dx.doi.org/10.1126/science.1173540 [ ] .
kim , j. h. wise , m. a. alvarez , t. abel , galaxy formation with self - consistently modeled stars and massive black holes .
i. feedback - regulated star formation and black hole growth , 738 ( 2011 ) 54+ .
http://arxiv.org/abs/1106.4007 [ ] , http://dx.doi.org/10.1088/0004-637x/738/1/54 [ ] .
a. aykutalp , m. spaans , the complexity that the first stars brought to the universe : fragility of metal - enriched gas in a radiation field , 737 ( 2011 ) 63+ .
http://arxiv.org/abs/1105.5158 [ ] , http://dx.doi.org/10.1088/0004-637x/737/2/63 [ ] .
l. iapichino , w. schmidt , j. c. niemeyer , j. merklein , turbulence production and turbulent pressure support in the intergalactic medium , 414 ( 2011 ) 22972308 .
[ ] , http://dx.doi.org/10.1111/j.1365-2966.2011.18550.x [ ] .
a. t. myers , m. r. krumholz , r. i. klein , c. f. mckee , metallicity and the universality of the initial mass function , 735 ( 2011 ) 49+ . http://arxiv.org/abs/1102.2023 [ ] , http://dx.doi.org/10.1088/0004-637x/735/1/49 [ ] .
d. c. collins , p. padoan , m. l. norman , h. xu , mass and magnetic distributions in self - gravitating super - alfvnic turbulence with adaptive mesh refinement , 731 ( 2011 ) 59+ .
[ ] , http://dx.doi.org/10.1088/0004-637x/731/1/59 [ ] .
a. v. kravtsov , a. a. klypin , a. m. khokhlov , adaptive refinement tree : a new high - resolution n - body code for cosmological simulations , 111 ( 1997 ) 73+ .
http://arxiv.org/abs/arxiv:astro-ph/9701195 [ ] , http://dx.doi.org/10.1086/313015 [ ] .
m. r. krumholz , r. i. klein , c. f. mckee , j. bolstad , equations and algorithms for mixed - frame flux - limited diffusion radiation hydrodynamics , 667 ( 2007 ) 626643 . http://arxiv.org/abs/arxiv:astro-ph/0611003 [ ] , http://dx.doi.org/10.1086/520791 [ ] .
r. teyssier , cosmological hydrodynamics with adaptive mesh refinement .
a new high resolution code called ramses , 385 ( 2002 ) 337364 .
[ ] , http://dx.doi.org/10.1051/0004-6361:20011817 [ ] .
e. j. hallman , b. w. oshea , j. o. burns , m. l. norman , r. harkness , r. wagner , the santa fe light cone simulation project . i. confusion and the warm - hot intergalactic medium in upcoming sunyaev - zeldovich effect surveys , 671 ( 2007 ) 2739 .
http://arxiv.org/abs/0704.2607 [ ] , http://dx.doi.org/10.1086/522912 [ ] .
r. kaehler , j. wise , t. abel , h .- c .
hege , gpu - assisted raycasting for cosmological adaptive mesh refinement simulations , in : proceedings of the international workshop on volume graphics 2006 , eurographics / ieee vgtc 2006 , boston , 2006 , pp .
103110 .
k. m. gorski , e. hivon , a. j. banday , b. d. wandelt , f. k. hansen , m. reinecke , m. bartelman , http://www.citebase.org/abstract?id=oai:arxiv.org:astro-ph/0409513[healpix a framework for high resolution discretization , and fast analysis of data distributed on the sphere ] , the astrophysical journal 622 ( 2005 ) 759 .
http://arxiv.org/abs/astro-ph/0409513 b. d. smith , e. j. hallman , j. m. shull , b. w. oshea , the nature of the warm / hot intergalactic medium .
i. numerical methods , convergence , and o vi absorption , 731 ( 2011 ) 6+ .
http://arxiv.org/abs/1009.0261 [ ] , http://dx.doi.org/10.1088/0004-637x/731/1/6 [ ] .
a. mignone , g. bodo , s. massaglia , t. matsakos , o. tesileanu , c. zanni , a. ferrari , pluto : a numerical code for computational astrophysics , 170 ( 2007 ) 228242 .
[ ] , http://dx.doi.org/10.1086/513316 [ ] .
j. c. hayes , m. l. norman , r. a. fiedler , j. o. bordner , p. s. li , s. e. clark , a. ud - doula , m. mac low , simulating radiating and magnetized flows in multiple dimensions with zeus - mp , 165 ( 2006 ) 188228 .
[ ] , http://dx.doi.org/10.1086/504594 [ ] .
|
the usage of the high - level scripting language python has enabled new mechanisms for data interrogation , discovery and visualization of scientific data .
we present ` yt ` , an open source , community - developed astrophysical analysis and visualization toolkit for data generated by high - performance computing ( hpc ) simulations of astrophysical phenomena . through a separation of responsibilities in the underlying python code , ` yt ` allows data generated by incompatible , and sometimes even directly competing , astrophysical simulation platforms to be analyzed in a consistent manner , focusing on physically relevant quantities rather than quantities native to astrophysical simulation codes .
we present on its mechanisms for data access , capabilities for mpi - parallel analysis , and its implementation as an _ in situ _ analysis and visualization tool .
|
quantization is a phenomenon that changes our bright classical perspective into a bit uncertain and at first sight rather nonintuitive picture .
this picture , however , is more rigorous than the classical one , possesses many fascinating features and has produced a lot of successful predictions .
the subtle problem of transition from classical to quantal attracts attention from the early days of quantum mechanics . over the years
various techniques and methods for solving this puzzle have been invented .
our aim is not to trace back the complete ( hi)story of the milestone ideas in this field ( for the review we refer to @xcite ) .
what we want to do is to give a concise exposition of the method we have developed .
however , since we have generalized the original feynman s path integral approach , we will recapitulate this approach shortly in section [ 2 ] .
the main goal of our paper is to obtain a functional integral formula for the quantum propagator which would not refer to lagrangian and/or hamiltonian function .
quantum propagator is the probability amplitude @xmath3 for the transition of the system from the initial configuration @xmath4 to the final configuration @xmath5 .
we will derive a closed expression for this quantity starting from the given set of classical dynamical equations of motion .
the proposed method uses functional integration in the extended phase space , but instead of integration over path histories we introduce integration over stringy surfaces .
this crucial element of our approach is explained in full detail in sections [ 3 ] and [ 4 ] and in appendices [ a ] and [ b ] .
we also make sure that whenever the system under consideration becomes lagrangian , the stringy description reduces to the standard one with the feynman path integral . in sections [ 4.1 ] and [ 4.2 ]
some simple examples are scrutinized .
it is well known that there are classical dynamical systems that can not be described within the traditional lagrangian or hamiltonian framework .
the stringy approach enables us to quantize them straightforwardly . in section
[ 4.1 ] the quantization of a weakly non - lagrangian system is performed .
the transition amplitude is computed for a particle in a conservative field , whose motion is damped by a general @xmath0-power friction force @xmath6 .
the stringy results for @xmath7 are compared with the results obtained in the approaches by caldirola - kanai @xcite , bateman @xcite and kostin @xcite .
one can argue against the stringy quantization of dissipative systems that it is unable to describe decoherence phenomena .
this is true , but the same objection can be raised against the generally accepted heuristic approaches by the authors cited above , as well as those by dekker @xcite , razavy @xcite , geicke @xcite and others , simply because their kinematical and dynamical prerequisites are different from the prerequisites of the particle - plus - environment quantum models .
however , a possible argument for the stringy quantization is that the particle - plus - environment models are not able to handle satisfactorily the case with the friction force @xmath1 for the general power @xmath0 .
moreover , they describe a rather different phenomenon , namely the quantum brownian motion for which the total force equals @xmath8 . section [ 4.2 ] contains the analysis of a curious two - dimensional douglas system @xcite which is strongly non - lagrangian , i.e. not derivable from any sort of lagrangian or hamiltonian .
this causes a fundamental problem for all conventional quantization methods , but can be dealt with in a rather transparent way in the stringy approach . in section
[ 5 ] conclusion , discussion and outlook are collected .
the section also includes some comments on the relation between the stringy quantization and the caldeira - leggett model of quantum brownian motion @xcite , as well as the influence functional technique by feynman and vernon @xcite .
some rather technical material is left to appendices .
appendix [ a ] is devoted to the stringy variational principle , which plays an important role in the motivation of our approach . in appendix [ b ]
computational details concerning the surface functional integral for quantum friction force systems are presented .
historically the first attempt at quantization based on the dynamical equations of motion belongs to feynman , see @xcite .
a similar problem was considered by wigner , yang @xmath9 feldman , nelson , okubo and others , see @xcite . among the recent investigations in this field
let us mention the work of lyakhovich @xmath9 sharapov @xcite and gitman @xmath9 kupriyanov @xcite .
they consider the same problem as us , but their strategy is different . in our opinion , their approach fits much better the context of gauge field dynamics .
as ludwig faddeev noted during the edward witten s talk at the * * mathematical physics conference : from xx to xxi century * * , _ quantization is not a science , quantization is an art_. let us believe that the quantization method proposed here will fit the ludwig s dictum and will be meaningful enough to be considered artistic .
according to feynman @xcite , the probability amplitude of the transition of the system from the space - time configuration @xmath4 to another space - time configuration @xmath5 is @xmath10 \exp{\bigl\{\frac{i}{\hslash}\int\limits_{\tilde{\gamma}}\hspace{-1mm}pdq - hdt\bigr\}}\,.\ ] ] here the integral is taken over all paths ( histories ) @xmath11 in the extended phase space , , the canonically conjugated momenta @xmath12 and the time coordinate @xmath13 ( the index @xmath14 runs from 1 to the number of degrees of freedom @xmath15 ) .
one assumes the standard poisson brackets @xmath16 . in a more elevated language , the extended phase space is the space @xmath17 , where @xmath18 stands for the configuration space and @xmath19 for time . ]
satisfying the conditions @xmath20 and @xmath21 .
the preexponential factor @xmath22 in the expression ( [ feynmani ] ) serves just the normalization . to fix it
properly we impose two physical conditions on the transition amplitude .
first we introduce an integral condition that ensures that the total probability is conserved , [ normalization ] @xmath23 this specifies the absolute value of @xmath24 .
then we add a constraint that expresses the obvious fact that no evolution takes place if the final time @xmath25 approaches the initial time @xmath26 , @xmath27 this determines the phase of @xmath24 . a miraculous consequence of the definition of the transition amplitude ( [ feynmani ] ) ( not an additional requirement ! )
is that it satisfies the evolutionary chain rule , or chapman - kolmogorov equation , @xmath28 the infinitesimal version of this formula is the celebrated schrdinger equation .
quantum states of the system are described by the square integrable functions with the standard hilbert space structure .
physical observables are hermitian operators acting on such functions . given the state @xmath29 at the initial moment @xmath26 one is able to predict the state @xmath30 at any later moment @xmath31 according to the formula @xmath32 in what follows it is assumed that the above concept of states , observables and quantum evolution is valid for the stringy quantization of non - lagrangian systems as well . the only new element is a modified prescription for the evolutionary integral kernel @xmath3 .
the same kinematical prerequisites can be found also in other phenomenological approaches @xcite-@xcite .
a possible step beyond the theory summarized above consists in the elimination of the hamiltonian function @xmath33 from formula ( [ feynmani ] ) .
the price to be paid is the replacement of the path integration by the surface functional integration .
our aim is to construct the amplitude for the transition between @xmath4 and @xmath5 starting from the classical equations of motion ( and not from the hamiltonian function which provides them ) @xmath34 in the first set of equations , @xmath35 is the mass of the particle .
we restricted ourselves to the simplest case of one particle , although it is trivial to generalize the theory to a system with an arbitrary number of particles .
note also that if the particle is unconstrained and we make use of cartesian coordinates , the momenta @xmath36 defined in ( [ eq . ] ) reduce to @xmath12 .
suppose that there exists a unique classical trajectory in the extended phase space @xmath37 , connecting the points @xmath4 and @xmath5
. then we can assign to any other trajectory @xmath11 , which enters the path integral in ( [ feynmani ] ) , two auxiliary curves @xmath38 the curves are parameterized by the parameter @xmath39 $ ] and live in the momentum subsectors of the extended phase space with fixed @xmath4 and @xmath5 , see figure [ umbilic ] .
and @xmath40 which connect the classical history @xmath41 with the given history @xmath42 in the extended phase space .
@xmath43-curves are located in the @xmath15-dimensional subspaces of the extended phase space in which the momenta are varying while the coordinates and time are kept fixed .
the contour @xmath44 forms a boundary for plenty of extended phase space surfaces .
one of them , denoted as @xmath45 , is drawn in the figure.,width=321,height=422 ] using these definitions one can write and @xmath46 give trivially zero contributions to the contour integral , since on both @xmath43 s @xmath47 and @xmath48 . ]
@xmath49 where @xmath44 is a contour in the extended phase space consisting of four curves @xmath42 , @xmath41 , @xmath50 , @xmath40 .
the first integral on the right hand side is the classical action @xmath51 ( its equivalent for the non - lagrangian case will be specified later ) , while the second integral can be rewritten as @xmath52 where @xmath45 is a surface spanning the contour @xmath53 , i. e. a map from the parametric space @xmath54\times[0,1]$ ] to the extended phase space , @xmath55 satisfying @xmath56 the surface @xmath45 can be viewed as a worldsheet of a string , therefore we will call the quantization method using such surfaces `` stringy . ''
partial derivatives of the hamiltonian function entering the integral over @xmath45 can be eliminated with the help of the equations of motion ( [ eq . ] ) . by doing so ,
one transforms ( [ loop ] ) into the following form : @xmath57 where the two - form @xmath58 is defined as @xmath59 this two - form is an object in the extended phase space and its structure can be read out from the underlying equations of motion .
note that for non - potential forces the expression @xmath60 does not reduce to @xmath61 , so that the two - form @xmath58 is not closed .
this becomes essential in the next subsection .
it is obvious that for a given pair of histories @xmath62 there exist infinitely many @xmath45-surfaces such that @xmath63 .
all of them form a set which we will call @xmath64 . and
@xmath65 in the extended phase space whose projections @xmath66 and @xmath67 on the extended configuration space connect @xmath4 with @xmath5 .
then @xmath68 contains all continuous and oriented surfaces @xmath45 in the extended phase space such that @xmath69 if @xmath70 , the orientation of @xmath45 is supposed to be such that @xmath71 is a part of the boundary of @xmath45 .
the @xmath72ection is simply the momentum forgetting map , or to put it in a more elevated way , the canonical projection from the cotangent bundle to the base manifold .
a special case is the set @xmath73 , comprised of all closed surfaces in the extended phase space containing @xmath74 . in what follows
, we consider any history @xmath74 as a degenerated ( shrunk ) closed surface , hence @xmath75 by definition . ] since no @xmath45 is preferred and @xmath76 is only boundary dependent , it is natural to average the exponent of ( [ loop2 ] ) over the whole stringy set @xmath64 .
after doing so we obtain the identity @xmath77 \exp{\bigl\{\frac{i}{\hslash}\int\limits_{\sigma}\omega\bigr\}},\ ] ] where @xmath78 is the cardinality of the stringy set @xmath64 , @xmath79 , and @xmath80 $ ] is a functional integration measure specified in appendix [ b ] .
if no topology - related problems arise in the extended phase space , the infinite constant @xmath78 is independent of the history @xmath74 . taking all this into account we can rewrite ( [ feynmani ] ) as @xmath81 \exp{\bigl\{\frac{i}{\hslash}\int\limits_{\sigma}\omega\bigr\}},\hspace{2 mm } } \ ] ] where the set @xmath82 over which the functional integration is carried out contains all strings in the extended phase space which are anchored to the given classical trajectory @xmath83 .
some surfaces from @xmath82 are depicted in figure [ string ] . in ( [ feynmaniii ] ) , the undetermined constant @xmath84
was absorbed into the overall preexponential factor @xmath22 and the path integral over @xmath74 s was converted into the surface functional integral , as promised earlier , using the identities @xmath85\hspace{-3.5mm}\int\limits_{\ \ \ \mathscr{u}_{(\tilde{\gamma},\gamma_{cl } ) } } \hspace{-5mm}[\mathscr{d}\sigma]\cdots\ = \hspace{-5.5mm}\int\limits_{\ \ \ \bigcup_{\tilde{\gamma}}\mathscr{u}_{(\tilde{\gamma},\gamma_{cl } ) } } \hspace{-6.7mm}[\mathscr{d}\sigma]\cdots\ = \hspace{-1.6mm}\int\limits_{\mathscr{u}_{\gamma_{cl}}}\hspace{-1.0mm}[\mathscr{d}\sigma]\cdots\ \,.\ ] ] ) of @xmath82 , anchored to the given classical history @xmath83 in the extended phase space .
the two open surfaces belong to two different stringy classes @xmath64 , while the closed and the shrunk surface are both elements of the same stringy class @xmath86 .
as the worldsheet @xmath45 varies , the front and rear boundary curves , denoted in the text @xmath87 and @xmath46 , vary as well.,width=321,height=422 ]
in case the hamiltonian function @xmath33 is given , the formula ( [ feynmaniii ] ) for the propagator is equivalent to the feynman formula ( [ feynmani ] ) we have started with .
the surface functional integral ( [ feynmaniii ] ) is however completely free of @xmath33 and requires just the knowledge of the classical equations of motion .
this observation allows us to postulate ( [ feynmaniii ] ) as a quantization tool in situations in which one can not use the standard hamiltonian approach .
we just have to relax the requirement of the closedness of the two - form @xmath58 , following from the definition of @xmath58 for hamiltonian systems .
this relaxation is what is hidden behind the slightly provocative phrase _
one more step beyond _ in the title of this subsection .
the problematic part here is the definition of the classical action @xmath51 for non - lagrangian systems .
later we will see that in some specific situations this quantity can be read out from the structure of the surface integral @xmath88\exp{\bigl\{\frac{i } { \hslash}\int\limits_{\sigma}\omega\bigr\}}\,.\ ] ] another possibility which comes to mind is to use the integrating factor of the generally non - closed two - form @xmath58 .
this means that one will look for a function @xmath89 on the extended phase space such that @xmath90 .
if the structure of the dynamical equations allows for such a function ( i.e. the forces satisfy the helmholtz condition ) then we can define a local auxiliary hamiltonian @xmath91 such that @xmath92 . having @xmath91 we can define the auxiliary classical action @xmath93 . the alternative approach which tries to identify @xmath94 with @xmath93
has , however , several substantial disadvantages .
first of all , the procedure of finding @xmath95 ( and subsequently @xmath91 and @xmath93 ) is highly ambiguous .
furthermore , @xmath93 lacks symmetries that were originally present in the equations of motion . because of these findings we do not follow this strategy hereinafter . from the physical point of view the dynamical equations seem to be more fundamental than their compact but ambiguous precursors , hamiltonian and/or lagrangian function , see @xcite .
since ( [ feynmaniii ] ) requires just the knowledge of the equations of motion , it determines the transition amplitude in a completely new way .
our proposal for the transition amplitude has appeared here out of thin air . actually , we were rewriting ( [ feynmani ] ) in terms of stringy surfaces and then , when realizing that the integrated function can be written without any reference to the hamiltonian , we postulated formula ( [ feynmaniii ] ) to be valid in general . however , this is not the whole story .
the stringy functional quantization can be motivated also by the stringy variational principle .
having the dynamical equations and initial and final endpoints , one can form @xmath82 and @xmath58 . then using these objects
one can introduce the stringy action functional @xmath96 this is a variational problem with varying boundaries , therefore the total variation has two terms .
first one specifies the boundary and determines the initial equations of motion for @xmath83 .
the second one specifies the bulk of the stationary world - sheet @xmath45 ( which turns out to be shrunk into @xmath83 itself ) .
moreover , one immediately verifies that in the special case when @xmath97 , the stringy variational principle reduces ( up to an additive constant ) to the celebrated hamilton least action principle .
of course , many subtleties were omitted here , but all of them can be found in @xcite , or in appendix [ a ] .
thus , our variational principle enables us to perform the limit @xmath98 , in which we recover the original classical dynamics as required .
to examine the functionality of the proposed quantization method let us first analyze the simplest friction force system .
it consists of a particle with unit mass , moving in one dimension under the action of the conservative force @xmath99 and the friction force @xmath100 .. the universally applicable expression is @xmath1010.5 mm sgn@xmath102 . ] thus , @xmath103 in this example the surface functional integral can be calculated explicitly ( for more detail see @xcite , or appendix [ b ] ) . in the course of calculation , the surface functional integral in the extended phase space reduces to path integral in the configuration space , @xmath104\exp{\bigl\{\frac{i}{\hslash } \int\limits_{\sigma}\omega\bigr\}}\propto\nonumber\label{dissipative}\\ & & \propto\exp{\bigl\{-\frac{i}{\hslash}\int\limits_{t_0}^{t_1}\bigl(\tfrac{1}{2}\dot{q}_{cl}^2-u(q_{cl})- \kappa q_{cl}p^{a}_{cl}\bigr)dt\bigr\}}\times\nonumber\\ & & \times\int[\mathscr{d}q]\exp{\bigl\{\frac{i}{\hslash}\int\limits_{t_0}^{t_1}\bigl(\tfrac{1}{2}\dot{q}^2-u(q)- \kappa qp^{a}_{cl}\bigr)dt\bigr\}}\,.\end{aligned}\ ] ] this combined with the formula ( [ feynmaniii ] ) suggests that it will be convenient to define the classical action as in the subscript @xmath105 , as well as in the subscript @xmath106 introduced later , refers to the power in the expression for the friction force . ]
@xmath107 when doing so , we obtain a controlled cancelation of @xmath108 with the preexponential factor arising from ( [ disi ] ) .
the final probability amplitude then assumes a compact and reasonable form : @xmath109\exp{\bigl\{\frac{i}{\hslash } \hspace{-0.5mm}\int\limits_{t_0}^{t_1}\hspace{-0.5mm}\bigl(\tfrac{1}{2}\dot{q}^2-\hspace{-0.2 mm } u(q)-\hspace{-0.2mm}\kappa qp^{a}_{cl}\bigr)dt\bigr\}}\,.\end{aligned}\ ] ] the preexponential factor @xmath110 can in principle be obtained by subjecting @xmath111 to the conditions ( [ normalization ] ) .
let us explain why one should consider the propagator formula ( [ propagator ] ) reasonable
. in the path integral above there appears an unconventional `` external source '' term @xmath112 .
its appearance guarantees that the quantum dynamics governed by @xmath111 transforms into classical mechanics in the limit @xmath113 .
this follows from the simple fact that the unique solution of the saddle point equation for the path integral ( [ propagator ] ) , @xmath114 which satisfies the given initial and final conditions @xmath115 and @xmath116 , is the classical trajectory @xmath117 .
the external source term in ( [ propagator ] ) breaks the validity of the chapman - kolmogorov ( memoryless ) equation . )
can be viewed as a sort of additional time - dependent potential , therefore one would expect that the chapman - kolmogorov equation ( [ ch - k ] ) will be satisfied . however
, this argument is not applicable in the present situation .
the point is that if we merge two paths _ kl _ and _ lm _ into a new path _ km _ , then the time - dependent potentials @xmath118 and @xmath119 differ , for a general position of @xmath120 , from the potential @xmath121 in the corresponding time ( sub)intervals , @xmath122 \atop -\kappa qp^a_{cl - lm}(t)\ \mbox{for}\ t\in[t_l , t_m]}\ ] ] since this differs substantially from the standard case , the argument about the validity of the chapman - kolmogorov equation can not be used . ] from the physical point of view it is a desired phenomenon .
the microscopic origin of friction is some environmental interaction .
this , however , was not accounted for here explicitly .
what has been considered is some effective ( macroscopic , phenomenological ) interaction emerging on the classical level only .
microscopically the system is a part of a larger system and hence it should be affected by the memory effect .
it is not difficult to compute @xmath111 for the general power @xmath0 and the potentials @xmath123 and @xmath124 . in both cases we can carry out the path integration in ( [ propagator ] ) explicitly to obtain @xmath125 where the non - lagrangian action @xmath126 is given by the expression ( [ a - action ] ) .
an open problem is to determine the preexponential factor @xmath127 for all powers @xmath0 except for @xmath2 ( at least the author is incapable to do that ) .
if @xmath128 , the preexponential factor is apparently dependent not only on the time difference @xmath129 , but also on the endpoint positions @xmath130 and @xmath131
. this hypothesis is supported by the results obtained in @xcite , where the path integral for a non - conservative force quadratic in velocity is calculated . the normalization prefactor which appears there is explicitly dependent on the endpoints positions as well as on @xmath129 . in what follows
we will restrict ourself to the case when the power @xmath0 is equal to 1 .
this simplified setup enables us to compare the stringy approach with the caldirola - kanai , bateman and kostin approaches .
the dissipative system under consideration is weakly non - lagrangian .
this means that @xmath132 , but @xmath58 possesses a local integrator @xmath133 . let us define the auxiliary lagrangian function ( one of plenty ) as @xmath134.\label{l - ck } % a&=2 & l_{hnt}(q,\dot{q})&=e^{2\kappa q}\,\frac{\dot{q}^2}{2}-\int^q\hspace{-2mm}d\tilde{q}\ , % e^{2\kappa\tilde{q}}\,\frac{du(\tilde{q})}{d\tilde{q}},\label{l - hnt}\end{aligned}\ ] ] this lagrangian is usually called ck - lagrangian after its inventors p. caldirola and e. kanai @xcite .
