Datasets:

ccdv
/

arxiv-summarization

Dataset card Data Studio Files Files and versions Community

Subset (2)

section · 216k rows

Split (3)

train · 203k rows

article stringlengths 0 745k	abstract stringlengths 0 92.7k
the study of massive binary systems in young clusters plays a key role in our understanding of how massive stars form . in addition to providing the binary fraction for massive stars , massive binary systems provide information about their formation environment in the form of quasi - preserved parameters such as eccentricity , separation , period , and angular momentum ( in the form of rotational velocities ; * ? ? ? in addition , an initial mass function ( imf ) composed of the secondary component masses can indicate whether the companions are randomly drawn from a @xcite ( or equivalent field star ) imf or not . this can tell us whether the binary systems may have randomly formed by way of gravitational capture after the formation process , or preferentially paired through a more complicated formation process such as competitive accretion @xcite . cyg ob2 provides one of the best regions for indirectly examining the formation of massive stars as it houses @xmath126070 o - type stars @xcite , including an o3if ( schulte 7 ; * ? ? ? * ) and an o4if ( schulte 22 ; * ? ? ? * ) , and possibly more than 2000 b stars @xcite . @xcite presented six new spectroscopic binary systems and summarized the then - known 11 ob binary systems in cyg ob2 . between then and this work , three additional ob binary systems have been uncovered , including schulte 9 @xcite , 2mass j20302730 + 4113253 ( * ? ? ? * submitted ) , and gsc 03161 - 00815 @xcite . six of the total 14 systems , mt421 , mt429 , mt696 , schulte 3 , schulte 5 , and gsc 03161 - 00815 are identified as eclipsing binaries ( where notation for the first five are from @xcite and @xcite ) . two of the 14 systems , mt059 and mt258 , are single - lined spectroscopic binaries ( sb1s ) , and eight of the 14 , mt252 , mt696 , mt720 , mt771 , schulte 3 , schulte 8a , schulte 9 , and 2mass j20302730 + 4113253 , are double - lined spectroscopic binaries ( sb2s ) . in part three of this ongoing study , we add to the 14 and present the newly uncovered spectroscopic binaries , mt145 ( sb1 ) , 2mass j20294666 + 4105083 ( listed as a candidate sb2 in * ? ? ? * ) , schulte 73 ( sb2 ) , and mt372 ( sb1 ) . we also present the first spectroscopic solution to the eclipsing system , gsc 03161 - 00815 ( sb2 ) . for simplicity , we will use @xcite notation for gsc 03161 - 00815 , 2mass j20294666 + 4105083 , and 2mass j20302730 + 4113253 ( a36 , a45 , and b17 respectively ) . these new systems bring the total number of ob binaries in cyg ob2 to 18 , constituting one of the highest numbers of massive binary solutions of any open cluster . section 2 of this work provides observational details of the new spectroscopic datasets . section 3 discusses the measurement of radial velocities , the search for periods in the radial velocity data , and the determination of orbital elements via radial velocity curve fitting . section 4 discusses the orbital solutions to the sb1 , mt145 , and the sb2s , a36 , a45 , and schulte 73 ( cyg ob2 no . 73 ) . section 5 presents the partial solution to the sb1 , mt372 . finally , section 6 summarizes the results of the survey to date , including the total number of ob binaries uncovered in the cyg ob2 core region , the total number of o star binary solutions , and the emerging distribution of orbital periods . @xcite & @xcite ( papers i & ii ) detail the observations of this survey through 2007 september . we have obtained additional datasets with the wiro - longslit spectrographamonson / wiro / long_slit.html ] on the wyoming infrared observatory ( wiro ) 2.3 m telescope and the hydra spectrograph on the wiyn 3.5 m telescope . table [ obs.tab ] lists the observing runs at each facility , the corresponding spectral coverages , and mean spectral resolutions . observations at wiyn took place over six nights on 2008 june 1015 . we used the hydra spectrograph with the red camera , 2blue fibers , and the 1200 l mm@xmath13 grating in second order to obtain four 1500 s exposures in each of two fiber configurations yielding a maximum signal - to - noise ratio ( snr ) of 80:1 for the brightest stars . the spectral coverage was 38204500 at a mean resolution of @xmath144500 . copper - argon lamps were used between each exposure to calibrate the spectra to an rms of 0.03 ( 2 at 4500 ) , and the typical resolution was 1.0 fwhm at 3900 and 0.82 fwhm at 4400 . spectra were doppler corrected to the heliocentric frame and checked against the radial velocity standards hd131156 ( g8v ) , hd146233(g2v ) , hd161096(k2iii ) , hd161797(g5iv ) , and hd171391(g8iii ) from @xcite before comparison to previous datasets . observations using the wiro - longslit spectrograph with the 1800 l mm@xmath13 grating in first order took place over 37 nights between 2007 october 23 and 2008 september 19 to examine the h@xmath15 , , and absorption lines in suspected sb2s . exposure times varied from 600 s to 4500 s ( in multiples of 600900 s ) depending on weather conditions and yielded a maximum snr of 200:1 for the brightest stars . the spectral coverages were 55006750 ( 2007 october 23 through 2007 november 5 ) and 52506750 ( 2008 june 23 through 2008 september 19 ) . copper - argon lamp exposures were taken after each star exposure to wavelength calibrate the spectra to an rms of 0.03 ( 1.4 at 6400 ) . the typical spectral resolution was 1.5 fwhm across the chip . spectra were doppler corrected to the heliocentric frame and checked against the same radial velocity standards taken during the 2008 june observations at wiyn before comparison to previous datasets . in addition , we also cross - correlated the wiro - longslit spectra with a composite interstellar line spectrum to look for systematic relative radial velocity shifts . the interstellar line spectrum was created by extracting the interstellar lines present in the 2008 spectra of a36 , doppler shifting them to the same velocity and combining them . interstellar lines and diffuse interstellar bands were identified with @xcite , @xcite , @xcite , and @xcite . relative velocity shifts were generally less than 6 . observations with relative shifts larger than 6 were corrected to the average absolute radial velocity offset before computing orbital solutions . all new datasets were reduced using standard iraf reduction routines as outlined in paper i. we obtained radial velocities , @xmath16 , for the single - lined system , mt145 via the iraf cross - correlation task xcsao in the rvsao package @xcite , using a model stellar atmosphere ( * ? ? ? * tlusty ) of the appropriate effective temperature and gravity as discussed in paper i. errors within the xcsao task are calculated using , @xmath17 where @xmath18 is the fwhm of the correlation peak and @xmath19 is the ratio of the correlation peak height to the amplitude of antisymmetric noise @xcite . to obtain rough @xmath16 measurements for the double - lined systems , a36 , a45 , and schulte 73 , we deblended the @xmath204471 , 5876 ( a36 & schulte 73 ) or h@xmath15 ( a45 & mt372 ) lines by fitting simultaneous gaussian profiles ( at fixed gaussian widths ) with the splot routine in iraf . we repeated this method 10 times while varying the baseline region used to define the continuum each time . we then used the mean of each component s list of gaussian centers to compute the velocity . utilizing this method , a number of factors contribute to the radial velocity uncertainty . these factors include , but are not limited to the errors associated with gaussian profile fitting , wavelength calibrations , emission present in the line cores , and lower signal - to - noise of the less - luminous component s spectral profiles ( especially while in a partially or wholly blended state ) . we estimated the contribution of the gaussian profile fitting uncertainty by adopting the rms of each component s list of gaussian centers . unfortunately , this only produces a lower limit on the uncertainty because the initial gaussian width estimates also affect the center measurements . for the total one - sigma uncertainties , we adopted these rms values combined in quadrature with the typical error in the wavelength calibration ( 1.4 for wiro and 2 for wiyn ) and the typical rms observed in the list of velocities obtained from the correlation with the interstellar line template . the total error given for each system reflects a lower limit as we can not characterize uncertainty contributions from emission present in the line cores ( e.g. , a36 ) or the lower snr of the secondary spectral features ( e.g. , a45 ) . we used the method outlined in @xcite to determine the orbital parameters of each system . we obtained an estimate of the orbital period through an idl program written by a. w. fullerton which makes use of the discrete fourier transform and clean deconvolution algorithm of @xcite . the strongest peaks in the power spectrum of each star were examined by folding the data at the corresponding period and inspecting the @xmath16 curve visually . orbital elements were then procured by using the best period as an initial estimate in the nonlinear , least - squares curve fitting program of @xcite . the best solutions were attained by manually varying the initial guesses of key orbital parameters until a minimum in the @xmath21 of the fit was found . weights for each point were assigned as the inverse of the 1@xmath22 @xmath16 error . while it is not uncommon to quote orbital period errors to one ten thousandth of a day ( e.g. , @xcite ) , we performed two additional tests to evaluate the appropriateness of the orbital period uncertainties for each system . in the first test , we explored the confidence range for each orbital parameter by fitting the radial velocity data 1000 times , with the orbital parameters as free variables , while adding gaussian noise to the data . the noise added to each data point was characterized by the one - sigma uncertainty . we compared the resultant rms deviations to the single - fit error estimates for each orbital parameter . the result of this test suggested that the majority of the single - fit parameter uncertainties were reasonable , as the monte carlo test routinely produced similar uncertainties . the disadvantage to this test , however , is that we could not examine each fit to see if it was reasonable ( i.e. , whether the newly folded radial velocities had a reasonable rms ) . therefore , with the error in the period strictly in mind , we ran the orbital fitting procedure 50100 times while incrementing the period by the single - fit error estimate and monitored how the rms changed . the result of this test suggests that we , at most , underestimate the orbital period errors by a factor of 10 . with this in mind , the errors for the orbital period throughout this work have been increased by a factor of ten . the complete list of orbital elements for each binary system appears in table [ orbparms.tab ] . listed within the table are the period in days ( _ p _ ) , eccentricity of the orbit ( _ e _ ) , longitude of periastron in degrees ( _ @xmath23 _ ) , systemic radial velocity ( _ @xmath24 _ ) , epoch of periastron ( _ @xmath25 _ ) , primary and secondary semi - amplitudes ( _ @xmath26 _ & _ @xmath27 _ ) , adopted or calculated minimum primary and secondary masses in solar masses ( _ @xmath28 _ & _ @xmath29 _ ) , primary and secondary mass functions in solar masses ( _ f(m)@xmath30 _ & _ f(m)@xmath31 _ ) , spectral classifications from this survey ( _ s.c.@xmath30 _ & _ s.c.@xmath31 _ ) , the minimum primary and secondary semi - major axes in solar radii ( @xmath32sin @xmath33 & @xmath34sin @xmath33 ) , and finally , the rms of the fits ( _ rms@xmath30 _ & _ rms@xmath31 _ ) . systemic radial velocities are known to vary depending on which lines are used to measure the radial velocity . this is especially evident with massive stars and evolved stars where the stellar atmosphere velocity gradient may be higher ( see @xcite for examples with a few well - studied o binaries ) . to minimize such discrepancies , radial velocities are normally averaged over many lines . in the case of mt145 , where we obtain radial velocities by cross correlation over many lines , this is not an issue . however , in the case of the remaining 4 systems , we only have 12 lines ( @xmath20 5876 , 6678 or h@xmath15 ) available for radial velocity measurement in the majority of spectra . this places an added uncertainty on the systemic velocity and semi - amplitudes measured for a36 , a45 , schulte 73 , and mt372 . in the simplest case , these effects may cause a systematic blueshift to the systemic velocity ( such as the case with a non - eclipsing system with equal and symmetric stellar winds ) . eclipses , uneven stellar winds , and uneven surface temperatures complicate the matter , and these effects can affect the components unequally , producing systemic radial velocities that vary from primary to secondary . we see this effect in two of our systems , a45 and schulte 73 . the component orbital solutions produce two different systemic velocities . for both of these systems , we adopted the mean of the two values as the systemic velocity of the system for table [ orbparms.tab ] . we observed mt145 a total of 54 times , including 6 at lick , 2 at keck , 12 at wiyn , and 34 at wiro . excluding two low - snr spectra from lick and one from wiro , a total of 51 spectra were of sufficient snr to obtain the solution for this o9iii star . the strongest signal in the cleaned power spectrum corresponds to a period of 25.12 days . an alias at 20 days is also present . however , the 20 day period produced a folded @xmath16 curve having a much larger rms than the 25.12 day period . the best - fitting orbital solution and @xmath16 curve for mt145 is shown in figure [ mt145curve ] , corresponding to a period of @xmath35 days with an eccentricity of @xmath0 and solution @xmath36 . with this period , a semi - amplitude of @xmath37 , and an assumed primary mass of @xmath38 @xcite , we estimate a minimum secondary mass of @xmath39 , corresponding to a mid b star ( interpolated from * ? ? ? this yields a mass ratio lower limit of @xmath40 . to constrain the mass ratio upper limit , we examined the spectra taken while the system was near quadrature for evidence of line - doubling or asymmetries originating from the secondary . neither is present in any of the spectra . given snr ratios of up to 150:1 for these spectra , we would expect to see evidence of the secondary star s spectral features at velocity separations of @xmath41 if the component luminosity ratio is larger than @xmath42 . this maximum luminosity ratio is determined by combining synthetic spectra at various luminosity ratios and line separations . the absence of secondary spectral features therefore conservatively limits the luminosity ratio to @xmath43 and implies a secondary less luminous than a b1v ( interpolated from * ? ? ? we thus constrain the mass of the secondary to @xmath44 @xmath45 and the mass ratio to @xmath46 ( where the range in mass ratio incorporates both observational and inclination uncertainties ) . these limits loosely require @xmath47 given the uncertainty in o star masses @xcite . additionally , the solution produces a minimum primary semi - major axis of @xmath48sin@xmath49 and a longitude of periastron of @xmath50 the ephemerides for mt145 and mt372 are listed in table [ oc ] and include the date ( _ @xmath51 _ ) , phase ( _ @xmath52 _ ) , measured radial velocity ( _ @xmath16 _ ) , cross - correlation @xmath53 error ( _ @xmath53 err _ ) , and the observed minus calculated velocity ( _ @xmath54 _ ) . a36 is originally introduced as a probable massive member of cyg ob2 in @xcite and listed as a possible binary based on the asymmetry in the absorption lines of a high snr spectrum in ( * ? ? ? * figure 1 ) . @xcite provided a partial photometric solution to this system and listed it as an eclipsing binary of the algol type with evidence of an oconnell effect ( a light curve having unequal out - of - eclipse maxima ; * ? ? ? we provide the first spectroscopic solution to this double - lined system . we observed a36 a total of 23 times between 2007 october and 2008 august , including 5 times with hydra at wiyn and 18 times with the wiro - longslit spectrograph . with the 19 higher snr spectra we obtained a period estimate of 4.680 days from the cleaned power spectrum with no aliases . from this estimate we obtained the solution for the secondary ( rms=10.0 ) shown in figure [ a36curve ] with a period of @xmath55 and relatively low eccentricity of @xmath2 . the period , eccentricity , epoch of periastron , and systemic velocity were applied as fixed parameters to obtain the solution for the primary ( rms=29.3 ) . the filled symbols correspond to the primary @xmath16 measurements and the open symbols correspond to the secondary @xmath16 measurements . the combined solutions provide minimum semi - amplitudes of @xmath56 and @xmath57 for the primary and secondary , respectively , and produce semi - major axes of @xmath32sin @xmath58 and @xmath34sin @xmath59 . the computed lower limit masses of @xmath28sin@xmath60 and @xmath29sin@xmath61 yield a mass ratio of @xmath62 . table [ oc2 ] provides the ephemerides for a36 , a45 , and schulte 73 , listing the date ( hjd@xmath632,400,000 ) , phase ( @xmath52 ) , measured radial velocities ( @xmath64 and @xmath65 ) , and observed minus calculated velocities ( @xmath66 and @xmath67 ) . error estimates are shown in parentheses . based on two of the spectra obtained during the 2008 wiyn run , we tentatively classify the components of a36 as a b0ib and b0iii for primary and secondary respectively . the exposures obtained on 2008 june 10 & 15 show the components at nearly the same phase and in quadrature . additionally , the light curve shown in @xcite , tells us that the temperatures of the components are similar , and the double - lined nature of the spectra tell us that the luminosities are similar . we can therefore conclude from the 2008 june 10 spectrum shown first from the top in figure [ threeplot ] that the spectral types of the components are similar with subtle differences , such as stronger @xmath20 4089 , 4116 and weaker @xmath20 4026 , 4121 , 4387 , 4471 for the primary . both primary and secondary have no @xmath68 4481 ( confirmed in figure 1 in * ? ? ? * ) , moderately strong @xmath20 4089 , 4116 , strong @xmath20 4026 , 4471 , weak @xmath20 4121 , 4387 , and strong @xmath68 4070 . after smoothing the wiyn spectra to approximately the resolution of the spectra in the @xcite digital atlas , we combined a b0ib ( hd164402 ) spectrum with a b0iii ( hd48434 ) spectrum at the appropriate radial velocities to make a comparison . the resulting combined spectrum ( second from top ) is also shown in figure [ threeplot ] along with the 2001 september 8 spectrum of schulte 73 ( second from bottom ) obtained at wiyn and the associated composite spectrum for schulte 73 ( bottom ) . intrinsic , , , , , and hydrogen absorption are labeled . the composite spectrum is not an exact match to a36 ( note that 4089 is stronger than that of the primary ) . however , the mass ratio obtained from @xcite for a b0ib primary and b0iii secondary is @xmath69 and agrees with our calculated mass ratio of @xmath62 . an apparent weakening of the primary s lines between phases 0.218 and 0.609 , seen in panels 1 and 3 of the time series presented in figure [ a36progress ] , at least partially explains the large residuals at these epochs and may originate from a struve - sahade effect @xcite . this apparent weakening also coincides with an absence of well - defined absorption profiles in h@xmath15 ( panel 2 of figure [ a36progress ] ) between secondary and primary eclipse . these spectral phenomena coupled with the continuous variation of the light curve presented in @xcite associated with algol eclipsing systems , an oconnell effect ( which is generally associated with contact or near contact systems ; * ? ? ? * ) and a seemly close separation between components ( @xmath70sin@xmath71 ) , implies that a36 is either a contact or near - contact system . in the very least , it is likely that one or both components are filling their roche lobes . because a36 is composed of b stars , one may question whether this is a be binary . the system does not hold the standard qualifications for a be binary however , as the mass ratio is much greater than 0.1 and the primary is evolved @xcite . * figure 3 ) also show that hot supergiants can have hydrogen emission that very closely resembles the double - peaked hydrogen emission seen in be stars . is a36 a member of cyg ob2 ? @xcite places doubt on its status as a member of the cluster based on its disagreement with the 2 myr isochrone and evolutionary tracks best fit for cyg ob2 stars . with a systemic velocity offset @xmath72 from the mean systemic velocity for cluster stars ( @xmath73 ; paper i ) , a36 is likely either a runaway binary system or a background system masquerading as a member of cyg ob2 . @xcite found observational evidence for several runaway binary systems , but concluded that they are rare . these systems can be produced by an asymmetric supernova of one of the components ( producing an sb1 ) or by dynamic interactions within a dense cluster . with a stellar density of @xmath74 @xcite it is certainly possible for a36 to have originated this way , but not probable . however , as a background object , a36 s systemic velocity inferred from a galactic rotation curve , would place it at a distance of 6.6 kpc , an interarm region where ob stars are unlikely to form . therefore , without a proper motion to trace its trajectory , or until a complete solution is computed for the photometric and spectroscopic solutions for this star , the status of a36 s membership will remain uncertain . we obtained 24 observations from 2007 october through 2008 september on this double - lined system . @xcite classified this star a b0.5v , but , as with a36 , she also suggests it may be an sb2 based on a high snr spectrum ( see figure 1 in * ? ? ? we present the first confirmation of the binary nature of a45 with a spectroscopic solution to this system . of the 24 observations , we used 21 to obtain rough @xmath16 measurements with the h@xmath15 @xmath686562.80 line . we did not use the @xmath205875.75 , 6678.15 lines owing to either the weaker absorption of the secondary or low snr of the spectra near @xmath685876 . the 18 measurements produced a cleaned power spectrum with a period estimate of 2.885 days and aliases of 2 days and 1.55 days . the latter is a @xmath75 alias ( where @xmath76 is the signal frequency per day ) . neither alias produced a folded @xmath16 curve with as low of an rms as 2.885 days . the best solution for the primary results in a period of @xmath77 and eccentricity of @xmath4 . the period , eccentricity , and epoch of periastron were applied as fixed parameters to obtain the solution for the secondary . the rms of 47 indicates there is room for improvement in the secondary s solution and also explains the high uncertainty on the calculated masses , @xmath28sin@xmath78 and @xmath29sin@xmath79 ( @xmath80 ) . with minimum semi - amplitudes of @xmath81 and @xmath82 for the primary and secondary respectively , the solution also produces minimum semi - major axes of @xmath32sin @xmath83 and @xmath34sin @xmath84 and a systemic velocity of @xmath85 . we have limited information for classification of the components of a45 owing to the limited spectral coverage of the wiro - longslit datasets ( @xmath1253006700 ) . no attempt to obtain exposures of this system was made during the 2008 wiyn run because of the system s angular separation from the rest of the sample . examination of the existing spectral coverage and the spectrum in figure 1 of @xcite suggests that her classification is more or less accurate in the sense that both components appear to be early b stars . there is no evidence of in either component , is moderately strong , is weak , and there appears to be no . an absence of the numerous additional metal lines , such as , , , and seen in evolved b stars suggests that neither component is very evolved and both are likely still dwarves . additionally , a double - lined system with a mass ratio of @xmath80 restricts the secondary component to within one or two temperature classes of the primary component . therefore , we adopt the classification of b0.5v for the primary and b2v ? b3v ? for the secondary component based on the aforementioned line strengths and mass ratio . is a45 a member of cyg ob2 ? although a45 has the largest angular separation from the core of cyg ob2 , the systemic velocity agrees with the mean systemic velocity of cluster members and the system does show the same diffuse interstellar band features characteristic of cyg ob2 stars . additionally , though the combined luminosity and temperature of the system disagrees slightly with its predicted placement on the h - r diagram presented for cyg ob2 stars in figure 12 in @xcite , individual component luminosities and temperatures shift a45 s position down on the diagram and into better agreement with other cyg ob2 members . schulte 73 is included in our original sample but excluded from paper i owing to the less than 3 observations then obtained . spectra taken during the 2008 june wiyn run , however , revealed the double - lined nature of this star , and re - examination of the 2001 september 8 spectrum confirmed it . the successive 2008 wiro runs cemented its status as a double - lined spectroscopic binary with a period of @xmath1217.3 days and a mass ratio nearly equal to unity . the solution computed for schulte 73 utilized 27 of the 36 spectra obtained . seven spectra from wiyn in 2008 june were excluded for low snr owing possibly to a fiber position discrepancy . two wiro spectra ( 2008 july 26 & 27 ) were excluded because of the extremely blended state of the components . for the 27 higher snr spectra , the initial period estimate from the cleaned power spectrum is @xmath1216.6 days with an alias