henceforward a damped free particle is considered only , i.e. the potential energy @xmath135 is supposed to be zero .
it is immediately clear that the transition amplitude computed from ( [ propagator1 ] ) differs from the ck - amplitude @xmath136\exp{\bigl\{\frac{i}{\hslash}\int\limits_{t_0}^{t_1}\hspace{-1mm}l_{ck}(q , t)dt\bigr\}}\\ & = & \hspace{-1mm}\dfrac{1}{\mathbf{n}_{ck}}\exp\{\tfrac{i}{\hslash}\,\mathbf{s}_{ck}(q_1,t_1|q_0,t_0)\ } \,.\end{aligned}\ ] ] indeed , the stringy and ck actions which enter the corresponding transition amplitudes are different , @xmath137 since both actions are quadratic functions of the endpoints @xmath130 and @xmath131 , and since we require that the total probability is conserved , the preexponential factors for both transition amplitudes can be computed from the van vleck formula @xmath138using this formula we arrive at the normalized transition amplitudes @xmath139 which trivially satisfy ( [ normalization ] ) .
moreover , one immediately verifies that in the frictionless limit @xmath140 both transition amplitudes coincide with the free particle propagator .
a short inspection shows that the stringy propagator depends on @xmath26 and @xmath31 only through @xmath129 , but this is not the case for the ck propagator .
the same observation holds for @xmath141 and @xmath142 .
this is a typical feature for all auxiliary actions defined in terms of integrators of @xmath58 , see @xcite .
we pay attention to this point because the classical equation of motion @xmath143 which we have started with is invariant with respect to time translations , and one would naturally expect the same invariance on the quantum level .
since only @xmath144 has this feature , we obtain an efficient argument for the stringy functional integral ( [ feynmaniii ] ) when compared to the ck one .
the evolution of gaussian wave packets @xmath145 whose dynamics is governed by ( [ damp.a ] ) and ( [ damp.b ] ) , is visualized in figure [ ck - characteristics ] . for more detail , see @xcite . ) and ( [ damp.b ] ) , respectively .
initial wave packet characteristics are @xmath146 and @xmath147 , and the friction constant @xmath148 is set to @xmath149 .
one realizes that at the time @xmath150 the stringy mean momentum becomes zero .
classically , for @xmath151 the physical relevancy of the solution breaks down .
this explains the time bound on the applicability of stringy propagator ( [ damp.a]).,width=321,height=422 ] from the figure we can see that the stringy averaged momentum @xmath152 becomes negative ( i.e. meaningless ) for @xmath153 . let us try to explain this peculiarity .
it is well known that the relevancy of the classical solution @xmath117 of @xmath143 breaks down when the time @xmath129 exceeds the relaxation time @xmath154 .
since @xmath155 is used in the derivation of the formula ( [ damp.a ] ) for the stringy propagator , its applicability is automatically restricted as well . after this restriction
is taken into account , the stringy evolution of @xmath152 becomes more acceptable from the physical point of view then the evolution in the ck approach .
the key element of the bateman(-morse - feshbach ) approach @xcite are new subsidiary degrees of freedom introduced in addition to the initial ones , which are amplified rather than damped ; thus , they evolve according to the time reversed dynamical equations . in our case
we have : @xmath156 these equations of motion can be derived from the least action principle with the quadratic and time independent bateman(-morse - feshbach ) lagrangian : @xmath157 the canonical as well as path integral quantization of the theory encounters various difficulties because non - normalizable states of the amplified system must be employed @xcite .
this problem will be discussed later , when the treatment of the auxiliary @xmath158-degrees of freedom in this approach will be described .
the path integral evaluation of the transition probability amplitude based on @xmath159 is straightforward @xcite .
the result is : @xmath160\bigr\}\,,\end{aligned}\ ] ] where @xmath161 our aim , however , is to find the transition amplitude for the damped ( sub)system only . in order to obtain it we must project out the nonphysical @xmath158-degrees of freedom .
when using the standard formula @xmath162 we do not reproduce the free particle propagator for @xmath163 as desired . to overcome this trouble the following `` repairing prescription ''
is introduced ( for more detail see @xcite ) : the amplitude for the damped ( sub)system to pass from @xmath164 to @xmath165 within the time @xmath166 is equal to the amplitude for the whole system to pass from the nonphysical state @xmath167 to the nonphysical state @xmath168 , @xmath169 where the non - normalizible @xmath158-system states @xmath170 are chosen as @xmath171 substituting ( [ bateman propagator ] ) into ( [ remedy ] ) we obtain the following effective propagator for the damped ( sub)system : @xmath172\bigr\}\,.\end{aligned}\ ] ] the form of the auxiliary states @xmath173 and @xmath174 guarantees that the effective bateman propagator for the damped ( sub)system satisfies the normalization condition ( [ normalization ] ) . ) and ( [ damp.c ] ) , respectively . the initial wave packet characteristics and the meaning of solid and dashed color lines are the same as in figure [ ck - characteristics].,width=321,height=422 ] the bateman effective propagator ( [ damp.c ] ) depends only on the time difference @xmath166 , which favors it in comparison with the ck - propagator ( [ damp.b ] ) .
on the other hand , the artificial `` repairing procedure '' involving nonphysical states discredits the bateman approach with respect to the stringy one .
the gaussian wave packet characteristics obtained from ( [ damp.a ] ) and ( [ damp.c ] ) are drawn in figure [ bateman - characteristics ] . as seen from the figure
, the stringy quantization possesses again better qualitative features than the approach to which we have compared it .
a possible incorporation of the classical dynamics into the quantum one can be obtained by using heisenberg equations for the position and momentum operators : @xmath175 where @xmath176 and @xmath177 obey @xmath178=i\hslash$ ] for all @xmath13 . in the special case when @xmath179 , kostin found an equivalent description of the system by the schrdinger equation @xcite : @xmath180\bigr]\psi(q , t)\,,\ ] ] where @xmath181 $ ]
is the @xmath182-dependent kostin potential defined as @xmath183=i\frac{\kappa}{2}\bigl[\ln{\frac{\overline{\psi}(q , t)}{\psi(q , t)}}- \int\hspace{-1.0 mm } dq'\,\overline{\psi}(q',t)\ln{\frac{\overline{\psi}(q',t)}{\psi(q',t)}}\psi(q',t)\bigr].\ ] ] this is known as kostin - schrdinger(-langevin ) equation . in the kostin s paper a complete set of solutions of this equation in the special case of free particle ( @xmath123 ) is presented .
the set consists of the states @xmath184 labeled by the continuous quantum number @xmath185 , the initial momentum of the particle .
the state @xmath184 starts as a momentum eigenstate with the momentum @xmath185 , and remains the momentum eigenstate also later , but with the decreasing momentum , @xmath186 from the ensemble of non - localized kostin s states one would like to form normalized wave packet solutions , localized in configuration as well as momentum space .
this is , however , impossible since the kostin - schrdinger equation is nonlinear ; thus , the kostin approach belongs to nonlinear quantum mechanics and lacks superposition principle . in figure [ kostin - characteristics ] we plot the expectation value of the momentum for the kostin state @xmath187 and the stringy evolved state @xmath188 for various values of @xmath148 .
is not a square integrable function , we evolved by @xmath189 a modified ( normalized ) state @xmath190 ( @xmath191 ) .
after the stringy evolution of @xmath192 , the expectation value of the momentum was computed and the regularization parameter @xmath193 was set to zero . ] as @xmath148 tends to zero , the stringy and kostin results match each other . , drawn for the stringy propagator ( [ damp.a ] ) ( blue dashed lines ) and the propagator given by the kostin - schrdinger equation ( [ kostin ] ) ( red dashed lines ) , respectively .
the values of the friction constant @xmath148 are indicated next to the curves .
the initial momentum is @xmath194.,width=321,height=422 ]
after we have dealt with the weak non - lagrangeanity , let us consider the simplest strongly non - lagrangian system .
it was proposed by jesse douglas ( one of the two winners of the first fields medals awarded in 1936 ) when studying the inverse problem of variational calculus @xcite .
the system is governed by the following dimensionless dynamical equations : @xmath195 a quick calculation shows that the associated two - form @xmath196 does not possess a non - trivial local integrator @xmath95 .
consequently , equations ( [ eqs . ] ) can not be obtained as euler - lagrange ( hamilton ) equations . from the standard point of view , this curious situation is stalemate : _ no lagrangian @xmath197 no quantum mechanics_. however , the surface integral method offers a way to overcome this deadlock . using the stringy functional integration in the extended phase space
we get @xmath198 \exp{\bigl\{\frac{i}{\hslash}\int\limits_{\sigma}\omega\bigr\}}\propto\nonumber\label{douglas}\\ & & \hspace{-1cm}\propto \exp{\bigl\{-\frac{i}{\hslash}\int\limits_{t_0}^{t_1 } \bigl(\tfrac{1}{2}\dot{x}_{cl}^2+\tfrac{1}{2}\dot{y}_{cl}^2-\tfrac{1}{2}y_{cl}^2-x_{cl}p_{ycl } \bigr)dt\bigr\}}\hspace{-1mm}\times\nonumber\\ & & \hspace{-1cm}\times\hspace{-1mm}\int[\mathscr{d}x\,\mathscr{d}y]\exp{\bigl\{\frac{i}{\hslash } \int\limits_{t_0}^{t_1}\hspace{-1mm}\bigl(\tfrac{1}{2}\dot{x}^2+\tfrac{1}{2}\dot{y}^2-\tfrac{1}{2}y^2-xp_{ycl } \bigr)dt\bigr\}}.\end{aligned}\ ] ] here , obviously , @xmath199 and @xmath200 stand for the classical solutions of ( [ eqs . ] ) matching the initial and final endpoints for which the quantum transition probability amplitude is sought for , @xmath201 , @xmath202 and @xmath203 , @xmath204 . in the classical limit @xmath98
we want the stringy propagator ( [ feynmaniii ] ) to maintain the properties of the original dynamical system .
this guides us to define : @xmath205 the exponential @xmath206 of the classical action defined in such a way cancels the prefactor in ( [ douglas ] )
. finally we arrive at the following transition amplitude for the douglas system : @xmath207 \exp{\bigl\{\frac{i}{\hslash}\hspace{-1mm}\int\limits_{t_0}^{t_1}\bigl(\tfrac{1}{2}\dot{x}^2+\tfrac{1}{2}\dot{y}^2 -\tfrac{1}{2}y^2-xp_{ycl}\bigr)dt\bigr\}}.\end{aligned}\ ] ] the path integral in the configuration space we have constructed is quadratic and hence can be computed explicitly .
the normalized transition amplitude is @xmath208 where @xmath209\\ & -&\frac{(t_1-t_0)}{4\sin^2{(t_1-t_0)}}\bigl[(y_1 ^ 2+y_0 ^ 2)-2y_1y_0\cos{(t_1-t_0)}\bigr]\,.\end{aligned}\ ] ]
in the paper we have developed a new quantization method that generalizes the conventional path integral approach . throughout the paper we considered only the nonrelativistic quantum mechanics of spinless systems .
however , the generalization to the field theory is rather straightforward .
we have just to pass from the space of particle positions to the space of field configurations .
the mathematical language of our exposition respects the vladimir arnod `` principle of minimal generality . ''
formulas are mostly written down in one ( local ) chart . from this , however , one can ascend to a global , coordinate free description employing bundles and jet prolongations . in order that we did not cloud up the main idea of the paper
, we did not follow such fluffy approach
. however , in the future it could be useful to analyze obstructions to the implementation of our method which arise from the global properties of the underlying space - time geometry .
special attention was paid to the stringy quantization of dissipative systems and of the douglas system .
the latter can not be quantized by any known quantization technique while the former can .
in fact , dissipative ( friction force ) systems were studied extensively in the past and several approaches to their quantization were proposed .
the stringy quantization was compared with three of them , caldirola - kanai , bateman and kostin , and it was shown that it either gives better results ( in the first two cases ) or is simpler to apply ( in the third case ) . as mentioned in the introduction , the quantization proposed here shares a common property with the well established approaches of caldirola - kanai , bateman , kostin and others : they all fail to cover decoherence phenomena . on the other hand , as we have seen ,
the stringy quantization produces a quantum propagator that does not respect the memoryless chapman - kolmogorov equation .
we believe that this essential feature of the theory is measuring / reflecting the non - lagrangeanity of the system on the quantum level . from the author s point of view one can rise a conceptual objection also against the particle - plus - environment model by caldeira and leggett @xcite .
their approach is microscopical except the way in which the spectral density @xmath210 for the reservoir degrees of freedom is introduced .
this density is not obtained from any kind of microscopical theory and the form it assumes is motivated solely by the necessity to obtain the dissipative term @xmath211 in the effective theory . on the top of it ,
the interaction hamiltonian is such that it produces , after integrating out the reservoir degrees of freedom , an additional langevin stochastic force @xmath212 .
thus , the effective classical motion of the particle is governed by the brownian equation of motion @xmath213 , which is conceptually different from the physical situation considered in section [ 4.1 ] .
it is also worth to point out that the caldeira - leggett model can not describe satisfactorily the friction force proportional to @xmath214 with @xmath128 .
there is no doubt that for @xmath7 the model gives better results than all effective theories mentioned before including the stringy one , but it is fair to say that it describes a different physical phenomenon , namely the quantum brownian motion of a particle which is in thermal equilibrium with a reservoir .
the goal of our paper was to write down a functional integral formula for the quantum transition amplitude in terms of the underlying classical equation of motion .
however , we did not eliminate the artificial notions of pure states and classical action for non - lagrangian systems . in the surface
functional integral ( [ feynmaniii ] ) , these notions were implicitly present . a possible alternative to our approach consists in expressing the transition probability ( not the probability amplitude ! ) using the functional integral .
the probability that the system evolves from the mixed state @xmath215 at the time @xmath26 to the mixed state @xmath216 at some later time @xmath31 is @xmath217[\mathscr{d}\tilde{\gamma}^\prime]\exp\bigl\{\frac{i}{\hslash } \bigl(\int\limits_{\tilde{\gamma}}-\int\limits_{\tilde{\gamma}^\prime}\bigr)(pdq - hdt)\bigr\}\,.\end{aligned}\ ] ] here @xmath74 and @xmath218 are curves in the extended phase space whose @xmath219-projections connect @xmath130 with @xmath131 and @xmath220 with @xmath221 respectively .
using the stokes theorem one is able to convert the difference of the line integrals of the one - form @xmath222 into the surface integral of the two - form @xmath223 .
explicitly , @xmath224 where @xmath45 represents again a map from the parametric space @xmath54\times[0,1]$ ] to the extended phase space , @xmath225 such that @xmath226 . the sideways @xmath43-boundary curves .
however , there is one substantial difference between the two cases . in the present case
, we allow the @xmath43-curves to vary in both coordinates @xmath219 and @xmath227 , only the time @xmath13 must stay unchanged along them . in the case considered in section [ 3 ] we were more stringent : the @xmath43-curves were allowed to vary only with respect to @xmath227 and had to stay unchanged with respect to both @xmath219 and @xmath13 . ] of @xmath45 @xmath228 live in the instant phase spaces of @xmath229 $ ] , i.e. in the submanifolds @xmath230 $ ] and @xmath231 $ ] .
moreover , as the intrinsic parameter @xmath232 varies from @xmath233 to @xmath234 , the @xmath235-components of @xmath50 and @xmath40 vary from @xmath130 to @xmath220 and from @xmath131 to @xmath221 respectively .
let us denote by @xmath236 the space of all @xmath45-maps for which @xmath237 , \atop \lambda_1(s)\subset\mathbb{r}^{2n}[q , p , t_1].}\ ] ] since the right hand side of ( [ loop3 ] ) depends only on @xmath45 and its two boundaries , the double path integral @xmath238[\mathscr{d}\tilde{\gamma}^\prime]\cdots$ ] entering @xmath239 can be rewritten as a surface functional integral @xmath240\cdots$ ] , @xmath241\exp\bigl\{-\frac{i}{\hslash}\bigl [ \int\limits_\sigma\omega+ \bigl(\int\limits_{\lambda_1}\hspace{-1mm}-\hspace{-1mm}\int\limits_{\lambda_0}\bigr ) pdq\bigr]\bigr\}\,.\end{aligned}\ ] ] the formula we have just arrived at was derived under the assumption that @xmath223 .
however , it is clear that the expression in the exponent is free of any reference to the hamiltonian and requires just the classical equations of motion .
therefore it seems reasonable to postulate the probability formula ( [ feynman - vernon ] ) also for non - lagrangian systems .
the approach we shortly presented here does not use the notion of pure states of a non - lagrangian system .
it resembles the feynman - vernon approach @xcite , in which one introduces the influence functional in the presence of dissipative forces .
our probability formula ( [ feynman - vernon ] ) uses the surface functional integral in the extended phase space , while in the feynman - vernon approach one just has to calculate a double path integral in the configuration space .
there is a chance that after a certain discretization we will be able to convert the surface functional integral into the double path integral in the configuration space ( see appendix [ b ] ) , and as a result , we will find the explicit form of the influence functional .
work on this topic is in progress .
this research was supported by m sr cern - fellowship program and vega grant 1/1008/09 .
special thanks go to vladimr balek for his interest and fruitful discussions we have had in the course of the work .
+ + @xmath242 @xmath243 @xmath244 @xmath245
in section [ 3 ] we shortly outlined how the transition amplitude ( [ feynmaniii ] ) can be related to the stringy variational principle , with the action @xmath246 defined in ( [ action ] ) . in what follows
we will perform a complete variation of @xmath246 , with the domain extended from @xmath82 to a wider class @xmath247 .
the stringy set @xmath236 contains all surfaces in the extended phase space trapped between the submanifolds with fixed @xmath4 and @xmath5 ( including separate histories , regarded as shrunk surfaces ) . in the course of variation the classical dynamics ( [ eq . ] ) we have started from will be recovered .
suppose we have an extremal surface @xmath248 and a variational vector field @xmath249 defined in its neighborhood such that the flow of @xmath249 preserves @xmath236 . by definition ,
@xmath249 moves @xmath250 to some other worldsheet @xmath251 , where @xmath252 is an infinitesimal increment of the parameter of the flow generated by @xmath249 .
the situation is depicted schematically in figure [ flow ] .
acting infinitesimally in the extended phase space . as a result ,
the initial surface @xmath250 moves to some nearby surface @xmath251 , and its boundary @xmath253 is transported to a new boundary @xmath254 . since the variational vector field is assumed to preserve @xmath236
, @xmath249 stays tangential to the momentum submanifolds with fixed @xmath4 and @xmath5.,width=321,height=422 ] the extremality of @xmath255 means that for all variational @xmath249-fields it holds @xmath256 where @xmath257 stands for the lie derivative . let us compute the surface integral on the left hand side : @xmath258 where the the symbol @xmath259 denotes the inner product ( contraction ) of a vector with a differential form .
the first term on the right hand side depends only the values of @xmath58 on the boundary of @xmath255 and can be recast into the form @xmath260 the last two terms here give individually zero contributions , as can be seen from the following quick consideration : the vector field @xmath249 is assumed to preserve @xmath236 and therefore its restrictions to the @xmath43-boundaries are @xmath261 the one - form we are integrating is @xmath262 , but since both @xmath219 and @xmath13 stay unchanged on @xmath43 , the integral is zero . as a result , to annihilate the boundary term for all variational fields
@xmath249 we are forced to chose the curves @xmath263 and @xmath264 in the extended phase space in such a way that their instant tangent vectors annihilate @xmath58 at any moment @xmath265 $ ] .
thus , it must hold [ boundary ] @xmath266 or , equivalently , @xmath267 for both @xmath268 and @xmath269 . here
one recognizes the dynamical equations ( [ eq . ] ) as desired . from the boundary term analysis and the assumption about the uniqueness of the history @xmath83
which we adopted from the very beginning we can conclude that @xmath270 .
hence @xmath255 has the topology of a closed string attached to the classical trajectory @xmath83 . to specify its shape the second term in ( [ variation ] )
must be employed .
the resulting variational equation , written in a coordinate - free notation , is [ bulk ] @xmath271 this is equivalent to the following system of partial differential equations for the unknown functions @xmath272 and @xmath273 ( indices @xmath14 and @xmath274 run from @xmath275 ) : @xmath276 one solution of these equations , satisfying all boundary conditions , is trivial .
it is the shrunk surface @xmath277 . after recalling the assumption of uniqueness of @xmath83 once more
, we can see that this is the _ only _ solution of ( [ bulk ] ) .
if there existed a closed unshrunk extremal surface @xmath255 , the initial set of equations of motion ( [ eq . ] ) would have at least one one - parametrical family of classical solutions between the given pair of endpoints .
the variational principle we described above operates on a wider stringy class @xmath236 than it is in fact necessary .
the surface integral formula ( [ feynmaniii ] ) requires just the restricted subset @xmath278 .
the transition to @xmath82 is advisable for two reasons .
first , in the lagrangian case with @xmath223 we obtain an equivalence ( modulo additive constant ) between the least action principle using @xmath246 and the standard hamilton least action principle .
explicitly , @xmath279 second , the wider class @xmath236 contains plenty of degenerate ( shrunk ) surfaces , the histories @xmath74 .
these are obviously stationary surfaces of @xmath246 when being varied within @xmath236 , since @xmath280 for any variational vector field @xmath249 .
however , only one of these histories , namely @xmath83 , satisfies both equations ( [ boundary ] ) and ( [ bulk ] ) at the same time .
this fictitious problem is avoided when one works from the beginning with the stringy subclass @xmath278 .
let us explain here in some detail how the surface functional integral is computed . to be as tangible as possible consider a one - dimensional system only , so that the extended phase space will be the three dimensional space @xmath281 $ ] ( more dimensions represent only a technical problem ) .
the particle is supposed to move under the combined action of the potential @xmath135 and the friction force with a general @xmath0-power law .
the dynamical equations are @xmath282 and in addition to them , we require that the particle satisfies the boundary conditions @xmath115 and @xmath116 . according to the definition ( [ omega ] ) , the two - form @xmath58 is @xmath283 our aim is to compute the surface functional integral ( [ feynmaniii ] ) over the stringy set @xmath284 . the direct application of ( [ surface ] ) and ( [ alpha ] ) together with the stokes theorem yields @xmath285\exp{\bigl\{\frac{i}{\hslash } \int\limits_{\sigma}\omega\bigr\}}=\label{giuseppe}\\ & & \hspace{-2mm}=\int[\mathscr{d}\tilde{\gamma}]\exp{\bigl\{\frac{i}{\hslash}\bigl(\,\int\limits_{\tilde{\gamma}}- \int\limits_{\gamma_{cl}}\,\bigr)\bigl(pdq-\frac{1}{2}p^2dt - u(q)dt\bigr)\bigr\}}\times\nonumber\\ & & \hspace{-2mm}\times\hspace{-4mm}\int\limits_{\ \ \
\mathscr{u}_{(\tilde{\gamma},\gamma_{cl})}}\hspace{-5 mm } [ \mathscr{d}\sigma]\exp{\bigl\{-\frac{i}{\hslash}\kappa\int\limits_{\sigma}p^{a}dq\wedge dt\bigr\}}\,.\nonumber\end{aligned}\ ] ] the nontrivial part of this is the functional integral over the stringy subset @xmath64 . as mentioned earlier
, @xmath286 is a map @xmath287\times[0,1]\rightarrow\mathbb{r}^3[q , p , t]\,,\ \ \
( t , s)\mapsto\sigma(t , s)\,,\ ] ] such that for @xmath288 $ ] and @xmath289 $ ] there holds @xmath290 \atop \sigma(t_1,s)\in\mathbb{r}^3[q_1,p , t_1]}\,.\ ] ] to proceed further with the functional integral in question , introduce a set of regularly distributed nodal points in the parameter space , @xmath291\times[0,1]\}\,.\ ] ] the points of the set are labeled by two discrete indices , the time index @xmath292 and the space index @xmath293 , see figure [ grid ] . in this way we obtain for any @xmath45-map from @xmath64
a discretized @xmath294-tuple @xmath295 the boundary values are required to be consistent with ( [ constraints ] ) .
explicitly , for all indices @xmath296 and @xmath297 it must hold @xmath298 the discretization of @xmath45 enables us to approximate the integral of @xmath299 as follows : for all discretized integrals . in the continuum limit
@xmath300 and @xmath301 approach independently infinity and @xmath302 .
however , we must keep in mind that we must perform simultaneously the limits @xmath303 and @xmath304 so that the quantities @xmath305 and @xmath306 stay finite . ]
@xmath307 \times[0,1]$ ] of a surface @xmath286 .
each elementary tile encloses the area @xmath308 ( at the end , the numbers @xmath300 and @xmath301 will be sent to infinity ) .
the points marked by crosses are constrained by the conditions ( [ constraints]).,width=321,height=422 ] formally , the discretized functional integral over all stringy configurations from @xmath64 is a multiple integral over all unconstrained variables @xmath309 which are needed to specify @xmath45 .