at @xmath1266 days . we dismiss the alias on the basis that progression of the absorption lines from a blended to deblended state is observed over approximately 45 nights . the best orbital solutions yield a period of @xmath86 days for both components . the exact period , eccentricity , and date of periastron are held constant for the primary in order to facilitate a matching solution with the secondary . coincidentally , this also provides a better - fitting solution to the primary @xmath16 curve . we attribute the larger error on the period ( a magnitude higher than the other three systems ) to the smaller number of velocity measurements obtained around apastron ( @xmath87 ) . the solutions and corresponding @xmath16 curves for schulte 73 are shown in figure [ s73curve ] . the filled and open symbols correspond to the primary and secondary @xmath16 measurements respectively . the errant wiyn measurements in the primary solution ( triangles ) reflect the lower snr of the 2008 june dataset and the error in the computed period . with an eccentricity of @xmath6 and semi - amplitudes of @xmath88 and @xmath89 the solutions produce masses of @xmath90sin@xmath91 and @xmath92sin@xmath93 . this results in a mass ratio of @xmath8 and our first confirmed `` twin '' in the cluster . the resulting semi - major axes are @xmath48sin @xmath94 and @xmath95sin @xmath96 ( i.e. , a separation of @xmath70sin @xmath97 ) , and the calculated systemic velocity is @xmath98 . @xcite classify this star as an o8v . the components each show spectral signatures indicative of a later o type star , namely moderate to weak and stronger absorption . the primary classification criteria for later o type stars is @xmath684542 : @xmath684387 and @xmath684200 : @xmath684144 @xcite . the @xmath684200 : @xmath684144 ratio ( also sensitive to luminosity class ) and a visual comparison with the 2001 september 8 spectrum from wiyn seen second from bottom in figure [ threeplot ] agree best with an o8iii spectral type for both components . the schulte 73 spectrum has been smoothed to approximately the resolution of the @xcite digital atlas . however , without a separation of the spectral components ( e.g. , tomographic separation ) , there is uncertainty in this assessment . the components also closely match an o7iii and o9iii spectral type . the very weak @xmath684144 however , precludes a later temperature class . the o8iii ( hd36861 ) and o8iii composite spectrum is shown at the bottom of figure [ threeplot ] . we also tried various combinations of an o8v , o8iii and o8ii . these did not fit the data as well and left larger residuals when subtracted from the schulte 73 spectrum . mt372 is a single - lined spectroscopic binary and an eclipsing system , likely of algol type , with a period of @xmath99 days . with a visual magnitude of nearly 15 , only 5 of our 21 spectroscopic observations of this b0v star are of sufficient snr for meaningful analysis . therefore , to assist in the period determination , we acquired the available photometric data for mt372 from the public data release of the northern sky variability survey @xcite , and ran it through our clean idl program to search for a period . while excluding points with errors greater than 0.2 magnitudes , we computed a period of @xmath100 days from the power spectrum . in the case that the light curve exhibits only a primary eclipse , the peak in the power spectrum corresponds to the real period of the system . in the case that the light curve exhibits both a primary and secondary eclipse , the peak corresponds to half the real period . we adopt the latter case because a period of 1.114 days produces an radial velocity orbital solution with an unacceptably large rms ( where the radial velocities are determined from the @xmath101 line in a similar manner to the sb2s ) . a period of @xmath102 produces the folded @xmath16 curve and solution shown in figure [ mt372curve ] . in order to obtain this solution , we fixed the period and the eccentricity ( @xmath103 ) . allowing the period and/or eccentricity to vary did not produce a viable solution , and we did not expect it would because of the very small number of radial velocities to work with . although very noisy , this new period also produced a light curve with two eclipses ( primary and secondary ; seen in figure [ mt372phot ] ) as opposed to the single primary eclipse present at a folding of 1.114 days . it should be noted , that the period of 1.114 days also produced a single eclipse that was asymmetric , suggesting that the time between primary and secondary eclipses and secondary and primary eclipses is not equal ( i.e. , the system seems to have a slight eccentricity ) . the orbital solution shown in figure [ mt372curve ] , also produces a semi - amplitude of @xmath104 , and a semi - major axis of @xmath105 . as this system has little evidence for a double - lined nature , we stay with the previous spectral classification of this star , b0v . a small change in velocity width seems to be visible in @xmath101 in figure [ mt372series ] ( panel 2 ) , but the snr of the spectra ( especially evident with the 5876 and 6678 lines shown in panels 1 & 3 of figure [ mt372series ] ) makes it difficult to produce an accurate determination . the minimum secondary mass determined from the solution is @xmath106 , assuming a primary of @xmath107 @xcite . interpolating from @xcite and reasonably assuming that the secondary is also unevolved , this results in a minimum spectral class of b8v . however , a b8v is unlikely to be the companion spectral class as it has too low of an effective temperature to produce the light curve in figure [ mt372phot ] . with a primary effective temperature of @xmath108 k @xcite and a period of @xmath102 days , a b2v at a relatively high inclination ( @xmath1275 degrees ) would have the correct effective temperature and stellar radius to produce a light curve that resembles the one for mt372 . however , given a b2v companion and a mass ratio of @xmath109 , the computed semi - major axis at approximately 75 degrees is not large enough to fully separate the two components . this may be explained by returning to the possible velocity width variation present in the h@xmath15 lines . if indeed there is a small amount of blending of component profiles present , then fitting a single gaussian profile to the blended h@xmath15 line will systematically produce a smaller primary velocity amplitude and , in turn , smaller semi - amplitude and semi - major axis . we thus report the listed primary velocity semi - amplitude as a lower limit . in this work , the third part of our ongoing survey to determine the distribution of binary orbital parameters in the cyg ob2 association , we presented three new binary systems ( mt145 , schulte 73 , & mt372 ) and confirmed the binary status of one massive binary candidate ( a45 ) . we also provided the first spectroscopic orbital solution to the eclipsing binary , a36 and a partial solution to the eclipsing , single - lined system , mt372 . while mt145 , schulte 73 , a45 , and mt372 are likely members of the cluster based on their systemic velocities ( deviating by only @xmath120.615 from the cluster mean of @xmath73 obtained in paper i and well within the association s radial velocity dispersion ) the status of a36 , with a systemic velocity of @xmath110 , remains uncertain . the system could be a runaway binary or a background object ( though both do not seem very probable ) . however , if we do include a36 , the current total for ob binaries in cyg ob2 is now 18 . of these 18 , 14 have period and mass ratio estimates . in addition , with six partial spectroscopic solutions , eight full spectroscopic solutions , three partial photometric solutions , and four full photometric solutions , cyg ob2 is on a track to become the cluster with the highest number of ob binary solutions . at seven it currently ties ngc 6231 @xcite for the highest number of full o - star binary solutions . the locations of all 18 systems are shown in figure [ binaries ] . no evidence of grouping is apparent . table [ binaries ] lists several key parameters for all 18 systems , including the star designation , photometric and/or spectroscopic binary type ( _ type _ ) , spectral classifications ( _ s.c . _ ) , period ( _ p _ ) , mass ratio when available ( _ q _ ) , and literature references ( _ ref _ ) . with the exception of schulte 9 ( @xmath111 years ) , periods for all systems are less than 25.2 days , with a mean of 9.6 days . eleven systems have a period less than 5 days . the 17 systems with periods less than 25.2 days also appear to obey pik s law @xcite , which states that the distribution of @xmath11 is flat . to avoid histogram binning selection effects , we choose to show the normalized cumulative distribution of the logarithm of the periods ( open diamonds ) in figure [ oepik ] . the flat relation of pik s law translates to a linear relation in a cumulative distribution . for comparison purposes , we also include the normalized cumulative distributions of o binaries in ngc 6231 ( * ? ? ? * solid triangles ) and the 54 ob binaries from the 9th spectroscopic binary catalog ( * ? ? ? * asterisks ) . all 3 cumulative distributions contain systems with periods less than 25.2 days and were normalized at a period of 8.9 days . to test the linearity of the cyg ob2 distribution , we computed its linear pearson correlation coefficient . the distribution yields a coefficient of @xmath112 ( where @xmath113 is a perfectly linear relation ) . the linearity is illustrated by the best fit line ( solid line ) in figure [ oepik ] . the dashed line in the figure represents a perfect pik distribution ( i.e. , a perfectly flat distribution within the range of periods in cyg ob2 ) . a two sided kolmogorov - smirnov ( @xmath114 ) test between the pik and cyg ob2 distributions gives a 38% probability that both were drawn from the same parent distribution . @xmath114 tests between cyg ob2 and ngc 6231 on the one hand and cyg ob2 and the 9th spectroscopic binary catalog ( @xmath115 days ) on the other , yield 91% and 62% probabilities of being drawn from the same parent distribution , respectively . interestingly , the cumulative distributions of cyg ob2 and ngc 6231 have a high correlation and are very similar in appearance , including an upturn near @xmath116 . one wonders if this upturn is a true deviation from a linear relation ( i.e. , a preference for periods around 5 days ) , just a coincidence related to small number statistics , or observational bias . though the latter case is a possible explanation for the upturn in ngc 6231 , we do not believe this is the case with cyg ob2 because several of the systems with periods around 5 days come from other programs with different observational cadences ( e.g. , mt421 and b17 ) . figure [ pvse ] displays the relation between the orbital periods and eccentricities for eight of the systems in cyg ob2 ( open diamonds ) , six of the systems in ngc 6231 ( solid triangles ) and 66 of the systems in the 9th spectroscopic binary catalog ( asterisks ) . the eccentricities in cyg ob2 range from circular to moderately eccentric ( 0.291 ) , while the total distribution in figure [ pvse ] has an eccentricity range of zero to @xmath120.75 for periods ranging from @xmath120.5 days to 100 days . systems with periods less than a few days tend toward more circular orbits , while systems with larger periods ( @xmath117 days ) display a full range of eccentricities . this agrees with previous studies such as @xcite . there is a paucity of highly eccentric systems longer than 30 days . this likely stems from an incompleteness of the surveys at longer periods and is not a true trend . the cyg ob2 star , a45 , falls on the high - eccentricity side of the shorter period systems with an eccentricity of 0.273 and a period of 2.884 days . it does not appear to be particularly close to the majority of ob stars in the cluster center ( see figure [ binaries ] ) , and so dynamical interaction with a massive neighbor does not seem a likely source of its higher eccentricity ( unless it is a runaway system ) . it is possible that further observations of this system will constrain the eccentricity to a slightly more circular orbit , but its position in the figure is hardly removed from the rest , and therefore , its moderately elliptical orbit is not extraordinary . the remainder of cyg ob2 stars fall well within the remaining distribution . with the exception of mt421 and mt429 , all the cyg ob2 systems have early - type companions and mass ratios widely ranging from @xmath118 ( mt258 ) to @xmath8 ( schulte 73 ) . one may see this as an indication that the companions are not randomly drawn from a @xcite or similar field star initial mass function ( imf ) . however , given that the current sample encompasses a number of systems that are double - lined spectroscopic binaries ( i.e. , they are easier to detect ) this assessment may be premature and biased . only time will tell , as we have a number of single - lined spectroscopic binary candidates still to investigate and a spectral database that has not been completely tapped . we continue to analyze these candidates with the long term goal of providing accurate distributions for binary orbital parameters of massive stars with an assumed shared formation history . we thank the time allocation committees of the lick , keck , wiyn , and wiro observatories for granting us observing time and making this project possible . we are also grateful for support from the national science foundation through research experience for undergraduates ( reu ) program grant ast 03 - 53760 and through grant ast 03 - 07778 , and the support of the wyoming nasa space grant consortium through grant nng05g165h . we would also like to graciously thank , in no particular order , christopher rodgers , emily may , megan bagley , michael dipompeo , michael alexander , jessie runnoe , and sabrina cales for their generous help in observing at wiro through good skies and bad , and jerry bucher and james weger for their ample help and upkeep of wiro facilities . bonnell , i. a. , bate , m. r. , & zinnecker , h. 1998 , mnras , 298 , 93 comern , f. , et al . 2002 , a&a , 389 , 874 comern , f. , & pasquali , a. 2007 , a&a , 467 , 23 conti , p. s , leep , e. m. 1974 , apj , 193 , 113 contreras , m. e. , rodriguez , l. f. , tapia , m. , cardini , d. , emanuele , a. , badiali , m. , & persi , p. 1997 , apj , 488 , 153 de becker , m. , rauw , g. , & manfroid , j. 2004 , a&a , 424 , l39 drilling , j. s. , & landolt , a. u. 2000 , in astrophysical quantities , ed . a. n. cox ( 4th ed . ; new york ; springer ) , 381 galazutdinov , g. a. , musaev , f. a. , krelowski , j. , & walker , g. a. h. 2000 , pasp , 112 , 648 giuricin , g. , mardirossian , f. , & mezzetti , m. 1984 , a&a , 134 , 365 hanson , m. m. 2003 , apj , 597 , 957 harmanec , p. , bisikalo , d. v. , boyarchuk , a. a. , & kuznetsov o. a. 2002 , a&a , 396 , 937 herbig , g. h. 1975 , apj , 196 , 129 herbig , g. h. , & leka , k. d. 1991 , apj , 382 , 193 hillwig , t. c. , gies , d. r. , bagnuolo , w. g. , jr . , huang , w. , mcswain , m. v. , & wingert , d. w. 2006 , apj , 639 , 1069 kiminki , d. c. , et al . 2007 , apj , 664 , 1120 kiminki , d. c. , mcswain , m. v. , & kobulnicky , h. a. 2008 , apj , 679 , 1478 kinemuchi , k. , et al . 2008 , in prep kndlseder , j. 2000 , a&a , 360 , 539 kurtz , m. j. , mink , d. j. , wyatt , w. f. , fabricant , d. g. , torres , g. , kriss , g. a. , & tonry , j. l. 1991 , in astronomical data analysis software and systems i , asp conf . worrall , c. biemesderfer , and j. barnes , p. 432 larson , r. b. 2001 , iau symposium , 200 , 93 lanz , t. , & hubeny , i. 2003 , apjs , 146 , 417 linder , n. , rauw , g. , sana , h. , de becker , m. , & gosset , e. 2007 , a&a , 474 , 193 liu , q .- y . ; yang , y .- 2003 , chinese journal of astronomy and astrophysics , 3 , 142 martins , f. , schaerer , d. , & hillier , d. j. 2005 , a&a , 436 , 1049 massey , p. , & thompson , a. b. 1991 , aj , 101 , 1408 massey , p. , puls , j. , pauldrach , a. w. a. , bresolin , f. , kudritzki , r. p. , & simon , t. 2005 , apj , 627 , 477 mcswain , m. v. 2003 , apj , 595 , 1124 mcswain , m. v. 2007 , apj , 655 , 473 miczaika , g. r. 1953 , pasp , 65 , 141 milone , e. f. 1968 , aj , 73 , 708 morbey , c. l. , & brosterhus , e. b. 1974 , pasp , 86 , 455 morton , d. c. 1991 , apj , 77 , 119 naz , y. , de becker , m. , rauw , g. , & barbieri , c. 2008 , a&a , 483 , 543 negueruela , i. , marco , a. , herrero , a. , & clark , j. s. 2008 , a&a , 487 , 575 pik , e. j. 1924 , tartu obs . publ . , 25 otero , s. 2008a , open european journal on variable stars , 83 , 1 otero , s. 2008b , open european journal on variable stars , 91 , 1 penny , l. r. , ouzts , c. , & gies , d. r. 2008 , 681 , 554 pourbaix d. , tokovinin a. a. , batten a. h. , fekel f. c. , hartkopf w. i. , levato h. , morrell n. i. , torres g. , udry , s. 2004 , a&a , 424 , 727 pigulski , a. , & kolaczkowski , z. 1998 , mnras , 298 , 753 rauw , g. , vreux , j. m. , & bohannan , b. 1999 , apj 517 , 416 rios , l. y. , & degioia - eastwood , k. 2004 , baas , 205 , no . 09.05 roberts , d. h. , lehr , j. , & dreher , j. w. 1987 , aj , 93 , 968 romano , g. 1969 , mmsai , 40 , 375 salpeter , e. e. 1955 , apj , 121 , 161 sana , h. , gosset , e. , naz y. , rauw , g. , & linder , n. 2008 , mnras , 386 , 447 schulte , d. h. 1956 , apj , 123 , 250 stefanik , r. p. , latham , d. w. , torres , g. 1999 , aspc , 185 , 354 stroud , v. e. , clark , j.s . , negueruela , i. , roche , p. , & norton , a.j . 2009 , a&a , submitted walborn , n. r. 1973 , apj , 180 , l35 walborn , n. r. , & fitzpatrick , e. l. 1990 , pasp , 102 , 379 walborn , n. r. , & howarth , i. d. 2000 , pasp , 112 , 1446 williams , et al . 2008 , apj , 682 , 492 wilson , o. c. 1948 , pasp , 60 , 385 wilson , o. c. , & abt , a. 1951 , apj , 144 , 477 wozniak , p. r. , et al . 2004 , aj , 127 , 2436 , northern sky variability survey : public data release 1999 jul 45 & keck / hires & 38906270 in 35 orders & 31.6 & 0.1 & 2,451,3632,451,364 + 1999 jul 2123 & lick / hamilton & 36507675 in 81 orders & 31.6 & 0.1 & 2,451,3812,451,383 + 1999 aug 2123 & lick / hamilton & 36507675 in 81 orders & 31.6 & 0.1 & 2,451,4112,451,413 + 1999 oct 1415 & keck / hires & 37005250 in 29 orders & 31.6 & 0.1 & 2,451,4662,451,467 + 2000 jul 1011 & lick / hamilton & 36507675 in 81 orders & 31.6 & 0.1 & 2,451,7362,451,737 + 2000 sep 1819 & keck / hires & 37005250 in 29 orders & 31.6 & 0.1 & 2,451,8052,451,806 + 2001 aug 24 & wiyn / hydra & 38004490 in order 2 & 1200 & 0.9 & 2,452,146 + 2001 sep 89 & wiyn / hydra & 38004490 in order 2 & 1200 & 0.9 & 2,452,1612,452,162 + 2004 nov 2830 & wiyn / hydra & 38004490 in order 2 & 1200 & 0.9 & 2,453,3382,453,340 + 2005 jul 1821 & wiro / wiro - spec & 38004490 in order 1 & 2400 & 2.5 & 2,453,5702,453,573 + 2005 jul 1820,22 & wiro / wiro - spec & 38004490 in order 1 & 2400 & 2.5 & 2,453,6322,453,635 + 2005 oct 13 & wiro / longslit & 40506050 in order 2 & 600 & 2.5 & 2,453,657 + 2006 jun 1620 & wiro / longslit & 39005900 in order 2 & 600 & 2.5 & 2,453,9032,453,907 + 2006 jul 1516,20 & wiro / longslit & 39005900 in order 2 & 600 & 2.5 & 2,453,9322,453,935 + 2006 sep 811 & wiyn / hydra & 38004490 in order 2 & 1200 & 0.9 & 2,453,9872,453,990 + 2006 oct 7 & wiro / longslit & 39005900 in order 2 & 600 & 2.5 & 2,454,016 + 2007 jun 2830 & wiro / longslit & 39005900 in order 2 & 600 & 2.5 & 2,454,2802,454,282 + 2007 jul 46 & wiyn / hydra & 38204510 in order 2 & 1200 & 0.9 & 2,454,2852,454,287 + 2007 aug 28sep 4 & wiro / longslit & 55506850 in order 1 & 1800 & 1.5 & 2,454,3412,454,348 + 2007 oct 23,25,2729,31 & wiro / longslit & 55506850 in order 1 & 1800 & 1.5 & 2,454,3972,454,405 + 2007 nov 1,35 & wiro / longslit & 55506850 in order 1 & 1800 & 1.5 & 2,454,4062,454,410 + 2008 jun 1015 & wiyn / hydra & 38204510 in order 2 & 1200 & 0.9 & 2,454,6282,454,633 + 2008 jun 2330 & wiro / longslit & 52106680 in order 1 & 1800 & 1.5 & 2,454,6412,454,648 + 2008 jul 21,2227 & wiro / longslit & 52506740 in order 1 & 1800 & 1.5 & 2,454,6692,454,675 + 2008 aug 1722 & wiro / longslit & 52506740 in order 1 & 1800 & 1.5 & 2,454,6962,454,701 + 2008 sep 12,1316,19 & wiro / longslit & 52506740 in order 1 & 1800 & 1.5 & 2,454,7222,454,729 + 2008 oct 1418 & wiro / longslit & 52506740 in order 1 & 1800 & 1.5 & 2,454,7542,454,758 + lrrrrr @xmath119 ( days ) & 25.140 ( 0.008 ) & 4.674 ( 0.004 ) & 2.884 ( 0.001 ) & 17.28 ( 0.03 ) & 2.228 ( fixed ) + @xmath120 & 0.291 ( 0.009 ) & 0.10 ( 0.01 ) & 0.273 ( 0.002 ) & 0.169 ( 0.009 ) & 0.0 ( fixed ) + @xmath23 ( deg ) & 158.3 ( 0.6 ) & 359.2 ( 0.7 ) & 188.8 ( 0.1 ) & 7.3 ( 0.8 ) & 344 ( 146 ) + @xmath24 ( ) & -19.3 ( 0.4 ) & -47 ( 2 ) & 5 ( 1 ) & -11 ( 2 ) & 3 ( 7 ) + @xmath25 ( hjd-2,400,000 ) & 51789.08 ( 0.04 ) & 54693.836 ( 0.009 ) & 54728.209 ( 0.001 ) & 54700.89 ( 0.03 ) & 54755 ( 9 ) + @xmath121 ( ) & 48.9 ( 0.7 ) & 169 ( 9 ) & 126.1 ( 0.3 ) & 117 ( 2 ) & 75 ( 20 ) + @xmath122 ( ) & & 243 ( 2 ) & 273 ( 17 ) & 117.9 ( 0.8 ) & + @xmath90 ( ) & 22.6 ( 0.5 ) & @xmath12319.8 ( 0.9 ) & @xmath12312 ( 1 ) & @xmath12311.2 ( 0.2 ) & 17.5 + @xmath92 ( ) & 6.0 ( 0.1)13.8 ( 0.6 ) & @xmath12313.8 ( 0.9 ) & @xmath1235.3 ( 0.5 ) & @xmath12311.1 ( 0.2 ) & @xmath1210.0 + @xmath124 ( ) & 0.267 ( 0.012 ) & 6.9 ( 0.2 ) & 0.534 ( 0.004 ) & 2.8 ( 0.1 ) & 0.099 ( 0.080 ) + @xmath125 ( ) & & 2.3 ( 0.4 ) & 5 ( 1 ) & 2.81 ( 0.06 ) & + s. c.@xmath30 & o9iii & b0ib & b0.5v & o8iii & b0v + s. c.@xmath31 & mid b & b0iii & b2v ? b3v ? & o8iii & b2?v + @xmath48sin @xmath33 ( ) & 23.2 ( 0.3 ) & 15.6 ( 0.8 ) & 6.91 ( 0.02 ) & 39.4 ( 0.5 ) & 3.4 ( 0.9 ) + @xmath95sin @xmath33 ( ) & & 22.345 ( 0.003 ) & 14.96 ( 0.01 ) & 39.675 ( 0.004 ) & + @xmath126 ( ) & 8.9 & 29.3 & 25 & 5.1 & 14.9 + @xmath127 ( ) & & 10.0 & 47 & 6.7 & + ccrrr 51,381.50 .......................... & 0.788 & 1.7 & 10.3 & 1.8 + 51,383.50 .......................... & 0.867 & -24.7 & 15.1 & -1.0 + 51,467.90 .......................... & 0.225 & -21.5 & 4.4 & 9.2 + 51,736.50 .......................... & 0.909 & -33.5 & 11.3 & 7.8 + 51,737.50 .......................... & 0.948 & -56.4 & 13.7 & 3.5 + 51,805.76 .......................... & 0.664 & 22.8 & 3.2 & 7.9 + 52,146.70 .......................... & 0.225 & -23.6 & 3.6 & 6.9 + 52,161.81 .......................... & 0.826 & -7.4 & 4.1 & 2.4 + 52,162.67 .......................... & 0.860 & -17.8 & 3.8 & 3.3 + 53,338.63 .......................... & 0.637 & 14.0 & 4.4 & -1.9 + 53,340.59 .......................... & 0.715 & 10.7 & 4.2 & -0.4 + 53,570.50 .......................... & 0.860 & -40.0 & 8.3 & -18.9 + 53,572.50 .......................... & 0.940 & -66.4 & 10.0 & -10.3 + 53,632.50 .......................... & 0.327 & -10.5 & 11.5 & -2.7 + 53,633.50 .......................... & 0.367 & -10.6 & 6.7 & -9.3 + 53,636.50 .......................... & 0.486 & 11.5 & 5.2 & -0.3 + 53,657.50 .......................... & 0.321 & -0.8 & 6.8 & 8.0 + 53,903.75 .......................... & 0.116 & -71.7 & 8.4 & -7.1 + 53,903.76 .......................... & 0.117 & -74.7 & 9.1 & -10.3 + 53,904.72 .......................... & 0.155 & -33.1 & 7.8 & 18.8 + 53,904.73 .......................... & 0.155 & -41.0 & 7.6 & 10.7 + 53,904.90 .......................... & 0.162 & -65.6 & 7.7 & -16.1 + 53,905.77 .......................... & 0.197 & -49.4 & 6.8 & -10.6 + 53,905.90 .......................... & 0.202 & -48.1 & 7.3 & -11.0 + 53,906.74 .......................... & 0.235 & -19.3 & 8.5 & 8.6 + 53,906.89 .......................... & 0.241 & -42.2 & 10.6 & -15.8 + 53,907.78 .......................... & 0.277 & -20.5 & 10.2 & -2.7 + 53,907.88 .......................... & 0.280 & -28.4 & 7.1 & -11.4 + 53,932.75 .......................... & 0.270 & -17.0 & 7.9 & 2.5 + 53,932.85 .......................... & 0.274 & 2.5 & 8.8 & 20.9 + 53,932.86 .......................... & 0.274 & 6.4 & 10.3 & 24.8 + 53,935.89 .......................... & 0.395 & -14.7 & 12.8 & -17.3 + 53,987.70 .......................... & 0.456 & 1.8 & 4.6 & -7.6 + 53,988.75 .......................... & 0.497 & 9.9 & 4.4 & -2.8 + 53,989.66 .......................... & 0.534 & 13.7 & 4.7 & -1.0 + 53,990.86 .......................... & 0.581 & 17.3 & 4.3 & 1.1 + 54,280.83 .......................... & 0.116 & -61.7 & 7.0 & 3.0 + 54,280.87 .......................... & 0.117 & -56.5 & 6.6 & 7.8 + 54,281.70 .......................... & 0.150 & -60.2 & 8.5 & -6.7 + 54,281.78 .......................... & 0.153 & -50.9 & 7.8 & 1.4 + 54,281.85 .......................... & 0.156 & -51.5 & 7.0 & -0.0 + 54,282.70 .......................... & 0.190 & -44.8 & 7.9 & -4.1 + 54,282.77 .......................... & 0.193 & -45.6 & 6.0 & -5.7 + 54,282.83 .......................... & 0.195 & -45.9 & 6.9 & -6.8 + 54,285.93 .......................... & 0.318 & -6.0 & 4.3 & 3.3 + 54,286.66 .......................... & 0.348 & -7.6 & 5.2 & -3.4 + 54,287.73 .......................... & 0.390 & 5.3 & 4.3 & 3.2 + 54,397.71 .......................... & 0.765 & 5.0 & 16.7 & 0.7 + 54,399.69 .......................... & 0.843 & -22.0 & 15.3 & -6.9 + 54,401.67 .......................... & 0.922 & -47.9 & 16.0 & -0.3 + 54,402.71 .......................... & 0.964 & -54.1 & 14.6 & 12.3 + 54,754.62 .......................... & 0.949 & 75.4 & 7.4 & 9.8 + 54,755.60 .......................... & 0.388 & -36.6 & 6.3 & 1.7 + 54,756.60 .......................... & 0.840 & 16.0 & 3.0 & -8.9 + 54,757.59 .......................... & 0.283 & 6.9 & 3.9 & -1.9 + 54,758.72 .......................... & 0.788 & 6.9 & 3.4 & 6.3 + lcrrrr ptabularcr ptabularcr ptabularcr 54,403.61 .......................... & 0.910 & -163.4 ( 5.6 ) & 27.7 & 167.5 ( 4.9 ) & 4.2 + 54,403.70 .......................... & 0.928 & -174.9 ( 5.0 ) & 30.9 & 184.7 ( 4.3 ) & 1.0 + 54,405.71 .......................... & 0.357 & 103.4 ( 5.3 ) & 47.6 & -194.2 ( 3.5 ) & 3.1 + 54,406.63 .......................... & 0.555 & 105.4 ( 3.7 ) & 5.6 & -261.1 ( 3.4 ) & -5.3 + 54,406.76 .......................... & 0.582 & 108.6 ( 3.7 ) & 16.2 & -244.5 ( 3.7 ) & 0.2 + 54,409.73 .......................... & 0.218 & -23.6 ( 5.5 ) & 45.3 & -23.0 ( 3.7 ) & -3.3 + 54,410.67 .......................... & 0.419 & 98.4 ( 3.7 ) & 10.5 & -243.8 ( 3.4 ) & -1.8 + 54,628.88 .......................... & 0.103 & -163.3 ( 2.6 ) & 24.6 & 147.2 ( 2.5 ) & -5.3 + 54,633.74 .......................... & 0.143 & -150.6 ( 3.1 ) & -0.8 & 108.1 ( 3.3 ) & 11.0 + 54,669.84 .......................... & 0.867 & -116.5 ( 6.2 ) & 34.0 & 113.4 ( 4.1 ) & 7.4 + 54,672.78 .......................... & 0.495 & 100.3 ( 3.8 ) & -4.4 & -280.6 ( 3.7 ) & -16.1 + 54,673.88 .......................... & 0.730 & -37.7 ( 12.9 ) & -33.6 & -92.2 ( 7.1 ) & 11.6 + 54,696.69 .......................... & 0.610 & 109.0 ( 14.4 ) & 27.6 & -236.8 ( 3.5 ) & -8.4 + 54,696.96 .......................... & 0.669 & 125.2 ( 4.3 ) & 79.0 & -155.7 ( 3.9 ) & 21.1 + 54,697.66 .......................... & 0.818 & -129.9 ( 5.9 ) & -32.7 & 23.1 ( 3.4 ) & -6.8 + 54,697.96 .......................... & 0.882 & -163.1 ( 4.3 ) & 2.4 & 129.6 ( 3.7 ) & 2.4 + 54,698.75 .......................... & 0.052 & -198.2 ( 4.4 ) & 24.4 & 200.8 ( 7.1 ) & -2.9 + 54,700.87 .......................... & 0.505 & 103.7 ( 3.7 ) & -1.3 & -254.3 ( 3.4 ) & 10.3 + 54,701.90 .......................... & 0.724 & 1.7 ( 3.6 ) & 0.8 & -114.0 ( 3.8 ) & -2.9 + 54,403.66 ........................... & 0.447 & 106.3 ( 14.89 ) & -6.7 & -257.5 ( 5.33 ) & -28.2 + 54,403.74 ........................... & 0.476 & 127.4 ( 12.48 ) & 14.6 & -225.1 ( 6.17 ) & 3.9 + 54,405.72 ........................... & 0.161 & 7.9 ( 5.40 ) & 3.8 & -33.0 ( 25.14 ) & -39.6 + 54,405.74 ........................... & 0.169 & -56.9 ( 6.78 ) & -67.9 & -12.9 ( 16.19 ) & -4.5 + 54,406.64 ........................... & 0.483 & 87.3 ( 15.39 ) & -25.3 & -200.5 ( 8.72 ) & 28.0 + 54,406.65 ........................... & 0.486 & 87.3 ( 20.36 ) & -25.1 & -165.5 ( 11.95 ) & 62.6 + 54,406.77 ........................... & 0.527 & 135.4 ( 13.52 ) & 26.4 & -171.1 ( 7.89 ) & 49.6 + 54,409.74 ........................... & 0.557 & 126.6 ( 9.00 ) & 21.8 & -226.0 ( 5.93 ) & -14.5 + 54,410.68 ........................... & 0.882 & -84.1 ( 14.51 ) & -9.6 & 233.8 ( 6.10 ) & 57.1 + 54,641.84 ........................... & 0.046 & -126.0 ( 5.87 ) & -14.6 & 226.6 ( 6.80 ) & -30.0 + 54,647.81 ........................... & 0.116 & -28.5 ( 8.80 ) & 11.4 & 4.0 ( 8.07 ) & -97.9 + 54,672.90 ........................... & 0.817 & 9.7 ( 15.16 ) & 26.3 & 8.3 ( 19.51 ) & -43.0 + 54,724.70 ........................... & 0.784 & -17.0 ( 17.09 ) & -26.4 & 4.0 ( 23.56 ) & 9.0 + 54,724.78 ........................... & 0.810 & 23.8 ( 7.31 ) & 34.1 & -1.4 ( 23.90 ) & -39.0 + 54,725.72 ........................... & 0.135 & -17.0 ( 10.74 ) & 3.9 & 6.8 ( 8.43 ) & -53.8 + 54,726.72 ........................... & 0.483 & 115.3 ( 10.66 ) & 2.7 & -250.6 ( 7.71 ) & -22.2 + 54,726.79 ........................... & 0.508 & 111.6 ( 13.80 ) & 0.6 & -228.2 ( 6.83 ) & -3.2 + 54,729.70 ........................... & 0.517 & 139.9 ( 11.19 ) & 29.8 & -234.4 ( 6.08 ) & -11.2 + 54,747.59 ........................... & 0.721 & 21.5 ( 5.86 ) & -27.6 & -4.1 ( 6.10 ) & 87.0 + 54,748.60 ........................... & 0.071 & -76.2 ( 6.34 ) & 11.4 & 285.4 ( 26.42 ) & 80.1 + 54,754.77 ........................... & 0.212 & -15.6 ( 10.59 ) & -59.0 & 3.2 ( 44.57 ) & 81.7 + 52,161.65 .......................... & 0.071 & -133.5 ( 2.0 ) & -1.9 & 110.7 ( 2.0 ) & 1.5 + 54,628.75 .......................... & 0.825 & -118.3 ( 2.9 ) & -43.6 & 86.5 ( 2.5 ) & 31.3 + 54,628.88 .......................... & 0.833 & -112.4 ( 3.6 ) & -31.8 & 89.1 ( 2.1 ) & 27.9 + 54,629.87 .......................... & 0.890 & -157.5 ( 3.1 ) & -34.0 & 88.6 ( 2.1 ) & -15.3 + 54,630.74 .......................... & 0.941 & -156.9 ( 2.9 ) & -4.9 & 121.3 ( 2.1 ) & -10.5 + 54,630.85 .......................... & 0.947 & -166.7 ( 2.5 ) & -12.3 & 122.3 ( 2.1 ) & -11.9 + 54,633.86 .......................... & 0.121 & -115.2 ( 2.4 ) & -22.0 & 95.6 ( 2.1 ) & 25.5 + 54,696.75 .......................... & 0.760 & -31.4 ( 6.8 ) & -5.4 & 0.5 ( 2.0 ) & -5.7 + 54,696.96 .......................... & 0.772 & -13.4 ( 2.9 ) & 21.2 & 16.1 ( 3.8 ) & 1.1 + 54,697.77 .......................... & 0.819 & -67.9 ( 5.3 ) & 1.9 & 46.8 ( 1.8 ) & -3.5 + 54,697.85 .......................... & 0.824 & -81.4 ( 3.0 ) & -8.0 & 60.9 ( 2.0 ) & 6.9 + 54,697.92 .......................... & 0.828 & -70.8 ( 2.9 ) & 5.9 & 69.9 ( 2.4 ) & 12.7 + 54,698.71 .......................... & 0.874 & -122.4 ( 1.9 ) & -10.5 & 88.7 ( 1.8 ) & -3.7 + 54,698.91 .......................... & 0.885 & -125.2 ( 3.1 ) & -5.0 & 107.1 ( 2.8 ) & 6.5 + 54,699.72 .......................... & 0.932 & -148.7 ( 1.4 ) & -0.6 & 127.3 ( 1.4 ) & -0.8 + 54,699.93 .......................... & 0.944 & -142.4 ( 1.7 ) & 10.9 & 138.2 ( 1.6 ) & 5.1 + 54,700.72 .......................... & 0.990 & -150.0 ( 2.1 ) & 11.8 & 131.3 ( 1.6 ) & -9.5 + 54,701.73 .......................... & 0.048 & -155.0 ( 1.8 ) & -9.8 & 126.9 ( 1.5 ) & 3.7 + 54,701.94 .......................... & 0.061 & -142.0 ( 2.7 ) & -3.8 & 124.4 ( 2.2 ) & 8.4 + 54,724.69 .......................... & 0.377 & 66.8 ( 3.3 ) & 6.6 & -91.7 ( 2.8 ) & -8.2 + 54,724.81 .......................... & 0.384 & 73.1 ( 1.8 ) & 11.2 & -88.7 ( 1.4 ) & -3.6 + 54,725.69 .......................... & 0.435 & 69.4 ( 1.5 ) & -1.0 & -93.3 ( 1.4 ) & -0.1 + 54,725.80 .......................... & 0.441 & 70.8 ( 1.6 ) & -0.2 & -93.3 ( 1.5 ) & 0.4 + 54,726.70 .......................... & 0.493 & 69.0 ( 1.7 ) & -3.0 & -90.4 ( 1.6 ) & 3.9 + 54,726.82 .......................... & 0.500 & 67.5 ( 1.6 ) & -4.2 & -87.9 ( 1.5 ) & 5.9 + 54,729.66 .......................... & 0.664 & 27.0 ( 2.0 ) & -2.4 & -52.7 ( 1.9 ) & -2.8 + 54,729.84 .......................... & 0.675 & 22.2 ( 2.1 ) & -2.2 & -52.9 ( 2.1 ) & -8.1 + lccccl mt059 & sb1 & o8v & b & 4.8527 ( 0.0002 ) & 0.220.67 & 1 + mt145 & sb1 & o9iii & mid b & 25.140 ( 0.008 ) & 0.260.63 & 2 + mt252 & sb2 & b2iii & b1v & 1819 & 0.8 ( 0.2 ) & 1 + mt258 & sb1 & o8v & b & 14.660 ( 0.002 ) & 0.180.89 & 1 + mt372 & ea?/sb1 & b0v & b2?v & 2.228 ( fixed ) & @xmath120.6 & 2,3 + mt421 & ea & o9v & b9v a0v & 4.161 & @xmath120.160.19 & 4 + mt429 & ea & b0v & ? ? & 2.9788 & & 4 + mt696 & sb2/ew / ke & o9.5v & early b & 1.46 & & 5 + mt720 & sb2 & early b & early b & @xmath128 5 & & 1 + mt771 & sb2 & o7v & o9v & 1.5 : & 0.8 ( 0.1 ) & 1 + schulte 3 & sb2/ea ? & o6iv ? & o9iii & 4.7464 ( 0.0002 ) & 0.44 ( 0.08 ) & 1,6 + schulte 5 & eb & o7ianfp & ofpe / wn9 ( & b0v ? ) & 6.6 & 0.28 ( 0.02 ) & 7,8,9,10,11,12 + schulte 8a & sb2 & o5.5i & o6 ? & 21.908 & 0.86 ( 0.04 ) & 13,14 + schulte 9 & sb2 & o5 ? & o67 ? & 2.355 yr & & 15 + schulte 73 & sb2 & o8iii & o8iii & 17.28 ( 0.03 ) & 0.99 ( 0.02 ) & 2 + a36 & sb2/ea & b0ib & b0iii & 4.674 ( 0.004 ) & 0.70 ( 0.06 ) & 2,16,17 + a45 & sb2 & b0.5v & b2v ? b3v ? & 2.884 ( 0.001 ) & 0.46 ( 0.02 ) & 2,17 + b17 & sb2 & o7 ? & o9 ? & 4.0217 ( 0.0004 ) & @xmath121 ? & 18,19 +	we present the orbital solutions for four ob spectroscopic binaries , mt145 , gsc 03161 - 00815 , 2mass j20294666 + 4105083 , and schulte 73 , and the partial orbital solution to the b spectroscopic binary , mt372 , as part of an ongoing study to determine the distribution of orbital parameters for massive binaries in the cygnus ob2 association . mt145 is a new , single - lined , moderately eccentric ( @xmath0 ) spectroscopic binary with period of @xmath1 days . gsc 03161 - 00815 is a slightly eccentric ( @xmath2 ) , eclipsing , interacting and double - lined spectroscopic binary with a period of @xmath3 days . 2mass j20294666 + 4105083 is a moderately eccentric ( @xmath4 ) double - lined spectroscopic binary with a period of @xmath5 days . schulte 73 is a slightly eccentric ( @xmath6 ) , double - lined spectroscopic binary with a period of @xmath7 days and the first `` twin '' in our survey with a mass ratio of @xmath8 . mt372 is a single - lined , eclipsing system with a period of @xmath9 days and low eccentricity ( @xmath10 ) . of the now 18 known ob binaries in cyg ob2 , 14 have periods and mass ratios . emerging evidence also shows that the distribution of @xmath11 is flat and consistent with `` pik s law '' .
black hole is one of the most important key objects in theoretical physics . though its quantum behavior and thermodynamic property have played great roles in the path to understand yet unknown quantum theory of gravity ( see e.g. refs . @xcite ) , a direct experimental test had been believed almost impossible . recently , the scenarios of large @xcite and warped @xcite extra dimension(s ) have led to an amazing possibility of producing black holes at future colliders with distinct signals @xcite ( see also refs . @xcite for studies before the observation @xcite that black holes radiate mainly on the brane ) . when the center - of - mass ( cm ) energy of a collision exceeds the planck scale , which is of the order of tev here , the cross section is dominated by a black hole production @xcite , which is predicted to be of the order of the geometrical one @xcite , increasing with the cm energy . in this trans - planckian energy domain , the larger the cm energy is , the larger the mass of the resulting black hole is , and hence the better its decay is treated semi - classically via hawking radiation @xcite . main purpose of this series of work is to discuss such decay signals in the hope that these will serve as the basis to pursue stringy or quantum gravitational corrections to them . in previous publications , we have pointed out that the production cross section of a black hole increases with its angular momentum , so that the produced black holes are highly rotating @xcite ( see also ref . @xcite for an earlier attempt ) . the form factor for the production cross section , taking this rotation into account @xcite , is larger than unity and increases with the number of dimensions @xmath3 . the result is in good agreement with an independent numerical simulation of a classical gravitational collision of two massless point particles @xcite . we note that this form factor is hardly interpretative without considering the angular momentum . it is indispensable to take into account the angular momentum of the black hole when we perform a realistic calculation of its production and evaporation . black holes radiate mainly into the standard model fields that are localized on the brane @xcite . in the previous paper @xcite , we have shown that the massless brane field equations with spin zero , one - half and one , i.e. for all the standard model fields , are of variables separable type in the rotating black hole background . then we obtained the analytic expressions for the greybody factor of @xmath4 ( randall - sundrum 1 ) black hole by solving the master equation under the low energy approximation of radiating field . in this work we present a generalized result to higher dimensional black hole in @xmath5 for brane localized scalar field without relying on the low energy approximation and discuss its physical implications . this paper is the longer version of the brief report @xcite . we also note that the related works by harris @xcite and by harris and kanti @xcite appeared more recently . a brane - localized scalar field @xmath6 in the higher @xmath7-dimensional rotating black hole background @xcite can be decomposed into the radial and angular parts @xmath8 and @xmath9 , respectively @xcite @xmath10 where the boyer - lindquist coordinate @xmath11 reduces to the spherical coordinate at spatial infinity and @xmath12 are the angular quantum numbers . the resultant equations are shown to be separable @xcite . the angular part @xmath13 obeys the equation for the spheroidal harmonics while the radial equation becomes @xmath14 ^ 2\over\delta } + 2ma\omega - a^2\omega^2-a \right]r & = 0 . \label{eq : teukolsky } \end{aligned}\ ] ] where @xmath15 with @xmath16 and @xmath17 being the horizon radius and the rotation parameter of the black hole , respectively . note that @xmath18 . the power spectrum of the hawking radiation is governed by , for each scalar mode , @xmath19 where @xmath20 and @xmath21 are the angular velocity and the hawking temperature of black hole @xmath22 with @xmath23 . the @xmath24 is the greybody factor that determines the departure from black body spectrum , which is the main object of this paper . one immediate observation is that the contribution from @xmath25 modes dominates over that from @xmath26 modes in rapidly rotating case : @xmath27 . the asymptotic forms of the radial wave function at the near horizon ( nh ) and far field ( ff ) limits , @xmath28 and @xmath29 respectively , are @xcite @xmath30 where @xmath31 and the tortoise coordinate @xmath32 is defined by @xmath33 for @xmath29 and @xmath34 we obtain the greybody factors in the following steps . 1 . put the purely ingoing boundary condition @xmath35 by imposing @xmath36 at the nh region @xmath37 . 2 . numerically integrate the master equation from the nh region to ff regime @xmath38 . 3 . perform a least squares fit to the obtained data by the function around @xmath38 to get @xmath39 and @xmath40 . finally greybody factor for the ( @xmath12 ) mode is given by the absorption rate @xmath41 for the angular eigenvalue @xmath42 , we employed the small @xmath43 expansions up to 6th order in @xcite . ( the last 6th order term in the expansion is less than a few percent of the leading order term at @xmath44 for the @xmath45 and @xmath46 modes and at @xmath47 for all the other modes . ) we have also performed above procedure in the ingoing kerr - newman coordinate as a cross - check . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50 , respectively . ] black hole for the brane scalar emission into the @xmath51 modes with @xmath52 and @xmath0 for the upper - left , upper - right and lower graphs , respectively . the @xmath53 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath51 modes with @xmath52 and @xmath0 for the upper - left , upper - right and lower graphs , respectively . the @xmath53 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath51 modes with @xmath52 and @xmath0 for the upper - left , upper - right and lower graphs , respectively . the @xmath53 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath55 modes with @xmath56 for the upper - left , upper - right , @xmath52 for the middle - left , middle - right , and @xmath57 for the lower graphs , respectively . the @xmath25 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath55 modes with @xmath56 for the upper - left , upper - right , @xmath52 for the middle - left , middle - right , and @xmath57 for the lower graphs , respectively . the @xmath25 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath55 modes with @xmath56 for the upper - left , upper - right , @xmath52 for the middle - left , middle - right , and @xmath57 for the lower graphs , respectively . the @xmath25 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath55 modes with @xmath56 for the upper - left , upper - right , @xmath52 for the middle - left , middle - right , and @xmath57 for the lower graphs , respectively . the @xmath25 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] black hole for the brane scalar emission into the @xmath55 modes with @xmath56 for the upper - left , upper - right , @xmath52 for the middle - left , middle - right , and @xmath57 for the lower graphs , respectively . the @xmath25 mode shows the superradiance , namely a negative greybody factor , in the low - frequency region @xmath54 . the red , green and blue curves correspond to @xmath48 , @xmath49 and @xmath50.,title="fig : " ] the numerical results of the greybody factors for the @xmath4 black hole is shown in figs . [ greybody_0]-[greybody_2 ] . they are in good agreement with the previous analytic expression in @xcite in the region @xmath59 . at @xmath60 with @xmath61 the bose statics factor diverges , but this divergence is regularized to give a finite emission rate due to the zero absorption rate at this point , where greybody factor cross the zero . as we claimed in ref . @xcite , our analytic expression in ref . @xcite correctly shows for which @xmath62 there emerges the super - radiance with the negative greybody factor @xmath63 . and @xmath64 . ( a ) , the green curve denotes the total power spectrum . dotted curves correspond to @xmath65 and @xmath66 modes , respectively . figure ( b ) shows that when the hole is rotating the total power spectrum is essentially determined by @xmath67 modes , with each peak corresponding to each angular mode ( see also fig . [ elmmodes]).,title="fig : " ] and @xmath64 . in fig . ( a ) , the green curve denotes the total power spectrum . dotted curves correspond to @xmath65 and @xmath66 modes , respectively . figure ( b ) shows that when the hole is rotating the total power spectrum is essentially determined by @xmath67 modes , with each peak corresponding to each angular mode ( see also fig . [ elmmodes]).,title="fig : " ] black hole with @xmath68 . the small concave curve denote the contribution from the sum of other than @xmath67 modes . we can safely neglect the other modes than @xmath67 modes . ] , 5 , 7 , 10 for the upper - left , upper - right , lower - left and lower - right plots , respectively . the red , yellow , green , blue and purple curves denote @xmath48 , @xmath69 , @xmath70 , @xmath71 and @xmath72 respectively . each peak corresponds to each angular mode with @xmath67 , see fig . [ powerspectrumtoshowmodes].,title="fig : " ] , 5 , 7 , 10 for the upper - left , upper - right , lower - left and lower - right plots , respectively . the red , yellow , green , blue and purple curves denote @xmath48 , @xmath69 , @xmath70 , @xmath71 and @xmath72 respectively . each peak corresponds to each angular mode with @xmath67 , see fig . [ powerspectrumtoshowmodes].,title="fig : " ] , 5 , 7 , 10 for the upper - left , upper - right , lower - left and lower - right plots , respectively . the red , yellow , green , blue and purple curves denote @xmath48 , @xmath69 , @xmath70 , @xmath71 and @xmath72 respectively . each peak corresponds to each angular mode with @xmath67 , see fig . [ powerspectrumtoshowmodes].,title="fig : " ] , 5 , 7 , 10 for the upper - left , upper - right , lower - left and lower - right plots , respectively . the red , yellow , green , blue and purple curves denote @xmath48 , @xmath69 , @xmath70 , @xmath71 and @xmath72 respectively . each peak corresponds to each angular mode with @xmath67 , see fig . [ powerspectrumtoshowmodes].,title="fig : " ] and @xmath64 for the upper - left , upper - right , lower - left , lower - right plots , respectively . the pink , red , yellow , green , cyan and blue curves correspond to @xmath73 and @xmath74 , respectively.,title="fig : " ] and @xmath64 for the upper - left , upper - right , lower - left , lower - right plots , respectively . the pink , red , yellow , green , cyan and blue curves correspond to @xmath73 and @xmath74 , respectively.,title="fig : " ] and @xmath64 for the upper - left , upper - right , lower - left , lower - right plots , respectively . the pink , red , yellow , green , cyan and blue curves correspond to @xmath73 and @xmath74 , respectively.,title="fig : " ] and @xmath64 for the upper - left , upper - right , lower - left , lower - right plots , respectively . the pink , red , yellow , green , cyan and blue curves correspond to @xmath73 and @xmath74 , respectively.,title="fig : " ] in fig . [ powerspectrumtoshowmodes ] we show the power spectrum for @xmath58 black hole . this shows that the spectrum is dominated by the @xmath67 modes . the result show that the domination of the @xmath67 mode is more significant for highly rotating case . as a check we plot the @xmath67 contribution and that from the other modes in fig . [ elmmodes ] . [ powerspectrums ] we plot using only the @xmath67 modes . we take the angular modes from @xmath75 to @xmath76 . in the low energy region ( @xmath77 ) the spectrum of the highly rotating black hole is suppressed compared with the non - rotating power spectrum , confirming the analytic result in low energy expansions @xcite , while in the higher energy regime the spectrum is greatly enhanced . this tendency is stronger for larger dimensions . to see that more explicitly , we also plot the power spectrum varying the number of dimensions in fig . [ fullpowerspectrum ] . black hole . the @xmath78 , @xmath49 , @xmath50 , @xmath64 from fig . ( a ) to ( d).,title="fig : " ] black hole . the @xmath78 , @xmath49 , @xmath50 , @xmath64 from fig . ( a ) to ( d).,title="fig : " ] black hole . the @xmath78 , @xmath49 , @xmath50 , @xmath64 from fig . ( a ) to ( d).,title="fig : " ] black hole . the @xmath78 , @xmath49 , @xmath50 , @xmath64 from fig . ( a ) to ( d).,title="fig : " ] in fig . [ angularpowerspectrum ] we plot the angular distribution for @xmath4 black holes . we approximate the spheroidal harmonics by the spherical harmonics with the assumption @xmath79 . we confirm the previous result in the low frequency approximation @xcite that the anisotropy is greatly enhanced for a highly rotating black hole , due to the @xmath80 mode . we have explained the importance of the angular momentum when one considers tev scale black hole production and evaporation . new numerical results are shown : greybody factor for the brane scalar emission from a general @xmath3 dimensional rotating black hole without relying on the low frequency expansions . the greybody factors are obtained for general @xmath81 dimensional cases and the various angular modes ( @xmath12 ) . we confirmed the nontrivial angular dependence of the scalar emission at the middle energy region @xmath82 and found that it is even more enhanced at higher energy region . to understand the actual evolution of a black hole and to predict the collider signature , we need further investigations . it is important to determine the greybody factors for spinor and vector fields . the evolution of the angular momentum and the mass can be determined once all the greybody factors are determined @xcite . in particular , the spin - down phase , whose time evolution has been impossible to determine so far , can be precisely described . one can in principle determine the angular momentum of the produced black hole from the nontrivial angular distribution of the signals . 99 j. r. david , g. mandal and s. r. wadia , `` microscopic formulation of black holes in string theory , '' phys . rept . * 369 * , 549 ( 2002 ) [ arxiv : hep - th/0203048 ] . t. padmanabhan , `` gravity and the thermodynamics of horizons , '' phys . rept . * 406 * , 49 ( 2005 ) [ arxiv : gr - qc/0311036 ] . g. veneziano , `` string - theoretic unitary s - matrix at the threshold of black - hole production , '' jhep * 0411 * , 001 ( 2004 ) [ arxiv : hep - th/0410166 ] . n. arkani - hamed , s. dimopoulos and g. r. dvali , `` the hierarchy problem and new dimensions at a millimeter , '' phys . b * 429 * , 263 ( 1998 ) [ arxiv : hep - ph/9803315 ] . l. randall and r. sundrum , `` a large mass hierarchy from a small extra dimension , '' phys . lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] . s. b. giddings and s. thomas , `` high energy colliders as black hole factories : the end of short distance physics , '' phys . d * 65 * , 056010 ( 2002 ) [ arxiv : hep - ph/0106219 ] . s. dimopoulos and g. landsberg , `` black holes at the lhc , '' phys . lett . * 87 * , 161602 ( 2001 ) [ arxiv : hep - ph/0106295 ] . p. c. argyres , s. dimopoulos and j. march - russell , `` black holes and sub - millimeter dimensions , '' phys . b * 441 * , 96 ( 1998 ) [ arxiv : hep - th/9808138 ] . t. banks and w. fischler , `` a model for high energy scattering in quantum gravity , '' arxiv : hep - th/9906038 . r. emparan , g. t. horowitz and r. c. myers , `` black holes radiate mainly on the brane , '' phys . rev . * 85 * , 499 ( 2000 ) [ arxiv : hep - th/0003118 ] . g. t hooft , `` graviton dominance in ultrahigh - energy scattering , '' phys . b * 198 * ( 1987 ) 61 . s. d. h. hsu , `` quantum production of black holes , '' phys . b * 555 * , 92 ( 2003 ) [ arxiv : hep - ph/0203154 ] . d. m. eardley and s. b. giddings , `` classical black hole production in high - energy collisions , '' phys . rev . d * 66 * , 044011 ( 2002 ) [ arxiv : gr - qc/0201034 ] . h. yoshino and y. nambu , `` black hole formation in the grazing collision of high - energy particles , '' phys . d * 67 * , 024009 ( 2003 ) [ arxiv : gr - qc/0209003 ] . s. b. giddings and v. s. rychkov , `` black holes from colliding wavepackets , '' phys . d * 70 * , 104026 ( 2004 ) [ arxiv : hep - th/0409131 ] . s. w. hawking , `` particle creation by black holes , '' commun . phys . * 43 * , 199 ( 1975 ) . d. ida , k. oda and s. c. park , `` rotating black holes at future colliders : greybody factors for brane fields , '' phys . d * 67 * , 064025 ( 2003 ) [ erratum - ibid . d * 69 * , 049901 ( 2004 ) ] [ arxiv : hep - th/0212108 ] . d. ida , k. oda and s. c. park , `` anisotropic scalar field emission from tev scale black hole , '' arxiv : hep - ph/0501210 . s. c. park and h. s. song , `` production of spinning black holes at colliders , '' j. korean phys . 43 * , 30 ( 2003 ) [ arxiv : hep - ph/0111069 ] . r. c. myers and m. j. perry , `` black holes in higher dimensional space - times , '' annals phys . * 172 * , 304 ( 1986 ) . e. seidel , `` a comment on the eigenvalues of spin weighted spheroidal functions , '' class . * 6 * , 1057 ( 1989 ) . c. m. harris , `` physics beyond the standard model : exotic leptons and black holes at future colliders , '' arxiv : hep - ph/0502005 . c. m. harris and p. kanti , `` hawking radiation from a ( 4+n)-dimensional rotating black hole , '' arxiv : hep - th/0503010 . d. n. page , `` particle emission rates from a black hole . ii . massless particles from a rotating hole , '' phys . rev . d * 14 * , 3260 ( 1976 ) . s. a. teukolsky and w. h. press , perturbation of a rotating black hole . interaction of the hole with gravitational and electromagnetic radiation , astrophys . j. * 193 * , 443 ( 1974 ) .	this is the sequel to the first paper of the series , where we have discussed the hawking radiation from five - dimensional rotating black holes for spin @xmath0 , @xmath1 and @xmath2 brane fields in the low frequency regime . we consider the emission of a brane localized scalar field from rotating black holes in general space - time dimensions without relying on the low frequency expansions .