the only problematic part , as usual , is the choice of an appropriate integration measure . for the reason that will become clear in a moment , we choose the measure as @xmath310:=\hspace{-2 mm } \int\limits_{-\infty}^{+\infty}\ \prod\limits_{\tau=1}^{k-1}\prod\limits_{\sigma=1}^{l-1}\,\bigl(\frac{a\kappa\vardelta}{2\pi i\hslash}\,p^{a-1}_{(\tau,\sigma)}\bigr)\ , dq_{(\tau,\sigma)}\,dp_{(\tau,\sigma)}.\hspace{-5mm}\ ] ] the first step ,
when dealing with the discretized functional integral @xmath311 \exp{\bigl\{-\frac{i}{\hslash}\kappa\fint\limits_{\sigma}p^{a}dq\wedge dt\bigr\}}\,,\ ] ] consists in the integration over the internal stringy positions @xmath312 . after this integration
, the integrand transforms into a chain of delta functions , @xmath313\bigr\}}\,.\end{aligned}\ ] ] the next step is the integration over the internal stringy momenta @xmath314 . after performing this trivial integration
we arrive at the expression @xmath315\bigr\}}\,.\ ] ] in the exponent there appears the discretized version of the integral @xmath316 where @xmath317 stands for the @xmath219-projection of @xmath42 .
consequently , after returning back to the continuum limit we obtain the following important result : @xmath318 \exp{\bigl\{-\frac{i}{\hslash}\kappa\int\limits_{\sigma}p^{a}dq\wedge dt\bigr\}}=\nonumber\\ & & \hspace{-5mm}=\exp{\bigl\{-\frac{i}{\hslash}\kappa \int\limits_{t_0}^{t_1}\,dt\bigl[\tilde{q}(t)\,p^a_{cl}(t)-q_{cl}(t)\,p^a_{cl}(t ) \bigr]\bigr\}}\,.\end{aligned}\ ] ] after substituting ( [ int . ] ) into the initial formula ( [ giuseppe ] ) we recover a path integral in the extended phase space .
the integral is quadratic in momenta , with the discretized ( standard ) liouville measure @xmath319\ \cdots = \int\limits_{-\infty}^{+\infty}\frac{d\tilde{p}_k}{2\pi\hslash}\,\prod\limits_{\tau=1}^{k-1 } \frac{d\tilde{p}_{\tau}\,d\tilde{q}_{\tau}}{2\pi\hslash}\ \cdots\,,\ ] ] the integration over momenta can be carried out explicitly and we finally obtain a path integral in the configuration space only ( tildes are removed from the position variables ) , @xmath320 \exp{\bigl\{\frac{i}{\hslash}\int\limits_{\sigma}\omega\bigr\}}=\nonumber\\ & & \hspace{-5mm}=\exp{\bigl\{-\frac{i}{\hslash}\int\limits_{t_0}^{t_1}\bigl(\frac{1}{2 } \dot{q}_{cl}^2-u(q_{cl})-\kappa q_{cl}\,p^a_{cl}\bigr)dt\bigr\}}\times\nonumber\\ & & \hspace{-5mm}\times\int[\mathscr{d}q]\exp{\bigl\{\frac{i}{\hslash}\int\limits_{t_0}^{t_1 } \bigl(\frac{1}{2}\dot{q}^2-u(q)-\kappa q\,p^a_{cl}\bigr)dt\bigr\}}\,.\end{aligned}\ ] ] this formula served us as a guide when introducing the classical action ( [ a - action ] ) .
let us make one final comment . in our functional measure ( [ discr.2 ] )
no integration over the momenta @xmath321 and @xmath322 was prescribed .
these momenta are however needed when the boundary of the discretized surface @xmath323 is specified .
they define the auxiliary curves @xmath50 and @xmath40 introduced in section [ 3 ] .
the curves were chosen completely arbitrarily , but as seen from ( [ discr.1 ] ) , the result is not affected by them at all . hence , since everything substantial is independent of @xmath321 and @xmath322 , it is justifiable to discard these quantities from the functional measure @xmath80 $ ] .
if we would not do that for some reason , they will integrate into an artificial infinite factor , which will be removed anyway after we apply the normalization conditions ( [ normalization ] ) .
99 s. t. ali , m. engli : _ quantization methods : a guide for physicists and analysts _ , rev . math .
* 17 * ( 2005 ) , 391 - 490 , arxiv : math - ph/0405065 .
caldirola : _ forze non conservative nella meccanica quantistica _ , nuovo cim .
* 18 * ( 1941 ) , 393 - 400 .
e. kanai : _ on the quantization of the dissipative systems _ ,
. phys . * 3 * ( 1948 ) , 440 - 442 .
h. bateman : _ on dissipative systems and related variational principles _
, phys . rev .
* 38 * ( 1931 ) , 815 - 819 .
m. d. kostin : _ on the schrdinger - langevin equation _ , j. chem
* 57 * ( 1972 ) , 3589 - 3591
. h. dekker : _ on the quantization of dissipative systems in the lagrange - hamilton formalism _ , z. phys .
b * 21 * ( 1975 ) 295 - 300 .
m. razavy : _ on the quantization of dissipative systems _ , z. physik b * 26 * ( 1977 ) 201 - 206 .
j. geicke : _ semi - classical quantisation of dissipative equations _ ,
a : math . gen .
* 22 * ( 1989 ) 1017 - 1025 .
j. douglas : _ solution of the inverse problem in the calculus of variations _ , trans . am .
soc . * 50 * ( 1941 ) , 71 - 128 .
a. o. caldeira , a. j. leggett : _ path integral approach to quantum brownian motion _
, physica * 121a * ( 1983 ) , 587 - 616 .
u. weiss : _ quantum dissipative systems _ ( second edition ) , world scientific 1999 , singapore .
r. p. feynman , f. l. vernon : _ the theory of a general quantum system interacting with a linear dissipative system _ , annals of physics * 24 * ( 1963 ) , 118 - 173 .
f. j. dyson : _
feynman s proof of the maxwell equations _
, am . j. phys . * 58 * ( 1990 ) , 209 - 211 .
e. p. wigner : _ do the equations of motion determine the quantum mechanical commutation relations ?
_ , phys . rev .
* 77 * ( 1950 ) , 711 - 712 .
s. okubo : _ does the equation of motion determine commutation relations ? _ , phys .
d * 22 * ( 1980 ) , 919 - 923 .
p. o. kazinski , s. l. lyakhovich , a. a. sharapov : _ lagrange structure and quantization _ , jhep 0507 ( 2005 ) 076 , arxiv : hep - th/0506093 .
s. l. lyakhovich , a. a. sharapov : _ quantizing non - lagrangian gauge theories : an augmentation method _ ,
jhep 0701 ( 2007 ) 047 , arxiv : hep - th/0612086 .
d. m. gitman , v. g. kupriyanov : _ canonical quantization of so - called non - lagrangian systems _ , eur .
j. c * 50 * ( 2007 ) , 691 - 700 , arxiv : hep - th/0605025 .
r. p. feynman : _ space - time approach to non - relativistic quantum mechanics _
, rev . of mod .
* 20 * ( 1948 ) , 367 - 387 .
r. p. feynman , a. r. hibbs : _ quantum mechanics and path integrals _ , mcgraw - hill inc .
, new york , 1965 .
m. henneaux : _ equations of motion , commutation relations and ambiguities in the lagrangian formalism _ , annals phys .
* 140 * ( 1982 ) , 45 - 64 .
w. sarlet : _ the helmholtz condition revisited
. a new approach to the inverse problem of lagrangian dynamics _
, j. phys .
* 15 * ( 1982 ) , 1503 - 1517 .
d. kochan : _ direct quantization of equations of motion _
, acta polytechnica * 47 * no . 2 - 3 ( 2007 ) , 60 - 67 , arxiv : hep - th/0703073 .
d. kochan : _ quantization of non - lagrangian systems : some irresponsible speculations _
* 956 * ( 2007 ) , 3 - 8 .
c. stuckens , d. h. kobe : _ quantization of a particle with a force quadratic in the velocity _ , phys . rev .
a * 34 * ( 1986 ) 3565 - 3567 .
h. dekker : _ classical and quantum mechanics of the damped harmonic oscillator _ ,
phys . rep .
* 80 * ( 1981 ) , 1 - 112 .
m. razavy : _ classical and quantum dissipative systems _ ; imperial colleges press , london , 2005 .
i. c. moreira : _
propagators for the caldirola - kanai - schr " odinger equation _ , lett .
nuovo cim .
* 23 * ( 1978 ) , 294 - 298 .
u. das , s. ghosh , p. sarkar , b. talukdar : _ quantization of dissipative systems with friction linear in velocity _ , physica scripta * 71 * ( 2005 ) , 235 - 237 .
d. kochan : _ quantization of non - lagrangian systems _ ,
a * 24 * nos .
28 @xmath9 29 ( 2009 ) , 5319 - 5340 .
g. ghosh , r. w. hasse : _ coherent state and the damped harmonic oscillator _ , phys .
a * 24 * ( 1981 ) , 1621 - 1623 . l. chetouani et al : _ path integral for the damped harmonic oscillator coupled to its dual _ , j. math
* 35 * ( 1994 ) , 1185 - 1191 .
m. c. nemes , a. f. r. de toledo piza : _ a quantum phenomenology of viscosity _ , revista brasileira de fsica * 7 * no.2 ( 1977 ) , 261 - 270 .
|
a novel functional integral formulation of quantum mechanics for non - lagrangian systems is presented .
the new approach , which we call `` stringy quantization , '' is based solely on classical equations of motion and is free of any ambiguity arising from lagrangian and/or hamiltonian formulation of the theory .
the functionality of the proposed method is demonstrated on several examples .
special attention is paid to the stringy quantization of systems with a general @xmath0-power friction force @xmath1 .
results for @xmath2 are compared with those obtained in the approaches by caldirola - kanai , bateman and kostin .
relations to the caldeira - leggett model and to the feynman - vernon approach are discussed as well .
dedicated to my father on the occasion of his 60th birthdays .
|
positronium ( ps ) , the electron - positron bound state , was discovered experimentally in 1951 @xcite . since that time
a lot of attention has been paid to the determination of its properties , including lifetime , decay modes , and spectroscopy .
the experimental and theoretical accuracies achieved by now being quite high , there is little doubt that quantum electrodynamics ( qed ) is the only interaction in this system .
in fact , thanks to the smallness of the electron mass @xmath2 relative to typical hadronic mass scales , its theoretical description is not plagued by strong - interaction uncertainties and its properties , such as decay widths and energy levels , can be calculated perturbatively in non - relativistic qed ( nrqed ) @xcite , as expansions in sommerfeld s fine - structure constant @xmath3 , with very high precision .
ps comes in two ground states , @xmath4 parapositronium ( @xmath5-ps ) and @xmath6 orthopositronium ( @xmath7-ps ) , which decay to two and three photons , respectively .
here we are concerned with the lifetime of @xmath7-ps , which has been the subject of a vast number of experimental and theoretical investigations .
its first measurement @xcite was performed later in the year 1951 and agreed well with its lowest - order ( lo ) prediction of 1949 @xcite .
its first precision measurement @xcite , of 1968 , had to wait 9 years for the first correct one - loop calculation @xcite , which came two decades after the analogous calculation for @xmath5-ps @xcite being considerably simpler owing to the two - body final state . in the year 1987 , the ann arbor group @xcite published a measurement that exceeded the best theoretical prediction available then by more than 8 experimental standard deviations .
this so - called @xmath7-ps lifetime puzzle triggered an avalanche of both experimental and theoretical activities , which eventually resulted in what now appears to be the resolution of this puzzle .
in fact , the 2003 measurements at ann arbor @xcite and tokyo @xcite , @xmath8 agree mutually and with the present theoretical prediction , @xmath9 the latter is evaluated from @xmath10 , \label{gamma}\ ] ] where @xcite @xmath11 is the lo result .
the leading logarithmically enhanced @xmath12 and @xmath13 terms were found in refs .
@xcite and ref .
@xcite , respectively .
the coefficients @xmath14 @xcite , @xmath15 @xcite , and @xmath16 @xcite were evaluated numerically in a series of papers .
comprehensive reviews of the experimental and theoretical status of ps may be found in refs .
@xcite .
we note in passing that high - precision tests make ps also a useful probe of new physics beyond he standard model . at present
, there is strong interest in models with extra dimensions @xcite , which may provide a solution of the gauge hierarchy problem @xcite ( see ref .
@xcite for a review ) .
some time ago , a peculiar feature of matter in brane world was observed in ref .
@xcite , where it was shown that massive particles initially located on our brane may leave the brane and disappear into extra dimensions .
the experimental signature of this effect is the disappearance of a particle from our world , _
i.e. _ its invisible decay .
the case of the electromagnetic field propagating in the randall sundrum type of metric in the presence of extra compact dimensions @xcite was considered in ref .
@xcite , where it was shown that the transition rate of a virtual photon into extra dimensions is non - zero .
this effect could result in the disappearance of a neutral system . in the case of @xmath7-ps , such estimations for the invisible decay branching fraction @xmath17-ps@xmath18_invisible _ )
@xcite range just one order of magnitude below the presently best experimental upper bound of @xmath19 at 90% confidence level established by badertscher _
et al._@xcite .
thus , this decay is of great interest for the possible observation of effects due to extra dimensions . in order to reduce the theoretical uncertainty in the @xmath7-ps total decay width @xmath20(theory ) , it is indispensable to increase the precision in the coefficients @xmath21 , @xmath22 , and @xmath23 in eq .
( [ gamma ] ) .
this is most efficiently done by avoiding numerical integrations altogether , _
i.e. _ by establishing the analytic forms of these coefficients .
the case of @xmath22 is beyond the scope of presently available technology , since it involves two - loop five - point functions to be integrated over the three - particle phase space . in the following , we thus concentrate on @xmath21 and @xmath23 .
the quest for an analytic expression for @xmath21 has a long history .
about 25 years ago , some of the simpler contributions to @xmath21 , due to self - energy and outer and inner vertex corrections , were obtained analytically @xcite , but further progress then soon came to a grinding halt . in our recent letter @xcite , this task was completed for @xmath21 as a whole .
the purpose of the present paper is to explain the most important technical details of this calculation and to collect mathematical identities that may be useful for similar calculations . an analytic expression for @xmath23
is then simply obtained from that for @xmath21 through the relationship @xcite @xmath24 which may be understood qualitatively by observing that the @xmath1 correction in eq .
( [ gamma ] ) receives a contribution from the interference of the relativistic @xmath0 term from the hard scale with non - relativistic @xmath12 terms from softer scales .
the structure of this paper is as follows .
section [ sec : app ] contains the well - known integral representation of the @xmath7-ps total decay width as given in ref .
@xcite . in sec .
[ sec : use ] , we show how to transform the contributing integrals to forms appropriate for analytic evaluation , which is carried out for the most complicated integrals , which are plagued by singularities , in sec . [ sec : eval ] .
more examples are studied in sec .
[ sec : some ] .
the final results for the coefficients @xmath21 and @xmath23 are presented in sec .
[ sec : res ] .
section [ sec : con ] contains a summary . in appendix
[ sec : a ] , we present the analytic results for all parts of the integral representation given in sec . [
sec : app ] .
appendix [ sec : b ] contains useful representations of the @xmath25 function and the expansion of the @xmath20 function about half - integer - valued arguments . in appendix
[ sec : c ] , transformation formulas for generalized polylogarithms of weight four with different arguments are collected .
feynman diagrams contributing to the total decay width of @xmath7-ps at @xmath0 .
self - energy diagrams are not shown . dashed and solid lines represent photons and electrons , respectively.,scaledwidth=45.0% ] the @xmath0 contribution in eq .
( [ gamma ] ) , @xmath26 , is due to the feynman diagrams where a virtual photon is attached in all possible ways to the tree - level diagrams , with three real photons linked to an open electron line , and the electron box diagrams with an @xmath27 annihilation vertex connected to one of the photons being virtual ( see fig . [
fig : dia ] ) . taking the interference with the tree - level diagrams , imposing @xmath27 threshold kinematics , and performing the loop and angular integrations
, one obtains the two - dimensional integral representation @xcite @xmath28 , \label{eq : org}\ ] ] where @xmath29 , with @xmath30 , is the energy of photon @xmath31 in the @xmath7-ps rest frame normalized to its maximum value , the delta function ensures energy conservation , and perm .
stands for the other five permutations of @xmath32 . the function @xmath33 is given by @xmath34 where @xmath35 are ratios of polynomials , which are listed in eqs .
( a5a)(a5h ) of ref .
@xcite , and @xmath36 , \nonumber\\ h_4(x_1 ) & = & \frac{1}{4x_1 } [ 3\zeta(2)-2\theta_1 ^ 2],\qquad h_5(x_1 ) = \frac{1}{2\overline{x}_1 } \theta_1 ^ 2 , \\
h_6(x_1,x_3 ) & = & \frac{1}{\sqrt{x_1\overline{x}_1x_3\overline{x}_3 } } \ , \left[{\mathop{\mathrm{li}}\nolimits}_2\left(r^+_a,\overline{\theta}_1\right ) -{\mathop{\mathrm{li}}\nolimits}_2\left(r^-_a,\overline{\theta}_1\right)\right ] , \label{h6}\\ h_7(x_1,x_3 ) & = & \frac{1}{2\sqrt{x_1\overline{x}_1x_3\overline{x}_3 } } \ , \left [ 2{\mathop{\mathrm{li}}\nolimits}_2(r^+_b,\theta_1 ) - 2{\mathop{\mathrm{li}}\nolimits}_2(r^-_b,\theta_1 ) - { \mathop{\mathrm{li}}\nolimits}_2(r^+_c,0 ) + { \mathop{\mathrm{li}}\nolimits}_2(r^-_c,0)\right ] , \label{h7}\end{aligned}\ ] ] where @xmath37 and @xmath38 here , @xmath39 and @xmath40 is the real part of the dilogarithm [ see line below eq .
( [ eq : s ] ) ] of complex argument @xmath41 @xcite . since we are dealing here with a single - scale problem , eq .
( [ eq : org ] ) yields just a number .
although bose symmetry is manifest in eq .
( [ eq : org ] ) , its evaluation is complicated by the fact that , for a given order of integration , individual permutations yield divergent integrals , which have to cancel in their combination . in order to avoid such a proliferation of terms
, we introduce an infinitesimal regularization parameter @xmath42 in such a way that the symmetry under @xmath43 for any pair @xmath44 is retained . in this way , eq .
( [ eq : org ] ) collapses to @xmath45 where @xmath46 .
note that we may now exploit the freedom to choose any pair of variables @xmath29 and @xmath47 @xmath48 as the arguments of @xmath49 and as the integration variables .
obviously , the functions @xmath50 and @xmath51 in eqs . ( [ h6 ] ) and ( [ h7 ] ) , respectively , give the most complicated contributions to @xmath52 . in order to perform integrations involving these terms , it is useful to apply the integral representation of eq .
( [ 1.4 ] ) to @xmath53 , @xmath54 , and @xmath55 .
let us first consider @xmath56 .
we see from eq .
( [ 1.3 ] ) that @xmath57 and thus @xmath58 , \label{1.5}\ ] ] where @xmath59 .
then , the term @xmath60 on the r.h.s . of eq .
( [ h7 ] ) , after the change @xmath61 , can be rewritten as @xmath62 .
\label{1.6}\ ] ] finally , substituting @xmath63 , we obtain @xmath64 .
\label{1.7}\ ] ] the residual term on the r.h.s . of eq .
( [ h7 ] ) , @xmath65 , can be transformed in the same way yielding @xmath66 . \label{1.8}\ ] ] we thus obtain the following integral representation for @xmath51 : ) corrects a misprint in eq .
( 15 ) of ref .
@xcite , which is , however , inconsequential because the difference between these expressions cancels due to the @xmath67 symmetry . ]
@xmath68 . \label{1.5a}\ ] ] exploiting the @xmath67 symmetry of the coefficient @xmath69 multiplying @xmath51 , eq .
( [ 1.5a ] ) can be effectively replaced by @xmath70 .
\label{1.6a}\ ] ] next , this expression , multiplied by @xmath69 , is to be integrated over @xmath71 , @xmath72 , and @xmath73 .
observing that the logarithmic terms in eq .
( [ 1.6a ] ) are independent of @xmath71 , we first integrate over @xmath71 ( for a similar approach , see ref .
@xcite ) . in order to avoid the appearance of complicated functions in the intermediate results , the integration over @xmath73 in eq .
( [ 1.6a ] ) is performed last . using the same technique ,
we obtain the following representation for the function @xmath50 : @xmath74 , \label{1.7a}\ ] ] in which the part proportional to @xmath75 and the complementary one are first integrated over @xmath72 and @xmath71 , respectively .
the @xmath73 integration is again performed last . in secs .
[ sec : eval ] and [ sec : some ] , we discuss in more details how these integrations can be performed .
[ sec : eval ] we now discuss the evaluation of the most complicated integrals , namely those involving the functions @xmath50 and @xmath51 .
we denote the corresponding integrated expressions as @xmath78 and @xmath79 , respectively .
they are both singular for @xmath80 , so that the regularization of eq .
( [ a.1 ] ) is indispensable .
let us first consider the contribution of the coefficient @xmath69 without the function @xmath51 .
it can be decomposed into two parts , as @xmath81 where @xmath82 gives rise to the singularity upon integration over @xmath71 and @xmath72 , while @xmath83 remains finite , so that the limit @xmath84 can be taken .
a similar decomposition can be made also for @xmath85 .
specifically , performing the integrations over @xmath71 and @xmath72 and taking the limit @xmath84 , we have @xmath86 observing that the presence of the functions @xmath50 and @xmath51 does not change the singularity structure of the integrals over the variables @xmath72 , @xmath71 , and @xmath73 in this order , the decomposition of eq .
( [ 1b.1 ] ) leads to @xmath87 with @xmath88 .
our evaluation yields @xmath89 where is to indicate that all its parts carry weight three . ]
@xmath90 . \label{1b.9}\end{aligned}\ ] ] as can be seen from eqs .
( [ 1b.11a ] ) and ( [ 1b.7a ] ) , @xmath91 cancels in the sum @xmath92 . here and in the following
, we use the short - hand notations @xmath93 furthermore , @xmath94 is the generalized nielsen polylogarithm , @xmath95 the polylogarithm of order @xmath96 , and @xmath97 , with @xmath98 being riemann s zeta function @xcite .
the result of eq .
( [ 1b.9 ] ) , which is the most complicated part arising from the terms with @xmath99 and 7 in eq .
( [ a.2 ] ) , assumes a rather simple form when written as an infinite series , @xmath100,\ ] ] where @xmath101 is the @xmath102-th logarithmic derivative of the @xmath20 function , @xmath103 .
we can now apply the well - known relations for @xmath20 and @xmath25 functions , @xmath104 , \label{1b.14a}\end{aligned}\ ] ] where @xmath105 is the harmonic sum . using eqs .
( [ 1b.14 ] ) and ( [ 1b.14a ] ) , the constant @xmath106 is rewritten in terms of so - called inverse central binomial sums , _
i.e._sums of the form @xmath107 where @xmath108 is some combination of harmonic sums and factors like @xmath109 , and @xmath110 is some number .
sums of such type were studied in great detail in refs .
@xcite .
it is known that , for the series in eq .
( [ icbs ] ) , there exists a nonlinear transformation , @xmath111 which leads to great simplifications in many cases .
the series in the new variable @xmath112 does not have a binomial coefficient and can be summed , yielding expressions involving generalized polylogarithms @xmath113 .
now we can explain the appearance of the prefactor @xmath114 in front of @xmath106 in eqs .
( [ 1b.11x ] ) and ( [ 1b.8 ] ) .
such a prefactor has not appeared in single - scale calculations so far .
the point is that all inverse binomial series involving products of the factor @xmath109 and some function @xmath115 that is a combination of the @xmath25 function and its derivatives have the form ( see , for example , ref .
@xcite ) @xmath116 where @xmath117 is some combination of generalized polylogarithms and @xmath112 is defined by eq .
( [ ydefinition ] ) . note that eq .
( [ 1b.14a ] ) contains the binomial sum @xmath118 , which is related to the basic one , @xmath119 , via @xmath120 thus , the last term on the r.h.s . of eq .
( [ 1b.15b ] ) leads to @xmath121 in eq .
( [ 1b.13 ] ) , which translates to @xmath122 via eq .
( [ ydefinition ] ) .
this term then cancels the term @xmath123 on the r.h.s . of eq .
( [ 1b.13 ] ) . for the term @xmath124 on the r.h.s . of eq .
( [ 1b.14a ] ) , we have @xmath125 so that the variable @xmath112 from eq .
( [ ydefinition ] ) assumes the value @xmath126 this explains the appearance of the factor @xmath114 in eqs .
( [ 1b.11x ] ) and ( [ 1b.8 ] ) , since @xmath127 .
finally , we can rewrite eq .
( [ 1b.9 ] ) as @xmath128 where @xmath129 .
let us now consider several typical integrals that arise upon the first integration .
our first example of the remaining two - fold integrals reads @xmath130\ln(1-x).\ ] ] direct integration over @xmath73 or @xmath131 would lead to rather complicated functions in the remaining variable .
instead , we taylor expand the first logarithm using @xmath132 to obtain @xmath133^n \int\limits_0 ^ 1 \frac{dx}{x}(1-x)^n\ln(1-x).\ ] ] now , the two integrals are separated and can be solved in terms of euler s @xmath20 function . using @xmath134
we finally have @xmath135 . \label{ires}\ ] ] clearly , in the cases of @xmath136 and @xmath137 , the argument @xmath110 in eq .