in the canonical model , t tauri systems comprise the central star , a rotating disk of gas and dust , a jet or outflow and possibly a residual circumstellar envelope ( see e.g. @xcite ) . in many cases , the central star is still accreting material and this process , as well as the mechanisms driving the outflow , are dependent on and influence the properties of the inner disk ( @xmath01 au ) . several groups ( e.g. @xcite and @xcite ) have proposed models in which the stellar magnetic field truncates the disk at a few stellar radii . matter from the disk flows along the field lines and onto the star producing hot spots or rings that can explain observed ultraviolet photometric variability @xcite . in the last several years , the technique of long - baseline infrared interferometry has been applied to the study of circumstellar material around young stellar objects . these observations are sensitive to hot material near the star itself . given the milliarcsecond resolution capability of the current generation of interferometers , these observations can in many cases spatially resolve the emission from the hot ( a few thousand kelvin ) material and are well suited for observations of the inner regions of young stellar objects . the first young stellar object to be observed using this technique was fu ori @xcite , followed by herbig ae / be stars @xcite and t tauri stars @xcite(hereafter paper 1 ) . the fu ori results were consistent with accretion disk models , while both the t tauri and herbig star results found characteristic sizes larger than expected from geometrically flat accretion disk models . more recent observations of herbigs @xcite have found earlier spectral type objects which are consistent with accretion disk predictions . measurements of the spectral energy distribution ( sed ) at optical through radio wavelengths probe a range of processes in young stellar objects including the stellar photosphere , accretion onto the star or disk , emission from gas and dust in the disk and emission from the outflow . in many sources , continuum emission from circumstellar or accreting material adds to the stellar spectrum , decreasing the stellar spectral features in an effect called veiling . for t tauri stars , the veiling in the infrared can very high , indicating substantial excess emission ( see e.g @xcite ) . in paper 1 we presented observations showing that the infrared emission from the t tauri stars t tau n and su aur is resolved . the visibilities from t tauri stars can be difficult to model given the substantial stellar component , infrared variability and the possible presence of a significant extended component . in this paper , we present further interferometric observations of the t tauri stars t tau n , su aur , dr tau and ry tau using the palomar testbed interferometer ( pti ) and infrared photometry from the pomona college 1-meter telescope . in [ model ] , we present geometric models to constrain the emission size and orientation . in [ scatter ] , we present detailed source models which include the scattered light and reprocessing of starlight and dissipation of viscous accretion energy in the disk to fit both the sed and the infrared visibilities . all four sources are located in the taurus - auriga molecular cloud ( distance @xmath2 140 pc ) and are well studied t tauri objects . source spectral types and stellar properties given in table [ table : source ] are taken from recent references using infrared spectroscopy . due to the sensitivity restrictions of pti , we have chosen sources which are among the most infrared luminous t tauri objects . as the pti acquisition system works in the optical , there is a selection effect against highly inclined , optically obscured sources . .stellar parameters for the observed sources . [ table : source ] [ cols="<,^,<,<,<",options="header " , ] all four systems have significant emission in excess of the stellar photosphere from near infrared through millimeter wavelengths and all are believed to have circumstellar disks . the t tau system comprises the optically visible star t tau n and its infrared companion t tau s , which is itself a binary @xcite . the pti observations are of t tau n , the component which dominates the millimeter emission @xcite . su aur has an sed similar to that of t tau n , although @xcite classified su aur separately from other t tauri s due to its high luminosity and broad absorption lines . ry tau is associated with a reflection nebulosity @xcite and has millimeter - wave molecular line emission consistent with a keplerian disk @xcite . dr tau is one of the most heavily veiled t tauri stars and is highly variable in the optical @xcite and near - infrared @xcite . infrared interferometry data were taken at the palomar testbed interferometer ( pti ) , which is described in detail by @xcite . pti is a long - baseline , direct detection interferometer which utilizes active fringe tracking in the infrared . data presented here were obtained in the k band ( 2.2 @xmath3 m ) in all three pti baselines : ns ( 110 meter ) , nw ( 85 meter ) and sw ( 85 meters ) . in our analysis below , we also use the su aur observations described in @xcite and paper 1 . a summary of the new observations is given in table [ table : obs ] . these data were acquired over a period from 24 september 2001 to 16 october 2003 . the data in the ns and nw baselines were taken with a 20 millisecond fringe integration time , while the sw data were taken with a 50 millisecond time , providing better snr for these data . lllllll + & & & + & nights & ints & nights & ints & nights & ints + t tau n & & & & & 1 & 6 + su aur & & & & & 1 & 6 + dr tau & 3&5 & 1 & 3 & 1 & 4 + ry tau & 4&27 & 3 & 14 & 2 & 8 + + calibrator & size est.(mas ) & sources + hd 28024 & 0.68 & + hd 30111 & 0.60 & + hd 30122 & 0.11 & + hd 28677 & 0.34 & + hd 26737 & 0.24 & + the data were calibrated using the standard pti method @xcite . briefly , a synthetic wide - band channel is formed from five spectrometer channels ( @xmath4 ) . the system visibility , the response of the interferometer to an unresolved object , is measured using calibrator stars . the calibrator star sizes were estimated using a blackbody fit to photometric data from the literature and were checked to be internally consistent . calibrators were chosen for their proximity to the source and for small angular size , minimizing systematic errors in deriving the system visibility . all calibrators used here have angular diameters @xmath5 milliarcsecond ( mas ) and were assigned uncertainties of 0.1 mas ( table [ table : obs ] ) . the calibrated data are presented in normalized squared visibility ( v@xmath6=1 for an unresolved source ) , which we refer to as visibility in this paper . the calibrated visibility uncertainties are a combination of the calibrator size uncertainty and the internal scatter in the data . as dr tau is near the tracking limit for pti , the wide - band data are used rather than the synthetic wide - band ( spectral channel ) data . the main difference between these two channels is that the spectral channels are spatially filtered and the wide - band channel is not . the accuracy of the wide - band data were confirmed by comparing the wide - band and synthetic wide - band data for other sources observed on the same night as dr tau . the calibrated data were edited to remove integrations with very high jitter ( a measure of the phase noise ) and integrations for which the estimates of the system visibility from separate calibrator observations disagreed by more than 3@xmath7 . in general , the points eliminated were from entire nights with marginal weather or integrations taken at large hour angles . no more than 10% of the data for any given source was removed , except for dr tau on the ns baseline , and inclusion of these data points would not substantially change the results given below . the calibrated and edited data are shown in figure [ fig : data ] for each source as a function of projected baseline length and position angle . three of the four sources , t tau n , su aur and ry tau are clearly resolved . the new observations of t tau n are consistent with the results of @xcite . we have not calculated models for t tau n here ; scattered light models of t tau n that reproduce the observed asymmetry @xcite are detailed in @xcite . a sample of young stellar objects , including dr tau , su aur and ry tau and a sequence of photometric standard stars taken from @xcite were observed over seven nights from december 2003 to march 2004 ( december 16 , january 10 , 13 , 15 , 22 , 23 , and march 9 ) using the pomona college 1-m telescope with the clircam infrared camera in the j and k bands . for each object , a series of at least five dithered exposures was used to create individual sky images for each field . the sky background and instrumental noise was subtracted from all the images , and repeat exposures were median combined after shifting to a common astrometric reference frame to remove a majority of the background noise . instrumental magnitudes were converted to standard j and k magnitudes using a combination of the published magnitudes and the j and k magnitudes from bright 2mass stars in the image frames . the magnitudes given in table [ irphot : table ] are the average magnitudes over all six nights using the average calibration zero points from the complete sample of photometric standards and 2mass stars . the three sources observed showed no statistically significant variability over the nights observed , and were constant in magnitude within the photometric error of 0.15 magnitudes in j and 0.15 magnitudes in k. for comparison , the 2mass j and k magnitude and observation date are also given . together , the pomona and 2mass data bracket the pti observations . at k band , only ry tau shows a significant difference between the 2mass measurement and our more recent observations ; however , past observations of these sources have shown infrared variability ( particularly dr tau ; @xcite ) . additional information on the infrared observations from table [ irphot : table ] is presented in @xcite . lllllll source & # nights & mag & rms & 2mass mag & 2mass rms & 2mass date + + su aur & 6 & 7.24 & 0.143 & 7.20 & 0.020 & 1/30/98 + dr tau & 6 & 8.75 & 0.195 & 8.84 & 0.024 & 10/10/97 + ry tau & 5 & 7.52 & 0.226 & 7.15 & 0.019 & 10/29/97 + + su aur & 6 & 6.17 & 0.114 & 5.99 & 0.022 & 1/30/98 + dr tau & 6 & 6.87 & 0.183 & 6.87 & 0.017 & 10/10/97 + ry tau & 6 & 5.76 & 0.168 & 5.39 & 0.022 & 10/29/97 + in this section , we discuss geometric models for su aur , ry tau and dr tau . as pti is a direct detection interferometer , any emission within the 1 gaussian ( fwhm ) field of view will contribute to the measured visibility . as discussed in @xcite there are many scenarios that could produce a visibility of less than 1 . these include additional point sources within the field of view , a resolved source of emission , or extended ( over - resolved ) emission which will contribute incoherently . any possible incoherent contribution ( in this case any emission on scales greater than 10 mas and within the 1 fov ) is hard to assess for our sources , given that many observations of envelopes or reflection nebulae do not include the central arcsecond due to contamination from the star itself . none of these three sources has a known companion within 1 . dr tau was included in lunar occultation observations of @xcite and no detection was reported with a point source limiting magnitude of @xmath8 = 2.5 . models including scattered light contributions are presented in [ scatter ] and extended components are discussed in [ extended ] . for the model fitting in this section , we adopt a configuration of an unresolved point source ( these stars have diameters @xmath9 0.1 mas and therefore a v@xmath10 at pti ) and a resolved component . we take the contribution of the stellar component from measurements of the infrared veiling . for su aur and dr tau we use the k band veiling measurements of @xcite . for ry tau the @xcite value of @xmath11 ( where @xmath12 = f@xmath13/f@xmath14 ) is much less than the lower limit of @xmath15 from @xcite . for our adopted model ( point source + resolved component ) , the pti data and a value of @xmath16 are incompatible ( i.e. the point source contribution can not be that large and still produce the pti measurement ) and we therefore use @xmath17 for ry tau in the geometric fits . simple geometric models of the emission are used to characterize the source size and inclination . the two models presented here use a uniform disk and a thin ring to represent the emission profile . for the uniform disk , only the measured visibility was used to determine the disk radius . for the ring model , visibilities were calculated for a range of inner diameters and compared to the observed visibilities . for each ring diameter considered , the width was determined by matching the excess flux , derived using the measured k - band veiling , with a blackbody emission source at a temperature of 1600 k , the assumed dust destruction temperature @xcite . in these models , the dust destruction temperature controls the width of the ring , but affects the fit radius only through the shape of the model visibility curve . for example , changing the blackbody temperature of the ring from 1200 to 2000 k would change the fit radius for ry tau by 30% . both face - on and inclined geometries were fit to the data ( table [ table : fits ] ) . the uncertainties in the model fits due to the uncertainty in the stellar contribution are also given . llll & su aur & dr tau & ry tau + f@xmath18 & @xmath19 & 0.8 @xmath20 & @xmath21 + k@xmath22 ( 2mass ) & 5.99 & 6.87 & 5.40 + + + radius ( au ) & 0.20@xmath23 & 0.10@xmath24 & 0.29@xmath25 + @xmath26 ( au ) & 0.042 & 0.004 & 0.080 + @xmath27 & 2.5 & 0.85 & 2.9 + + inner radius ( au ) & 0.13@xmath28 & 0.057@xmath29 & 0.17@xmath30 + width ( au ) & 0.050 & 0.028 & 0.035 + @xmath26 ( au ) & 0.036 & 0.010 & 0.059 + @xmath27 & 2.5 & 0.85 & 4.6 + + + radius ( au ) & 0.27@xmath31 & 0.11@xmath32 & 0.30@xmath33 + pa ( degr ) & 112 @xmath34 & 160 @xmath35 & 98 @xmath36 + incl ( degr ) & 51 @xmath37 & 40 @xmath38 & 19 @xmath39 + @xmath27 & 0.9 & 0.77 & 2.3 + + inner radius ( au ) & 0.18 @xmath40 & 0.070 @xmath41 & 0.19@xmath30 + width ( au ) & 0.008 & 0.019 & 0.029 + pa ( degr ) & 114 @xmath42 & 160@xmath35 & 110@xmath43 + incl ( degr ) & 52 @xmath44 & 40 @xmath38 & 25 @xmath45 3 + @xmath27 & 0.9 & 0.78 & 3.0 + the ring model fits are graphically shown in figure [ fig : udpoint ] . in this sky plane representation , the radial coordinate for each data point is the inner ring size corresponding to the measured visibility and accounting for the stellar component listed in table [ table : fits ] . the error bars include the errors on the data points but not the uncertainty in the stellar contribution . the polar coordinate is determined by the projected baseline position angle . in this way , the constraint provided by the data on both the size and the inclination are visible . the best fit face - on and inclined ring models are also plotted . using the simple geometric models , we find source sizes ranging from 0.04 to 0.3 au in radius . as discussed in paper 1 and by @xcite , the measured sizes for t tauri stars and herbig ae stars were larger than expected from simple disk models . an explanation for this discrepancy in herbig stars was proposed independently by two groups based on sed modeling @xcite and aperture masking observations @xcite . in these models the inner edge of the dust disk is located at the radius where the dust reaches the sublimation temperature ( r@xmath46 ) . this configuration produces a vertically extended inner wall , reproducing both the sed and the interferometry observations for the herbig sources . @xcite also applied this model to t tauri stars . further work by @xcite extended the model to include the accretion luminosity as well as the stellar luminosity in determining the dust destruction radius for several t tauri stars , including the three shown in figure [ fig : udpoint ] . in all these models , optically thin gas may be present within r@xmath46 ( we discuss this point further in [ results ] ) . we chose a ring distribution as a simple representation of a model in which the infrared emission arises from the inner wall of the dust disk . the values for r@xmath46 predicted by @xcite are larger by roughly a factor of 2 than our fit ring radii . we note that the presence of extended emission which was not included in our model would decrease the fit radius , and would therefore increase this discrepancy . the fit ring radii for su aur and ry tau correspond to 10 r@xmath47 and 11 r@xmath47 , much larger than the expected magnetic truncation radius ( 3 5 r@xmath47 ; @xcite ) . in [ scatter ] we show that emission from gas between the magnetic truncation radius and r@xmath46 can reconcile accretion disk models with our observations . the position angle coverage of the pti data allow us to constrain the inclination of infrared emission . the @xmath27 improves substantially for su aur and ry tau for the inclined models as compared to the face - on models , but the dr tau data do not provide a good constraint on the inclination given the large error bars and because the source is at best marginally resolved . for ry tau , our inclination angle of 19 25 agrees with that derived by @xcite from resolved millimeter emission ( 25 ) . however , the position angle is not well constrained by our data ( 98 @xmath45 40 for the uniform disk and 110@xmath45 22 for the ring ) and does not agree with the pa of the millimeter emission ( 48 @xmath45 5 , @xcite ; 27 @xmath45 7 , @xcite ) and is not orthogonal to the jet pa of 110 from @xcite . our inclination angle of 52@xmath45 10 agrees with the 60 estimate of @xcite based on the photometric period and line widths . @xcite find high ( 86 ) inclination values for both ry tau and su aur , which are not supported by the pti data , particularly for ry tau , and are also inconsistent with the low visible extinctions ( a@xmath48=2.1 and 0.9 , respectively ) . at such high viewing angles , the star would be occulted by the flared circumstellar disk . if there is a large incoherent component for any of the sources , then our simple geometric fits will underestimate the inclination angle as an incoherent contribution is independent of baseline . however , in our detailed models ( [ scatter ] ) for these three objects , the extended light contribution is less than 10% , which is insufficient to change the measured ry tau inclination by 60 . one of the major uncertainties in the simple fits presented above is the assumption of no extended emission within the 1 pti field of view . to address this issue directly , we have calculated radiative transfer models for su aur , dr tau and ry tau . the input properties are given in table [ table : source ] and the goal is to match the pti data and the sed . we use the monte carlo radiative equilibrium technique of @xcite , updated by @xcite , to self - consistently model each of our target sources . this code iteratively solves for the disk density structure , assuming the dust and gas are well - mixed with a standard gas to dust ratio of 100:1 and the system is in vertical hydrostatic equilibrium . in addition to stellar irradiation , the code includes accretion and shock / boundary layer luminosity calculations according to @xcite . multiple scattering is treated alongside the heating and reprocessing of photons in the disk . output data can be used to produce synthetic seds and multi - wavelength images for any viewing angle of the disk system . for more detailed description of the code and its updates see @xcite and references therein . the code computes the flared density structure of a steady accretion disk extending from the inner dust destruction radius to a specified outer radius ( figure [ density : plot ] ) . the monte carlo technique naturally accounts for radiation transfer effects and the heating and hydrostatic structure of the inner wall of the dust disk . the vertical height of the inner wall of dust is not preset , but rather calculated as part of the modeling process . for these models the scale height of the density distribution is 0.3 to 0.7 r@xmath47 at the inner edge . the position of the inner dust disk edge , r@xmath49 is determined from the destruction temperature of silicates , taken to be 1600 k @xcite . within the disk we adopt the dust - size distribution used for the modeling of hh30 irs and gm aur @xcite . with a distribution of grain sizes or compositions , the dust destruction may take place over a range of radii , but this is beyond the scope of our work . @xcite discuss the constraints on the dust properties from infrared interferometry observations . in order to match the new pti observations , r@xmath46 for some sources was large enough ( @xmath50 0.2 au ) that continuum emission from gas within r@xmath46 becomes significant . the structure and temperature of the gas disk is not computed self - consistently in our models . instead , accretion luminosity is emitted following the temperature structure of an optically thick accretion disk @xmath51 ( e.g. , @xcite ) where @xmath52 is the radial distance in the mid - plane . the gas disk is assumed to be infinitely thin , so after being emitted , the `` accretion photons '' do not encounter any opacity in the gas , but may be scattered and absorbed , and produce heating in the dust disk . clearly this is a simplification for the gas emission , but is sufficient for our models . the assumed geometry of the gas disk is supported by recent modeling by @xcite of herbig ae / be sources in which the gas disk is geometrically thin , allowing direct radiation of the inner dust disk . future work will investigate the effects of possible shielding of the dust disk by a flared and possibly optically thick , inner gas disk . the gas disk extends down to the magnetic truncation radius ( r@xmath53 ) at which point , material is thought to be channeled along magnetic field lines onto the star at a high latitude shock zone ( e.g. @xcite , @xcite ) . we assume the gas disk is truncated at a magnetospheric radius dependent on the stellar radius , mass , accretion rate and surface magnetic field @xcite . for dr tau and ry tau we assume kilogauss magnetic fields and truncate the gas disk at 5 r@xmath47 . for su aur , thought to be more weakly magnetic and with inconclusive evidence for hot spots @xcite we use 2 r@xmath47 for r@xmath54 . we assume photons emitted from the shock / boundary layers have a spectrum of an 8000 k planck function @xcite and are emitted along with stellar photons as in @xcite . for each model , the stellar luminosity has been fixed as detailed in table 1 and for input stellar spectra we use the appropriate kurucz ( 1994 ) model atmosphere . we also used fixed stellar masses of 2.25 m@xmath55 , 2m@xmath55 , and 1 m@xmath55 for su aur , ry tau and dr tau , respectively @xcite ; note that the stellar mass is not a critical parameter in the near - infrared . the disk properties such as mass , accretion rate and inclination were varied in order to produce a grid of synthetic seds . these models allowed us to explore likely parameter configurations . to calculate model visibilities , a simulated k band image was created using the monte carlo models with pixel size 0.05 mas and a width of 12.5 mas . the pixel size was chosen to be much smaller than the fringe spacing of 4 mas and the total size was a compromise between calculation time ( large images are computationally intensive ) and capturing the relevant structure . the outer size is large enough to contain any component which would contribute substantially to the model visibility . for example , a thin ring with an inner radius of 3 mas has v@xmath6=0.01 on the shortest pti baseline . the k band emission in the models is dominated by structures a few mas in size or less ( figure [ model : plot ] ) . as discussed in [ model ] , any emission within the 1 fov will contribute incoherently . to calculate the extended component in the model , a larger image is also constructed with 2 mas pixels and a 1 field . the emission outside the central 12 mas is calculated and included in the visibility calculation as an incoherent contribution . the effects of the 1 gaussian field of view and the finite fringe envelope are also included in the visibility calculation . for each pti baseline , the model visibility , including the incoherent flux , was calculated for the average baseline length and position angle using the fourier transform of the image , assuming a position angle for the disk as given in [ model ] . we thus `` observe '' the models as they would be at pti . the model visibilities and full seds were then compared to the data presented in [ observations ] and sed data taken from the literature . the optical and infrared sed data are taken from the compilation of @xcite and are not contemporaneous but instead represent an average for each source and the millimeter data are taken from @xcite and @xcite . for each object a set of models was calculated to explore the disk parameters to find a viable model and an example selection of models for each object is given in table [ model : table ] . the parameter space chosen for the inclination angle was restricted using the results from the geometric fits . the model with the total lowest @xmath56 is shown for each object in fig [ model : plot ] and the corresponding sed fit in fig [ fig : sed ] . llllllllll model & @xmath57 & r@xmath58 & @xmath59 & incl & lacc+shk & @xmath60 & @xmath61 & @xmath62 & notes + & m@xmath55/yr & au & m@xmath55 & deg & l@xmath55 + + su - a & @xmath63 & 0.21 & 0.001 & 60 & 0.02 & 7 & 118 & 125 + su - b & @xmath63 & 0.21 & 0.001 & 50 & 0.02 & 29 & 101 & 130 + su - c & @xmath63 & 0.22 & 0.001 & 50 & 0.02 & 60 & 198 & 258 + su - d & @xmath64 & 0.21 & 0.001 & 50 & 0.02 & 48 & 163 & 211 + su - e & @xmath65 & 0.24 & 0.005 & 50 & 0.02 & 99 & 893 & 992 + + dr - a & @xmath66 & 0.09 & 0.16 & 30 & 1.3 & 12 & 24 & 36 & r@xmath67 + dr - b & @xmath66 & 0.09 & 0.16 & 30 & 1.3 & 30 & 19 & 49 + dr - c & @xmath66 & 0.09 & 0.12 & 30 & 1.3 & 30 & 59 & 89 + dr - d & @xmath68 & 0.09 & 0.15 & 60 & 0.97 & 9 & 55 & 63 + dr - e & 0 & 0.08 & 0.08 & 60 & 0 & 3 & 82 & 85 & no accretion + + ry - a & @xmath69 & 0.27 & 0.015 & 25 & 4.28 & 8 & 37 & 45 + ry - b & @xmath70 & 0.27 & 0.015 & 25 & 5.13 & 1 & 294 & 295 + ry - c & @xmath71 & 0.27 & 0.012 & 25 & 3.42 & 135 & 31 & 166 + ry - d & @xmath69 & 0.27 & 0.015 & 25 & 4.28 & 4 & 145 & 149 & no gas + ry - e & @xmath69 & 0.27 & 0.015 & 25 & 4.28 & 36 & 313 & 349 & envelope + before discussing the results of the monte carlo models , we discuss other possible physical components which have not been included in our models . as discussed in [ model ] , a binary companion will contribute coherently or incoherently to the measured visibility depending on the separation from the primary . ry tau was classified as a `` variability induced mover '' from hipparcos data and @xcite found a solution in which the possible companion had a minimum separation of 24 mas . however , the k band speckle interferometry survey of @xcite did not detect a companion for ry tau in the angular range of 01 to 10 and hst archival images of ry tau from wfpc2 show only a single point source . for the incoherent contribution from a companion to account entirely for the measured visibility the k band flux ratio would be 0.81 to 1.44 ( secondary / primary ) , but