( [ icbs ] ) is equal to @xmath138 and @xmath139 , respectively .
the case of @xmath136 is simpler and leads to a smaller number of constants .
indeed , we can use the results of refs .
@xcite to obtain @xmath140 and so on .
according to ref .
@xcite , after transformation to the variable @xmath112 of eq .
( [ ydefinition ] ) , we arrive at polylogarithms of argument @xmath141 , which are expressed in terms of alternating and non - alternating euler
zagier sums , such as @xmath142 , @xmath143 , etc .
let us now turn to the case of @xmath137 in eq .
( [ ires ] ) .
the argument @xmath125 gives @xmath144 and leads to a new type of constants .
again using formulas from ref .
@xcite , we have @xmath145 \nonumber\\ & & { } + 4 l_r [ { \mathop{\mathrm{li}}\nolimits}_3(r)- { \mathop{\mathrm{s}}\nolimits}_{1,2}(r)-l_2 { \mathop{\mathrm{li}}\nolimits}_2(r ) ] - l_r^2 [ 2{\mathop{\mathrm{li}}\nolimits}_2(r)+l_2 ^ 2].\end{aligned}\ ] ] with the help of the relations listed in appendix [ sec : c ] , @xmath146 can be alternatively expressed as @xmath147 \nonumber\\ & & { } - 4 { \mathop{\mathrm{re}}\nolimits}\left [ { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{1-\sqrt{2}}{2}\right ) + { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{1+\sqrt{2}}{2}\right ) \right ] .
\label{5f.4a}\end{aligned}\ ] ] it has been observed empirically that , at weight four , the terms that are not expressed through the usual riemann zeta function @xmath148 often come in the combination @xmath149 introduced by broadhurst in ref .
examples include the three - loop qcd correction to the electroweak @xmath150 parameter @xcite , the electron anomalous magnetic moment at three loops @xcite , critical exponents in high orders of perturbation theory @xcite , the heavy - quark contribution to the vacuum polarization function at four loops in qcd @xcite , and the matching conditions for the strong - coupling constant at five loops @xcite .
our result for @xmath136 in eq .
( [ 5f.4 ] ) exhibits a violation of this empirical observation .
in fact , the non - zeta terms form some different combination there . another class of typical integrals yields sums involving @xmath25 functions of half - integer arguments ( see appendix b ) , _
e.g._@xmath151\ln(1-x)}{x-2 } \nonumber\\ & = & \sum\limits_{n=1}^\infty \frac{(\pm 4)^n}{8n}\ , \frac{\gamma^2(n)}{\gamma(2n ) } \left[\psi^\prime\left(\frac{n+2}{2}\right ) -\psi^\prime\left(\frac{n+1}{2}\right ) \right ] \nonumber\\ & = & \sum\limits_{n=1}^\infty \frac{1}{\left ( 2n \atop n \right ) } \ , \frac{(\mp 4)^n}{n^2 } \left [ -\frac{1}{2}\zeta_2 - s_{-2}(n-1 ) - \frac{(-1)^n}{n^2 } \right ] .
\label{ires1}\end{aligned}\ ] ] following a similar strategy as above and using formulas from sec .
[ sec : eval ] , we may express @xmath152 in terms of known irrational constants , as @xmath153 + 4 \left [ { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{2-\sqrt{2}}{4}\right ) + { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{2+\sqrt{2}}{4}\right ) \right ] \nonumber\\ & = & -\frac{5}{2 } \zeta_2 l_2 ^ 2 + \frac{17}{48 } l_2 ^ 4 + \frac{21}{4 } \zeta_4 -g_4 , \nonumber\\ j_- & = & \frac{1}{2 } \zeta_2l_2 ^ 2- \frac{49}{48 } l_2 ^ 4 -\zeta_4 + 6{\mathop{\mathrm{li}}\nolimits}_4\left(\frac{1}{2}\right ) + { l_r}^2 \left(3{l_2}^2- \frac{11}{2 } \zeta_2 \right ) - \frac{7}{4 } { l_r}^4 \nonumber\\ & & { } + { l_r}\left\{\frac{1}{3 } { l_2}^3 + 5 \zeta_2{l_2}+ \frac{7}{4 } \zeta_3 -16 { \mathop{\mathrm{li}}\nolimits}_3\left(\frac{1}{\sqrt{2}}\right ) - 5 { \mathop{\mathrm{re}}\nolimits}\left [ { \mathop{\mathrm{li}}\nolimits}_3\left(\frac{1-\sqrt{2}}{2}\right ) - { \mathop{\mathrm{li}}\nolimits}_3\left(\frac{1+\sqrt{2}}{2}\right ) \right ] \right\ } \nonumber\\ & & { } + 5 { \mathop{\mathrm{re}}\nolimits}\left [ { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{1-\sqrt{2}}{2}\right ) + { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{1+\sqrt{2}}{2}\right ) \right ] - 4 \left [ { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{2-\sqrt{2}}{4}\right ) + { \mathop{\mathrm{li}}\nolimits}_4\left(\frac{2+\sqrt{2}}{4}\right ) \right ] , \nonumber\\\end{aligned}\ ] ] where @xmath154 , expressed with the help of the variable @xmath155 defined in eq .
( [ 1b.15c ] ) , is given in eq .
( [ aa.5 ] ) .
these results again contain various contributions of polylogarithms with argument @xmath122 , arising from terms of the form @xmath156 on the r.h.s . of eq .
( [ ires1 ] ) for @xmath157 and terms of the form @xmath158 on the r.h.s . of eq .
( [ ires1 ] ) for @xmath159 , and with argument @xmath144 , arising from the residual terms .
unfortunately , not all integrals can be computed so straightforwardly . in more complicated cases ,
the integrations are not separated after expansion to infinite series .
we then rely on the pslq algorithm @xcite , which allows one to reconstruct the representation of a numerical result known to very high precision in terms of a linear combination of a set of irrational constants with rational coefficients , if that set is known beforehand .
the experience gained with the explicit solution of the simpler integrals helps us to exhaust the relevant sets . in order for the pslq algorithm to work in our applications , the numerical values of the integrals
must be known up to typically 150 decimal figures . however , for some integrals more accurate determinations are required .
the success of the application of the pslq algorithm also relies on the fact that only certain combinations of polylogarithms , like @xmath106 in eqs .
( [ 1b.9 ] ) and ( [ 1b.16 ] ) , @xmath154 in eq .
( [ aa.5 ] ) , and @xmath160 in eq .
( [ aa.4 ] ) are incorporated as independent structures .
finally , to get rid of complex polylogarithms , such as @xmath161 $ ] , in the above formulas , we transform all polylogarithms to arguments of value below unity . to this end
, we need transformation formulas through weight four . some of these formulas are listed in appendix [ sec : c ] . after a laborious calculation
, we obtain the final result for the one - loop correction @xmath162 where the constants @xmath106 , @xmath154 , and @xmath160 are specified in eqs .
( [ 1b.16 ] ) , ( [ aa.5 ] ) , and ( [ aa.4 ] ) , respectively . transforming the polylogarithmic functions by means of the formulas given in appendix [ sec : c ] , we arrive at the form of ref .
@xcite , @xmath163 \ , , \label{ares1}\end{aligned}\ ] ] where @xmath155 is given in eq .
( [ 1b.15c ] ) . from eqs .
( [ ares1 ] ) and ( [ cres ] ) , @xmath21 and @xmath23 can be numerically evaluated with arbitrary precision , @xmath164 these numbers agree with the best existing numerical evaluations @xcite within the quoted errors .
we presented the details of our evaluation of the @xmath0 and @xmath1 corrections to the total decay width of @xmath7-ps , _
i.e. _ of the coefficients @xmath21 and @xmath23 in eq .
( [ gamma ] ) , respectively , which had been presented in our previous letter @xcite in closed analytic form .
we discussed the nature and the origin of new irrational constants that appear in the final results .
they were shown to be some particular cases of inverse central binomial sums and corresponding generalized polylogarithms .
these constants enlarge the class of the known constants in single - scale problems .
the @xmath165 correction @xmath22 in eq .
( [ gamma ] ) still remains analytically unknown .
we are grateful to g.s .
adkins for providing us with the computer code employed for the numerical analysis in ref .
the work of b.a.k . was supported in part by the german federal ministry for education and research bmbf through grant no .
the work of a.v.k . was supported in part by the german research foundation dfg through grant no .
inst 152/4651 , by the heisenberg - landau program through grant no . 5 , and by the russian foundation for basic research through grant no .
the work of o.l.v . was supported in part by the helmholtz association hgf through grant no .
0 in this appendix , we present separate results for the integrals @xmath166 of eq .
( [ 1b.7 ] ) with @xmath167 .
note , that not all of them a finite in the limit @xmath84 .
we have : @xmath168 where @xmath106 is given in eqs .
( [ 1b.9 ] ) and ( [ 1b.16 ] ) , and @xmath169 + \frac{83}{256 } { l_3}^4 \nonumber\\ & & { } -\frac{119}{12 } \zeta_3 { l_2 } , \label{aa.4}\\ g_{4 } & = & \frac{15}{16}{l_2}^4 + \frac{1}{4 } { l_2}l_r^3 + \frac{5}{96 } l_r^4 + 5 \zeta_4 + \zeta_2 \left ( - 3{l_2}l_r - \frac{7}{4 } l_r^2 \right ) + \zeta_3 \left ( - 3{l_2}- \frac{5}{2 } l_r \right ) \nonumber\\ & & { } + \left ( 9{l_2}+ \frac{9}{2 } l_r \right ) { \mathop{\mathrm{li}}\nolimits}_3(r ) + ( -6{l_2}- 2 l_r ) { \mathop{\mathrm{li}}\nolimits}_3(-r ) - 9 \left [ { \mathop{\mathrm{li}}\nolimits}_4(r ) - { \rm s}_{2,2}(r ) \right ] \nonumber\\ & & { } + 4 \left [ { \mathop{\mathrm{li}}\nolimits}_4(-r ) - { \rm s}_{2,2}(-r ) \right ] .
\label{aa.5}\end{aligned}\ ] ] from eqs .
( [ aa.3 ] ) it is clear that @xmath91 cancels in the sum @xmath170 .
0 in this appendix , we present some useful relations between derivatives of the @xmath25 function with half - integer arguments and the @xmath25 and @xmath171 functions with integer arguments , and consider the expansion of the @xmath20 function in the vicinity of half - integer arguments . starting from the well - known relations between the @xmath25 and @xmath171 functions , @xmath172 + { l_2 } , \nonumber \\
\beta(2z ) & = & \frac{1}{2 } \left[\psi\left(z+\frac{1}{2}\right)-\psi(z ) \right ] , \label{b.1 } \end{aligned}\ ] ] and differentiating them @xmath2 ( @xmath173 ) times , we have @xmath174 where @xmath175 denotes the @xmath2-th derivative of @xmath176 etc .
we can combine eqs .
( [ b.1 ] ) and ( [ b.2 ] ) as @xmath177 -\delta_{0 m } { l_2 } , \nonumber \\
\psi^{(m)}\left(z+\frac{1}{2}\right ) & = & 2^m \left [ \psi^{(m)}(2z ) + \beta^{(m)}(2z ) \right ] -\delta_{0 m } { l_2 } , \label{b.3 } \end{aligned}\ ] ] where @xmath178 is the kronecker symbol . using the series representations of the @xmath25 and @xmath171 functions @xcite , @xmath179 we obtain the following relations : @xmath180 , \nonumber \\
\beta(n+1 ) & = & ( -1)^n [ l_2 + { \mathop{\mathrm{s}}\nolimits}_{-1}(n ) ] , \nonumber\\ \beta^{(m)}(n+1 ) & = & ( -1)^{m+n } m ! [ { \mathop{\mathrm{s}}\nolimits}_{-(m+1)}(n ) - { \mathop{\mathrm{s}}\nolimits}_{-(m+1)}(\infty ) ] , \label{b.4 } \end{aligned}\ ] ] where @xmath181 is defined in eq .
( [ 1b.15 ] ) .
thus , eqs .
( [ b.3 ] ) and ( [ b.4 ] ) lead to the following results for the `` sums '' @xmath182 with half - integer arguments : ) as definition of the `` sums '' @xmath182 with half - integer arguments . ]
@xmath183 l_2 , \nonumber \\
s_m\left(\frac{n}{2}\right ) & = & 2^{m-1 } [ s_m(n)+ ( -1)^n s_{-m}(n ) ] + [ 1-(-1)^n](1 - 2^{m-1 } ) \zeta_m \qquad ( m\geq 2).\ \label{b.7}\end{aligned}\ ] ] these equations are useful for expansions of the @xmath20 function in the vicinities of half - integer arguments . indeed , using a well - known formula for the expansions of the @xmath20 function about integer values , which was used , _
e.g. _ , in ref .
@xcite , @xmath184 , \label{b.11}\ ] ] where @xmath185 is euler s constant , we find the corresponding expansions about half - integer values to be @xmath186 , \label{b.12}\end{aligned}\ ] ] where @xmath187 is given by eq .
( [ b.7 ] ) .
such expansions are useful in many applications , including those in ref .
@xcite and references cited therein .
0 in this appendix , we present relations between the generalized polylogarithms @xmath188 of weight four ( @xmath189 ) with different arguments .
transformations at lower weights can be found in the literature @xcite .
although the derivation of these formulas is straightforward , we present them here for the convenience of interested readers . at weight four ,
there are three independent nielsen polylogarithms , which we choose to be @xmath190 , @xmath191 , and @xmath192 .
* relations for the functions with arguments @xmath193 and @xmath112 : @xmath194 + \frac{1}{2 } \ln^2(1-y ) [ \zeta_2 - { \mathop{\mathrm{li}}\nolimits}_{2}(y ) ] \nonumber \\ & & { } - \frac{1}{6 } \ln^3(1-y ) \ln y , \nonumber\\ { \mathop{\mathrm{s}}\nolimits}_{2,2}(1-y ) & = & \frac{1}{4 } \zeta_4 - { \mathop{\mathrm{s}}\nolimits}_{2,2}(y ) + \ln y { \mathop{\mathrm{s}}\nolimits}_{1,2}(y ) + \ln(1-y ) [ \zeta_3 - { \mathop{\mathrm{li}}\nolimits}_{3}(y ) + \ln y { \mathop{\mathrm{li}}\nolimits}_{2}(y ) ] \nonumber \\ & & { } + \frac{1}{4 } \ln^2(1-y ) \ln^2 y , \nonumber \\
{ \mathop{\mathrm{s}}\nolimits}_{1,3}(1-y ) & = & \zeta_4 - { \mathop{\mathrm{li}}\nolimits}_{4}(y ) + \ln y { \mathop{\mathrm{li}}\nolimits}_{3}(y ) - \frac{1}{2 } \ln^2 y { \mathop{\mathrm{li}}\nolimits}_{2}(y ) - \frac{1}{6 } \ln^3 y \ln(1-y ) .
\label{app1}\end{aligned}\ ] ] * 2 . * relations for the functions with arguments @xmath195 and @xmath196 : @xmath197 + \frac{1}{24 } \ln^4 y , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{1,3}\left(-\frac{1}{y}\right ) & = & - { \mathop{\mathrm{s}}\nolimits}_{1,3}(-y ) + { \mathop{\mathrm{s}}\nolimits}_{2,2}(-y ) - { \mathop{\mathrm{li}}\nolimits}_{4}(-y ) - \zeta_4 - \ln y [ { \mathop{\mathrm{s}}\nolimits}_{1,2}(-y ) - { \mathop{\mathrm{li}}\nolimits}_{3}(-y ) ] \nonumber \\ & & { } - \frac{1}{2 } \ln^2 y { \mathop{\mathrm{li}}\nolimits}_{2}(-y ) - \frac{1}{24 } \ln^4 y .
\label{app2 } \end{aligned}\ ] ] * 3 .
* relations for the functions with arguments @xmath198 and @xmath112 : @xmath199 \nonumber \\ & & { } + \frac{1}{2 } \ln^2(1-y ) { \mathop{\mathrm{li}}\nolimits}_{2}(y ) - \frac{1}{2 } \zeta_2 \ln^2\frac{1-y}{y } + \frac{1}{24 } \ln^4(1-y ) \nonumber \\ & & { } - \frac{1}{24 } \ln^4\frac{1-y}{y } , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{2,2}\left(\frac{y-1}{y}\right ) & = & 2{\mathop{\mathrm{li}}\nolimits}_{4}(y ) - { \mathop{\mathrm{s}}\nolimits}_{2,2}(y ) - \frac{7}{4 } \zeta_4 + \ln y { \mathop{\mathrm{s}}\nolimits}_{1,2}(y ) + \ln\frac{1-y}{y } \zeta_3 - \ln[(1-y)y ] { \mathop{\mathrm{li}}\nolimits}_{3}(y ) \nonumber \\ & & { } + \ln y\ln(1-y ) { \mathop{\mathrm{li}}\nolimits}_{2}(y ) + \frac{1}{4 } \ln^2 y \ln^2(1-y ) - \frac{1}{6 } \ln^3 y \ln(1-y ) \nonumber \\ & & { } + \frac{1}{24 } \ln^4 y , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{1,3}\left(\frac{y-1}{y}\right ) & = & { \mathop{\mathrm{li}}\nolimits}_{4}(y ) - \zeta_4 - \ln y { \mathop{\mathrm{li}}\nolimits}_{3}(y ) + \frac{1}{2 } \ln^2 y { \mathop{\mathrm{li}}\nolimits}_{2}(y ) + \frac{1}{6 } \ln^3 y \ln(1-y ) \nonumber \\ & & { } -\frac{1}{24 } \ln^4 y .
\label{app3}\end{aligned}\ ] ] * 4 . * relations for the functions with arguments @xmath200 and @xmath112 : @xmath201 \nonumber \\ & & { } - \frac{1}{2 } \ln^2(1-y ) { \mathop{\mathrm{li}}\nolimits}_{2}(y ) - \frac{1}{24 } \ln^4(1-y ) , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{2,2}\left(\frac{y}{y-1}\right ) & = & { \mathop{\mathrm{s}}\nolimits}_{2,2}(-y ) - 2 { \mathop{\mathrm{s}}\nolimits}_{1,3}(y ) - \ln ( 1-y ) { \mathop{\mathrm{s}}\nolimits}_{1,2}(y ) + \frac{1}{24 } \ln^4(1-y ) , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{1,3}\left(\frac{y}{y-1}\right ) & = & - { \mathop{\mathrm{s}}\nolimits}_{1,3}(y ) - \frac{1}{24 } \ln^4(1-y ) . \label{app4}\end{aligned}\ ] ] * 5 . * relations for the functions with arguments @xmath202 and @xmath196 : @xmath203 + \frac{1}{2 } \ln^2(1+y ) [ { \mathop{\mathrm{li}}\nolimits}_{2}(-y)+ \zeta_2 ] \nonumber \\ & & { } + \frac{1}{6 } \ln^3(1+y ) \ln y - \frac{1}{24 } \ln^4(1+y ) , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{2,2}\left(\frac{1}{1+y}\right ) & = & 2 { \mathop{\mathrm{s}}\nolimits}_{1,3}(-y ) - { \mathop{\mathrm{s}}\nolimits}_{2,2}(-y ) + \frac{1}{4 } \zeta_4 + + \ln [ y(1+y ) ] { \mathop{\mathrm{s}}\nolimits}_{1,2}(-y ) \nonumber \\ & & { } - \ln(1+y ) [ { \mathop{\mathrm{li}}\nolimits}_{3}(-y ) + \zeta_3 ] + \ln(1+y ) \ln y { \mathop{\mathrm{li}}\nolimits}_{2}(-y ) + \frac{1}{4 } \ln^2(1+y ) \ln^2 y \nonumber \\ & & { } - \frac{1}{6 } \ln^3(1+y ) \ln y + \frac{1}{24 } \ln^4 ( 1+y ) , \nonumber \\ { \mathop{\mathrm{s}}\nolimits}_{1,3}\left(\frac{1}{1+y}\right ) & = & { \mathop{\mathrm{s}}\nolimits}_{1,3}(-y ) - { \mathop{\mathrm{s}}\nolimits}_{2,2}(-y ) + { \mathop{\mathrm{li}}\nolimits}_{4}(-y ) + \zeta_4 + \ln y [ { \mathop{\mathrm{s}}\nolimits}_{1,2}(-y ) - { \mathop{\mathrm{li}}\nolimits}_{3}(-y ) ] \nonumber \\ & & { } + \frac{1}{2 } \ln^2 y { \mathop{\mathrm{li}}\nolimits}_{2}(-y ) + \frac{1}{24 } \ln^4 y -\frac{1}{24 } \ln^4\frac{1+y}{y } .
\label{app5}\end{aligned}\ ] ] equations ( [ app1 ] ) and ( [ app2 ] ) were directly obtained from ref .
@xcite , where they are presented for the generalized polylogarithms @xmath204 with arbitrary values of @xmath205 and @xmath206 , but in some complicated form less convenient for applications .
equations ( [ app3])([app5 ] ) were found by iterated application of eqs .
( [ app1 ] ) and ( [ app2 ] ) and equations from ref .
they are simple and useful for applications together with equations for @xmath204 from ref .
@xcite , with the constraints @xmath207 or @xmath208 .
i. harris and l. m. brown , phys .
rev . * 105 * , 1656 ( 1957 ) . c. i. westbrook , d. w. gidley , r. s. conti , and a. rich , phys .
* 58 * , 1328 ( 1987 ) ; * 58 * , 2153(e ) ( 1987 ) ; phys .
rev . a * 40 * , 5489 ( 1989 ) .
r. s. vallery , p. w. zitzewitz , and d. w. gidley , phys .
* 90 * , 203402 ( 2003 ) .
m. a. stroscio and j. m. holt , phys .
rev . a * 10 * , 749 ( 1974 ) ; m. a. stroscio , phys .
* 22 * , 215 ( 1975 ) ; g. s. adkins , ann .
( n.y . ) * 146 * , 78 ( 1983 ) ; g. s. adkins , a. a. salahuddin , and k. e. schalm , phys . rev . a * 45 * , 7774 ( 1992 ) ; g. s. adkins , phys
. rev .
lett . * 76 * , 4903 ( 1996 ) [ arxiv : hep - ph/0506213 ] .
g. s. adkins , r. n. fell , and j. r. sapirstein , phys .
lett . * 84 * , 5086 ( 2000 ) [ arxiv : hep - ph/0003028 ] ; phys .
a * 63 * , 032511 ( 2001 ) .
g. s. adkins , phys .
a * 72 * , 032501 ( 2005 ) .
b. a. kniehl and a. a. penin , phys .
lett . * 85 * , 1210 ( 2000 ) ; * 85 * , 3065(e ) ( 2000 ) [ arxiv : hep - ph/0004267 ] ; r. j. hill and g. p. lepage , phys . rev .
d * 62 * , 111301(r ) ( 2000 ) [ arxiv : hep - ph/0003277 ] ; k. melnikov and a. yelkhovsky , _ ibid . _ * 62 * , 116003 ( 2000 ) [ arxiv : hep - ph/0008099 ] .
g. s. adkins , r. n. fell , and j. r. sapirstein , ann .
( n.y . ) * 295 * , 136 ( 2002 ) ; d. sillou , int . j. mod .
a * 19 * , 3919 ( 2004 ) .
a. rubbia , int .
j. mod .
a * 19 * , 3961 ( 2004 ) [ arxiv : hep - ph/0402151 ] ; s. n. gninenko , n. v. krasnikov , v. a. matveev , and a. rubbia , phys .
nucl . * 37 * , 321 ( 2006 ) ; s. n. gninenko , n. v. krasnikov , and v. a. matveev , phys .
d * 78 * , 097701 ( 2008 ) [ arxiv:0811.0974 [ hep - ph ] ] .
k. akama , lect .
notes phys .
* 176 * , 267 ( 1983 ) [ arxiv : hep - th/0001113 ] ; v. a. rubakov and m. e. shaposhnikov , phys .
b * 125 * , 136 ( 1983 ) ; m. visser , _ ibid .
_ * 159 * , 22 ( 1985 ) [ arxiv : hep - th/9910093 ] ; i. antoniadis , _ ibid . _ * 246 * , 377 ( 1990 ) ; n. v. krasnikov , _ ibid . _ * 273 * , 246 ( 1991 ) . n. arkani - hamed , s. dimopoulos , and g. r. dvali , phys .
b * 429 * , 263 ( 1998 ) [ arxiv : hep - ph/9803315 ] ; i. antoniadis , n. arkani - hamed , s. dimopoulos , and g. r. dvali , _ ibid .
_ * 436 * , 257 ( 1998 ) [ arxiv : hep - ph/9804398 ] ; l. randall and r. sundrum , phys . rev .
* 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] ; _ ibid . _ * 83 * , 4690 ( 1999 ) [ arxiv : hep - th/9906064 ] .
v. a. rubakov , phys .
usp . * 44 * , 871 ( 2001 ) [ usp .
nauk * 171 * , 913 ( 2001 ) ] [ arxiv : hep - ph/0104152 ] . s. l. dubovsky , v. a. rubakov , and p. g. tinyakov , phys . rev .
d * 62 * , 105011 ( 2000 ) [ arxiv : hep - th/0006046 ] . i. oda , phys .
b * 496 * , 113 ( 2000 ) [ arxiv : hep - th/0006203 ] .
s. l. dubovsky , v. a. rubakov , and p. g. tinyakov , jhep * 0008 * , 041 ( 2000 ) [ arxiv : hep - ph/0007179 ] ; s. l. dubovsky and v. a. rubakov , in _ proceedings of the xxxviith rencontres de moriond : 2002 electroweak interactions and unified theories _
, edited by j. trn thanh vn ( th gii publishers , vietnam , 2003 ) , p. 367
[ arxiv : hep - th/0204205 ] .
s. n. gninenko , n. v. krasnikov , and a. rubbia , phys .
d * 67 * , 075012 ( 2003 ) [ arxiv : hep - ph/0302205 ] .
a. badertscher , p. crivelli , w. fetscher , u. gendotti , s. gninenko , v. postoev , a. rubbia , v. samoylenko , and d. sillou , phys .
d * 75 * , 032004 ( 2007 ) [ arxiv : hep - ex/0609059 ] .
m. a. stroscio , phys .
* 48 * , 571 ( 1982 ) ; g. s. adkins , phys .
a * 27 * , 530 ( 1983 ) ; b. a. kniehl , a. v. kotikov , and o. l. veretin , phys .
* 101 * , 193401 ( 2008 ) [ arxiv:0806.4927 [ hep - ph ] ] .
l. lewin , _ polylogarithms and associated functions _ ( elsevier , new york , 1981 ) .
j. fleischer , a. v. kotikov , and o. l. veretin , phys .
b * 417 * , 163 ( 1998 ) [ arxiv : hep - ph/9707492 ] ; j. fleischer , m. yu . kalmykov , and a. v. kotikov , _ ibid .
_ * 462 * , 169 ( 1999 ) [ arxiv : hep - ph/9905249 ] ; a. kotikov , j. h. khn , and o. veretin , nucl .
* b788 * , 47 ( 2008 ) [ arxiv : hep - ph/0703013 ] .
a. i. davydychev and m. yu . kalmykov , nucl .
* b699 * , 3 ( 2004 ) [ arxiv : hep - th/0303162 ] . b. a. kniehl , a. v. kotikov and , o. l. veretin , nucl .
b ( proc . suppl . ) * 183 * , 14 ( 2008 ) [ arxiv:0811.0306 [ hep - ph ] ] .
d. j. broadhurst , arxiv : hep - th/9604128 ; eur .
j. c * 8 * , 311 ( 1999 ) [ arxiv : hep - th/9803091 ] .
l. avdeev , j. fleischer , s. mikhailov , and o. tarasov , phys .
b * 336 * , 560 ( 1994 ) [ arxiv : hep - ph/9406363 ] ; * 349 * , 597(e ) ( 1995 ) ; j. fleischer and o. v. tarasov , nucl .
b ( proc . suppl . ) * 37 * , 115 ( 1994 ) [ arxiv : hep - ph/9407235 ] ; k. g. chetyrkin , j. h. khn , and m. steinhauser , phys .
b * 351 * , 331 ( 1995 ) [ arxiv : hep - ph/9502291 ] .
s. laporta and e. remiddi , phys .
b * 379 * , 283 ( 1996 ) [ arxiv : hep - ph/9602417 ] .
d. j. broadhurst and a. v. kotikov , phys .
b * 441 * , 345 ( 1998 ) [ arxiv : hep - th/9612013 ] .
b. a. kniehl and a. v. kotikov , phys .
b * 642 * , 68 ( 2006 ) [ arxiv : hep - ph/0607201 ] .
b. a. kniehl , a. v. kotikov , a. i. onishchenko , and o. l. veretin , phys .
lett . *
97 * , 042001 ( 2006 ) [ arxiv : hep - ph/0607202 ] .
h. r. p. ferguson and d. h. bailey , rnr technical report no . rnr-91 - 032 ; h. r. p. ferguson , d. h. bailey and s. arno , nasa technical report no .
nas-96 - 005 .
d. i. kazakov and a. v. kotikov , nucl .
. * b307 * , 721 ( 1988 ) ; * b345 * , 299(e ) ( 1990 ) ; theor . math
. phys . * 73 * , 1264 ( 1988 ) [ teor
. mat .
fiz . * 73 * , 348 ( 1987 ) ] ; a. v. kotikov , theor .
phys . * 78 * , 134 ( 1989 ) [ teor . mat .
fiz . * 78 * , 187 ( 1989 ) ] .
m. yu . kalmykov and b. a. kniehl , nucl .
* b809 * , 365 ( 2009 ) [ arxiv:0807.0567 [ hep - th ] ] .
|
recently , the @xmath0 and @xmath1 radiative corrections to the orthopositronium lifetime have been presented in closed analytical form , in terms of basic irrational numbers that can be evaluated numerically to arbitrary precision [ phys .
rev .
lett .
* 101 * , 193401 ( 2008 ) ] . here
, we present the details of this calculation and reveal the nature of these new constants . we also list explicit transformation formulas for generalized polylogarithms of weight four , which may be useful for other applications .
|
to explain the formation of active galactic nucleus ( agn ) jets and other astrophysical jets , various models have been proposed . among them , the magnetohydorodynamic ( mhd ) model is one of the most promising models , since it can explain both the acceleration and the collimation of the jets ( see , e.g. , meier , koide , & uchida 2001 , and references therein ) .
lovelace ( 1976 ) and blandford ( 1976 ) first proposed the theoretical model of the magnetically driven jet from accretion disks , and blandford & payne ( 1982 ) discussed magneto - centrifugally driven outflow from a keplerian disk in steady , axisymmetric and self - similar situation .
time - dependent , 2-dimensional axisymmetric simulations were performed by uchida & shibata ( 1985 ) , shibata & uchida ( 1986 ) , uchida & shibata ( 1986 ) .
they pointed out that large amplitude torsional alfvn waves ( taws ) generated by the interaction between the accretion disk and a large scale magnetic field play an important role .
the toroidal magnetic field propagates along the large scale magnetic field while squeezing it into a collimated jet - shape by the pinching effect of the lorentz force . in this paper
, we refer this model as a `` sweeping magnetic twist model '' . after these papers ,
many authors have performed time - dependent , 2-dimensional axisymmetric simulations ( e.g. stone & norman 1994 , ustyugova et al .
1995 , matsumoto et al .
1996 , ouyed & pudritz 1997 , kudoh , matsumoto , & shibata 1998 ) . using the numerical data of mhd models , observational quantities such as faraday rotation measure ( frm ) or stokes parameters
have been derived to be compared with observations of agn jets : laing ( 1981 ) computed the total intensity , the linear polarization , and the projected magnetic field distributions , assuming some simple magnetic field configurations and high energy particle distributions in the cylindrical jet .
clarke , norman , & burns ( 1989 ) performed 2-dimensional mhd simulations in which a supersonic jet with a dynamically passive helical magnetic field was computed , and derived distributions of the total intensity , the projected electric field , and the linear polarization .
hardee & rosen ( 1999 ) calculated the total intensity and the projected magnetic field distributions , using 3-dimensional mhd simulations of strongly magnetized conical jets .
hardee & rosen ( 2002 ) calculated the frm distribution and discussed that the observation of the radio source 3c465 in abell cluster a2634 ( eilek & owen 2002 ) suggests helical twisting of the flow .
frm is given by the integral of @xmath0 along the line - of - sight between the emitter and the observer ( where @xmath1 is the line - of - sight component of the magnetic field , and @xmath2 is the electron density there ) .
it is , in principle , not possible to specify which part on the line - of - sight the contribution comes from .
however , in recent high - resolution radio observations ( e.g. eilek & owen 2002 , asada et al .
2002 ) , the frm distribution seems to have good correlation with the configuration of the jet ; this suggests that the frm variation is due to the magnetized thermal plasma surrounding the emitting part of the jet .
in fact , sharp frm gradients seen in 3c273 can not be produced by a foreground faraday screen ( taylor 1998 , asada et al .
if this is the case , we can get a new information , that is , the line - of - sight component of the magnetic field , and thus can predict the 3-dimensional configuration of the magnetic field around the jet , together with the projected magnetic field .
uchida et al . ( 2004 ) ( hereafter paper i ) carried out a 2-dimensional axisymmetric simulation , and investigated the model counterpart distributions of frm and the projected magnetic field in the basic straight part of agn jets .
it was described how a systematically helical field configuration is produced in the `` sweeping magnetic twist model '' .
it was also shown as a result that the model can reproduce fairly well the characteristic distribution of frm having the gradient across the jet axis ( perley , bridle , & willis 1984 , asada et al .
2002 , gabuzda & murray 2003 ) .
the systematic frm gradient was caused by the gradient of the line - of - sight component of the magnetic field .
this means that the existence of the helical magnetic field ( and the propagation of taws ) is plausible .
on the basis of this success , in this paper the treatment is advanced to the non - axisymmetric situation , the wiggled structure , in agn jets .
the morphological structures in agn jets , such as `` wiggles ( kinks ) '' or `` bends '' are frequently seen not only on kpc scales , but also on pc scales and smaller ( hummel et al .
such a helical distortion might be caused either by plasma instabilities or precession of jet ejection axis due to the gravitational interaction between binary black holes ( bbhs ) ( begelman , blandford , & rees 1980 ) , bh / disk , or galaxies .
magnetically driven jets possess a toroidal field component , which is equivalent to an axial electric current , and such `` current - carrying '' jets are susceptible to mhd instabilities , and moreover , kelvin - helmholtz instabilities must be also taken into account for jet dynamics .
mhd instabilities are usually divided into pressure - driven instabilities and current - driven instabilities ( bateman 1980 ) .
kelvin - helmholtz instabilities have been considered for past decades by many theoretical and numerical work ( for reviews , see birkinshaw 1991 ; ferrari 1998 , and references therein ) .
one of the promising possibilities is a magnetic mechanism based on the `` sweeping magnetic twist model '' .
it was shown by 3-dimensional mhd simulations that the formation of the wiggled structure can be explained by the current - driven helical kink instability ( todo et al . 1993 ) . in these simulations ,
the magnetic field was a force - free helical field from the beginning and the propagation of taws was not dealt with .
nakamura , uchida , & hirose ( 2001 ) extended the treatment of the `` sweeping magnetic twist model '' to the part far from the gravitator and the accretion disk .
they investigated the behavior of taws propagating far from the agn core .
they showed that the current - driven helical kink instability can explain the production of the observed wiggles .
if the mechanism of the wiggle formation is clarified , it can become a clue to the physics of the jet formation . in this paper , we calculate frm , the projected magnetic field , and the total intensity from the numerical data of a mhd simulation based on our `` sweeping magnetic twist model '' , and discuss these model counterparts comparing with an observation . here
we consider the wiggled structure of the jet , and thus use the same kind of 3-dimensional simulation as in nakamura et al .
( 2001 ) . in section 2 ,
we describe the application of the model to the distant part of the jet from the agn core , and show the formation of a helical structure of the jet by the current - driven helical kink instability .
we introduce the method to calculate model counterparts of observational quantities in section 3 , and show the results in section 4 .
we discuss comparison between the calculated distributions and observed ones in section 5 .
the conclusion is summarized in section 6 .
the `` sweeping magnetic twist model '' was proposed for star - forming jets ( uchida & shibata 1985 , shibata & uchida 1986 ) , and it has since then been extended to the case of agn jets ( uchida & shibata 1986 , matsumoto et al .
1996 , uchida et al . 1999 ,
uchida et al .
2000 , nakamura et al .
2001 ) as described in paper i in some detail . in paper
i , we confined ourselves to the straight part of the jet as a first step , by using the 2-dimensional axisymmetric model . in the present paper , we extend our treatment to the wiggled part of our 3-dimensional jet model , in order to compare the model counterpart with the observation . in this section
, we describe the formation of the wiggled structure due to the current - driven helical kink instability .
figure [ fig01 ] is the schematic picture of a physical situation we treat in this paper .
we consider a primordial large scale magnetic field , and that it is squeezed under an assumption of frozen - in magnetic flux due to the gravitational contraction to form the central agn core ( figure [ fig01]a , b ) . during the contracting process of the magnetized gas , the toroidal component of the magnetic field
is continuously created by the rotation of the intergalactic medium having the angular momentum . as a result ,
large amplitude non - linear torsional alfvn waves ( taws ) begin to propagate out along the large scale poloidal field .
we expect that this process reaches its maximum as the accretion disk is formed , and the interaction of the rotating accretion disk and the poloidal magnetic field penetrating it reaches its maximum phase ( figure [ fig01]c ) . and also
, the mhd jets powered by taws are emitted in opposite directions along the rotation axis of the disk which is the direction of the original large scale poloidal magnetic field . in the present paper ,
we concentrate our attention on the distant part of the jet from the agn core .
the taws finally encounter the ambient medium with a higher density surrounding the large spherical `` cavity '' from which the mass fell toward the center in the gravitational contraction ( figure [ fig01]d ) .
the hst optical and merlin radio observations of a few hundreds pc - scale 3c264 jet indicate that the jet has a wiggled structure as a result of interacting with a circular optical cold ( dense ) `` ring '' , which may be a true ring , a shell , or a filled spheroid ( baum et al . 1997 ) .
this observational result indicates the interaction between the jet and the ambient medium will affect the jet dynamics and morphology .
the existence of denser material ahead of the propagating taws causes the deceleration of mhd jets due to the decreasing local alfvn velocity ; the accumulation of the toroidal component of the magnetic field occurs , which may lead to the current - driven helical kink instabilities through this non - linear process ( nakamura et al . 2001 ) . even in the case of decreasing density atmospheres
, the mhd jets would be subject to the current - driven kink distortions ( m. nakamura & d. l. meier , in preparation ) .
the computational domain we report here is also shown in figure [ fig01](d ) , and it includes a part of `` squeezed '' poloidal field near the edge of the density `` cavity '' .
we assume the poloidally magnetized intergalactic medium with a varying alfvn velocity distribution ( _ i.e. _ , gradually decreasing as the taws propagate ) due to the existence of a denser shell which represents the edge of the `` cavity '' .
a continuous cylindrical mhd inflow , powered by taws into the `` evolved '' region of the domain , is specified for all time in the lower `` boundary zone '' .
we note that a quasi - stationary poynting - flux - dominated flow is injected into our calculated region throughout the time evolution .
we adopt the ideal mhd as the governing equations .
@xmath3 @xmath4 @xmath5 @xmath6 where @xmath7 are the density , pressure , and velocity of the gas , and @xmath8 is the magnetic field .
@xmath9 represents the ratio of specific heats and is equal to 5/3 in this paper . in paper
i , we introduced the effect of gravity in eq .
( 2 ) in order to consider the launching and accelerating of mhd jets near the agn core .
we neglect the gravity term in this equation , and concentrate on the dynamics of the propagating mhd jets in this paper .
we normalize all physical quantities by the typical length @xmath10 , typical density @xmath11 , and typical velocity @xmath12 .
@xmath10 is ten times as long as the radius with the maximum velocity of the circular motion . @xmath11 and @xmath12 are the density and the alfvn velocity at the origin respectively . the total computational domain is taken to be @xmath13 , @xmath14 , and @xmath15 , where @xmath16 , @xmath17 , @xmath18 , and @xmath19 . the numbers of grid points in the simulation reported here are ( @xmath20 ) @xmath21 @xmath22 , where the grid points are distributed non - uniformly in the @xmath23 , @xmath24 , and @xmath25 directions .
the grid spacing is uniform , ( @xmath26 ) @xmath21 ( @xmath27 ) for @xmath28 , @xmath29 , and @xmath30 , and then stretched by 5 % per each grid step for the regions @xmath31 , @xmath32 , and @xmath33 .
initial conditions are almost the same as those of nakamura et al .
( 2001 ) , but the radial profiles of @xmath34 and @xmath35 ( the circular motion and inflow imposed at the lower `` boundary zone '' ) in this paper are different from those in nakamura et al .
the profiles in this paper are shown in figure [ fig02 ] .
the initial distributions of the density and the pressure of the gas , and the 3-dimensional configuration of selected magnetic lines of force , are shown in figure [ fig03 ] .
this figure displays the distributions of the range of @xmath36 , @xmath37 .
@xmath38 decreases gradually along the @xmath25-axis and tend to @xmath39 for the upper region , and @xmath40 also decreases in the same way , but increases again toward the upper boundary .
we assume a constant pressure throughout the computational domain and the plasma @xmath41 ( the ratio of gas pressure to magnetic pressure ) at the origin is @xmath42 .
the value of the plasma @xmath41 increases along the @xmath25-axis and exceeds unity at @xmath43 . in the decreasing region ( @xmath44 ) in the lower part , we have the constant alfvn velocity ( @xmath45 as @xmath40 increases .
the formation of the wiggled structure is universal independent of the imposed velocity profiles .
the clearest difference in the results between nakamura et al .
( 2001 ) and this paper is the time which it takes before the wiggled structure is formed .
we discuss the dynamical behavior of a typical case in our performed 3-dimensional simulations : the growth of a current - driven instability in a mhd jet .
figure [ fig04 ] , [ fig05 ] , and [ fig06 ] show the time evolution of the density , pressure , and 3-dimensional configuration of the selected magnetic lines of force in the @xmath46 plane ( @xmath47 , @xmath37 ) at @xmath48 . in the early stage ( @xmath49 ) , a strongly magnetized helical jet powered by taws advances with a constant velocity into the decreasing poloidally magnetized medium .
our mhd jets have a fast - mode mhd wave front , and a slow - mode mhd wave front closely beyond a contact surface . the front of the taw ( a fast - mode mhd wave ) begins to be decelerated due to the gradually decreasing @xmath50 ; the accumulation of @xmath51 occurs behind this wave front ( @xmath52 ) .
the toroidal magnetic pressure gradient force @xmath53 becomes large and works effectively at this front , strongly compressing the external medium along this propagating taw front .
the initially `` hourglass - shaped '' magnetic field becomes `` tucked up '' behind this wave front ( @xmath54 ) . at
later stages the front becomes super - fast magnetosonic and a strong bow shock ( fast - mode mhd shock ) is formed .
the accumulation of @xmath51 causes a concentration of the axial current density @xmath55 near the central ( @xmath25 ) axis ( figure [ fig07 ] _ left _ ) , and the magnetic pitch ( @xmath56 ) becomes large inside the jet .
the lorentz force breaks the quasi - equilibrium balance in the radial direction , and the jet is disrupted ( @xmath57 ) .
we have an asymmetric distribution of the current density and projected ( in a @xmath58 plane ) velocity field seen in figure [ fig07 ] _
right_. there is a strong correlation between the 3-dimensional configuration of the magnetic field and the distribution of specific power due to the radial ( @xmath59 ) component of the lorentz force : @xmath60 ( figure [ fig08 ] ) .
therefore it is confirmed that the distortion of our mhd jet is driven by the current - driven helical kink instability ( @xmath61 ) ( nakamura et al . 2001 ) .
using the numerical data of the 3-dimensional simulation explained in the previous section , we computed the distributions of frm , the projected magnetic field , and the total intensity seen from different viewing angles .
we computed the frm distribution by integrating @xmath1 along the line - of - sight although in paper i we integrated @xmath0 . in this paper , we intend to consider the effect of the deformed magnetic field configuration and to omit the effect of the dense shell from consideration . to calculate stokes parameters , we assume the following : ( 1)radiation processes are dominated by synchrotron radiation , ( 2)synchrotron self absorption is negligible , ( 3)the spectral index , @xmath62 , is equal to unity , and ( 4)the projected magnetic field is perpendicular to the projected electric field .
the emissivity of the synchrotron radiation is given by @xmath63 , where @xmath64 is the local magnetic field strength , @xmath65 is the angle between the local magnetic field and the line - of - sight , and @xmath66 is the gas pressure . in our simulation the relativistic particles are not explicitly tracked , therefore we assume that the energy and number densities of the relativistic particles are proportional to the energy and number densities of the thermal fluid ( clarke et al .
1989 , hardee & rosen 1999 , 2002 ) .
the total intensity is then given by the integration of the emissivity along the line - of - sight as @xmath67 .
other stokes parameters are given by @xmath68 and @xmath69 , where the local polarization angle @xmath70 is determined by the direction of the local magnetic field and the direction of the line - of - sight .
using these @xmath71 and @xmath72 , the polarization angle @xmath73 is given by @xmath74 . finally the projected magnetic field is determined from the polarization angle @xmath73 and the polarization intensity @xmath75 . here
we separate the faraday rotation screen and the emitting region , and we performed the integrations only in the emitting region for stokes parameters , and only in the faraday rotation screen for frm .
we assumed this separation on the basis of the fact that linear dependence of the observed polarization angle on wavelength - squared holds in some observations ( perley et al .
1984 , feretti et al . 1999 ,
asada et al .
2002 , gabuzda & murray 2003 ) ; this would not be the case if the faraday rotation is caused in the emitting region ( burn 1966 ) .
the emitting region and the faraday rotation screen are deformed into the 3-dimensional shape in the helical kink instability .
therefore we trace the flux tube in its time development by the initial footpoints of magnetic lines of force at the upper boundary of the simulation region .
this is reasonable because the footpoints do nt move until a wave front arrives .
the region which is defined as the faraday rotation screen is almost corresponding to the magnetic tube to which the circular motion is imposed .
figure [ fig09 ] shows the frm distributions at the time before the formation of the wiggled structure ( @xmath76 ) .
black in the map corresponds to a maximum and white to a minimum in each viewing angle case .
the frm distributions have the gradient in the direction perpendicular to the jet axis independent of the viewing angle .
this gradient is caused by the toroidal magnetic field propagating as taws .
figure [ fig10 ] shows the total intensity distributions . when the angle between the jet axis and the line - of - sight , @xmath77 , is equal to @xmath78 , total intensity has the highest value on the jet axis since both the emissivity and the integration depth of the emitter are maximum .
as the viewing angle becomes small , the asymmetry appears owing to the asymmetry of the emissivity , this is equivalent to the asymmetry of the magnetic field component perpendicular to the line - of - sight .
the jet has the helical magnetic field , therefore it is opposite in both sides of the jet whether the emissivity increases or decreases when the viewing angle changes .
figure [ fig11 ] shows the distributions of the projected magnetic field .
it is parallel to the jet axis in almost all region independent of the viewing angle because the pitch angle of the helical magnetic field is not so large .
however , the polarization intensity becomes lower as @xmath77 becomes small .
the frm distributions at the time after the formation of the wiggled structure ( @xmath79 ) are shown in figure [ fig12 ] .
the gradient across the jet axis is partly seen , but not seen in the wiggled structure because the toroidal component of the magnetic field is smaller compared to the field component along the structural axis ; i.e. , the magnetic field is less twisted around the wiggled structure . therefore frm is large in the structure coming toward us and small in the structure going backward .
this means that the frm distribution in the wiggled structure can strongly reflect its 3-dimensional shape .
frm changes suddenly where different parts of the wiggled jet overlap each other as seen by the observer .
figure [ fig13 ] shows the total intensity distributions . behind the fast - mode mhd shock front , toroidal magnetic field
is accumulated and magnetic field is strengthened .
it leads the emissivity to increase and the total intensity to become high . on the other hand ,
the intensity is not so high in the wiggled structure , where the magnetic field is less twisted around the wiggled structure .
however , as the viewing angle becomes small , the intensity in the wiggled structure increases because the structure which lies on this side and that on the back side overlap on the same ray .
this means that the integration depth of the emitter increases . for the same reason , the polarization intensity gets higher in the wiggled structure ( figure [ fig14 ] ) .
the projected magnetic field is along the structural axis of the jet on the whole . in the wiggled structure
, the projected field appears transverse to the global jet axis ( @xmath25-axis ) .
in this section , we discuss the comparison between the calculated distributions shown in the previous section and those in the 3c449 jet .
figure [ fig15 ] shows the distributions of frm and the projected magnetic field in part of the 3c449 jet ( feretti et al .
as we go along the jet , the color in the frm distribution starts from blue , changes to deep blue , jumps to yellow , and changes to red at an edge of the jet ( a bending point ) on the right hand side .
the parallel projected magnetic field exists widely where the color in the frm distribution is yellow or red .
figure [ fig16 ] shows the color frm distribution of the figure [ fig11](c ) and the projected magnetic field distribution ( @xmath77 is equal to @xmath80 ) in the wiggled structure .
the structure takes the form of a 3-dimensional helix .
the structure going backward corresponds to blue , while that coming toward us corresponds to red .
the structure going upward almost perpendicularly to the line - of - sight corresponds to the yellow part .
the projected magnetic field in the wiggled structure is almost parallel .
this is caused by the following ; the structure going from left to right on the back side and that from right to left on this side appear to overlap partly ( cf . figure [ fig17 ] ) . as stated in section 3
, we omitted the effect of the dense shell from consideration . here
we demonstrate the effect of the dense shell .
figure [ fig18 ] shows the frm distribution in the wiggled structure calculated by the integration of @xmath0 .
the characteristics of the distribution are same so that it can be said that the dense shell do not affect the distribution qualitatively so much in this case .
feretti et al .