this could not account for the change in visibility with baseline length and orientation . another likely contributor of infrared emission is a circumstellar envelope . an envelope can be a source of scattered and thermally reprocessed starlight and can also veil emission from the central star and accretion disk ( see e.g. the models of @xcite ; @xcite ) . however , su aur , dr tau and ry tau are all class ii t tauri stars and have visual extinctions @xmath72 2 . in general , class ii sources are thought to have little or no envelope remaining ( see e.g. @xcite ) . ry tau shows near - infrared co lines in absorption @xcite which @xcite cite as evidence of no substantial envelope for other class ii sources . to assess the possible presence of emission within the 1 pti fov , we examined hst archival images from the standard imaging pipeline . for each source , we found wfpc2 images taken with the f814w filter on the planetary camera ccd . no extended emission or additional sources were apparent in the images , however they are dominated by the central point source . azimuthal brightness averages were computed for comparison to a star extracted from the psf archive . the core of ry tau was saturated to such an extent that we were unable to find a matching saturated psf for comparison . although the scatter in the psf averages do not allow a precise comparison , especially in the core , su aur and dr tau are dominated by a central source ( figure [ psf ] ) . this is in contrast with the images of class i sources , such as the sample imaged by @xcite which show images dominated by scattered light from circumstellar material hundreds of au in extent . coronagraphic techniques have revealed some extended emission in ry tau @xcite and su aur @xcite but as detailed in the next section , this emission is unlikely to contribute substantially in the near - infrared . we have also calculated an example model with a disk and an envelope for ry tau ( [ rytau ] ) . as the wavelength range used for the sed comparison ranges from 0.365 microns to 3 millimeters , the model sed is sensitive to many model parameters , from the extinction to the outer disk size and mass . as expected , the infrared visibility is sensitive to only a few model parameters , particularly the inner radius , the inclination angle and the luminosity . each object is considered separately below , but the general conclusion is that these models , which include the contribution from extended emission , support the simple geometric models in the large value of r@xmath46 found for ry tau and su aur . in part this is because the models contain incoherent contributions at 2 microns ( here defined as flux from scales greater than 10 mas ) which were less than 6% for all sources , and could still reproduce both the sed and the infrared interferometry observations . high resolution infrared imaging observations would further constrain the extended emission component of these models . the k band excess flux from the models is also close to the veiling values used in [ model ] , with f@xmath18/f@xmath73 values of 0.4 , 0.74 and 0.68 respectively for su aur , dr tau and ry tau . the second general conclusion is that emission from gas within r@xmath49 is a significant component of the near - infrared emission if r@xmath46 is large . for our three objects this is most evident in the ry tau model . the relative flux of the gas and dust components for ry tau can be seen in figure [ fig : slice ] , which shows a cut through the model image with the inner dust wall facing the observer on the left in the plot . in comparison , for the dr tau model a smaller r@xmath46 is necessary to match the high visibilities measured at pti and so r@xmath46 and r@xmath54 are similar . for dr tau , a smaller value of r@xmath54 than estimated from the stellar properties ( 2 r@xmath47 instead of 5 r@xmath47 ) was necessary to match the data . we have not explored the value of r@xmath54 extensively in these models , so these values should be taken as approximate . observations by @xcite of co fundamental emission for several single t tauri stars similar to our targets ( e.g. bp tau ) found the inner co radius to be smaller than the calculated corotation radius for 5 out of 6 sources with co inner radii of 0.02 to 0.09 au . modeling of herbig ae / be sources by @xcite found emission from the inner gas exceeded the stellar emission for accretion rates @xmath74 m@xmath55/yr . to confirm the effect of the gas emission , a model was constructed for ry tau in which the gas was artificially removed from the region within r@xmath46 , which was set to match the measured pti visibilities . however , the sed fit for this model ( table [ model : table ] , model ry - d ) is not as good as for the model with gas emission . for su aur , both the sed and the measured visibilities are well fit by model su - a . figure [ model : plot ] shows the inner region of this model , with the star and the inner dust disk edge producing the k band emission . at an inclination angle of 60 the inner edge of the dust disk facing the observer is clearly brighter . we find r@xmath46 = 0.21 au , similar to the inclined ring model radius of 0.18 au from table [ table : fits ] . the gas emission is visible close to the star ( r@xmath75r@xmath47 ) but the emission is limited by the small surface area at this radius . recent observations by @xcite have traced extended emission at k out to radii of 26 . this study did not measure the scattered light within 1 so does not help constrain the pti data , but does suggest that a complete model for su aur would include an extended scattered component ; however , their measured k flux from 1 to 26 was only 4% of the 2mass k flux , so neglecting this component adds only a small error to the sed fit . it was not possible to fit both the pti data and the sed data well with these models . the best - fit model listed in table [ model : table ] underestimates the infrared visibility and underestimates the sed throughout the near and mid - infrared ( figure [ fig : sed ] ) . in order to produce an inner disk radius small enough to fit the pti data , the model must contain no accretion ( model dr - e ) , which drastically underestimates the sed and disagrees with the accretion diagnostics observed for the source @xcite . the infrared photometry ( table [ irphot : table ] ) does not reveal substantial variations recently , however the optical veiling for dr tau has been highly variable @xcite and as the sed data and pti measurements are not contemporaneous , there may be issues with source variability in our modeling . also note that the models in table [ model : table ] have a lower inclination ( @xmath76 ) than given by the geometric fits ( @xmath77 ) but within the uncertainty . we compared the photometry from @xcite which are averages of measurements from the literature to the optical and infrared contemporaneous sed from @xcite and the main deviation is slightly lower fluxes at @xmath78 for the contemporaneous sed , which does not improve our model fits . in the variability study by @xcite dr tau showed no trend in color with brightness changes , suggesting the variability is not due to large extinction changes . the variability of dr tau has been modeled as a hot spot on the stellar photosphere @xcite , but the models here do not attempt to model dr tau with that level of detail . ry tau has the largest r@xmath46 of the three sources and gas within r@xmath46 contributes substantially to the infrared emission . this second component in the disk emission means that the simple ring model is an _ underestimate _ of r@xmath46 . although this same gas component is present in the models for all three sources , the contribution to the k band flux is largest for ry tau as r@xmath46 is larger than for the other two sources . for ry tau , the best fit model was relatively close to the optimal parameters for both the pti and sed data . the pti data are best modeled by a higher total luminosity and accretion rate ( model ry - b ) than the sed data . a reflection nebulosity has been observed extending to @xmath240 from ry tau at visible wavelengths . however , this is unlikely to contribute substantially at k as the reflection component is only 2% of the total flux at 0.9 @xmath3 m and scattering decreases with increasing wavelength . to test the effect of an envelope on the predicted visibilities and sed , a model was calculated using the disk properties of model ry - a with an envelope using the same gas and dust radii , 0.01 times the disk mass and an infall rate of 1 @xmath79 m@xmath55/yr for the envelope . as seen in table [ model : table ] , this model does not fit either the sed or the visibilities as well . other models with higher envelope masses were also calculated and had even worse fits to the sed . it may be possible to better match the sed with a different disk and envelope combination , however we found no observational evidence for substantial near - infrared emission from an envelope . we note that @xcite have recently characterized ry tau as a g1 star , substantially earlier than previous spectral type determinations . however , their stellar properties ( r@xmath47 = 2.9 r@xmath55 , m@xmath47 = 2.0 m@xmath55 ) agree reasonably well with the values we used ( table [ table : source ] ) . we used a slightly lower effective temperature ( 5782 k compared to 5945 k ) and a higher luminosity ( 12.8 l@xmath55 compared to 9.6 l@xmath55 ) . a lower stellar luminosity would require a more massive disk and higher accretion to produce the same flux at longer wavelengths , but the general properties of a large r@xmath80 would not change . the accretion rate of the model presented here , @xmath69 m@xmath55/yr , is actually higher than the @xcite estimate of @xmath81 m@xmath55/yr . infrared interferometric observations of t tauri stars are used to constrain the inner disk properties . detailed models were presented for su aur , ry tau and dr tau to reproduce both the interferometry observations and the spectral energy distribution . for both the simple geometric fits to the interferometry data and the monte carlo disk models which include accretion and scattering , the inner dust radius ranges from 0.05 to 0.3 au . extended envelopes were not needed to reproduce the sed for these sources , although additional high resolution infrared images would help in constraining the extended emission on larger scales ( tens to hundreds of au ) . however , the significant variations in the visibility with baseline length and orientation seen for su aur and ry tau require a resolved component to be present as an extended component produces a constant visibility reduction . although the models parameters given here may not be a unique solution to the data set used , they are consistent with the domination of the near - infrared excess by thermal emission from the disk , as expected for class ii sources . the su aur model agrees well with the size derived from geometric fits to the interferometer data alone . the sed and pti visibilities for dr tau can not both be fit with these models , perhaps due to source variability . for ry tau , the model predicts significant emission in the k band from gas within the inner dust disk radius . this gas emission at infrared wavelengths is generally not considered in the simple models used to fit interferometric data and when this emission is present results in an underestimate of the dust radius when using a simple ring model . future work will extend the observational database to study more sources and probe the innermost regions of disks . interferometric observations at a second wavelength would add additional constraints on the inner disk structure . from a theoretical perspective , we are extending the monte carlo codes to include gas opacity in the inner disk and self - consistently calculate its structure . the data and models presented in this paper clearly show that inner gas disks can not be ignored and are required when fitting observations that probe the inner regions of disks . this work was performed at the michelson science center , california institute of technology , under a contract with the national aeronautics and space administration . data were obtained at the palomar observatory using the nasa palomar testbed interferometer . science operations with pti are possible through the support of the pti collaboration(http://huey.jpl.nasa.gov / palomar / ptimembers.html ) and the efforts of kevin rykoski . jae acknowledges support from a michelson graduate research fellowship . this work has made use of software produced by the michelson science center . pomona college would like to acknowledge the support of the nsf ari and ccli grants in providing funds for development of the infrared camera and the pomona college 1-meter telescope . this work has made use of the simbad database , operated at cds , strasbourg , france , and the nasa / ipac infrared science archive , operated by the jpl under contract with nasa . this work utilizes observations made with the nasa / esa hubble space telescope , obtained from the data archive at the space telescope science institute . stsci is operated by the association of universities for research in astronomy , inc . under nasa contract nas 5 - 26555 .	we present observations of four t tauri stars using long baseline infrared interferometry from the palomar testbed interferometer . the target sources , t tau n , su aur , ry tau and dr tau , are all known to be surrounded by dusty circumstellar disks . the observations directly trace the inner regions ( @xmath01 au ) of the disk and can be used to constrain the physical properties of this material . for three of the sources observed , the infrared emission is clearly resolved . we first use geometric models to characterize the emission region size , which ranges from 0.04 to 0.3 au in radius . we then use monte carlo radiation transfer models of accretion disks to jointly model the spectral energy distribution and the interferometric observations with disk models including accretion and scattering . with these models , we are able to reproduce the data set with extended emission arising from structures larger than 10 milliarcseconds contributing less than 6% of the k band emission , consistent with there being little or no envelope remaining for these class ii sources ( @xmath1 -2 to 0 in the infrared ) . the radiation transfer models have inner radii for the dust similar to the geometric models ; however , for ry tau emission from gas within the inner dust radius contributes significantly to the model flux and visibility at infrared wavelengths . the main conclusion of our modeling is that emission from inner gas disks ( between the magnetic truncation radius and the dust destruction radius ) can be a significant component in the inner disk flux for sources with large inner dust radii .
the electron transport through nanometric conductors attract a huge interest due to constant shrinkage of microelectronic devices.@xcite in particular , metal nanowires ( nw ) display interesting quantum conductance behavior even at room temperature.@xcite from a practical point of view , nw s can be easily generated by putting in contact two metal surfaces , which are subsequently pulled apart . during the nw elongation and just before rupture , the conductance displays flat plateaus and abrupt jumps , which for metals such as au , take a value of approximately one conductance quantum @xmath0 = 2@xmath1 ( where @xmath2 is the electron charge and @xmath3 is planck s constant ) . in spite of the simplicity of the experimental procedure , a new structure with a different evolution is observed for each nw generation and all conductance curves have plateaus and jumps , but they display disparate profiles.@xcite in order to overcome this difficulty , a simple statistical method has been usually applied to analyze the average behavior . instead of considering the conductance as a function of the elongation , the transport properties can be represented as a histogram of conductance occurrence , in such a way that a flat plateau generates a histogram peak.@xcite by linearly adding the histograms associated to each conductance curve , a global histogram is generated , which describes the general tendencies of an ensemble of nw s . the global histogram displays well defined peaks close to the integer multiples of the conductance quantum ; this fact has been adopted as a proof of the tendency to conductance quantization in metal nws.@xcite the statistical method , discussed above , provides information on the average behavior but it hinders the study of nw transport properties . for example , it is not possible to get the detailed information on how structural factors influence the conductance evolution . for several years , the structure evolution was derived from molecular dynamics simulations , where the high computer cost imposes the use of simplified potentials based on effective medium theory ; @xcite subsequently free electron methods were applied to estimate the conduction of the metallic neck ( or confining potential ) . more precise methods , considering the electronic structure , have also been applied to calculate the nw conductance , but for static atomic configurations.@xcite recently , _ in situ _ high resolution transmission electron microscopy ( hrtem ) experiments have provided a new insight in the field . for example , rodrigues _ _ @xcite have showed that just before rupture , gold nw s are crystalline and free of defects and they assume only three kinds of atomic arrangements : two of them form bipyramidal constrictions which evolve to one - atom - thick contacts , while the other one generates rod - like nw s that break when they are rather thick ( three - to - four atoms ) . by considering that et al._@xcite have already shown that atom size contacts display a conductance of 1 @xmath0 , it is possible to discriminate between the rod - like and pyramidal nw morphologies . further , the relationship between each nw type and electrical transport measurements was obtained by simple crystallographic arguments . validation of the initial assumption was obtained by statistically comparing the occurrence of observed curve profiles . although these important developments mostly focused on the last conductance plateau , a quantitative understanding of the correlation between atomic structure and conductance during the nanowire stretching is still lacking . in this work , we have addressed the connection between gold nw structure and the quantized conductance behavior during the nw elongation . we have used hrtem to obtain the detailed information of the atomic structure evolution of gold nw s during stretching , and using crystallographic arguments , we proposed the three dimensional structure of these nanostructures . the nw conductance was measured using an independent , dedicated experimental set - up : a mechanically controlled break junction operated in ultra - high - vacuum ( uhv - mcbj ) . in order to correlate both sets of data , we have used a semi - empirical atomistic theoretical technique based on the extended hckel theory@xcite ( eht ) that allows for the estimation of transport properties of systems with a large number of atoms.@xcite the results displayed an excellent agreement between observed structural and conductance experiments and theoretical calculations , enabling a complete modeling of the nw elongation process both from the structural and transport properties points of view . the remaining of this work is organized as follows : section [ experiment ] describes the experimental techniques used to obtain the structural and transport data , respectively ; section [ theory ] describes the theoretical model employed to calculate the electronic quantum transport through the metallic nw s ; the experimental results are presented in section [ expresul ] and the analysis of the data via theoretical analysis is presented in sections [ discuss1 ] for nw s oriented along the [ 100 ] , [ 111 ] and [ 110 ] directions as well as [ discuss2 ] , where the presence of impurities is investigated ; finally , in section [ conclu ] we draw our conclusions . we have generated nw s _ in situ _ in a hrtem ( jem 3010 uhr , operated at 300 kv , point resolution 1.7 ) using the method developed by takayanagi s group.@xcite the procedure consists in focusing the microscope electron beam ( current density 120 a/@xmath4 ) on the sample to perforate and grow neighboring holes until a nanometric bridge is formed between two of them . we have used a self - supported polycrystalline gold thin film ( 5 nm thick , deposited on a holey carbon grid ) in order to generate nw s between apexes of different orientations and elongate them in different directions.@xcite atomic resolution image acquisition has been performed after reducing the electron beam intensity to its usual value ( 30 a/@xmath4 ) . a high sensitivity tv camera ( gatan 622sc , 30 frames / s ) associated with a conventional video recorder was used to register nw real - time evolution . the procedure described above , allows us to generate nw s with a remarkable stability , because the nw , its apexes and the surrounding thin film regions , form a monolithic block . in consequence , the nw s formed by a few atomic layers usually show a long life time ( 1 - 10 minutes ) . despite this stability , the generated nw s elongate spontaneously , get thinner and then break due to the relative slow movement of the nw apexes . these apexes displacement are probably due to a film deformation induced by thermal gradients between parts of the sample , as usually observed in tem thin film work . a critical aspect when studying such tiny nanostructures is the presence of contaminants , the most critical one in tem is amorphous carbon.@xcite however , we must keep in mind , that the intense electron irradiation transforms carbon into bucky - onions and finally clean the gold surface.@xcite the electric transport properties of gold nw s were studied using an independent instrument specially designed for this purpose , namely an uhv - mcbj.@xcite in this approach , a macroscopic gold wire ( 99.99 @xmath5 pure , @xmath6 = 75 @xmath7 m ) is glued in a flexible substrate in two points ; then it is weakened in a point between the two fixing parts by an incomplete cut . by bending the substrate _ in situ _ in the uhv , we break the wire and produce two clean gold surfaces ; using the same bending movement , the fresh tips are put together and separated repeatedly in order to generated and deform nw s . it s important to remark that in this configuration the nw s are generated from surfaces obtained in uhv ( @xmath8 10@xmath9 pa ) , so it is expected to have a clean sample for few hours.@xcite this care with vacuum conditions is extremely important to generate reliable experimental measurements on nw properties.@xcite electrical measurements were done using a home - made voltage source , and current - voltage converter powered with isolated batteries to reduce electrical noise . data acquisition was based on an eight - bit digital oscilloscope ( tektronic tds540c , 2 gsample / s , bandwidth 500 mhz ) ; this electronic system was developed with the aim of improving the conductance measurement precision , yielding values with a relative error of ( @xmath10 ) @xmath11 10@xmath12 . the applied voltage was 100 mv , and the conductance was measured in the [ 0 , 2.8 ] @xmath13 range to improve the detection of the last two quantum conductance plateaus . we follow the approach introduced by emberly and kirczenow @xcite that is based on the extended hckel theory , which is a method traditionally used in quantum chemistry to calculate the electronic properties of clusters and molecules.@xcite one advantage of this approach lies in the possibility of treating the whole nanocontact both tips and the nanowire between them as a unique entity , described by molecular orbitals . the tips and the nanowire , upon which we base our transport calculations , are built on the basis of the crystallographic properties of the material as well as the obtained hrtem images ; no attempt was made to simulate the dynamics of the nw formation . in our calculations the nanocontacts are made up of as many as 150 gold atoms for some structures . in order to calculate the quantum electronic states of au clusters as big as these , a semi - empirical method is necessary to produce results within reasonable time . for this reason the eht is employed ; the method includes overlap terms and matrix elements that are not restricted to first neighbors ; moreover , all the valence states 5d , 6s and 6p , in the case of au are considered . one drawback is the fact that , in its standard form , the eht is a one - electron formalism that does not take into account explicitly electron - electron interaction effects . however , we believe that this problem can be partially circumvented , because of the parameterization of the orbital wavefunctions by self - consistent hartree - fock states and the use of semi - empirical energy parameters . such effects are more relevant for the analysis of the structural stability and relaxation of nanowires , where the calculation of energy manifolds is necessary . nonetheless , tight - binding together with molecular dynamics methods , similar to the present one , have been successfully used to simulate the dynamics of the nanowire formation.@xcite let us assume that a nanocontact is formed by two apexes ( or tips ) and the nanowire itself that connects them . we start the description by defining the molecular hamiltonian of the nanocontact , @xmath14 and its molecular orbitals ( mo s ) @xmath15 : @xmath16 in ( [ mo ] ) the sum over @xmath17 extends over the @xmath18 , @xmath19 and @xmath20 orbitals and the index @xmath21 covers all the au atoms in the nanocontact . the molecular orbitals @xmath15 , written as a linear combination of atomic orbitals in ( [ mo ] ) , obey the orthonormality relation @xmath22 . equation ( [ e_nc ] ) defines the mo energies . so that these mo s bear quantum phase coherence along the entire nanocontact . to account for electron transport the nanocontact is coupled to infinite au leads on both sides ( left ( @xmath23 ) and right ( @xmath24 ) of the nanocontact ) , which behave as charge reservoirs ( see fig . [ drawing ] ) . the leads are divided into unit cells whose mo s are also obtained by the eht method : @xmath25 for the sake of simplicity only @xmath19 au orbitals are used in ( [ bloch ] ) . the hamiltonian @xmath26 includes overlap terms and is not restricted to first neighbor matrix elements within the unit cell . we consider this approximation because the contacts are expected to behave as bulk gold , hence s orbitals are adequate for describing its electronic properties . in the nanoconstriction , where the number of atoms is smaller , we expect a different behavior . although @xmath27 , it is noteworthy that the scalar product between both types of mo s define the overlap term : @xmath28 having defined the basic physical quantities , let us consider the total hamiltonian @xmath29 the term @xmath30 in ( [ h ] ) is responsible for coupling the nanocontact with the first unit cell of the lead on its left and @xmath31 , similarly , with respect to the first unit cell on the right . in principle @xmath32 and @xmath33 can be different from each other , as in the case of tips with distinct crystallographic orientations or defects . to account for the transport calculations consider bloch electrons that propagate through the leads : @xmath34 with @xmath35 as one of the mo s of the unit cell at position @xmath36 and @xmath37 . at the site of the nanocontact the incoming bloch wave is scattered giving rise to reflected and transmitted waves , according to the framework of the multi - channel landauer formalism.