( 1999 ) estimated the jet velocity and the inclination of the line - of - sight by the ratio of @xmath81 to @xmath82 , where @xmath81 is the really observed core radio power at 5ghz and @xmath82 is the power inferred from the total radio power at 408mhz . from such estimation , they concluded that an initial jet velocity is @xmath83 and a jet inclination to the line - of - sight is @xmath84 .
however , these values were obtained from the information about near the core .
it is likely that the jet is bent in some places .
birkinshaw , laing , & peacock ( 1981 ) estimated from the polarization intensity along the jet that the inclination of the north jet in 3c449 is @xmath85 .
this inclination is almost same with that at which we can reproduce the characteristics of the observation .
these similarities between the calculated distributions and the observed ones suggest that the 2-dimensional wiggle in the observation is due to the projection of a 3-dimensional helical structure onto the plane of the sky .
it is also suggested that the structure which lies on this side ( yellow ) hides the structure on the back side .
the reddish part corresponds to the part coming toward us in the helix . if this is the case , it is highly likely that the magnetic field ( especially the toroidal component ) plays an important role not only in the emission but also in the dynamics of the jet formation because the wiggle formation is caused by the accumulation of the toroidal field .
it is also suggested that the frm distribution can be a sensitive clue for telling which of the proposed models is the correct one . because it is not obvious that how much the faraday screen that is irrelevant to the jet itself contributes to the value of rotation measure , it is not possible to predict the value of faraday rotation measure .
it is , however , possible to estimate the difference between the maximum and the minimum of rotation measure . in our simulation ,
the scale of the wiggled structure is almost equal to the length unit of the simulation .
integration depth of faraday screen is about twice as long as the length unit .
the rotation measure is given by frm [ rad m@xmath86 = @xmath87 [ @xmath88 @xmath1 [ g ] @xmath89 [ pc ] , where @xmath2 is the electron density , @xmath1 is the line - of - sight component of the magnetic field and @xmath89 is the integration depth of the faraday screen . in our calculation , @xmath89 is twice as long as @xmath90 where @xmath90 is the scale of the wiggled structure . if the maximum of the rotation measure ( frm@xmath91 ) is made by @xmath92 and the minimum of the rotation measure ( frm@xmath93 ) by @xmath94 , then frm@xmath91 - frm@xmath93 @xmath95 . in the case of 3c449 jet ,
the scale of the wiggled structure is about @xmath96 ( from @xmath97 to @xmath98 in figure [ fig15 ] ) and it corresponds to 30kpc . finally frm@xmath91 - frm@xmath93 is equal to @xmath99 . in the case of 3c449 jet
, frm@xmath91 - frm@xmath93 is nearly equal to 100 [ rad m@xmath86 ( feretti et al . 1999 ) .
it is therefore predicted from our calculation that @xmath100 in the jet is equal to @xmath101 [ g @xmath88 .
except for the wiggled structure , there are some differences .
first , the total and polarized intensities enhance beyond the wiggled structure in the model counterparts while it is not seen in the observation .
the enhancement of the intensities in the model counterparts is due to the accumulation of the toroidal magnetic field .
this is equivalent to the enhancement of the current density .
it is therefore likely that the magnetic reconnection takes place .
the magnetic reconnection is the dissipation of the magnetic field so that the strength of the magnetic field decreases and the intensities can also decrease . for more exact discussion ,
resistive mhd simulations are necessary .
second , the projected magnetic field is longitudinal at the side near the core in the model counterparts while it is transverse in the observation .
it is remarkable that the transverse field was obtained in paper i in which the part of the jet near the core was dealt with .
we therefore think that the simulations simultaneously solving the formation , propagation and destabilization of the jet are necessary .
it can be said that the influence of the establishment of the validity of the magnetic model , especially the `` sweeping magnetic twist model '' , will be extremely large .
this claims that the dynamics of the jet formation , including the collimation and the destabilization in large scale , is due to the operation of the magnetic field .
since there is no local source for this large energy to do the job in the part of the space where the wiggles of the jets are seen ( closer to the hotspots at the tip of long jets ) , the energy should come all the way from the agn core .
we also claim that the energy dumped in the hotspots and the radio lobes is carried by the poynting flux . in mhd models ,
the toroidal component is generated and its propagation carries the poynting energy , therefore the models are plausible .
the generation of the toroidal magnetic field may be due to either the rotation of the accretion disk ( blandford & payne 1982 , uchida & shibata 1985 ) or the effect of the rotating black hole ( blandford & znajek 1977 , koide et al .
this means that the large amount of energy is flowing out from the core part of an agn ( though we can not yet say exactly whether that comes from the inner edge of the accretion disk , or from the black hole ) .
if we accept this , the major bearer of energy flowing out is , even in the agn core itself , large amplitude taws , and even very high lorentz factor phenomena will be a product of the energetic taws .
we performed a 3-dimensional mhd simulation based on our `` sweeping magnetic twist model '' , which was applied to the situation far from the agn core by nakamura et al .
( 2001 ) . in the `` sweeping magnetic twist model ''
, the disk rotation generates the toroidal magnetic field and it propagates into two directions along a large scale magnetic field as torsional alfvn waves ( taws ) . in this paper , the situation taws are propagating far from the gravitator was dealt with .
it was assumed that there is a lower alfvn velocity region ahead of the propagating taws .
the toroidal magnetic field becomes accumulated after taws enter the low alfvn velocity region .
this causes the growth of the current - driven helical kink instability and the wiggled structure is formed .
we also calculated the observational quantities ( frm , stokes , @xmath102 , @xmath72 , and @xmath71 parameters ) by integrating the numerical data along the line - of - sight .
the faraday rotation screen and the emitting region were defined separately .
the integration for frm was done only in the faraday rotation screen and that for stokes parameters only in the emitting region .
the projected magnetic field was determined from @xmath72 and @xmath71 . before the formation of the wiggled structure
, the frm distribution has a gradient across the jet axis .
an asymmetry in the total intensity distribution exists except the case where @xmath77 is equal to @xmath78 .
same features are seen in the result of paper i and these features can explain some observations ( perley et al .
1984 , asada et al .
2002 , gabuzda & murray 2003 ) .
the projected magnetic field is parallel to the jet axis because the pitch angle of the helical magnetic field is not so large .
after the formation of the wiggled structure , the frm distribution has a clear correlation with the large scale structure of the jet itself .
this is caused by the following ; the magnetic field becomes less twisted around the wiggled structure owing to the current - driven helical kink instability .
this is equivalent to that the magnetic field is almost along the structural axis of the jet in the wiggled structure .
the total and polarization intensities are low in the wiggled structure when the jet inclination to the line - of - sight is large .
however , they become higher as the jet inclination becomes small because the structure on the backside and that on the frontside of the 3-dimensional helix overlap on the same ray as seen by observer .
we found that when we see the jet at a certain angle , we can reproduce the characteristics of the observation of the 3c449 jet ( feretti et al .
this suggests that the frm distribution could be strongly affected by the magnetic field in or around the jet and that the jet has a helical structure .
if this is the case , it is also suggested that the magnetic field ( especially the toroidal component ) plays an important role in the formation of astrophysical jets .
finally , we would like to thank dr .
l. feretti for providing their precious data of frm , and for valuable discussions .
one of the authors ( h. k. ) thanks syuniti tanuma and troels haugblle for helpful comments .
we would hope that more observations of the frm distribution in agn jets will be performed in the near future because they are very important and scientifically rewarding .
this means that with those we can determine the correct model if our claim in the above is correct .
we would therefore urge high quality observations of the frm distribution , to seek for the clues of the mechanism of jet formation .
numerical computations were carried out on vpp5000 at the astronomical data analysis center of the national astronomical observatory , japan , which is an interuniversity research institute of astronomy operated by the ministry of education , culture , sports , science , and technology .
9999 asada , k. , inoue , m. , uchida , y. , kameno , s. , fujisawa , k. , iguchi , s. , & mutoh , m. 2002 , , 54 , l39 bateman , g. 1980 , mhd instabilities ( cambridge : mit press ) baum , s. a. , et al .
1997 , , 483 , 178 begelman , m. c. , blandford , r. d. , & rees , m. j. 1980 , nature , 287 , 307 birkinshaw , m. , laing , r. a. , & peacock , j. a. 1981 , , 197 , 253 birkinshaw , m. 1991 , in beams and jets in astrophysics , ed .
a. hughes ( cambridge , uk : cambridge univ . press ) , 278 blandford , r. d. 1976 , , 176 , 465 blandford , r. d. , & payne , d. g. 1982 , , 199 , 883 blandford , r. d. , & znajek , r. l. 1977 , , 179 , 433 burn , b. j. 1966 , , 133 , 67 clarke , d. a. , norman , m. l. , & burns , j. o. 1989 , , 342 , 700 eilek , j. a. , & owen , f. n. 2002 , , 567 , 202 feretti , l. , perley , r. , giovannini , g. , & andernach , h. 1999 , , 341 , 29 ferrari , a. 1998 , ara&a , 36 , 539 gabuzda , d. c. , & murray 2003 , preprint ( astro - ph/0309668 ) hardee , p. e. , & rosen , a. 1999 , , 524 , 650 hardee , p. e. , & rosen , a. 2002 , , 576 , 204 hummel , c. a. , et al .
1992 , , 257 , 489 koide , s. , meier , d. l. , shibata , k. , & kudoh , t. 2000 , , 536 , 668 kudoh , t. , matsumoto , r. , & shibata , k. 1998 , , 508 , 186 laing , r. a. 1981 , , 248 , 87 lovelace , r. v. e. 1976 , nature , 262 , 649 matsumoto , r. , uchida , y. , hirose , s. , shibata , k. , hayashi , m. , ferrari , a. , bodo , g. , & norman , c. 1996 , , 461 , 115 meier , d. l. , koide , s. , & uchida , y. 2001 , science , 291 , 84 nakamura , m. , uchida , y. , & hirose , s. 2001 , new astronomy , 6 , 61 ouyed , r. , & pudritz , r. e. 1997 , , 482 , 712 perley , r. a. , bridle , a. h. , & willis , a. g. 1984 , , 54 , 291 shibata , k. , & uchida , y. 1986 , , 38 , 631 stone , j. m. , & norman , m. l. 1994 , , 433 , 746 taylor , g. b. 1998 , , 506 , 637 todo , y. , uchida , y. , sato , t. , & rosner , r. 1993 , , 403 , 164 uchida , y. , & shibata , k. 1985 , , 37 , 515 uchida , y. , & shibata , k. 1986 , can.j.phys . , 64 , 507 uchida , y. , nakamura , m. , hirose , s. , & uemura , s. 1999 , , 264 , 195 uchida , y. , nakamura , m. , miyagoshi , t. , kobayashi , t. , mukawa , t. , & hirose , s. 2000 , in iau symp .
195 , highly energetic physical processes and mechanisms for emission from astrophysical plasmas , ed .
martens , & s. tsuruta ( san francisco : asp ) , 213 uchida , y. , kigure , h. , hirose , s. , nakamura , m. , & cameron , r. 2004 , , 600 , 88 ustyugova , g. v. , koldoba , a. v. , romanova , m. m. , chechetkin , v. m. , & lovelace , r. v. e. 1995 , , 439 , l39
|
distributions of faraday rotation measure ( frm ) and the projected magnetic field derived by a 3-dimensional simulation of mhd jets are investigated based on our `` sweeping magnetic twist model '' .
frm and stokes parameters were calculated to be compared with radio observations of large scale wiggled agn jets on kpc scales .
we propose that the frm distribution can be used to discuss the 3-dimensional structure of magnetic field around jets and the validity of existing theoretical models , together with the projected magnetic field derived from stokes parameters . in the previous paper , we investigated the basic straight part of agn jets by using the result of a 2-dimensional axisymmetric simulation .
the derived frm distribution has a general tendency to have a gradient across the jet axis , which is due to the toroidal component of the magnetic field generated by the rotation of the accretion disk . in this paper
, we consider the wiggled structure of the agn jets by using the result of a 3-dimensional simulation .
our numerical results show that the distributions of frm and the projected magnetic field have a clear correlation with the large scale structure of the jet itself , namely , 3-dimensional helix .
distributions , seeing the jet from a certain direction , show a good matching with those in a part of 3c449 jet .
this suggests that the jet has a helical structure and that the magnetic field ( especially the toroidal component ) plays an important role in the dynamics of the wiggle formation because it is due to a current - driven helical kink instability in our model .
|
the rotation of rigid bodies often displays interesting instability problems .
bodies with three different inertia moments are well known to have unstable rotation about the intermediate axis @xcite , as commonly observed from acrobatic jumps or dives .
throwing a tennis racket provides an easy illustration of this instability of the free rotation .
such purely inertial instability has no threshold , _
i.e. _ it can be observed for very low rotation rates .
in contrast , for axisymmetric bodies such as hoops , disks , rockets , eggs etc . , the free rotations remain stable , but different mechanisms may also lead to instability when external forces are present . under some circumstances , spectacular and unexpected instabilities may originate from frictional forces .
this is the case for the tippe - top , a popular toy that flips over and rotates on its stem , or for the hard - boiled egg problem , which has recently received a nice analysis @xcite .
more classically , the competition between gravity and centrifugal forces may lead to an instability with a finite rotation rate threshold @xcite , as illustrated by the simple experiment described in this paper .
a hoop is hung on a long wire , whose upper end is spun . at low rotation rate ,
the hoop is vertical and simply spins about its diameter .
increasing the rotation rate , the hoop progressively rises and becomes horizontal , spinning about its symmetry axis .
this situation may appear paradoxical , since the horizontal position maximizes both kinetic and potential energy .
it is similar to the conical pendulum problem @xcite , commonly illustrated in the classical demonstration experiment of the bead sliding along a vertically rotating hoop @xcite .
the popular ` hula hoop ' game , where the wire rotation is replaced by the hips oscillations of the player , is a common illustration of this phenomenon .
once the hoop spins horizontally , its rotation is maintained by a parametric oscillation mechanism @xcite . in this paper ,
the equation of motion is derived from the lagrange s equation .
an alternate derivation , from the angular momentum equation , is also presented as a good illustration of dynamics of a rigid body .
the hoop is shown to rise following a supercritical bifurcation for the wire rotation rate @xmath2 above a critical value @xmath3 , where @xmath4 is the gravitational acceleration and @xmath5 the hoop radius .
the stable solution @xmath6 coincides with that of the conical pendulum or the bead - on - a - hoop .
the nonlinear oscillations are briefly described by means of phase portraits , and compared to that of the bead - on - a - hoop .
simple considerations from wire torsion and air friction allow us to estimate the startup and damping timescales . finally some experimental measurements are reported , and are shown to compare well with the exact solution . with the vertical.,width=124 ] in this experiment , the angular rotation threshold being of order of a few rad / s , the wire can be simply spun with the fingers .
since it only requires a wire and a hoop , this experiment can be conveniently used as a classroom illustration of spontaneous symmetry breaking and bifurcation .
although the detail of the calculation is somewhat more subtle than that of the classical bead - on - a - hoop problem , since it deals with rigid body dynamics , the physics is basically the same and does not require the much heavier apparatus of the bead - on - a - hoop demonstration .
we consider a thin uniform hoop , of radius @xmath5 and mass @xmath7 , fixed at a point p of its periphery by a massless wire of length @xmath8 ( see figure [ fig : sketch ] ) .
the wire is spun from its other end o at a constant angular velocity @xmath2 .
the center of mass g is assumed to remain at the vertical of o. let us first consider a torsionless wire , so that the hoop angular velocity @xmath9 about the vertical axis @xmath10 instantaneously follows the imposed angular velocity @xmath2 .
let us consider the rotating frame of reference of the hoop @xmath11 .
we note @xmath12 the pitch angle between the hoop plane and the vertical axis @xmath10 .
we consider for simplicity a zero roll angle about @xmath13 ( _ i.e. _ we assume that the point p remains the highest point of the hoop ) , so that @xmath12 is the only degree of freedom for this problem .
the absolute angular velocity @xmath14 of the hoop has two contributions . the first one , of magnitude @xmath2 ,
is imposed from o through the torsionless wire , and is about @xmath15 ( the points o and g are at rest , so that og is the instantaneous axis of rotation when @xmath12 is kept constant ) .
the second one comes from the pitch variation @xmath16 , and is about the axis @xmath17 . we therefore obtain @xmath18 the potential energy is @xmath19 , where @xmath20 is the center of mass elevation to first order in @xmath21 .
the kinetic energy @xmath22 has two contributions : one from the rotation @xmath23 , where @xmath24 is the inertia matrix of the hoop relative to g , and one from the vertical translation of the center of mass , @xmath25 . in the reference frame of the hoop ,
the inertia matrix @xmath24 has diagonal components @xmath26 and @xmath27 .
the lagrangian function @xmath28 then finally writes @xmath29 + m g r \cos \theta.\ ] ] writing the lagrange s equation for the coordinate @xmath12 , we end up with the differential equation of motion @xmath30 where @xmath31 .
this equation of motion can also be obtained from the angular momentum equation and newton s law . in the rotating frame of reference
, the angular momentum equation reads @xmath32 where @xmath33 is the angular momentum and @xmath34 is the torque of the wire tension relative to the center of mass ( the gravitational torque vanishes ) .
the wire tension is obtained from newton s law , @xmath35 to first order in @xmath21 , @xmath36 is vertical , so its torque is simply @xmath37 . using @xmath38 , we obtain @xmath39 replacing into ( [ eq : eulv ] ) and projecting on @xmath40 , we finally recover the equation of motion ( [ eq : edc ] ) .
we are interested in the steady states and the natural frequency for small oscillations about them . to first order in @xmath12 , equation ( [ eq : edc ] )
reduces to @xmath41 it is worth noting that this linearized equation is exactly the same as the one from the bead - in - a - hoop problem @xcite .
the difference between equations ( [ eq : edc ] ) and ( [ eq : edl ] ) originates from the kinetic energy of the center of mass translation , which is not present in the bead - on - a - hoop .
in addition to the trivial solution @xmath42 , equation ( [ eq : edl ] ) has a non trivial solution for @xmath43 , @xmath44 following the usual terminology , the pitch angle @xmath45 is the order parameter , and we introduce the reduced angular velocity @xmath46 as the control parameter .
the linearization of ( [ eq : sol ] ) finally leads to the classical form of a supercritical pitchfork bifurcation for @xmath47 : @xmath48 stability and natural frequency for small oscillations are obtained by introducing in equation ( [ eq : edl ] ) a small perturbation in the form @xmath49 where @xmath45 stands for the trivial or non - trivial solution .
the trivial solution is stable for @xmath50 ( @xmath51 ) , with a natural frequency @xmath52 , and unstable for @xmath53 , with a growth rate given by @xmath54 ( and @xmath55 ) .
the non - trivial solution ( [ eq : sol ] ) is found to be always stable , with the same natural frequency . in terms of @xmath56
, one can see that the period of the oscillations , @xmath57 diverges as one approaches the transition from both side .
this critical slowing down is a usual signature of supercritical bifurcation@xcite . for the hula hoop ( solid lines )
compared to the bead - on - a - hoop system ( dashed lines ) .
( @xmath58 ) : @xmath59 .
( @xmath60 ) : @xmath61.,width=283 ] when the oscillation amplitudes about the steady states are not small , the nonlinear terms in ( [ eq : edc ] ) become important and may affect the dynamics of the system . the orbits in the phase space
@xmath62 then provide a useful tool to characterize the nonlinear dynamics of the hula hoop , and to compare it with the one of the bead - on - a - hoop ( [ eq : edl ] ) .
the equation of motion can be integrated using the painlev invariant , @xmath63 which takes a constant value along an orbit ( this invariant corresponds to the total energy @xmath64 in the case of a conservative system ) .
figure [ fig : orbits ] shows isolines of @xmath65 for two forcing frequencies , below ( @xmath66 , fig .
@xmath58 ) and above ( @xmath67 , fig .
@xmath60 ) the transition , together with the orbits of the bead - on - a - hoop problem ( dashed lines ) .
one can clearly see from these phase portraits the trivial solution @xmath42 for @xmath66 and the two non trivial solutions ( [ eq : sol ] ) separated by a saddle point at @xmath68 for @xmath67 .
note that we are only interested in the closed orbits in the domain @xmath69 : orbits that cross @xmath70 are not consistent with the assumption that the point p remains the highest point of the hoop . around the stable fixed points , both
the two systems show nearly elliptic orbits , as expected from small harmonic oscillations .
larger oscillations of the hula hoop show orbits with sharper corners around @xmath71 , associated to larger angular velocities @xmath72 .
this discrepancy originates from the vertical translation of the center of mass , which is responsible for an additional pitch angle acceleration when the hoop is nearly vertical .
two timescales are relevant for a practical experiment : the startup timescale of rotation @xmath73 and the damping timescale of oscillations @xmath74 .
the startup timescale may be obtained considering the wire torsion .
we start from an initially vertical hoop ( @xmath68 ) , and let now its rotation angle @xmath75 about @xmath76 be free .
the wire communicates a torque @xmath77 to the hoop , where @xmath78 is the torsion constant and @xmath79 the wire torsion .
the angular momentum equation about @xmath10 applied to the hoop then reads @xmath80 leading to a startup timescale @xmath81 ( which corresponds also to the period of the wire torsion oscillations if no damping were present ) .
the damping timescale @xmath74 is relevant both for the @xmath75 oscillations due to the wire torsion , and for the @xmath12 oscillations around the equilibrium states .
it can be obtained considering the dissipation with the surrounding air . with typical velocity of order m / s and hoop thickness @xmath82 of a few millimeters
, we can make use of a turbulent estimate for the drag force . neglecting @xmath16 compared to @xmath9 , a unit length @xmath83 of the hoop experiences a drag force @xmath84 , where @xmath85 is the air density .
integrating over the hoop perimeter leads to the frictional torque @xmath86 the angular momentum equation about @xmath10 reads @xmath87 where @xmath88 is the hoop density .
taking @xmath89 and @xmath90 , the damping timescale finally writes @xmath91 for values of practical interest , @xmath74 is of order of a few tens of the rotation period @xmath92 .
note here that in the case of the real hula hoop game , the rolling friction on the player hips is fortunately dominant , leading to a damping timescale of order @xmath92 .
( d ) @xmath93.,width=294 ] an experiment has been carried out using a wood hoop ( @xmath94 g.@xmath95 ) of mass @xmath96 g , radius @xmath97 mm and section @xmath98 mm@xmath99 .
the expected frequency threshold is then @xmath100 hz ( classical hula hoops have @xmath101 hz ) .
a simple cotton thread , of length @xmath102 m and mass less than 3% of the hoop mass , was fixed to a constant current motor .
the startup timescale for this wire can be estimated from the free oscillation period , @xmath103 s ( corresponding to a torsion constant @xmath104 n.m ) .
this wire is far from being torsionless , and during the early stage of the rotation , torsional energy is stored into the wire .
the wire progressively communicates rotation to the hoop , and the fluctations ( due to a varying imposed rotation rate at its upper end , or to pitch angle variation at the other end ) are smoothed down on a timescale @xmath74
. seen from the side , the hoop rotation appears as lissajous ellipses , whose principal axis makes an angle @xmath12 with the vertical axis .
pictures acquired from a simple ccd camera ( see figure [ fig : pic_hula ] ) allows us to easily measure the pitch angle @xmath12 within 1@xmath105 from these ellipses .
the time aperture of the camera may blur the picture , but facilitate in some case the angle measurement ( see for instance picture c ) . as a function of the normalized angular velocity @xmath106 .
the solid line is the calculated solution ( [ eq : sol ] ) , with @xmath0 adjusted to fit the experimental data.,width=317 ] the measured pitch angles @xmath12 , shown in figure [ fig : avsw ] , are in excellent agreement with the exact solution ( [ eq : sol ] ) , except for low rotation rate , where non zero angles are measured below the expected transition .
the experimental frequency threshold , obtained by extrapolating the curve down to @xmath71 , is @xmath107 hz , which agrees within 5% with the theoretical value .
the discrepancy at low rotation rate is typical of an imperfect bifurcation , where small asymmetry in the apparatus ( such as the position of the knot ) slightly anticipates the destabilization of the basic state .
an interesting observation is that , for @xmath108 ( see the dashed line in figure [ fig : avsw ] ) , the bifurcated state is no stable any more : a secondary instability appears , in the form of a slow precession of the center of mass , with a period of about ten rotation periods .
this new behavior is probably an effect of the non zero mass of the wire , and can clearly not be described in our calculation , where the hoop center of mass is constrained to remain aligned with the vertical axis .
is increased from 0 to 5.5 hz.,width=283 ] transient phenomena may also be investigated , by means of usual video processing .
an illustration is given in figure [ fig : spatio ] , showing a spatio - temporal diagram obtained by collecting the light intensity recorded on a horizontal line passing through the wire .
in this example , the imposed frequency @xmath109 has been suddenly increased from 0 to 5.5 hz @xmath110 ( first arrow ) .
one can see that , after a transient time of around 7 s ( second arrow ) during which the wire rotation propagates down to the hoop , the amplitude increases and saturates to a finite value .