@xcite therefore , for a rightward incoming electron we write the following wavefunction @xmath38 , where @xmath39 the incoming electron has energy @xmath40 in order to be a propagating state . in expression ( [ psi_nc ] ) , @xmath41 represents the @xmath6-mo s of the nanocontact , which occupies the site @xmath42 , and @xmath43 are unknown coefficients . similarly , in ( [ psi_l ] ) and ( [ psi_r ] ) , @xmath44 is the @xmath45-mo of the unit cell at position @xmath36 ; @xmath46 and @xmath47 are the channel reflection and transmission coefficients , to be determined in the sequence , which will be used to calculate the total transmission @xmath48 and reflection @xmath49 of electrons propagating through the nanocontact . the parameter @xmath50 is a dimensionless wave - vector for the bloch waves in the leads . in order to determine the coefficients @xmath51 a complete set of mo s @xmath52 is put together , where we project the equation @xmath53 . as a result we obtain a non - homogeneous system of coupled equations that can be solved numerically to yield these coefficients . a detailed description is provided in the appendix . concluding the procedure , we define the total transmission and reflection coefficients , which are written in terms of the incoming and outgoing coefficients and the electron velocity @xmath54 of the respective bloch wave mode : @xmath55 where the prime means that the sum is made exclusively over the propagating modes , that is , @xmath56 . finally , the landauer formula @xcite is used to determine the charge transport through the nanocontact . for a degenerate metallic system in the linear response regime , the electrochemical potential difference between source and drain takes the limit @xmath57 and the conductance @xmath58 becomes @xmath59 where @xmath60 is the fermi - dirac distribution and @xmath61 is the fermi energy . since the voltage applied across the leads is small compared to @xmath61 the conductance is to be considered when the electron energy is close to this value . the arrows show the existence of peaks below 3 @xmath0 . the insets of every figure show typical conductance curves for each direction.,width=321 ] figure [ hist](a ) shows the global histogram generated from a series of 500 collected conductance curves of gold nw s measured using the uhv - mcbj . three peaks can be identified in the histogram ; as expected the major one is close to 1 @xmath0 and , the two minor ones are at 1.8 and 2.8 @xmath0 . et al._@xcite have demonstrated that analyzing the occurrence of the last two conductance plateaus ( around 1 and 2 @xmath0 ) , the conductance curves can be grouped in three different classes : a ) only a plateau at 1 @xmath0 b ) plateaus at 1 and 2 @xmath0 ; and , finally , c ) no plateau at 1 @xmath0 . these characteristics are associated with the crystallographic direction along which the nw is being elongated , corresponding to the [ 100 ] , [ 111 ] and [ 110 ] , respectively . as the global histogram superposes and hides this structural information , we have built three histograms formed by each group of conductance curves ( fig . [ hist](b - d ) ) , examples of the conductance curves are presented at the insets ) . this approach allows us to obtain a statistical validation for the sequence of conductance plateaus appearing during the nw elongation along each axis , extending the range from 2 to 3 @xmath0 . in these terms , it can be deduced that the thinning of a nw along [ 100 ] axis ( hereafter denoted [ 100 ] nw ) generates a conductance curve with plateaus close to 3 and 1 @xmath0 ( see curves inset fig . 2(b ) ) ; a [ 111 ] nw , plateaus around 3 , 2 and 1 @xmath0 and ; finally , for [ 110 ] nw s , plateaus around 3 and 2 @xmath0 . -oriented nw ; note the existence of two misaligned planes just before the formation of the atc . ( e ) atc oriented along the [ 111 ] direction and ( f ) rod - like structure formed along the [ 110 ] direction.,width=321 ] figure [ hrtem ] shows typical hrtem micrographs of the three possible au nw structures ; a common feature of all the observed nw s is the fact that they are defect free crystals . figures [ hrtem](a - d ) show snapshots of the elongation process of a nw along the [ 100 ] direction , where the progressive thinning of the constriction can be followed with atomic resolution . we can see from these images that the nw stretches from a symmetrical bipyramidal nw with a two - atom wide minimal section ( fig . [ hrtem](b ) ) to a constriction formed by two misaligned truncated pyramids with a two atom wide top ( fig . [ hrtem](c ) ) . finally , a one - atom - thick contact is formed ( fig . [ hrtem](d ) ) ; this contact subsequently evolves to form a suspended chain of atoms ( atc ) before rupture ( not shown ) . figure [ hrtem](e ) allows the clear visualization of the bipyramidal morphology of a one - atom - thick contact formed for [ 111 ] nw s . in contrast , [ 110 ] nw s generate rod - like structures before rupture ( fig . [ hrtem](f ) ) . at this point , it is useful to remind some basic aspects of the reported _ in situ _ hrtem experiments . firstly , hrtem images of nw s can be described as a two - dimensional projection of the atomic potential , then good quality atomic resolution information is only obtained from crystals very well oriented in relation to the microscope electron beam . in our experiments , we can not easily control the orientation of the apexes forming the nanometric constrictions , so the possibility of getting sequences as shown in fig . [ hrtem](a - d ) ( [ 100 ] nw ) is statistical and unpredictable . unfortunately , this kind of detailed information on the atomic arrangement evolution could not be observed for the other two nw types of nanowires ( [ 111 ] and [ 110 ] ) . another limiting factor of our dynamical hrtem observations is the temporal sampling of our image acquisition system allowing the capture of 30 frames per second ; very rapid events ( @xmath8 1/30 s ) may go by undetected . despite these difficulties , hrtem images provide a wealth of remarkable information to derive the atomic structure of gold nw s from . by combining the atomic resolution micrographs ( two dimensional projections of the structure ) and the known surface energy properties of the analyzed systems it is possible to deduce the three dimensional atomic arrangement of nw s . this procedure is based on the so - called wulff construction that allows a quick geometrical determination of the faceting pattern of nanostructures;@xcite such approach has already been successfully applied to study au,@xcite ag@xcite and pt@xcite nw s . in the following sections , we present a careful and thorough analysis of the derived atomic arrangements and their evolution for the three kinds of au nw s ; the theoretically obtained electronic transport properties of such structures are compared with the conductance behavior observed with the uhv - mcbj experiment . the notation indicates the number of atoms in the planes perpendicular to the [ 100 ] axis in the constriction region : ( a ) 941 , ( b ) 949 and ( c ) 9449 tips . the dashed lines in ( a ) indicate the edge of the fcc unit - cell and a@xmath62 is the lattice parameter.,width=283 ] the [ 100 ] nanocontacts are generated by two pyramidal tips , each of them faceted by four low energy ( 111 ) planes ; they can be elongated until atomic - chains are formed , having generally two - to - four atoms in length.@xcite figure [ 100 ] illustrates three nanocontact geometries based on the crystallographic structure of face - centered - cubic ( fcc ) bulk gold and the wulff method . these structures simulate the steps of the nw evolution in the experiments . figure [ 100](a ) shows the cross - section ( [ 100 ] direction ) and side view ( [ 011 ] direction ) of an atomically sharp [ 100 ] nw tip ; the side view presented represents the structural arrangement as seen from hrtem images in fig . [ hrtem](a - d ) . this particular tip is formed by the stacking of a plane of 9 atoms ( black circles ) followed by a plane of 4 atoms ( dark gray circles ) and , finally , one atom at the end of the tip ( light gray circle ) . the notation used hereafter to designate the structures is given by the number of atoms that make up the alternate planes stacked along the nw orientation axis . the one - atom - thick contact ( 941149 ) which is the structure observed just before rupture ( example is presented in fig . [ hrtem](e ) ) is then formed by mirror imaging the tip arrangement described above . we can use this pyramidal tip arrangement to model the thicker nw s observed prior to the formation of the 941149 structure , _ i. e. _ , we can build thicker structures by removing planes of atoms from the tip arrangement in fig . [ 100](a ) . hence , in simple terms , we would expect that the second thinnest [ 100 ] constriction would be formed by pyramids where the one atom tip is removed leading to a minimum cross - section of four atoms . by joining two pyramids sharing the four - atom plane we generate the 949 neck shown in fig . [ 100](b ) and corresponding to the experimental results in fig . [ hrtem](b ) . however , hrtem data clearly show the existence of an intermediate structure ( fig . [ hrtem](c ) ) where the two truncated pyramids forming the apexes are not aligned with respect to each other . assuming that all the neck region is defect free ( _ i . e. _ a perfect arrangement of atoms in a fcc lattice ) and following the wullf morphology we must conclude that the nw is formed by two four - atom planes which are not not aligned with each other , rather the corner of one of these planes is located at the central axis of the other and vice - versa . this structure is presented in fig . [ 100](c ) and labeled 9449 . ) as a function of the electron energy for various nw structures oriented along the [ 100 ] direction : 941149 ( the one - atom - thick contact ) , the 9449 and 949 nanocontacts . the vertical line indicates the fermi energy ; the conductance is in units of the quantum @xmath0.,width=245 ] using the formalism of section [ theory ] , we have calculated the quantum conductance for the nw morphologies shown in fig . [ 100 ] as a function of the electron energy . the results are shown in fig . [ theo-100 ] , where the vertical line indicates the value of the fermi energy in the leads . at first , we notice that the one - atom - thick contact ( 941149 , distance between the atoms is 2.88 even at the contact ) shows a conductance curve that yields a very stable plateau at @xmath63 = 1 @xmath0 which corresponds to a peak around 1 @xmath0 in the histograms ( fig . [ hist](b ) ) . on the other hand , the 9449 nanocontact display a conductance of @xmath63 @xmath11 3 @xmath0 , and the 949 , 4.5 @xmath0 . in brief , the one - atom - thick ( 941149 ) and the atc s conductance ( 9411149 , also calculated but not shown here ) is @xmath11 1 @xmath0 while the 9449 structure shows a conductance of @xmath11 3 @xmath0 . each of these conductance values agrees very well with the peaks in the global conductance histogram for the [ 100 ] zone axis ( fig . [ hist](b ) ) . in other words , he have accounted for the peaks in the histogram corroborating the assumption that the sequence of atomic arrangements shown in fig . [ 100 ] can , in fact , represent the dynamics of nw evolution along the [ 100 ] direction . it is also worth noticing that the deduced structural evolution shows no contribution around 2 @xmath0 , in agreement with the conductance histogram . finally , the 949 morphology presents @xmath63 @xmath11 4.5 @xmath0 , which would represent a thicker structure however , our experiments have not included conductance measurements in that range . direction . figure ( a ) shows the arrangement of the tip . the notation indicates the number of atoms in the planes perpendicular to the [ 111 ] axis in the constriction region : ( b ) 731137 , ( c ) 737 , ( d ) 7337 , and ( e ) 732237,width=321 ] for the case of [ 111 ] nw s , we have no atomic resolution hrtem images that show the structural evolution during the thinning process . however , we can model the structures using the same procedure of section [ [ 100]],@xcite _ i. e. _ , we can start with the one - atom - thick nanowires and make our way up towards thicker structures . again , the tip morphology was built from a fcc crystal . it is worth noting that the stacking of planes in the [ 111 ] direction is made by three hexagonal planes placed at alternate intersite spaces ( usually labeled abcabc@xmath64 ) . the derived [ 111 ] pyramidal morphology is presented in fig . [ this arrangement is obtained from the smallest possible fcc gold nanoparticle , _ i. e. _ , the cuboctahedral with regular hexagonal faces based on the wulff method . @xcite as it was done in the previous section , the one - atom - thick nw is formed by two aligned [ 111 ] apexes oriented in opposite directions . we can label this structure 731137 by following the notation introduced earlier . this structure is depicted in fig . [ 111](b ) . in analogy with the procedure adopted for the [ 100 ] direction , we would initially expect that the removal of the last atom from the tip and then by joining the remaining truncated pyramids we would have the arrangement before the one - atom - thick nw is formed . this particular arrangement is sketched in fig . [ 111](c ) and is named 737 . however , this arrangement is defective because it forms a twin at the neck . this twin should be unfavorable in the sense that the nw s are free of defects in this size range.@xcite . in that case , we proposed a structure analogous to the 9449 [ 100 ] nw which was presented in section [ [ 100 ] ] . this new structure is formed by adjoining two truncated [ 111 ] pyramids ( 73 ) rotated 180@xmath65 with respect to each other . the resulting arrangement has a perfect fcc stacking as shown in fig . [ finally , molecular dynamics simulations using classical potentials@xcite indicate the existence of a thinner region with two misaligned dimmers . as a matter of fact , this arrangement is formed by simply removing atoms from the apexes of the 7337 nw . this structure , labeled 732237 , is presented in fig . [ 111](e ) for atoms in a perfect fcc arrangement . as a function of the electron energy for various proposed nanocontact structures oriented along the [ 111 ] direction : the 737 , 7337 , 732237 and the one - atom - thick ( 731137 ) nanocontacts.,width=8 ] in fig . [ theo-111 ] , we present the theoretical conductance calculations for each of the proposed structures for [ 111 ] nw s . as previously stated , the experimental results must be compared to the calculated conductance at the fermi energy level , represented by the vertical line in the figure . although the 737 arrangement ( fig . [ 111](c ) ) would be a possible configuration displaying a conductance of @xmath11 2.8 @xmath0 , we must bear in mind that it presents a twin defect . hence we have neglected it in our analysis . the calculated conductance for the 7337 structure , which presents a perfect fcc stacking , is below 2.6 @xmath0 as shown in fig . [ theo-111 ] . conductance measurements for this orientation ( fig . [ hist](c ) ) , show the existence of a peak close to 2 @xmath0 . in this sense , the 732237 nw , results in @xmath63 @xmath11 1.9 @xmath0 as can be seen in fig . [ theo-111 ] . finally , structure 731137 yields @xmath66 = 1 @xmath0 ( fig . [ theo-111](d ) ) . one might be led to believe that the relaxation of the thinnest region ( .. 22 .. ) of the 732237 structure will rotate the two dimmers to form a perfect tetrahedron in order to maximize the number of neighbors . this deformation generates a nw with two [ 111 ] apexes bounding a [ 100]-oriented neck . this deformation decreases the conductance by a significant amount from 1.9 to @xmath11 1.2 @xmath0 ; emphasizing that nanowires are particularly sensitive to breaking symmetries and to defects in the neck region . in other words , the 7337 morphology has a conduction close to 2.6 @xmath0 , the 732237 nw shows the calculated conductance value @xmath63 @xmath11 1.9 @xmath0 and at last , for the one - atom - thick contact , @xmath63 @xmath11 1 @xmath0 . each of these structures can be associated with one of the peaks of the global conductance histogram of fig . [ hist](c ) , and to the evolution steps of nanowires along this direction . nw s at various stages of the thinning process . the atomic arrangements are labeled from the number of atoms forming the alternate stacking planes generating the rods : ( a ) 4/3 , ( b ) 2/2 , ( c ) 1/1 and ( d ) 2/0 . see the text for a definition of the notation.,width=245 ] the hrtem results show that compared to the other two kinds of nanowires , the [ 110 ] nw s are rather brittle , _ i. e. _ , they break when they are two - to - three atoms thick without forming atomic contacts . figures [ 110](a - d ) show the proposed structure of rod - like [ 110 ] nw s as they decrease in diameter , by loosing atomic planes@xcite . again , the notation used in naming these structures comes from the number of atoms contained in each plane forming the nw ( black and gray circles ) . however , in this case , due to their large aspect ratio , we show only the rod - like part and leave out the apexes . the thickest nw ( fig . [ 110](a ) ) , labeled 4/3 , is formed by two families of low - energy ( 111 ) facets and two high - energy ( 100 ) ones . by removing 3 lines of atoms from the previous structure we reduce it to the 2/2 nw . finally we consider two possible scenarios : the 2/2 nw is subtracted of one of the ( 111 ) planes ( atoms a@xmath67 and a@xmath68 ) , giving rise to the alternate 2-atom chain , 1/1 , shown in fig . [ 110](c ) ) ; the other possibility is the elimination of the central plane of atoms ( atoms a@xmath68 and a@xmath69 ) , generating the parallel 2-atom chain 2/0 ( fig . [ 110](d ) ) . as a function of the electron energy for various rod - like nw s along the [ 110 ] direction : ( a ) 4/3 , ( b ) 2/2 , ( c ) 1/1 , ( d ) 2/0 structures . the vertical line indicates the @xmath61.,width=8 ] the calculated conductance for each of the [ 110 ] nw s presented in fig . [ 110 ] is shown in fig . [ theo-110 ] . we can see that the conductance curves presented show oscillations that arise from interference effects inside the contacts . this oscillations are sensitive to the atomic positions so we must consider an average around the value e@xmath70 . the calculated conductance for the 4/3 structure , as presented in fig . [ theo-110](a ) , is @xmath63 @xmath71 4.5 @xmath0 . meanwhile , for the 2/2 case it is close to 2.8 @xmath0 . the last two nw s structures , 1/1 and 2/0 , show essentially the same conductance pattern , @xmath63 @xmath11 2 @xmath0 . the last hrtem image - within the time resolution considered - for nw s formed along the [ 110 ] direction is the 2/2 structure , so it is not possible to resolve the final structure . however , the conductance histogram of fig . [ hist](d ) shows that the last plateau is approximately 1.8 @xmath0 , indicating that at least one of these structures must be present in uhv - mcbj experiments . we might consider that the elongation process will increase the interplanar distance in the 2/2 structure and bring atoms a@xmath68 and a@xmath69 together until they reach first neighbor distances . nonetheless the orientation of the stacking is changed to the [ 100 ] direction . as it happened for the 732237 structure of section [ [ 111 ] ] , the conductance calculated for this case shows slight decrease compared to the perfect structure ( from 2 to 1.5 @xmath0 ) . in the light of the discussion above , we can model the atomic rearrangements in this direction by removing lines of atoms , starting at the 4/3 structure , which has a conductance @xmath63 @xmath71 4.5 @xmath0 , down to the 2/2 that can be associated , aided by the calculations , to the peak at 2.8 @xmath0 on fig . [ hist](d ) . finally , the peak at 2.0 @xmath0 is related to either the 1/1 or 2/0 structures . which is not present in the first hours of experiment . inset : typical conductance curves ; note the existence of plateaus at @xmath66 @xmath11 0.2 @xmath0,width=245 ] the statistical correlation between the occurrence of conductance curve profiles and the multiplicity of zone axis has been hitherto described as a property of the au nanowires.@xcite however , the accordance is only observed during the first hours of measurement after the macroscopic wire was fractured inside the mcbj , while the gold surfaces are free of impurities ; passed this initial period , contamination occurs even in uhv conditions . that is then evidenced by the appearance of conductance plateaus at values ranging approximately from 0.1 to 0.2 @xmath0 and whose occurrence increases with time.@xcite similar observations have been reported for nw s seen using uhv scanning tunneling microscopy.@xcite the global histogram and typical experimental conductance curves illustrating this effect are shown in fig . [ imp - exp ] . likewise , another indication that the mcbj might be contaminated is the fact that the 1 @xmath0 plateaus become exceedingly stable and long , suggesting the formation of entire nanowires made of impurity atoms . one of the most probable contaminants is carbon . _ ab initio _ calculations@xcite have shown that the carbon atom naturally occupies a position between two gold atoms in atc s . since the presence of carbon can not be detected by hrtem images because of its low atomic number,@xcite the effect characterized by plateaus at 0.10.2 @xmath0 ( fig . [ imp - exp ] ) is therefore the only experimental evidence of its presence . we have used the formalism of section [ theory ] to address this question . for this purpose , consider the following gold atc s in the [ 100 ] direction : a chain of 2 suspended gold atoms between two [ 100 ] tips ( au - au - au - au , four au atoms forming the narrowest region ) , as well as the contaminated au - c - au - au chain . the assumed distance separating c and au atoms is 1.8,@xcite whereas the au - au distance is taken to be 2.88 . calculations of the conductance produced for the 2 cases are illustrated in fig . [ imp - theo ] . it can be verified that the presence of c atoms in the chain is responsible for decreasing the conductance of the nanocontact considerably from 1 to @xmath11 0.1 @xmath0 , in very good agreement with the experimental results and , moreover , corroborating the assumption that carbon can act as a contaminant of au nw s after a few hours of measurement . in summary , we have studied experimentally and theoretically the elongation of gold nw s , both from the structural evolution and the quantum conductance behavior . from the experimental point of view , we have used two independent set - ups : an uhv - mcbj operating at room temperature for conductance measurements and hrtem for real - time imaging of the nw atomic rearrangements . gold nw s adopt only three kinds of atomic structures where the elongation direction is parallel to one of the [ 100 ] , [ 111 ] and [ 110 ] axis . for each one of these nw types , we have derived the three dimensional structures during different stages of the elongation process , by combining hrtem images and the wulff method . subsequently conductance calculations , based on the extended hckel theory and the landauer formalism , were applied to the deduced atomic structures , showing an excellent accordance with the conductance measurements . this approach has provided a consistent correlation between the structural evolution of the three kinds of au nw s and the conductance behavior in the 03 @xmath0 conductance range . the eht , although rather simple , allows us to take into account the atomic arrangement of the nanocontacs and , also it includes the possibility of considering atoms of different species . as an example of its versatility , we have considered carbon impurities on suspended atom chains ; the results indicate that they act as a contaminant causing the conductance plateaus to decrease from 1 to 0.1 @xmath0 . this result agrees extremely well with our conductance data of contaminated au nw s and , also those from olesen _ et al_.@xcite we have , therefore , thoroughly analyzed the relationship between the structural modifications involved in the elongation of au nanowires and their transport behavior . the main conclusion that must be drawn from this work is the determinant role of an atomistic description when analyzing the conductance properties of nanoscopic metallic wires . the authors are grateful to fapesp , cnpq and lnls for financial support . this appendix presents the details of the theory , circumvented in section [ theory ] . when projected onto the basis states @xmath72 the equation @xmath53 yields the following set of coupled equations for the set of coefficients @xmath73 : with @xmath75 the term @xmath76 renders the mo energies of a given unit cell whenever @xmath77 and @xmath78 . it also accounts for the hoppingenergies between the nanocontact and the first unit cell of the leads , on both sides , that is , @xmath79 . the same idea applies to the overlap terms @xmath80 . within the leads we simplify the treatment by taking into account only the coupling between mo s of the same type , which yields @xmath81 and @xmath82 , with @xmath3 and @xmath83 being the same for all mo s . this is necessary for obtaining analytical expressions for the dispersion energies of the electrons in the leads , which allows us to consider explicitly only the sites @xmath84 in ( [ eq1 ] ) . thus the energy dispersion relation is @xmath85 and the velocity of the electron mode @xmath86 , with @xmath87 and @xmath88 is the length of the unit cell . the parameters @xmath89 and @xmath90 were chosen to yield a bandwidth of approximately @xmath91 , which is found in bulk au . l. olesen , k. hansen , e. lgsgaard , i. stensgaard , f. besenbacher , in _ nanowires _ , vol . 340 , pp . 191 , _ nato advanced studies institute , series e : applied science _ , edited by p.a . serena and n. garcia ( kluwer , amsterdam 1996 ) .	gold nanowires generated by mechanical stretching have been shown to adopt only three kinds of configurations where their atomic arrangements adjust such that either the [ 100 ] , [ 111 ] or [ 110 ] zone axes lie parallel to the elongation direction . we have analyzed the relationship between structural rearrangements and electronic transport behavior during the elongation of au nanowires for each of the three possibilities . we have used two independent experiments to tackle this problem , high resolution transmission high resolution electron microscopy to observe the atomic structure and a mechanically controlled break junction to measure the transport properties . we have estimated the conductance of nanowires using a theoretical method based on the extended hckel theory that takes into account the atom species and their positions . aided by these calculations , we have consistently connected both sets of experimental results and modeled the evolution process of gold nanowires whose conductance lies within the first and third conductance quanta . we have also presented evidence that carbon acts as a contaminant , lowering the conductance of one - atom - thick wires .
the determination of distances to astronomical objects is perhaps one of the oldest and most important problems in astronomy . accurate distances are necessary to establish physical properties of sources as well as the distribution of different classes of objects in the galaxy . in particular , for objects that are believed to be tracers of spiral structure , such as massive star formation regions , one may infer the number and location of spiral arms in the galaxy . for most sources located in the galactic plane , optical distance determination methods can not be used due to the large opacity of interstellar dust . the usual way to estimate distances to radio sources ( e.g. , molecular clouds , h ii regions , pns ) is based on the kinematic properties of the galactic plane . the velocity field of the galaxy can be approximated by axially - symmetric circular orbits in which the velocity is only a function of galactocentric distance ( e.g. , brand & blitz 1993 , and references therein ) . the distance to an object in the galactic plane can be obtained from its observed radial velocity with respect to the local standard of rest ( lsr ) reference frame by inversion of the mathematical relationship between the velocity and galactocentric distance ( i.e. , rotation curve ) . this procedure gives a unique solution for the kinematic distance ( for a given longitude and radial velocity ) in the outer galaxy , however the lsr distance is degenerate in the inner galaxy ( quadrants i and iv ) . in the inner galaxy , the line of sight to a given object intersects its circular orbit at two points that define the _ near _ and _ far _ positions . at these two locations an object has the same radial velocity , giving rise to a kinematic distance degeneracy which is referred to as _ the distance ambiguity problem_. only inner galaxy sources located at the _ tangent point _ have unambiguous distances . the tangent point is the point at which a line of sight to an object is tangent to its orbit . this work is the third in a series ( araya et al . 2002 , paper i ; watson et al . 2003 , paper ii ) intended to resolve the distance ambiguity of bright h ii regions located within the area of the spitzer / glimpse survey of the inner galaxy ( @xmath2l@xmath2 = 10@xmath3 @xmath4 65@xmath3 , @xmath2b@xmath6 1@xmath3 ) . the motivation for this work stems from the fact that : ( a ) accurate measurements of distances to h ii regions and their associated star forming regions are necessary to determine their bolometric luminosity , mass , and size . the physical parameters of massive star forming regions constrain theoretical models of star formation and evolution of molecular clouds ; and , ( b ) regions of massive star formation are expected to trace the spiral arms of the galaxy and therefore , the establishment of distances to these objects provides an opportunity to improve our understanding of the large - scale structure of the galaxy . in the first part of this project , 75 h ii regions with galactic longitudes ranging from @xmath730@xmath3 to @xmath770@xmath3 and galactic latitude between @xmath41@xmath3 and @xmath71@xmath3 , were observed in two surveys with the arecibo 305 m telescope ( the half - power beamwidth @xmath8 1@xmath9 ; paper i and paper ii ) . the arecibo antenna was used to simultaneously observe the h110@xmath0 line in emission and the @xmath10 line of h@xmath11co in absorption against h ii regions . the h110@xmath0 line was used to establish the velocity of the h ii regions and the h@xmath11co absorption components were used to distinguish between near and far distances . it was assumed that all h@xmath1co absorption lines are produced by absorption of radio continuum from a single background h ii region , thus neglecting absorption of the cosmic microwave background radiation . this assumption is supported by the fact that only seven h@xmath1co absorption features were observed in off - source integrations , whereas 134 were detected in on - source spectra . the rotation curve of brand and blitz ( 1993 ) was used to derive kinematic distances . distance ambiguities were resolved for @xmath895@xmath12 ( 20 sources ) and @xmath881@xmath12 ( 44 sources ) of the sample in the first ( paper i ) and second ( paper ii ) arecibo surveys , respectively . here we report observations of selected h ii regions in the northern spitzer / glimpse survey area that is not accessible from arecibo . the observations were made using the national radio astronomy observatory ( nrao ) 100 m green bank telescope ( gbt ) . distance ambiguities are resolved using the same method as in the arecibo surveys . a longitude - velocity diagram based on the arecibo and gbt surveys is constructed . the observations were made on 2002 november 1@xmath43 , 8@xmath49 , and 14 with the gbt . seventy two sources were selected from the iras point source catalog that satisfy the far - infrared color - color criteria of uc h ii regions ( i.e. , log(s@xmath13/s@xmath14)@xmath15 1.30 and log(s@xmath16/s@xmath14 ) @xmath15 0.57 ; wood and churchwell 1989 ) , and have 100 @xmath17 m flux densities @xmath15 1000 jy . in addition , we required that all candidate sources , visible with the gbt , lie within galactic longitude ranges , @xmath420@xmath3 to @xmath410@xmath3 and @xmath710@xmath3 to @xmath730@xmath3 , and have galactic latitudes between @xmath41@xmath3 and @xmath71@xmath3 . these areas are included in the spitzer / glimpse survey region ( @xmath2l@xmath2= 10@xmath18@xmath465@xmath18 and @xmath2b@xmath2 @xmath5 1@xmath18 ) and were not sampled by the arecibo surveys . the observed sources are listed in table 1 . the main goal of this survey is to establish reliable distances to massive star formation regions ( young ob associations and clusters ) that will be observed at mid - infrared wavelengths by the glimpse legacy science program with the spitzer space telescope . the gbt observations supplement the arecibo data which together cover the spitzer / glimpse survey area accesible from the northern hemisphere . the 1@xmath19 - 1@xmath20 transition of formaldehyde ( h@xmath11co , @xmath21 = 4829.6594 mhz ) was observed in absorption against h ii regions . simultaneously , the h110@xmath0 ( @xmath21 = 4874.1570 mhz ) radio recombination line was observed to establish kinematic distances . both spectral lines were measured using the c - band gregorian focus receiver . the half - power beamwidth ( hpbw ) of the gbt at 4.86 ghz was measured to be [email protected] @xmath22 [email protected] . the 4 if mode of the correlator was used to observe two transitions simultaneously in both senses of linear polarization with two subcorrelators centered on the h@xmath11co line , and the other two centered on the h110@xmath0 radio recombination line . each subcorrelator had 4096 channels , resulting in a resolution of 3.05 khz ( 0.188 km s@xmath23 ) , a total bandwidth of 12.5 mhz , and a total velocity coverage of @xmath8770 km s@xmath23 . the central velocity for each subcorrelator was set to v@xmath24 = 0 km s@xmath23 . the correlator used 9-level sampling . position switching was used with a noise - diode calibration signal injected into the system every second . the off position was chosen to trace the same path on the sky as the on@xmath4source position . depending on the continuum flux densities , the observations required 5 to 30 minutes on@xmath4source integration time ( 10@xmath460 minutes / source ) to achieve a signal to noise @xmath15 5 ( see table 1 ) . for both polarizations , the system temperature during the observations was 26 @xmath22 2 k. observations of the flux calibrator ngc 7027 with assumed flux density of 5.37 jy at 4.82 ghz indicate that the antenna gain remained constant at 1.83 k jy@xmath23 , independent of zenith and azimuth angles ( [email protected] @xmath4 [email protected] and @[email protected] @xmath4 @[email protected] , respectively ) . for each source both on and off positions were inspected individually to check for polarization , absorption and/or emission in the off position , and radio interference . the final spectra were obtained by averaging both polarizations and all scans for each source . the data were calibrated using the astronomical imaging processing system ( aips++ ) . subsequent spectral line analysis used the class software package . the h110@xmath0 and h@xmath25co line parameters were derived by fitting gaussian profiles to each velocity component along the line of sight . before the line parameters were measured , all spectra were hanning smoothed to the velocity resolution listed in table 2 . continuum emission was detected in all but one source ( [email protected] ) . the results of the spectral line analysis are presented in tables 3 , 4 , and 5 for the sources with single h110@xmath0 velocity components along a given line of sight , multiple h110@xmath0 velocity components ( see discussion below ) , and h@xmath11co absorption lines , respectively . the full - widths at half - power ( fwhp ) of the spectral lines are corrected for broadening introduced by the spectrometer and smoothing . the errors quoted in tables 3@xmath45 are the formal 1@xmath26 errors ( 68.3 % confidence level ) of the gaussian fits to the data . only the spectral lines with intensities greater than 3@xmath26 were included in the tables . figures [email protected] show both the h110@xmath0 emission and h@xmath11co absorption spectra for each source in our sample . flux densities throughout the paper are given in jy . seventy two massive star formation regions were observed . we detected formaldehyde absorption toward all the sources , and h110@xmath0 emission toward all but 6 sources ( [email protected] , [email protected] , [email protected] , [email protected] , [email protected] , and [email protected] ) . eighteen sight lines had multiple h110@xmath0 velocity components ( see table 4 ) , all of which are located in the longitude interval 12@xmath3 to 31@xmath3 . extensive tests of the local oscillator and the system were made , including observations of several lines of sight , to confirm that indeed the observed spectra are repeatable , thus faithful representations of multiple h ii regions along these lines of sight and not instrumental artifacts . injection of a test tone into the system did not show the multiple line effect . we also shifted the local - oscillator frequencies lo1 and lo2 by 1 mhz , and did not measure any change in the spacing of the two ( or three ) lines . we did not detect multiple h110@xmath0 radio recombination lines in our arecibo observations . this non - detection is probably primarily the result of the beam area of the gbt ( @xmath27 deg@xmath28 ; [email protected] ) being seven times larger than the beam area of the arecibo telescope ( @xmath29 deg@xmath28 ; hpbw@xmath81 ) , so the probability of having more than one h ii region in a beam is much higher . for the inner galaxy ( quadrants i and iv ) h ii regions with a h110@xmath0 line detection , the near / far distance classification was based on the presence of h@xmath1co absorption lines between the source velocity ( @xmath30 ) plus 10 km s@xmath23 and the tangential velocity ( velocity at the tangent point ; hereafter @xmath31 ) plus 10 km s@xmath23 . if the h@xmath1co absorption feature was found in this velocity range , we classified the source at the far distance ; otherwise , the near distance was assigned . the addition of 10 km s@xmath23 to the source and tangent point velocities account for typical departures from circular motions ( burton 1974 ) . we do not resolve the near / far distance ambiguity for h ii regions with velocities higher than @xmath31 @xmath4 10 km s@xmath23 and lower than @xmath31 , because we can not distinguish between near and far distances in this velocity range . if the velocity of an h ii region is higher than @xmath31 , we locate this source at the tangent point . the excursions of the data points to velocities higher than @xmath31 values are similar to the h i excursions associated with the frequently - discussed `` bumps '' in the rotation curve , first pointed out by kwee et al . 1954 . the distance ambiguity can also be resolved for the intervening molecular clouds . if the velocity of a h@xmath1co absorption feature ( @xmath32 ) is less than @xmath30 @xmath4 10 km s@xmath23 , we locate a molecular cloud at the near distance . however , if @xmath32 lies between @xmath30 @xmath7 10 km s@xmath23 and @xmath31 @xmath4 10 km s@xmath23 , the distance ambiguity can not be resolved , because in either case ( near or far position ) the molecular gas is in front of the h ii region . the velocities of many absorption features lie within 10 km s@xmath23 of @xmath30 , demonstrating that the h ii regions are closely associated with molecular gas . in this case , we give the same distance classification for both h ii regions and h@xmath1co clouds associated with them . as in the case of h ii regions , we do not assign distances to molecular clouds with velocities between @xmath31 @xmath4 10 km s@xmath23 and @xmath31 , and we locate them at the tangent point if their velocities are higher than @xmath31 . a brief summary of the criteria used to assign kinematic distances to h ii regions and h@xmath1co clouds is presented in table 6 . in table 7 we report the kinematic parameters of 48 h ii regions and 149 intervening h@xmath1co clouds . h ii regions with no h110@xmath0 line detection or multiple h110@xmath0 lines and h@xmath1co clouds observed in their direction , as well as h@xmath1co molecular clouds with ambiguous distances are not included . all distances in table 7 were calculated assuming the brand and blitz ( 1993 ) galactic rotation model . in table 7 , we report the radial velocities of the sources ( v@xmath24 ) , the near ( d@xmath33 ) and far ( d@xmath34 ) distances , tangent point velocities ( v@xmath35 ) , distances to the tangent point ( d@xmath35 ) , the adopted kinematic distances to the sources ( d@xmath24 ) , distances from the galactic plane ( z ) , and distances to the galactic center ( r@xmath36 ) . the errors of the kinematic distances were calculated assuming a velocity uncertainty due to peculiar motions of 10 km s@xmath23 . we resolved the distance ambiguity toward 44 out of 48 h ii regions with a single h110@xmath0 line velocity . fourteen h ii regions are classified at the near distance , 26 at the far distance , and 4 are placed at the tangent point . we do not give a distance to 4 h ii regions ( see table 6 for classification criteria ) . we classify 72 out of 149 h@xmath1co clouds listed in table 7 at the near distance . we do not assign a distance to 7 h@xmath1co clouds , and place 5 at the tangent point . 64 molecular clouds are associated with h ii regions , 16 of which are at the near distance , 37 at the far distance , 5 at the tangent point , and 6 are not classified . one molecular cloud ( mc 2 ) observed toward [email protected] has an unambiguous distance , because its velocity is positive and it lies outside the solar circle . to check our results , we observed two sources from the first ( [email protected] and [email protected] ; paper i ) and two from the second arecibo surveys ( [email protected] , [email protected] ; paper ii ) . we agree on the near - far distance ambiguity resolution for all four sources . we also observed a total of 11 h ii regions observed by downes et al . ( 1980 ) , kolpak et al . ( 2003 ) , and fish et al . we compared our results with their near / far distance classifications for the common sources . kolpak et al . ( 2003 ) and fish et al . ( 2003 ) resolved the distance ambiguity for 49 and 20 h ii regions with known radial velocities , respectively , by measuring the 21 cm h i absorption spectrum toward each source with the vla . we share one source ( [email protected] ) with both kolpak et al . ( 2003 ) and fish et al . kolpak et al . ( 2003 ) placed this source at the far distance , while fish et al . ( 2003 ) put it at the tangent point . we agree with kolpak at el . ( 2003 ) on the far distance . we share two more sources ( [email protected] and [email protected] ) with kolpak et al . ( 2003 ) and agree on all distance determinations , and one source ( [email protected] ) with fish et al . ( 2003 ) that we do not classify , because we did not detect the radio recombination line toward it . six of the sources from our sample were previously discussed by downes et al . ( 1980 ) who observed h110@xmath0 emission and h@xmath1co absorption lines toward h ii regions with the effelsberg 100 m telescope . downes et al . ( 1980 ) resolved the distance ambiguity for four out of six sources we have in common . of these , we agree on one ( [email protected] ) and disagree on two ( [email protected] and [email protected] ) . we do not classify the fourth one ( [email protected] ) , because multiple recombination lines were detected toward this source . the two sources on which we disagree with downes et al . ( 1980 ) , [email protected] and [email protected] , are part of the w31 complex of bright h ii regions and molecular clouds in the inner galaxy . wilson ( 1974 ) and kalberla et al . ( 1982 ) detected h@xmath1co , oh , and h i in absorption toward the h ii regions in the w31 complex and found absorption line velocities up to 33 km s@xmath23 greater than that of the h ii regions . they argued , however , that w31 is unlikely to be at the far distance because none of the absorbing clouds ( including h i ) achieved velocities as large as the tangent point velocity . they argued instead that w31 with a peculiar velocity of @xmath830 km s@xmath23 is likely to be located on the near side of the expanding `` 3-kpc arm '' at a distance of @xmath85.5 kpc from the sun . corbel et al . ( 1997 ) argue against the association of w31 with the 3-kpc arm because their co observations show that the co clouds associated with w31 departs from the well defined velocity of the 3-kpc arm by @xmath823 km s@xmath23 ( @xmath85 times the velocity dispersion of the arm ) ; they also claim that the 3-kpc arm is quite deficient in star formation while w31 contains some of the brightest h ii regions in the inner galaxy , and the co cloud mc 13 ( associated with w31 ) is far more massive than any cloud in the 3-kpc arm . corbel et al . ( 1997 ) therefore claim that w31 lies at the far distance . based on our h@xmath1co data and classification criteria , we locate w31 at the far distance ; however , non - detection of h i at the tangent point velocity is problematic . when viewed at optical wavelengths , the arms in spiral galaxies are traced by young massive stars and their h ii regions . a compelling explanation for this is that compression due to spiral density waves initiates star formation near the leading edge of spiral arms ( roberts 1969 ) . whatever the reason , we know empirically that young massive stars are good tracers of spiral structure in galaxies . it is , therefore , natural to ask if the kinematically determined distances of h ii regions , whose distance ambiguities have been resolved , indicate a recognizable spiral arm pattern in the milky way . in figures 2 and 3 we show two plots to assist in answering this question . figure 2 shows the projected location in the galactic plane of the h ii regions whose distance ambiguities have been resolved . in this figure , we view the galactic plane from the north galactic pole with the galactic center at ( 0 , 0 ) kpc and the sun at ( 0 , 8.5 ) kpc ; both are designated by a filled circle . in all surveys included in this figure , the rotation constants , the galactocentric distance of the sun ( r@xmath37 ) and its rotational velocity around the galactic center ( @xmath38 ) were assumed to be 8.5 kpc and 220 km s@xmath23 ( kerr and lynden - bell 1986 ) , respectively . in araya et al . ( 2002 ) , watson et al . ( 2003 ) and in the present study distances were derived assuming that the galaxy rotates according to the model of brand and blitz ( 1993 ) . for sources from kolpak et al . ( 2003 ) and fish et al . ( 2003 ) , who used the clemens ( 1985 ) and flat rotation curves , respectively , all distances were recalculated using the brand and blitz ( 1993 ) rotation curve for consistency with the rest of the data . the solid curves in figure 2 indicate the centroids of the four spiral arms proposed initially by georgelin and georgelin ( 1976 ) and modified by taylor and cordes ( 1993 ; hereafter tc93 ) . the tc93 model is based on the standard rotation constants ( r@xmath37 , @xmath38 ) = ( 8.5 kpc , 220 km s@xmath23 ) . figure 3 shows a longitude - velocity ( l@xmath4v ) plot with the same set of h ii regions as in figure 2 . we emphasize that the h ii region data points do not depend on any model assumptions about galactic rotation . we have superimposed the locus of points of the four spiral arms and the tangent point velocity from the tc93 model using the brand and blitz ( 1993 ) rotation curve for comparison with the distribution of h ii regions . what can we conclude from these plots ? first , no sharp , `` well defined '' spiral pattern is apparent when looking at the distribution of h ii regions projected onto the galactic plane . however , we have to remember that spiral arms are not like the thin lines drawn in figure 2 ; they have significant widths which will cause some smearing even if we could determine positions precisely . if there were no curves to guide the eye in figure 2 , one might come close to designating similar spiral arms as tc93 if the loci of highest density of points are followed . the preponderance of h ii regions tend to lie closer to the tc93 spiral arm centroids than in the regions between the centroids . exceptions to this are in the following longitude intervals : 10@xmath3@xmath4 35@xmath3 where a significant number of objects lie between the perseus and sagittarius centroids ; 5@xmath3@xmath415@xmath3 where a number of sources lie between the norma and scutum centroids ; near 0@xmath3 six or so sources clearly lie between the scutum and sagittarius centroids . at least 9 sources appear to be located in the `` local spur '' including the sun from 75@xmath3 @xmath5 l @xmath5 90@xmath3 . in the first quadrant of the galactic plane the tc93 perseus , sagittarius , and scutum spiral arm centroids have separations @xmath5 2 kpc . in those longitude intervals where this is the case , positions would have to be determined to very high accuracy to clearly distinguish between the three arms . most of the sources that lie well between the tc93 arm centroids have longitudes from near 0@xmath3 to @xmath5 20@xmath3 where kinematic distances have lower reliability than at higher longitudes because radial velocities are dominated by peculiar motions rather than by circular velocities . kinematic distances have an intrinsic uncertainty because they depend on a rotation model of the galaxy , that is known to have relatively large departures from the true motions of the galaxy at some longitudes . to first order , a circular rotation model fits the observed velocities