( left ) and rotation frequency @xmath111 ( right ) as measured from figure [ fig : spatio ] .
the origin time @xmath112 corresponds to the second arrow in that figure.,width=302 ] from this diagram , the instantaneous pitch angle as well as the instantaneous oscillation frequency @xmath111 may be extracted , as shown in figure [ fig : envel ] .
the frequency is obtained from averaging over 6 successive oscillations . due to the progressive torsion of the wire
, this frequency slowly approaches its imposed value 5.5 hz . as a consequence , the increase of the pitch angle towards its stationary value
@xmath113 rad is rather slow .
it is interesting to note that , when plotting the instantaneous pitch angle as a function of the instantaneous frequency ( see figure [ fig : bifinst ] ) , the bifurcation diagram of figure [ fig : avsw ] is recovered to a high degree of accuracy .
this suggests that the pitch angle follows ` adiabatically ' the instantaneous frequency , so that the apparent growth rate is essentially controlled by the wire torsion more than the intrinsic dynamics of the instability ( at least far from the transition ) . as a function of the instantaneous rotation frequency @xmath114 ( same data as in figure [ fig : envel ] ) .
the solid line is the calculated solution ( [ eq : sol ] ) , as in figure [ fig : avsw ] .
the cross indicates the stationary solution @xmath115 rad for @xmath116.,width=317 ]
this simple experiment provides an interesting alternative to the bead - on - a - hoop @xcite experiment as a mechanical analog to bifurcation and second - order phase transition in physics . the poor attention paid to
the choice of the material and the experimental conditions illustrates the robustness of the phenomenon , and makes this experiment easy to work out by undergraduate students .
further experiments can be performed , _
e.g. _ studying the transient phenomena resulting from a small perturbation . as a suggestion , restrain the point p on the vertical axis when @xmath117 by means of a small hook around the wire .
releasing the hook allows us to measure the growth rate and to characterize its divergence as @xmath118 .
this method can be hardly achieved with a bead - on - a - hoop apparatus due to its inherent difficult access .
another motion may compete with the bifurcation described in this paper : the hoop can rotate as a whole about the vertical axis , its center of mass remaining aligned with the wire .
this is the usual motion for the conical pendulum , which also leads to a supercritical bifurcation with a threshold simply given by the natural frequency @xmath119 .
the conical pendulum motion overcomes the hula hoop motion for @xmath120 of order of @xmath0 , _
i.e. _ for a wire length of order of the hoop radius .
moving away the center of mass from the vertical axis may allow us to observe the competition between the two regimes , even for a longer wire .
similar bifurcations as the result of a competition between centrifugal force and gravity are present in a number of situations .
spinning plates provides an interesting illustration : as for the real hula hoop , the motion here is forced by the precession of the rigid rod rather than its rotation .
the imposed precession frequency has to overcome the natural frequency for the plate to stand up , and then the horizontal state is maintained by a parametric oscillation mechanism @xcite .
lasso roping is another example , where the rigid hoop is replaced by a deformable loop . here
again the motion is maintained by the precession of the spoke rather than rotation . as a consequence ,
the knot on the loop makes the spoke to rotate as well , so that the cord has to be continuously untwisted at its other end .
it is worth pointing out that this experiment can be carried out with any rigid body , not necessarily axisymmetric . in this case , the angle of the equilibrium state ( [ eq : sol ] ) remains unchanged , and the angular velocity threshold just becomes @xmath121 , where @xmath122 is the distance between the center of mass and the knot .
note however that , for bodies with three different inertia moments , in addition to the bifurcation described here , inertial instabilities may also occur .
such instability involves the roll angle about @xmath13 as an additional degree of freedom , and is not described by our calculation .
|
the motion of a hoop hung on a spinning wire provides an illustrative and pedagogical example of a supercritical bifurcation . above a certain angular velocity threshold @xmath0 ,
the hoop rises , making an angle @xmath1 with the vertical .
the equation of motion is derived in the limit of a long massless wire , and the calculated steady states are compared to experimental measurements .
this simple experiment is suitable for classroom demonstration , and provides an interesting alternative to the classical experiment of the bead sliding on a rotation hoop .
|
observations of protostellar systems and their prominent jets and outflows suggest that accretion disks start to form very early during the main accretion phase .
these disks live much longer than the central protostar needs to build up most of its mass . when the initial protostellar core is dispersed by accretion and outflows , the central star still accretes matter at low rates from the surrounding disk . typical disk life
times around low - mass stars were shown to be at least @xmath4yrs and their masses ( typically 0.01 - 0.1m@xmath1 ) do not seem to decrease considerably during this time @xcite .
theoretical and laboratory studies show that the timescale for grain growth and planetesimal formation is shorter than the typical disk life time .
such disks should soon evolve into protoplanetary disks .
although the direct detection of planetary - mass bodies in such disks will be extremely difficult , indirect effects of larger bodies , such as a change in the dust opacity spectrum , should be observable with current or near - future techniques .
cb26 ( l1439 ) is a small ( @xmath5pc ) , slightly cometary - shaped bok globule at a distance of @xmath6pc @xcite .
located at the south - east rim of the globule is a small bipolar near - infrared ( nir ) nebula .
the central star is not visible , even at 2.2@xmath7 m .
the spectral energy distribution together with the bolometric luminosity of @xmath8l@xmath1 suggest the presence of a low - mass class i yso in cb26 .
we observed strong submm / mm dust continuum emission from this source showing a thin extended envelope with an unresolved condensation at 10 resolution @xcite , @xcite , @xcite .
no molecular outflow has been detected yet .
* a ) * k - band image of the cb26 nir nebula with polarization vectors superimposed .
the white star marks the center of the polarization pattern , i.e. , the location of the illuminating source . *
b ) * grey - scale b ) .
very high polarization degrees were detected in the lobes , presumably caused by single scattering at small dust grains .
the orientation of the polarization vectors corresponds to a system consisting of a young star surrounded by both a circumstellar disk and a thin envelope .
the polarization pattern indicates that the disk is seen almost edge - on causing the band of enhanced extinction in between the scattering lobes . at the very center ,
the polarization vectors are aligned linearly .
this could be either due to photons scattered back from the envelope onto the disk or because of multiple scattering in the outer disk regions .
the location of the illuminating source was derived to an accuracy of 03 by minimizing the mean square scalar product between polarization vectors in the lobes which are probably due to single scattering and their corresponding normalized radius vectors .
the central source is located behind the extinction lane , i.e. , at the center of the disk ( fig . [ fig_jmh]a ) .
dust continuum emission from the cb26 disk : ovro results ( from @xcite ) . a ) and c ) show the 3 and 1 mm maps synthesized from all uv data .
b ) and d ) show maps derived from long @xmath9 spacings only . the synthesized beams ( fwhm )
are shown as grey ellipses at the lower left corners .
contour levels are in steps of ( -2,2,4 to 22 by 3 ) times 1@xmath10rms .
a ) @xmath11mjy/ 12@xmath12084 beam .
b ) @xmath13mjy/ 10@xmath12063 beam .
c ) @xmath14mjy/ 058@xmath12039 beam .
d ) @xmath15mjy/ 050@xmath12032 beam . ]
cb26 was observed with the owens valley radio observatory ( ovro ) millimeter - wave array during 2000 .
we obtained wide - band continuum maps at 1.3 and 2.7 mm and a @xmath16co(10 ) map with a spectral resolution of 0.17km / s .
the angular resolution of the maps is @xmath312@xmath1208 at 2.7 mm and @xmath305@xmath1203 at 1.3 mm , respectively .
the observations are described in more detail in @xcite .
the dust continuum maps show an elongated source with a position angle of @xmath17 ( from n to e ; fig .
[ fig_mmcres ] ) .
the position and morphology of this source suggest that the emission originates from the circumstellar disk implied from the nir data @xcite .
figure [ fig_jmh]b shows that the dust continuum emission matches well the extinction lane at the center of the nir reflection nebula . at 3 mm
we derive a projected fwhm source size of @xmath18au@xmath12@xmath1960au with no signature of an extended envelope .
the 1 mm images show a narrow lane and a small envelope extending perpendicular to the disk .
the disk height remains unresolved at even the highest angular resolution of 03 , confirming that it must be seen almost edge - on .
the projected fwhm size of the 1.3 mm disk is @xmath20au@xmath12@xmath2120au ( @xmath22 of the hpbw in that direction ) .
the actual scale height of the inner disk where the mm emission arises is probably much smaller .
the disk has a symmetric 20 warp outside @xmath23100au and is traced out to @xmath3200au .
the total flux densities derived from the ovro maps are @xmath24mjy and @xmath25mjy . at 1.3 mm , the iram 30-m single - dish flux of the unresolved component is completely recovered ( 160mjy ; @xcite ) . a simple decomposition yields 1.3 mm flux densities for the disk and envelope of @xmath26mjy and @xmath27mjy , respectively .
assuming @xmath28k , @xmath29@xmath30g@xmath31 of dust and @xmath32 we derive an envelope mass of @xmath33m@xmath1 .
the total mass of the more extended envelope seen in the single - dish maps is @xmath34m@xmath1 ( @xcite , @xcite ) .
the disk appears to be considerably smaller in the 3 mm dust emission than at 1 mm . for the central @xmath35au
we derive a 13 mm spectral index @xmath36 ( envelope subtracted ) .
further out @xmath37 increases to reach 3.5 - 4 at @xmath38au . at larger radii
the 3 mm emission is not longer traced .
this implies that most of the dust mass in the outer warped part of the disk is contained in classical , @xmath7m mm size grains with a spectral index of the dust opacity of @xmath39 .
the mm dust emission from the inner @xmath3200au of the disk is highly optically thick .
therefore , no constraints can be made on @xmath40 without better constraints on the temperature distribution and extensive modeling .
assuming that the 3 mm emission is mostly optically thin , a lower limit to the disk mass can be derived . adopting @xmath41@xmath30g@xmath31 of ism , @xmath42 , @xmath43k at @xmath44au , @xmath45 , and surface density @xmath46 as typical disk parameters , we derive @xmath47m@xmath1 , a value which is typical for disks around t tauri stars @xcite .
@xmath16co(10 ) emission from the cb26 disk : ovro results .
a ) grey - scale : integrated intensity of the @xmath16co(10 ) emission .
the synthesized beam ( fwhm ) is shown as grey ellipse in the lower right corner .
overlayed are the contours of the 1.3 mm dust continuum emission from the disk .
the image is rotated by 30 .
b ) position - velocity diagram of @xmath16co along the plane of the disk ( dashed line in a ) .
thick grey contours show the observed velocity field ( 27 , 45 , 63 , 81 , 99% of max ) .
dashed contours show the modeled emission from a keplerian disk around a 0.35m@xmath1 star . ]
strong @xmath16co(10 ) emission ( @xmath48k ) was detected from the outer parts of the cb26 disk , but there seems to be a lack of emission from the central part ( fig .
[ fig_cores]a ) .
the line is double - peaked with the blue part coming from the n - e side and the red part from the s - w side of the disk .
emission at the systemic velocity of the envelope @xmath49kms@xmath31 ( @xmath500.6kms@xmath31 ) is self - absorbed and/or resolved out .
although the current spectral data cover only a small velocity range ( 3@xmath21@xmath51@xmath52km / s ) , they could be well - modeled by a keplerian disk rotating around a ( [email protected])m@xmath1 star ( fig .
[ fig_cores]b ) .
the high - velocity line wings from the inner parts of the disk expected for a keplerian disk are not recovered in the data because of the small band width .
there is some forbidden red - shifted emission from the blue side of the disk which may be due to infall or outflow .
a more detailed study of the kinematics of this disk - envelope system is subject to follow - up observations .
the cb26 disk compared to the solar system . the grey scale image and contours show the 1.3 mm dust continuum emission from the cb26 disk as observed with ovro . for comparison ,
the solar system is represented by the orbits of jupiter and neptune and the kuiper belt ( 4070au ) . ] from ovro millimeter interferometric observations we discovered and resolved the thermal dust continuum and @xmath16co line emission from the circumstellar disk in cb26 which was recently implied by high - resolution nir polarimetric observations .
the disk is seen edge - on as predicted by the nir observations and matches well the extinction lane at the center of the bipolar nir reflection nebula .
it has a fwhm diameter of ( 220@xmath5320)au and a maximum ( traced ) diameter of @xmath3400au .
the ( projected ) scale height remains unresolved ( @xmath5420au ) .
for the first time , we directly detect a warp in a young circumstellar disk .
the warp affects the disk outside @xmath55au . like in many extra - galactic disks , such a warp could be caused by a small external disturbance like , e.g. , an encounter with a nearby star .
however , the actual cause of the warp is not known .
the 1.3 mm dust emission from the inner @xmath56au of the disk is highly optically thick , and only a lower limit to the total disk mass @xmath47m@xmath1 can be derived from the 3 mm emission . due to the high optical depths , no constraints on the dust opacity spectral index @xmath40 and possible planetesimal growth in the inner undisturbed part of the disk can be drawn .
the dust emission from the outer parts of the disk is optically thin and arises from small classical grains .
however , since the outer disk is obviously disturbed as indicated by the warp and has much lower particle densities , this does not exclude that particle growth in the inner disk has already taken place .
the small disk envelope , which extends above and below the plane of the disk and may be related to an yet undetected outflow or a disk wind , contains @xmath57m@xmath1 .
further @xmath34m@xmath1 are contained in a more extended envelope ( @xmath58au in diameter ) which extends mainly to the north - east and may represent the remnant , nearly dispersed protostellar core from which this system has formed @xcite .
strong @xmath16co emission shows that the disk must be still gas rich , that it is rotating , and has a kinetic temperature of order 30k outside @xmath56au .
the rotation curve is consistent with keplerian rotation around a ( [email protected])m@xmath1 star . with a total luminosity of @xmath8l@xmath1 , its large disk and its close connection with the parental cloud core ,
the cb26 system resembles an intermediate m type ttauri star of age @xmath59yrs .
the lack of a dense , centrally peaked cloud core with spectroscopic infall signatures as well as of a prominent molecular outflow indicates that the system has already passed its main accretion phase .
therefore , we conclude that this is a young protoplanetary disk . if compared to the solar system today , the cb26 disk is @xmath60 times larger than the outer kuiper belt ( fig . [ fig_solsys ] ) .
however , the optically thick and undisturbed part inside the warp has approximately the size of the kuiper belt . the total mass in the cb26 disk ( if all the gas is still there ) is at least 10 times higher than the planetary mass in the solar system .
however , the dust mass in the cb26 disk could be comparable or lower than the rocky mass in the solar system .
these numbers depend critically on the dust properties which are not well - known .
altogether , the cb26 system appears like a 23 times lower - mass equivalent to the early solar nebula at the verge of forming planetesimals and probably planets .
beckwith , th .
henning , y. nakagawa : ` dust properties and assembly of large particles in protoplanetary disks ' . in : _ protostars and planets
iv _ , ed . by v. mannings , a.p .
boss , s.s .
russell ( univ . of arizona press 2000 )
|
we present new sub - arcsecond - resolution near - infrared polarimetric imaging and millimetre interferometry data on the circumstellar disk system in the bok globule cb26 .
the data imply the presence of a @xmath0m@xmath1 edge - on - seen disk of @xmath2au in diameter , being in keplerian rotation around a young @xmath30.35m@xmath1 star .
the mm dust emission from the inner 200au is highly optically thick , but the outer parts are optically thin and made of small dust grains .
planetesimal growth in the inner disk could neither be comfirmed nor excluded .
we argue that the cb26 disk is a very young protoplanetary disk and show that it is comparable to the early solar system .
|
several cobalt - based low - dimensional compounds and rare - earth - based pyrochlore oxides exhibit fascinating magnetic properties such as magnetic field ( @xmath9 ) induced spin order - disorder transition @xcite , 1/3 magnetization plateau in the @xmath12(@xmath9 ) curve @xcite , quantum phase transition @xcite and spin - structural change @xcite .
geometrical frustration due to the triangular or tetrahedral arrangement of the magnetic moments , bond frustration as a result of competing ferromagnetic ( fm ) and antiferromagnetic ( afm ) exchange interactions and large single ion anisotropy are the fundamental ingredients that eventually determine the complexity of the magnetic ground state and hence the new functionalities in these compounds . often , the ground state of the frustrated materials is extremely sensitive to external perturbations such as magnetic field .
though there are a number of systems showing such kind of magnetic ground state , the underlying mechanism responsible for the above features has not been well understood yet .
+ recently , the quasi - one - dimensional ( 1d ) spin - chain cov@xmath3o@xmath4 has received attention to the scientific community due to the observation of 1/3 magnetization plateau similar to that observed in a regular triangular lattice @xcite . in monoclinic @xmath2-cov@xmath3o@xmath4 and triclinic @xmath13-cov@xmath3o@xmath4 , the edge - shared coo@xmath4 octahedra form a magnetic chain along the @xmath14 axis , and the edge - shared vo@xmath15 square pyramids are located in between the magnetic chains . the much larger co - co interchain distance as compared to the intrachain co - co distance and the presence of the nonmagnetic v@xmath16 ion in between the chains weaken the interchain magnetic coupling considerably .
both @xmath2-cov@xmath3o@xmath4 and @xmath13-cov@xmath3o@xmath4 show large single ion anisotropy , undergo long - range afm transition below 15 and 6 k , respectively and exhibit field - induced metamagnetic transitions at two critical fields @xmath17 and @xmath18 @xcite .
the values of magnetic moment determined from the saturation magnetization ( @xmath19 ) in the field - induced fm state and susceptibility in the paramagnetic ( pm ) state are found to be significantly larger than the expected spin - only moment of high spin - state co@xmath20 ion @xcite . in @xmath2-cov@xmath3o@xmath4 , @xmath19 is as much as 1.5 @xmath21/co larger than the spin - only moment ( 3.0 @xmath21/co ) for @xmath9 parallel to @xmath22 axis .
this excess 1.5 @xmath21/co moment is thought to come from the orbital magnetic moment due to the strong spin - orbit or spin - lattice coupling @xcite .
+ magnetic , electric , and structural properties have been studied extensively to disclose the underlying mechanism responsible for the step - like jumps in @xmath12(@xmath9 ) of cov@xmath3o@xmath4 @xcite .
neutron diffraction studies have revealed that these jumps in @xmath12(@xmath9 ) curve are coupled to the field - induced magnetic phase transitions from afm to fm state through an intermediate ferrimagnetic state .
@xcite . to elucidate the nature of field - induced metamagnetic transition and the thermodynamic properties of frustrated systems ,
the measurement of specific heat ( @xmath0 ) in applied field is important . in this work
, we present the specific heat data of @xmath2-cov@xmath3o@xmath4 over a wide range of @xmath9 across the magnetic ordering at @xmath5 .
strong spin - lattice coupling has also motivated us to investigate the magnetostriction effect [ @xmath10(@xmath9)/@xmath11 in this system .
we observe that both @xmath0(@xmath7 ) and the linear thermal expansion coefficient @xmath2(@xmath7 ) [ @xmath6(@xmath23)d@xmath10(@xmath7)/d@xmath7 ] exhibit a single @xmath25-like peak at @xmath5 in absence of magnetic field but a sharp and symmetric peak appears below @xmath5 with applied field .
also , the system releases the ising - like entropy ( rln2 ) and exhibits very large magnetostriction effect below @xmath7@xmath26 20 k. for understanding the origin of the field - induced magnetic transition in @xmath2-cov@xmath3o@xmath4 , the present results have been compared and contrasted with well known frustrated systems where the field - induced metamagnetic transition occurs due to the triangular or tetrahedral crystal geometry of the magnetic moments .
polycrystalline @xmath2-cov@xmath27o@xmath28 samples were prepared by standard solid - state reaction method using high purity cobalt ( ii ) acetate tetrahydrate ( aldrich , 99.999@xmath29 ) and vanadium pentoxide ( alfa aesar , 99.995@xmath29 ) .
stoichiometric quantities of these compounds were mixed properly and the mixture was heated in air for 16 h at 650 @xmath30c and then at 725 @xmath30c for 48 h. after the heat treatment , the material was quenched in liquid nitrogen to obtain single phase @xmath2-cov@xmath27o@xmath28 .
phase purity was checked by powder x - ray diffraction ( xrd ) method with cuk@xmath31 radiation in a high resolution rigaku ttrax ii diffractometer .
we have not observed any impurity phase within the resolution ( @xmath262@xmath29 ) of xrd .
all the peaks in the diffraction pattern were fitted well to a monoclinic structure of space group @xmath322/@xmath33 using the rietveld method with an acceptance factor @xmath34(@xmath29)@xmath63.78 ( fig .
1 ) . the observed lattice parameters _ _ a__=9.2501 , _ _
b__=3.5029 , _ _
c__=6.6175 , and @xmath35 = @xmath36 are in good agreement with the reported values for @xmath2-cov@xmath27o@xmath28 @xcite .
we have determined the average grain size ( @xmath2635@xmath37 m ) in the sample using the scanning electron microscope ( fei quanta 250 ) .
the sintered powder was pressed into pellet and then the sample of dimensions [email protected]@xmath381.08 mm@xmath39 was cut from this pellet for the specific heat , magnetostriction and thermal expansion measurements .
the longitudinal magnetothermal expansion was measured by capacitive method using a miniature tilted - plates dilatometer with applied field parallel to thickness ( @xmath40=1.08 mm ) of the sample .
the magnetic measurements were done in both physical property measurement system and squid - vsm ( quantum design ) .
the specific heat was measured by conventional relaxation time method using physical property measurement system ( quantum design ) .
+ -cov@xmath3o@xmath4 at room temperature .
the black solid line corresponds to the rietveld refinements of the diffraction pattern.,scaledwidth=60.0% ]
figure 2(a ) displays the temperature dependence of dc magnetic susceptibility @xmath41 ( @xmath6@xmath42 ) of polycrystalline @xmath2-cov@xmath3o@xmath4 sample in a field of 0.1 t. @xmath41 increases with decreasing @xmath7 and shows a sharp peak at @xmath5@xmath615 k due to the transition from pm to afm state .
no other peak or anomaly is observed in @xmath41(@xmath7 ) in the measured temperature range ( 2 - 300 k ) .
this indicates that the studied sample does not contain any magnetic impurity .
figure shows that @xmath43(@xmath7 ) can be fitted well to the curie - weiss law [ @xmath41@xmath44 ) ] over a wide range of temperature . from this fit
, we have calculated the value of effective paramagnetic moment @[email protected] @xmath21/co ion and the weiss temperature @xmath46@xmath6 - 9.2 k. the negative value of @xmath46 implies that the predominant magnetic interaction is antiferromagnetic in nature .
the observed value of @xmath45 is significantly larger than the expected spin - only moment of high spin co@xmath20 ( 3.87 @xmath21 ) .
it may be mentioned that markkula et al .
have reported even larger value of @xmath45 ( 6.09 @xmath21/co ) for this system @xcite .
this huge discrepancy between the observed and expected spin - only moment is due to the strong spin - orbit coupling @xcite . + ( @xmath7 ) and @xmath43(@xmath7 ) for @xmath2-cov@xmath3o@xmath4 at 0.1 t field .
solid line is the curie - weiss fit to the @xmath43(@xmath7 ) .
( b ) the temperature dependence of the specific heat @xmath0 .
( c ) the temperature dependence of magnetic contribution to the specific heat ( @xmath47 ) and entropy removal ( @xmath48).,title="fig:",scaledwidth=55.0% ] ( @xmath7 ) and @xmath43(@xmath7 ) for @xmath2-cov@xmath3o@xmath4 at 0.1 t field .
solid line is the curie - weiss fit to the @xmath43(@xmath7 ) .
( b ) the temperature dependence of the specific heat @xmath0 .
( c ) the temperature dependence of magnetic contribution to the specific heat ( @xmath47 ) and entropy removal ( @xmath48).,title="fig:",scaledwidth=45.0% ] ( @xmath7 ) and @xmath43(@xmath7 ) for @xmath2-cov@xmath3o@xmath4 at 0.1 t field .
solid line is the curie - weiss fit to the @xmath43(@xmath7 ) .
( b ) the temperature dependence of the specific heat @xmath0 .
( c ) the temperature dependence of magnetic contribution to the specific heat ( @xmath47 ) and entropy removal ( @xmath48).,title="fig:",scaledwidth=50.0% ] the temperature dependence of zero - field specific heat is shown in fig .
with decreasing temperature , @xmath0 decreases down to @xmath2620 k and then increases rapidly and exhibits a @xmath25-like anomaly close to @xmath5 .
similar to the magnetization data , @xmath0(@xmath7 ) supports single magnetic phase transition in @xmath2-cov@xmath27o@xmath28 . for better understanding
the nature of magnetic ground state , the magnetic contribution to the specific heat ( @xmath47 ) in the vicinity of afm transition and beyond has been estimated . in the temperature range 50 - 222 k
, @xmath0 can be fitted well with the einstein model to determine the lattice contribution to the specific heat of the system @xcite ( please see figure s1 in the supporting information ) . after subtracting the lattice contribution from @xmath0 ,
we have plotted @xmath47 in fig .
it is clear from fig .
2(c ) that @xmath47 does not decrease rapidly with increasing @xmath7 in the pm state . the magnetic entropy ( @xmath48 ) obtained by integrating ( @xmath49 is also shown in fig .
@xmath48 does not saturate even at @xmath7 as high as 3@xmath5 , which reflects the highly anisotropic nature of the magnetic structure of @xmath2-cov@xmath3o@xmath4 .
we would like to mention that the observed value of @xmath48 ( = 8.3 j mol@xmath50 k@xmath50 at @xmath7@xmath63@xmath5 ) for the present sample is about 1.5 times larger than that reported by kim et al . @xcite .
the larger value of @xmath48 is an indication of superior ordering of spins because it means , more number of spins are participating in the magnetic transition .
the estimated value of @xmath48 , however , is significantly smaller than the expected 11.5 j mol@xmath50 k@xmath50 for the high spin co@xmath20 ( @xmath8@xmath63/2 ) .
it may be noted that @[email protected] j mol@xmath50 k@xmath50 at 20 k is almost in accord with the expected value ( rln2 ) for the degree of freedom of the ising moments .
thus , the entropy of the ising - like spins is mostly released below 20 k. as the intrachain ferromagnetic interaction in cov@xmath27o@xmath28 is much stronger than the interchain afm interaction , the intrachain spin degree of freedom is effectively frozen and the spins would behave like the ising spins and hence , a huge reduction in magnetic entropy occurs .
+ typical specific heat data ( @xmath0/@xmath7 ) in the vicinity of @xmath5 are shown in fig .
3 as a function of @xmath7 for different magnetic fields . the application of magnetic field suppresses and broadens the peak at @xmath5 and the peak position shifts slowly towards lower temperature .
apart from these usual changes , another important feature is emerging with increasing field strength . at a field of 2 t , a weak shoulder - like feature starts to appear in @xmath0(@xmath7 ) curve just below 11 k. when the applied field exceeds 2 t , this weak anomaly transforms into a sharp and symmetric peak .
the sharp nature of the peak suggests that the transition is first - order . as this phenomenon occurs in the neighborhood of strong afm ordering transition
, it is difficult to determine the exact position of the peak .
we have determined the positions of the peaks by fitting a pair of gaussian functions plus a background to the data and found that the peak positions do not shift with field .
also , @xmath0 in the low temperature region is gradually enhanced with increasing field up to 5 t and then decreases . in the antiferromagnetic state , initially magnetic entropy ( and hence @xmath0 ) increases with @xmath9 due to the increase of field - induced spin disordering in one of the magnetic sublattices which is antiparallel to applied field . at high fields ,
however , the volume fraction of fm phase increases significantly and as a result , magnetic entropy reduces . in several afm systems where field - induced afm - fm transition occurs , with increasing magnetic field , magnetic entropy initially increases up to a critical field and then decreases @xcite .
+ versus @xmath7 plots for @xmath2-cov@xmath3o@xmath4 for different fields in the vicinity of antiferromagnetic transition.,scaledwidth=70.0% ] we now compare the observed specific heat results of @xmath2-cov@xmath3o@xmath4 with that reported for several other frustrated magnetic systems .
the two - peak feature in the temperature dependence of the specific heat has been extensively investigated both in theory and experiments @xcite .
several ferrimagnets and quasi-1d afm spin chains exhibit field - induced peak in @xmath0 due to the opening of an energy gap at low temperature @xcite . in these cases
, however , the peak is not very sharp and symmetric .
also , the peak shifts and broadens with increasing field strength . in dy@xmath3ti@xmath3o@xmath51 pyrochlore , the zero - field @xmath0(@xmath7 ) exhibits a single peak but a new sharp and symmetric peak appears at very low temperature with the application of magnetic field @xcite .
we have magnetically aligned the polycrystalline powder sample and done detailed @xmath12(@xmath9 ) and @xmath12(@xmath7 ) measurements to understand the field and temperature induced transitions .
for this aligned sample , @xmath12(@xmath9 ) at 5 k and @xmath12(@xmath7 ) curves for three different fields are shown in the supporting information section ( please see figures s2 and s3 in the supporting information ) .
similar to randomly oriented powder sample , @xmath12(@xmath7 ) shows afm to pm transition at 15 k for applied field below @xmath17 but @xmath12(@xmath7 ) exhibits a sharp peak at 11 k in the field range 1.5 - 3.3 t due to the transition from ferrimagnetic to pm state . above 3.3 t where the system is in field - induced fm state ,
the nature of @xmath12(@xmath7 ) curve is very different . with increasing temperature , @xmath12 decreases very slowly up to about 8 k and then decreases at a faster rate .
this behavior of @xmath12(@xmath7 ) is typical of a fm system in presence of high magnetic field .
though , the transition region gets smeared , d@xmath12/d@xmath7 reveals a minimum at around 11 k. so it is clear that both ferrimagnetic - pm and fm - pm transitions occur almost at same temperature ( 11 k ) .
for this reason , we have not observed a third peak in @xmath0(@xmath7 ) curve at high fields above 3.3 t due to the fm - pm transition . in polycrystalline samples , however , for some crystallites the effective field along the easy axis of magnetization direction can be significantly smaller than the applied field due to their random orientations .
thus , when the applied field exceeds @xmath52 or @xmath53 , these crystallites remain in the afm or ferrimagnetic state and , as a result , different magnetically ordered phases coexist in the sample .
this observation is fully consistent with the reported low - temperature neutron diffraction results @xcite .
powder neutron diffraction studies have revealed the coexistence of different magnetically ordered phases in @xmath2-cov@xmath3o@xmath4 when the strength of applied magnetic field exceeds @xmath52 or @xmath53 @xcite .
for example , both afm and ferrimagnetic phases coexist with volume fractions 0.54 and 0.46 , respectively at 2.5 t. however , both the phases decrease substantially and fm phase appears at fields above @xmath53 . at 5 t , the volume fractions of afm ,
ferrimagnetic and fm phases are 0.20 , 0.14 and 0.66 , respectively . + in order to shed some more light on the nature and origin of the field - induced peak in @xmath0 , we have studied the linear expansion of sample length [ @xmath54 as functions of @xmath7 and @xmath9 .
figure 4(a ) presents @xmath10(@xmath7)/@xmath55 ( @xmath6[@xmath56(@xmath7)-@xmath55]/@xmath55 ) plots up to 45 k for some selected fields , where @xmath55 is the length of the sample at @xmath7@xmath64 k in an applied field @xmath9 .
the inset of fig .
4(a ) shows @xmath10(@xmath7)/@xmath55 plot up to 300 k for @xmath9@xmath60 . at zero field , though @xmath58 decreases monotonically with decreasing @xmath7 , it starts to decrease at a much faster rate as @xmath7 approaches @xmath5 . in the pm state
, @xmath58 is approximately linear in @xmath7 below 50 k for field up to 3 t but it develops a broad maximum close to @xmath5 and shows an upward curvature above @xmath5 for higher fields .
though , the anomaly at @xmath5 gets weakened with increasing field strength , it remains visible up to the highest applied field 9 t. at and above 1 t , @xmath10(@xmath7)/@xmath55 shows a step - like decrease below a critical temperature @xmath59@xmath68 k .
this decrease in @xmath58 at @xmath59 is much sharper in nature than that at @xmath5 . to keep track on the field - evolution of the anomalies at @xmath5 and @xmath59 ,
we have determined @xmath2(@xmath7 ) at different fields from the slope of the curves in fig .
4(a ) . figure 4(b ) illustrates the temperature variation of the respective @xmath2(@xmath7 ) . at zero field ,
@xmath2(@xmath7 ) is positive over the whole temperature range .
similar to @xmath0(@xmath7 ) , @xmath2(@xmath7 ) exhibits a sharp @xmath25-like peak close to @xmath5 and the peak shifts to lower temperature and broadens with increasing field . over a narrow temperature range in the pm state
, @xmath2 becomes small and negative for applied field above 5 t. the field - induced peak in @xmath2(@xmath7 ) at @xmath59 is extremely sharp and symmetric and its position is almost insensitive to @xmath9 .
this peak is well separated from afm transition and visible only below 5 t. in polycrystalline samples , the linear thermal expansion measured using dilatometer is just an average of the three crystallographic axis directions .
normally , the coefficient of linear thermal expansion measured using dilatometer is 1/3 of the volume expansion coefficient @xcite .
we have compared the present results on thermal expansion with that reported from the temperature variation of lattice parameters determined using powder neutron diffraction @xcite .
indeed , the nature of @xmath7 dependence of the linear thermal expansion is quite similar to the volume expansion determined from powder neutron diffraction data .
+ ) for @xmath2-cov@xmath3o@xmath4 for different fields .
inset : @xmath60t@xmath61 in zero field in the temperature range 4 - 300 k. ( b ) plots of the thermal expansion coefficient @xmath2(@xmath7 ) for different fields .
( c ) magnetostriction , @xmath62 , at several temperatures both below and above @xmath5 .
inset : the field dependence of magnetization and @xmath1 at 5 k.,title="fig:",scaledwidth=42.5% ] ) for @xmath2-cov@xmath3o@xmath4 for different fields .
inset : @xmath60t@xmath61 in zero field in the temperature range 4 - 300 k. ( b ) plots of the thermal expansion coefficient @xmath2(@xmath7 ) for different fields .
( c ) magnetostriction , @xmath62 , at several temperatures both below and above @xmath5 .
inset : the field dependence of magnetization and @xmath1 at 5 k.,title="fig:",scaledwidth=42.5% ] ) for @xmath2-cov@xmath3o@xmath4 for different fields .
inset : @xmath60t@xmath61 in zero field in the temperature range 4 - 300 k. ( b ) plots of the thermal expansion coefficient @xmath2(@xmath7 ) for different fields .
( c ) magnetostriction , @xmath62 , at several temperatures both below and above @xmath5 .
inset : the field dependence of magnetization and @xmath1 at 5 k.,title="fig:",scaledwidth=42.5% ] the magnetostriction , @xmath10(@xmath9)/@xmath63@xmath6[@xmath56(@xmath9)-@xmath11/@xmath63 , where @xmath63 is the length of the sample in absence of magnetic field , for some selected temperatures both below and above @xmath5 is depicted in fig .
figure shows that the field dependence of magnetostriction is very sensitive to temperature . at low temperatures ( @xmath7@xmath6410 k )
, @xmath10/@xmath63 is almost independent of @xmath9 up to @[email protected] t. above @xmath17 , @xmath10(@xmath9)/@xmath63 increases rapidly and exhibits a weak anomaly around @xmath18 .
thus , @xmath10(@xmath9)/@xmath63 exhibits anomaly at two critical fields that correspond to field - induced metamagnetic transitions in @xmath12(@xmath9 ) curve @xcite .
similar to @xmath12(@xmath9 ) , these anomalies are clearly visible at low temperatures but they become weak with increasing @xmath7 and disappear above @xmath7@xmath6510 k @xcite .
the disappearance of these anomalies in both @xmath12(@xmath9 ) and @xmath10(@xmath9)/@xmath63 curves above 10 k may be related to the observed field - induced first order transition in @xmath66(@xmath7 ) .
at low temperature , @xmath10(@xmath9)/@xmath63 approximately mimics the nature of @xmath12(@xmath9 ) curve [ inset of fig .
@xmath10(@xmath9)/@xmath63 develops a downward curvature with the increase of @xmath9 but it does not show any tendency of saturation up to the maximum applied field .
the absence of saturation in @xmath10(@xmath9)/@xmath63 is consistent with the nature of @xmath12(@xmath9 ) curve .
@xmath12 in single crystals or magnetically aligned sample with field along the @xmath22 axis saturates above 4 t while @xmath12 at 7 t is just 50@xmath29 of the expected value in magnetically random polycrystalline sample @xcite .
the downward curvature in @xmath10(@xmath9)/@xmath63 decreases with increasing @xmath7 and @xmath10(@xmath9)/@xmath63 becomes almost linear for temperatures close to or slightly above @xmath5 .
well above @xmath5 , @xmath10(@xmath9)/@xmath63 decreases rapidly with increasing @xmath7 and develops a weak upward curvature .
the magnetostriction effect is very small above 50 k. the unusually high value of the magnetostriction below @xmath5 is comparable to that reported for the frustrated spinels zncr@xmath3se@xmath67 and cdcr@xmath3o@xmath67 where the spin - lattice coupling is believed to be quite strong @xcite . in these spinels , however , the nature of @xmath9 dependence of @xmath10/@xmath63 is different from that for the @xmath2-cov@xmath3o@xmath4 system .
furthermore , zncr@xmath3se@xmath67 exhibits a strong nearly constant negative thermal expansion over a wide range of @xmath7 just above @xmath5 which has been attributed to the geometrical frustration of the lattice degrees of freedom @xcite . in @xmath2-cov@xmath3o@xmath4 , however , @xmath2(@xmath7 ) is small and negative only at high fields .
the appearance of the peak in @xmath0(@xmath7 ) and @xmath2(@xmath7 ) with applied field and the large positive magnetostriction effect below @xmath5 suggest that the field - induced magnetostructural transition occurs due to the strong spin - lattice coupling and the field - induced fm state occupy more space than the afm state .
we believe that the values of thermal expansion and magnetostriction can be significantly larger in single crystal .
in conclusion , @xmath0(@xmath7 ) and @xmath2(@xmath7 ) of the quasi-1d antiferromagnetic spin chain @xmath2-cov@xmath3o@xmath4 exhibit a sharp @xmath25-like peak around @xmath5 . the magnetic entropy calculated from @xmath0(@xmath7 ) data shows that the magnetic fluctuations persist well above @xmath5 but the system behaves like a spin-1/2 ising chain below 20 k. in the afm state , both @xmath0 and @xmath2 exhibit strong @xmath7 and @xmath9 dependence .
when the applied field exceeds @xmath17 , a sharp and symmetric peak emerges well below @xmath5 due to the ferromagnetic / ferrimagnetic - paramagnetic transitions .
the sharp nature of the peak indicates that the transition is first order . the huge magnetostriction effect below @xmath5 and its strong @xmath9 dependence along with the field - induced spin structure transition suggest that the spin - lattice coupling in this system is quite strong .
the measurements of @xmath0 and @xmath10/@xmath56 on single crystal with @xmath9 along different crystallographic axes may reveal interesting physical properties for @xmath2-cov@xmath3o@xmath4 .
supporting information + in the temperature range 50 - 222 k , @xmath0 of @xmath2-cov@xmath3o@xmath4 has been fitted with the einstein model of lattice specific heat [ figure s1 ] .
the temperature and magnetic field dependence of magnetization have been measured for the magnetically aligned polycrystalline powder sample [ figure s2 and s3 ] .
this material is available free of charge via the internet at http://pubs.acs.org .
+ 99 he , z. ; taniyama , t. ; k@xmath68omen , t. ; itoh , m. field - induced order - disorder transition in the quasi - one - dimensional anisotropic antiferromagnet baco@xmath3v@xmath3o@xmath69 .
b _ * 2005 * , _ 72 _ , 172403 1 - 4 .
maartense , i. ; yaeger , i. ; wanklyn , b. m. field - induced magnetic transitions of conb@xmath3o@xmath4 in the ordered state .
_ solid state commun . _ * 1977 * , _ 21 _ , 93 - 96 .
coldea , r. ; tennant , d. a. ; wheeler , e. m. ; wawrzynska , e. ; prabhakaran , d. ; telling , m. ; habicht , k. ; smeibidl , p. ; kiefer , k. quantum criticality in an ising chain : experimental evidence for emergent e@xmath69 symmetry .
_ science _ * 2010 * , _ 327 _ , 177 - 180 .
bramwell , s. t. ; gingras , m. j. p. spin ice state in frustrated magnetic pyrochlore materials .
_ science _ * 2001 * , _ 294 _ , 1495 - 1501 and references therein .
tabata , y. ; kadowaki , h. ; matsuhira , k. ; hiroi , z. ; aso , n. ; ressouche , e. ; fk , b. kagom@xmath70 ice state in the dipolar spin ice dy@xmath3ti@xmath3o@xmath51 .
_ * 2006 * , _ 97 _ , 257205 1 - 4 .
ramirez , a. p. ; hayashi , a. ; cava , r. j. ; siddharthan , r. ; shastry , b. s. zero - point entropy in ` spin ice ' .
_ nature ( london ) _ * 1999 * , _ 399 _ , 333 - 335 .
hiroi , z. ; matsuhira , k. ; takagi , s. ; tayama , t. ; sakakibara , t. specific heat of kagom@xmath70 ice in the pyrochlore oxide dy@xmath3ti@xmath3o@xmath51 . _ j. phys .
* 2003 * , _ 72 _ , 411 - 418 . kimura , s. ; takeuchi , t. ; okunishi , k. ; hagiwara , m. ; he , z. ; kindo , k ; taniyama , t. ; itoh , m. novel ordering of an s=1/2 quasi-1d ising - like antiferromagnet in magnetic field .
lett . _ * 2008 * , _ 100 _ , 057202 1 - 4 .
he , z. ; yamaura , j. ; ueda , y. ; cheng , w. cov@xmath3o@xmath4 single crystals grown in a closed crucible : unusual magnetic behaviors with large anisotropy and 1/3 magnetization plateau .
soc . _ * 2009 * , _ 131 _ , 7554 - 7755 .
lenertz , m. ; alaria , j. ; stoeffler , d. ; colis , s. ; dinia , a. magnetic properties of low - dimensional @xmath2 and @xmath13 cov@xmath3o@xmath4 .
_ j. phys .
* 2011 * , _ 115 _ , 17190 - 17196
. kimber , s. a. j. ; mutka , h. ; chatterji , t. ; hofmann , t. ; henry , p. f. ; bordallo , h. n. ; argyriou , d. n. ; attfield , j. p. metamagnetism and soliton excitations in the modulated ferromagnetic ising chain cov@xmath3o@xmath4 .
b _ * 2011 * , _ 84 _ , 104425 1 - 8 .
singh , k. ; maignan , a. ; pelloquin , d. ; perez , o. ; simon , ch .
magnetodielectric coupling and magnetization plateaus in @xmath2-cov@xmath3o@xmath4 crystals .
_ j. mater .
chem . _ * 2012 * , _ 22 _ , 6436 - 6440 .
markkula , m. ; arevalo - lopez , a. m. ; attfield , j. p. neutron diffraction study of monoclinic brannerite - type cov@xmath3o@xmath4 .
_ j. solid state chem . _ * 2012 * , _ 192 _ , 390 - 393 . markkula , m. ; arevalo - lopez , a. m. ; attfield , j. p. field - induced spin orders in monoclinic cov@xmath3o@xmath4 .
b _ * 2012 * , _ 86 _ , 134401 1 - 4 . lenertz , m. ; alaria , j. ; stoeffler , d. ; colis , s. ; dinia , a. magnetic structure of ground and field - induced ordered states of low - dimensional @xmath2-cov@xmath3o@xmath4 : experiment and theory _ phys . rev .
b _ * 2012 * , _ 86 _ , 214428 1 - 8 .
lenertz , m. ; colis , s. ; ulhaq - bouillet , c. ; dinia , a. epitaxial growth of @xmath13-cov@xmath3o@xmath4 thin films : structure , morphology , and magnetic properties . _ appl .
* 2013 * , _ 102 _ , 212407 1 - 4 .
yao , x. 1/3 magnetization plateau induced by magnetic field in monoclinic cov@xmath3o@xmath4 .
_ j. phys .
* 2012 * , _ 116 _ , 2278 - 2282 .
kim , b. ; kim , b. h. ; kim , k. ; choi , h. c. ; park , s. y. ; jeong , y. h. ; min , b. i. unusual magnetic properties induced by local structure in a quasi - one - dimensional ising chain system : @xmath2-cov@xmath3o@xmath4 _ phys .
rev . b _ * 2012 * , _ 85 _ , 220407(r ) 1 - 5 .
sa@xmath71l , a. ; vodenicarevic , d. theoretical study of the magnetic order in @xmath2-cov@xmath3o@xmath4 .
b _ * 2013 * , _ 87 _ , 024403 1 - 5 .
hermann , r. p. ; jin , r. ; schweika , w. ; grandjean , f. ; mandrus , d. ; sales , b. c. ; long , g. j. einstein oscillators in thallium filled antimony skutterudites .
lett . _ * 2003 * , _ 90 _ , 135505 1 - 4 .
long , y. ; liu , q. ; lv , y. ; yu , r. ; jin , c. various 3_d_-4_f _ spin interactions and field - induced metamagnetism in the cr@xmath16 system dycro@xmath67 .
b _ * 2011 * , _ 83 _ , 024416 1 - 6 .
midya , a. ; mandal , p. ; das , s. ; banerjee , s. ; sharath chandra , l. s. ; ganesan , v. ; roy barman , s. magnetocaloric effect in homno@xmath72 crystal .
lett . _ * 2010 * , _ 96 _ , 142514 1 - 4 .
midya , a. ; das , s. n. ; mandal , p. ; pandya , s. ; ganesan , v. anisotropic magnetic properties and giant magnetocaloric effect in antiferromagnetic @xmath73mno@xmath72 crystals ( @xmath73=dy , tb , ho , and yb ) .
b _ * 2010 * , _ 84 _ , 235127 1 - 10 . antonosyan , d. ; bellucci , s. ; ohanyan , v. exactly solvable ising - heisenberg chain with triangular xxz - heisenberg plaquettes .
b _ * 2009 * , _ 79 _ , 014432 1 - 11 .
li , w. ; gong , s. s. ; zhao , y. ; ran , s. j. ; gao , s. ; su , g. phase transitions and thermodynamics of the two - dimensional ising model on a distorted kagome lattice .
b _ * 2010 * , _ 82 _ , 134434 1 - 8 .
honda , z. ; katsumata , k. ; yamada , k. the spin gap in a quantum antiferromagnet on the kagom@xmath70 lattice .
_ j. phys . : condens
. matter _ * 2002 * , _ 14 _ , l625 - 629 .
yoshida , y. ; tateiwa , n. ; mito , m. ; kawae , t. ; takeda , k. ; hosokoshi , y. ; inoue , k. specific heat study of an s = 1/2 alternating heisenberg chain system : f@xmath15pnn in a magnetic field .
lett . _ * 2005 * , _ 94 _ , 037203 1 - 4 .
souza , j. a. ; neumeier , j. j. ; white , b.d . ;
yu , yi - kou .
analysis of the critical behavior associated with the antiferromagnetic transitions of lamno@xmath72 and camno@xmath72 .
b _ * 2010 * , _ 81 _ , 172410 1 - 4 .
hemberger , j. ; krug von nidda , h .- a .
; tsurkan , v. ; loidl , a. large magnetostriction and negative thermal expansion in the frustrated antiferromagnet zncr@xmath3se@xmath67 .
* 2007 * , _ 98 _ , 147203 1 - 4 .
ueda , h. ; katori , h. a. ; mitamura , h. ; goto , t. ; takagi , h. magnetic - field induced transition to the 1/2 magnetization plateau state in the geometrically frustrated magnet cdcr@xmath3o@xmath67 .
lett . _ * 2005 * , _ 94 _ , 047202 1 - 4 .
the temperature dependence total specific heat ( @xmath0 ) is shown in fig .
s1 . in the range 50 - 222 k ,
@xmath0 is fitted well by the einstein model of lattice specific heat . in order to extract the lattice contribution to @xmath0 in the paramagnetic ( pm ) as well as in the antiferromagnetic ( afm ) states ,
the einstein fitting curve has been extrapolated down to lowest measured temperature ( 2 k ) .
we have aligned the powder sample of @xmath2-cov@xmath3o@xmath4 in a polymeric glue by applying 7 t magnetic field at room temperature which freezes the particle orientation .
the field dependence of magnetization ( m ) at 5 k is shown in fig .
it is clear from the figure that m(h ) exhibits sharp field - induced transitions at two critical fields ( @xmath17 and @xmath18 ) and 1/3 magnetization plateau between @xmath17and @xmath18 , and the saturation value of m in the ferromagnetic ( fm ) state ( above @xmath18 ) is about 4.6 @xmath21/co .
figure s3 shows representative m(t ) curves at three different magnetic fields of aligned polycrystalline @xmath2-cov@xmath3o@xmath4 . at 0.1 t ,
an afm - pm transition occurs around 15 k ( upper panel ) . for @xmath17<h<@xmath18 , the peak due to afm - pm transition disappears but a new peak appears at lower temperature ( 11 k ) due to ferrimagnetic - pm transition ( middle panel ) .
however , above @xmath18 , the nature of m(t ) curve is typical of a fm system in high field with fm - pm transition @xmath2611 k ( lower panel ) .
this study shows that @xmath2-cov@xmath3o@xmath4 undergoes ferrimagnetic - pm and fm - pm transitions at same temperature ( 11 k ) .
|
we have investigated the temperature and magnetic field dependence of magnetization , specific heat ( @xmath0 ) , and relative sample length change ( @xmath1 ) for understanding the field - induced spin - structural change in quasi - one - dimensional spin chain @xmath2-cov@xmath3o@xmath4 which undergoes antiferromagnetic ( afm ) transition below @xmath5@xmath615 k. analysis of @xmath0(@xmath7 ) shows that an effective @xmath8@xmath61/2 ising state is realized below 20 k , though the magnetic fluctuations persist well above @xmath5 . @xmath0 and the coefficient of linear thermal expansion ( @xmath2 ) exhibit strong @xmath9 dependence in the afm state .
we also observe a huge positive magnetostriction [ @xmath10(@xmath9)/@xmath11 below 20 k which does not show any tendency of saturation up to 9 t. with increasing field , a sharp and symmetric peak emerges below @xmath5 in both @xmath0(@xmath7 ) and @xmath2(@xmath7 ) due to field - induced first order ferrimagnetic / ferromagnetic - paramagnetic transitions .
the large value of magnetostriction below @xmath5 suggests strong spin - lattice coupling in @xmath2-cov@xmath3o@xmath4 .
+ keywords : quasi - one - dimensional , spin chain , antiferromagnetic , ferrimagnetic , field - induced peak in specific heat , thermal expansion , spin - lattice coupling
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.