relatively well , but departures from this general pattern , such as radial motions , occur at all longitudes . a good example of this is illustrated in figure 3 where several h ii regions have velocities 10 to 25 km s@xmath23 in excess of the tangential velocity . it is therefore not surprising that there is a lot of scatter in the position - position plot of figure 2 , especially at low longitudes . until more precise galactic rotation models are available , this may be the limit to which spiral arms can be determined using kinematic distances . the l@xmath4v plot in figure 3 is an effort to examine possible spiral structure using only observed quantities of h ii regions ; the positions of the h ii region and h i data are independent of a galactic rotation model . we plot the tc93 spiral arms using the brand and blitz ( 1993 ) rotation curve and tangential velocity centroids to show where they lie in an l@xmath4v diagram , the location of the observed sources , however , do not depend on a galactic rotation model to be placed in this diagram . ignoring the tc93 arm and tangential velocity centroids , we see that sources lie in restricted areas of the l@xmath4v plane mostly bound by the tc93 tangential velocity centroids . also the sources are not uniformly distributed between the tc93 tangential velocity centroids , but are clumped in the northern plane around l @xmath39 30@xmath3 and v @xmath39 100 km s@xmath23 and l @xmath39 40@xmath3@xmath445@xmath3 and v @xmath39 50 km s@xmath23 . in the tc93 model one sees that in the l@xmath4v diagram these are locations where the spiral arm loci of tc93 are approaching the tangent point velocity locus ; one should expect an increase in the number of sources at these locations due to velocity crowding . there is also an obvious significant decrease in the number of sources between those two locations in the northern plane . the density of sources is generally higher in the regions where the tc93 spiral arm loci are located than away from them , but there are no obvious sharp continuous features that can be unambiguously traced in the l@xmath4v diagram that would clearly indicate a spiral arm . also , we see several sources that exceed the tc93 tangential velocities by @xmath15 10 km s@xmath23 ; however , their velocities are not higher than the observed h i cut - off velocities ( i.e. , the most extreme velocities on the h i spectra ; see figure 3 ) . why is a spiral pattern not obvious in the l@xmath4v plane ? this probably has several reasons . one , spiral arms have intrinsic widths both in space and velocity which will smear the distribution of sources about arm centroids . second , peculiar motions ( i.e. , departures from the motions included in the rotation model from which distances were determined ) will smear out the distribution around a spiral arm centroid even further . third , unlike the measurement of h i and co where continuous kinematic features can be traced over large ranges of galactic longitude , measurement of a limited number of discrete h ii regions does not provide the velocity continuity necessary to recognize large scale kinematic features . if the spiral arms in the inner galaxy are only separated by @xmath5 2 kpc as implied by the tc93 model , it is unlikely that kinematic distances , with their uncertainties due to distance ambiguities and peculiar motions , to a hundred or so discrete sources will be capable of delineating spiral arm centroids precisely enough to clearly distinguish between spiral arms , which is what figure 2 illustrates . we report simultaneous h110@xmath0 and h@xmath1co line observations toward 72 h ii regions using the nrao gbt . the main goal of this program was to obtain kinematic distances and to resolve the near / far distance ambiguity for inner galaxy h ii regions . the main motivation to do this was to obtain accurate distances to massive star formation regions in the inner galaxy in preparation for the spitzer / glimpse survey . the h@xmath1co 6 cm line was detected in absorption toward all 72 h ii regions and the h110@xmath0 line was detected toward all but 6 h ii regions . most lines of sight had multiple velocity components in h@xmath1co ( on average @xmath40 3 ) . of the 72 h ii regions observed in the h110@xmath0 line , 18 in the longitude range 12@xmath3 @xmath5 l @xmath5 31@xmath3 had multiple velocity components ; because we could not determine which lines were being absorbed against which h ii region , we could not resolve distance ambiguities for these sources . of the remaining 54 h ii regions , 6 were not detected in the h110@xmath0 line . we were able to resolve the distance ambiguity of 44 out of the remaining 48 h ii regions , 14 of which are at the near distance , 26 at the far distance , and 4 at the tangent point . of 149 h@xmath1co clouds ( i.e. , velocity components ) , we found that 72 are at the near distance , 5 are at the tangent point , 7 are not classified , 64 are associated with h ii regions , and one is located outside the solar circle ( [email protected] mc 2 ) . sixteen out of 64 molecular clouds associated with h ii regions are at the near distance , 37 are at the far distance , 5 are at the tangent point , and 6 are not classified . in those cases where we have sources in common with other groups who have resolved distance ambiguities , we generally agree on distance assignments with only a few exceptions that mostly arise from differences in the range of peculiar motions considered . the sources with resolved distance ambiguities were plotted in position - position and l@xmath4v plots to determine if spiral arms could be clearly discerned . it was found that the highest density of points in a position - position plot approximately follows the spiral arms proposed by tc93 ; however , the scatter about the spiral arm centroids is about as large as the separation between the arms of tc93 making the distinction between spiral arms indistinct . positions of the h ii regions in an l@xmath4v plot rely on observed quantities alone and are independent of galactic rotation models . comparing the locus of points of the tc93 spiral arms with the distribution of h ii regions whose distance ambiguities have been resolved shows an increase in the density of sources at the points where the loci of spiral arms are approaching the locus of tangent point velocities and a lower density between the arm loci . however , even in an l@xmath4v plot , the scatter is too large to clearly trace spiral arms over significant segments in galactic longitude . four reasons were discussed why h ii regions do not seem to clearly delineate spiral arms in our galaxy . they are : 1 ) spiral arms have intrinsic widths both in space and velocity which will smear the distribution of sources about arm centroids ; 2 ) peculiar motions ( i.e. , departures from the motions included in the rotation model from which distances were determined ) will smear out the distribution around a spiral arm centroid even further ; 3 ) a limited number of discrete h ii regions do not provide the velocity continuity necessary to recognize large scale kinematic features ; and , 4 ) errors in the near / far distance ambiguity assignment can cause large distance errors which will further confuse a position - position plot . thus , we conclude that without a very large number of h ii regions whose distance ambiguities have been resolved in combination with more sophisticated galactic rotation models that take into account non - circular motions both in the plane and perpendicular to it , it is unlikely that a more continuous spiral pattern will emerge from kinematic studies of h ii regions . resolution of distance ambiguities of h ii regions is not a fruitless exercise , however . accurate distances , which was the main motivation for undertaking this study , are crucial for the determination of physical properties ( e.g. , absolute sizes , luminosities , and masses ) of massive star formation regions . we thank f.d . ghigo and r.c . bignell for help during the observations and data reduction . we express our gratitude to butler burton for his thought - provoking comments on the manuscript that significantly improved this paper . we also thank him for providing the hi data that we used in our analysis . ms gratefully acknowledges support of a gbt student support fellowship . ebc acknowledges partial support from nsf grant ast-0303689 . ph acknowledges partial support from research corporation award cc4996 and nsf grant ast-0098524 . sk acknowledges partial support from conacyt grant e-36568 and dgapa - unam grant in118401 . [email protected] & 16 44 34.0 & @xmath445 16 18 & 16445@xmath44516 & 8.54 & 10 + [email protected] & 17 01 38.8 & @xmath441 24 59 & 17016@xmath44124 & 7.03 & 22 + [email protected] & 17 05 59.8 & @xmath441 32 09 & 17059@xmath44132 & 25.8 & 10 + [email protected] & 17 00 59.9 & @xmath440 42 19 & 17009@xmath44042 & 18.20 & 10 + [email protected] & 17 00 51.0 & @xmath440 40 18 & 17008@xmath44040 & 18.20 & 10 + [email protected] & 17 14 56.4 & @xmath439 16 02 & 17149@xmath43916 & 9.16 & 10 + [email protected] & 17 17 48.6 & @xmath437 42 22 & 17178@xmath43742 & 6.88 & 10 + [email protected] & 18 06 25.9 & @xmath420 20 04 & 18064@xmath42020 & 27.30 & 12 + [email protected] & 18 06 00.3 & @xmath420 05 57 & 18060@xmath42005 & 12.00 & 12 + [email protected] & 18 05 37.6 & @xmath419 52 40 & 18056@xmath41952 & 10.2 & 10 + [email protected] & 18 06 42.0 & @xmath419 27 03 & 18067@xmath41927 & 1.29 & 52 + [email protected] & 18 08 34.5 & @xmath419 31 24 & 18085@xmath41931 & 2.09 & 30 + [email protected] & 18 11 03.7 & @xmath418 54 18 & 18110@xmath41854 & 4.93 & 12 + [email protected] & 18 09 05.7 & @xmath418 32 42 & 18090@xmath41832 & 1.71 & 22 + [email protected] & 18 07 54.9 & @xmath417 56 33 & 18079@xmath41756 & 2.38 & 30 + [email protected] & 18 10 16.7 & @xmath418 00 26 & 18102@xmath41800 & 0.88 & 20 + [email protected] & 18 10 17.9 & @xmath417 57 04 & 18102@xmath41757 & 0.88 & 20 + [email protected] & 18 09 17.1 & @xmath417 42 42 & 18092@xmath41742 & 2.08 & 10 + [email protected] & 18 08 56.4 & @xmath417 32 14 & 18089@xmath41732 & 3.15 & 10 + [email protected] & 18 11 43.7 & @xmath417 53 01 & 18117@xmath41753 & 6.31 & 32 + [email protected] & 18 11 46.3 & @xmath417 38 02 & 18117@xmath41738 & 1.85 & 60 + [email protected] & 18 11 41.4 & @xmath416 46 19 & 18116@xmath41646 & 6.14 & 12 + [email protected] & 18 15 22.0 & @xmath416 51 52 & 18153@xmath41651 & 1.23 & 60 + [email protected] & 18 15 59.6 & @xmath416 48 55 & 18159@xmath41648 & 2.82 & 32 + [email protected] & 18 16 10.8 & @xmath416 26 31 & 18161@xmath41626 & 2.69 & 30 + [email protected] & 18 18 16.8 & @xmath414 33 19 & 18182@xmath41433 & 1.07 & 32 + [email protected] & 18 19 37.3 & @xmath413 31 45 & 18196@xmath41331 & 5.70 & 42 + [email protected] & 18 21 47.4 & @xmath412 52 47 & 18217@xmath41252 & 1.89 & 22 + [email protected] & 18 22 21.7 & @xmath412 43 59 & 18223@xmath41243 & 1.10 & 32 + [email protected] & 18 23 58.5 & @xmath412 28 23 & 18239@xmath41228 & 6.78 & 12 + [email protected] & 18 24 17.6 & @xmath412 27 34 & 18242@xmath41227 & 1.29 & 32 + [email protected] & 18 23 36.3 & @xmath412 05 36 & 18236@xmath41205 & 1.03 & 44 + [email protected] & 18 27 14.5 & @xmath412 17 32 & 18272@xmath41217 & 1.15 & 20 + [email protected] & 18 26 26.6 & @xmath411 52 27 & 18264@xmath41152 & 1.10 & 52 + [email protected] & 18 29 30.2 & @xmath410 30 50 & 18295@xmath41030 & 1.40 & 50 + [email protected] & 18 28 17.2 & @xmath409 51 39 & 18282@xmath40951 & 0.96 & 20 + [email protected] & 18 29 00.1 & @xmath409 24 21 & 18290@xmath40924 & 1.08 & 44 + [email protected] & 18 30 49.0 & @xmath408 41 39 & 18308@xmath40841 & 0.91 & 32 + [email protected] & 18 31 45.9 & @xmath408 45 47 & 18317@xmath40845 & 1.45 & 10 + [email protected] & 18 30 40.2 & @xmath408 35 53 & 18306@xmath40835 & 0.91 & 20 + [email protected] & 18 31 59.9 & @xmath408 34 50 & 18319@xmath40834 & 9.60 & 12 + [email protected] & 18 32 29.1 & @xmath408 20 14 & 18324@xmath40820 & 1.32 & 22 + [email protected] & 18 33 46.3 & @xmath407 43 06 & 18337@xmath40743 & 0.99 & 32 + [email protected] & 18 33 28.5 & @xmath407 33 44 & 18334@xmath40733 & 3.51 & 10 + [email protected] & 18 34 04.2 & @xmath407 38 15 & 18340@xmath40738 & 4.33 & 10 + [email protected] & 18 34 09.4 & @xmath407 27 23 & 18341@xmath40727 & 3.98 & 12 + [email protected] & 18 34 27.1 & @xmath407 25 04 & 18344@xmath40725 & 3.98 & 10 + [email protected] & 18 34 16.8 & @xmath406 55 43 & 18342@xmath40655 & 0.91 & 42 + [email protected] & 18 35 48.4 & @xmath406 47 12 & 18358@xmath40647 & 0.99 & 20 + [email protected] & 18 36 10.5 & @xmath406 27 44 & 18361@xmath40627 & 1.55 & 10 + [email protected] & 18 34 49.3 & @xmath406 16 52 & 18348@xmath40616 & 1.94 & 22 + [email protected] & 18 36 00.0 & @xmath405 37 49 & 18360@xmath40537 & 1.68 & 10 + [email protected] & 18 37 15.6 & @xmath405 41 33 & 18372@xmath40541 & 1.25 & 22 + [email protected] & 18 38 33.5 & @xmath405 11 51 & 18385@xmath40512 & 1.56 & 22 + [email protected] & 18 39 27.6 & @xmath404 25 17 & 18394@xmath40425 & 1.22 & 10 + [email protected] & 18 40 19.5 & @xmath404 17 06 & 18403@xmath40417 & 3.94 & 20 + [email protected] & 18 40 11.1 & @xmath404 12 52 & 18401@xmath40412 & 1.01 & 20 + [email protected] & 18 42 10.2 & @xmath404 04 35 & 18421@xmath40404 & 2.04 & 30 + [email protected] & 18 40 50.3 & @xmath403 53 22 & 18408@xmath40353 & 1.79 & 24 + [email protected] & 18 41 08.4 & @xmath403 38 34 & 18411@xmath40338 & 3.18 & 22 + [email protected] & 18 43 27.2 & @xmath402 42 36 & 18434@xmath40242 & 11.70 & 10 + [email protected] & 18 44 34.1 & @xmath402 22 27 & 18445@xmath40222 & 1.76 & 10 + [email protected] & 18 42 36.7 & @xmath402 04 23 & 18426@xmath40204 & 0.88 & 32 + [email protected] & 18 44 00.5 & @xmath401 48 40 & 18440@xmath40148 & 1.22 & 26 + [email protected] & 18 45 25.3 & @xmath401 58 12 & 18454@xmath40158 & 2.90 & 10 + [email protected] & 18 44 39.9 & @xmath401 50 30 & 18446@xmath40150 & 1.88 & 10 + [email protected] & 18 45 16.1 & @xmath401 41 26 & 18452@xmath40141 & 1.75 & 34 + [email protected] & 18 44 59.4 & @xmath401 16 04 & 18449@xmath40115 & 2.82 & 12 + [email protected] & 18 58 30.2 & @xmath704 07 48 & 18585@xmath70407 & 2.15 & 10 + [email protected] & 18 59 24.5 & @xmath704 08 27 & 18593@xmath70408 & 4.52 & 12 + [email protected] & 19 20 44.6 & @xmath714 10 50 & 19207@xmath71410 & 7.45 & 10 + [email protected] & 19 44 43.5 & @xmath725 05 22 & 19446@xmath72505 & 13.2 & 10 + [email protected] & 0.76 & 0.76 & 1.01 + [email protected] & 0.38 & 0.76 & 1.02 + [email protected] & 0.76 & 0.76 & 1.03 + [email protected] & 0.38 & 0.76 & 1.04 + [email protected] & 0.76 & 0.76 & 1.05 + [email protected] & 0.76 & 0.76 & 1.06 + [email protected] & 0.76 & 0.76 & 1.07 + [email protected] & 0.38 & 0.76 & 1.08 + [email protected] & 0.76 & 0.76 & 1.09 + [email protected] & 0.76 & 0.76 & 1.10 + [email protected] & 0.76 & 0.76 & 1.11 + [email protected] & 0.38 & 0.76 & 1.12 + [email protected] & 0.38 & 0.76 & 1.13 + [email protected] & 0.76 & 0.76 & 1.14 + [email protected] & 0.38 & 1.50 & 1.15 + [email protected] & 0.76 & 0.76 & 1.16 + [email protected] & 0.76 & 0.76 & 1.17 + [email protected] & 0.76 & 1.50 & 1.18 + [email protected] & 0.76 & 0.76 & 1.19 + [email protected] & 0.38 & 0.76 & 1.20 + [email protected] & 0.38 & 1.50 & 1.21 + [email protected] & 0.38 & 0.76 & 1.22 + [email protected] & 0.38 & 1.50 & 1.23 + [email protected] & 0.38 & 1.50 & 1.24 + [email protected] & 0.76 & 1.50 & 1.25 + [email protected] & 0.38 & 1.50 & 1.26 + [email protected] & 0.76 & 1.50 & 1.27 + [email protected] & 0.38 & 0.76 & 1.28 + [email protected] & 0.76 & 1.50 & 1.29 + [email protected] & 0.38 & 0.76 & 1.30 + [email protected] & 0.76 & 0.76 & 1.31 + [email protected] & 0.38 & 1.50 & 1.32 + [email protected] & 0.76 & 1.50 & 1.33 + [email protected] & 0.76 & 1.50 & 1.34 + [email protected] & 0.76 & 1.50 & 1.35 + [email protected] & 0.76 & 1.50 & 1.36 + [email protected] & 0.76 & 0.76 & 1.37 + [email protected] & 0.38 & 1.50 & 1.38 + [email protected] & 0.76 & 0.76 & 1.39 + [email protected] & 0.76 & 1.50 & 1.40 + [email protected] & 0.38 & 0.76 & 1.41 + [email protected] & 0.76 & 0.76 & 1.42 + [email protected] & 0.38 & 0.76 & 1.43 + [email protected] & 0.76 & 0.76 & 1.44 + [email protected] & 0.76 & 0.76 & 1.45 + [email protected] & 0.76 & 0.76 & 1.46 + [email protected] & 0.76 & 0.76 & 1.47 + [email protected] & 0.76 & 0.76 & 1.48 + [email protected] & 0.76 & 0.76 & 1.49 + [email protected] & 0.76 & 1.50 & 1.50 + [email protected] & 0.76 & 0.76 & 1.51 + [email protected] & 0.76 & 1.50 & 1.52 + [email protected] & 0.76 & 1.50 & 1.53 + [email protected] & 0.76 & 1.50 & 1.54 + [email protected] & 0.76 & 1.50 & 1.55 + [email protected] & 0.38 & 1.50 & 1.56 + [email protected] & 0.38 & 0.76 & 1.57 + [email protected] & 0.38 & 0.76 & 1.58 + [email protected] & 0.38 & 0.76 & 1.59 + [email protected] & 0.38 & 0.76 & 1.60 + [email protected] & 0.38 & 0.76 & 1.61 + [email protected] & 0.76 & 1.50 & 1.62 + [email protected] & 0.76 & 1.50 & 1.63 + [email protected] & 0.76 & 1.50 & 1.64 + [email protected] & 0.76 & 0.76 & 1.65 + [email protected] & 0.38 & 0.76 & 1.66 + [email protected] & 0.76 & 0.76 & 1.67 + [email protected] & 0.38 & 0.76 & 1.68 + [email protected] & 0.38 & 0.76 & 1.69 + [email protected] & 0.38 & 0.76 & 1.70 + [email protected] & 0.38 & 0.76 & 1.71 + [email protected] & 0.38 & 0.76 & 1.72 + [email protected] & & & & + [email protected] & & & & + [email protected] & & & & + [email protected] & 84.8 ( 6.0 ) & -22.3 ( 0.5 ) & 33.8 ( 1.3 ) & 3.05 ( 0.09 ) + [email protected] & & & & + [email protected] & 471.7 ( 10.4 ) & -17.9 ( 0.1 ) & 29.0 ( 0.2 ) & 14.54 ( 0.10 ) + [email protected] & 238.3 ( 7.7 ) & -25.7 ( 0.1 ) & 25.9 ( 0.3 ) & 6.57 ( 0.06 ) + [email protected] & 1493.7 ( 26.3 ) & 12.5 ( 0.1 ) & 32.9 ( 0.1 ) & 52.35 ( 0.19 ) + [email protected] & 468.7(10.7 ) & 9.4 ( 0.1 ) & 32.0 ( 0.2 ) & 15.96 ( 0.07 ) + [email protected] & 81.2 ( 6.1 ) & 70.2 ( 0.3 ) & 21.9 ( 0.7 ) & 1.89 ( 0.05 ) + [email protected] & 36.0 ( 2.8 ) & 17.7 ( 0.3 ) & 30.5 ( 0.8 ) & 1.17 ( 0.03 ) + [email protected] & 40.6 ( 3.2 ) & 2.1 ( 0.3 ) & 25.2 ( 0.9 ) & 1.09 ( 0.03 ) + [email protected] & 50.2 ( 3.9 ) & 40.3 ( 0.4 ) & 37.4 ( 1.0 ) & 1.99 ( 0.05 ) + [email protected] & & & & + [email protected] & 6.3 ( 1.3 ) & 34.1 ( 1.7 ) & 23.1 ( 3.7 ) & 0.16 ( 0.02 ) + [email protected] & 36.7 ( 4.3 ) & 33.5 ( 0.5 ) & 22.9 ( 1.1 ) & 0.90 ( 0.04 ) + [email protected] & 17.2 ( 4.8 ) & 15.8 ( 1.3 ) & 28.5 ( 3.3 ) & 0.52 ( 0.05 ) + [email protected] & & & & + [email protected] & 53.6 ( 6.4 ) & 31.7 ( 0.5 ) & 28.1 ( 1.4 ) & 1.63 ( 0.06 ) + [email protected] & 4.0 ( 1.2 ) & 40.6 ( 3.8 ) & 47.3 ( 10.9 ) & 0.20 ( 0.03 ) + [email protected] & 306.9 ( 7.6 ) & 51.0 ( 0.1 ) & 22.9 ( 0.2 ) & 7.47 ( 0.04 ) + [email protected] & 16.5 ( 3.4 ) & 27.6 ( 0.9 ) & 25.7 ( 2.3 ) & 0.45 ( 0.03 ) + [email protected] & 20.5 ( 2.6 ) & 27.5 ( 0.5 ) & 23.5 ( 1.3 ) & 0.51 ( 0.02 ) + [email protected] & 11.3 ( 3.0 & 34.7 ( 1.2 ) & 31.0 ( 3.2 ) & 0.37 ( 0.03 ) + [email protected] & 9.5 ( 2.2 ) & 25.2 ( 1.2 ) & 18.0 ( 2.8 ) & 0.18 ( 0.02 ) + [email protected] & 37.2 ( 3.7 ) & 56.5 ( 0.4 ) & 30.3 ( 1.0 ) & 1.20 ( 0.03 ) + [email protected] & 15.6 ( 2.7 ) & 44.1 ( 0.9 ) & 32.4 ( 2.2 ) & 0.54 ( 0.03 ) + [email protected] & 263.7 ( 8.1 ) & 64.2 ( 0.1 ) & 24.0 ( 0.2 ) & 6.73 ( 0.04 ) + [email protected] & 97.3 ( 4.8 ) & 63.5 ( 0.2 ) & 26.1 ( 0.4 ) & 2.70 ( 0.03 ) + [email protected] & 9.5 ( 1.2 ) & 51.1 ( 0.8 ) & 21.9 ( 1.9 ) & 0.22 ( 0.02 ) + [email protected] & 12.8 ( 2.2 ) & 38.7 ( 1.2 ) & 36.7 ( 2.8 ) & 0.50 ( 0.03 ) + [email protected] & 4.3 ( 1.0 ) & 50.9 ( 2.3 ) & 29.3 ( 4.8 ) & 0.14 ( 0.02 ) + [email protected] & 13.2 ( 2.0 ) & 70.2 ( 0.6 ) & 25.3 ( 1.8 ) & 0.36 ( 0.02 ) + [email protected] & 42.4 ( 3.9 ) & 26.9 ( 0.4 ) & 29.8 ( 0.9 ) & 1.35 ( 0.04 ) + [email protected] & 30.0 ( 2.6 ) & 82.5 ( 0.3 ) & 22.5 ( 0.7 ) & 0.72 ( 0.02 ) + [email protected] & 132.7 ( 12.5 ) & 76.3 ( 0.3 ) & 20.6 ( 0.8 ) & 2.90 ( 0.09 ) + [email protected] & 11.0 ( 2.3 ) & 78.2 ( 1.7 ) & 41.9 ( 4.9 ) & 0.49 ( 0.04 ) + [email protected] & 327.0 ( 8.3 ) & 101.1 ( 0.2 ) & 23.7 ( 0.2 ) & 8.24 ( 0.05 ) + [email protected] & 36.8 ( 3.6 ) & 76.5 ( 0.4 ) & 29.0 ( 1.0 ) & 1.14 ( 0.03 ) + [email protected] & 47.6 ( 8.0 ) & 108.1 ( 0.5 ) & 31.6 ( 1.1 ) & 1.58 ( 0.05 ) + [email protected] & 13.3 ( 2.8 ) & 119.5 ( 1.4 ) & 29.1 ( 3.4 ) & 0.41 ( 0.04 ) + [email protected] & 130.0 ( 6.4 ) & 112.1 ( 0.1 ) & 31.8 ( 0.3 ) & 4.40 ( 0.03 ) + [email protected] & 13.2 ( 3.0 ) & 107.0 ( 1.1 ) & 17.6 ( 2.3 ) & 0.25 ( 0.03 ) + [email protected] & 20.4 ( 4.3 ) & 76.6 ( 0.9 ) & 23.0 ( 1.9 ) & 0.50 ( 0.04 ) + [email protected] & 56.8 ( 3.7 ) & 89.0 ( 0.2 ) & 19.7 ( 0.4 ) & 1.19 ( 0.02 ) + [email protected] & 31.4 ( 4.5 ) & 101.7 ( 0.4 ) & 23.0 ( 1.0 ) & 0.77 ( 0.03 ) + [email protected] & 283.6 ( 9.7 ) & 91.3 ( 0.1 ) & 32.2 ( 0.3 ) & 9.72 ( 0.07 ) + [email protected] & 35.1 ( 4.7 ) & 99.8 ( 0.7 ) & 39.9 ( 1.6 ) & 1.49 ( 0.05 ) + [email protected] & 10.2 ( 2.0 ) & 22.5 ( 0.7 ) & 13.8 ( 2.0 ) & 0.14 ( 0.02 ) + [email protected] & 81.6 ( 4.8 ) & 100.9 ( 0.3 ) & 25.8 ( 0.6 ) & 2.24 ( 0.05 ) + [email protected] & 155.2 ( 7.1 ) & 65.8 ( 0.1 ) & 24.7 ( 0.3 ) & 4.09 ( 0.05 ) + [email protected] & 139.8 ( 9.5 ) & 59.2 ( 0.2 ) & 39.4 ( 0.5 ) & 5.86 ( 0.07 ) + [email protected] & 448.5 ( 10.0 ) & 67.9 ( 0.1 ) & 29.3 ( 0.2 ) & 13.98 ( 0.07 ) + [email protected] & 206.7 ( 8.8 ) & 26.5 ( 0.1 ) & 27.5 ( 0.3 ) & 6.04 ( 0.06 ) + [email protected] h1 & 21.3 ( 4.7 ) & 34.1 ( 0.7 ) & 30.4 ( 1.7 ) & 0.69 ( 0.03 ) + [email protected] h2 & & 109.0 ( 0.9 ) & & + & & & & + [email protected] h1 & 13.5 ( 3.4 ) & 19.7 ( 0.8 ) & 28.4 ( 2.0 ) & 0.41 ( 0.02 ) + [email protected] h2 & & 54.5 ( 1.7 ) & & + & & & & + [email protected] h1 & 14.3 ( 3.1 ) & 90.7 ( 0.7 ) & 28.3 ( 1.6 ) & 0.43 ( 0.02 ) + [email protected] h2 & & 22.4 ( 1.7 ) & & + & & & & + [email protected] h1 & 40.3 ( 3.4 ) & 99.9 ( 0.2 ) & 21.1 ( 0.6 ) & 0.90 ( 0.02 ) + [email protected] h2 & 8.7 ( 3.4 ) & 55.2 ( 1.0 ) & 22.1 ( 2.8 ) & 0.21 ( 0.02 ) + & & & & + [email protected] h1 & 126.4 ( 7.9 ) & 96.4 ( 0.2 ) & 23.0 ( 0.4 ) & 3.09 ( 0.04 ) + [email protected] h2 & 30.5 ( 7.9 ) & 44.3 ( 0.6 ) & 18.6 ( 1.3 ) & 0.60 ( 0.04 ) + & & & & + [email protected] h1 & 184.8 ( 7.5 ) & 111.1 ( 0.1 ) & 23.3 ( 0.3 ) & 4.58 ( 0.05 ) + [email protected] h2 & 49.2 ( 7.5 ) & 44.0 ( 0.3 ) & 12.2 ( 0.8 ) & 0.64 ( 0.04 ) + & & & & + [email protected] h1 & 70.2 ( 7.1 ) & 83.5 ( 0.3 ) & 20.1 ( 0.7 ) & 1.50 ( 0.04 ) + [email protected] h2 & 33.1 ( 7.1 ) & 43.7 ( 0.5 ) & 12.0 ( 1.2 ) & 0.42 ( 0.03 ) + [email protected] h3 & 19.8 ( 7.1 ) & 123.4 ( 0.9 ) & 14.6 ( 3.0 ) & 0.31 ( 0.04 ) + & & & & + [email protected] h1 & 31.6 ( 3.7 ) & 47.0 ( 0.4 ) & 25.8 ( 0.9 ) & 0.87 ( 0.03 ) + [email protected] h2 & 13.8 ( 2.6 ) & 105.2 ( 0.4 ) & 9.0 ( 0.8 ) & 0.13 ( 0.01 ) + & & & & + [email protected] h1 & 32.5 ( 4.3 ) & 113.2 ( 0.4 ) & 29.6 ( 0.9 ) & 1.03 ( 0.03 ) + [email protected] h2 & 17.7 ( 4.3 ) & 58.1 ( 0.8 ) & 34.8 ( 2.0 ) & 0.65 ( 0.03 ) + & & & & + [email protected] h1 & 12.7 ( 2.8 ) & 21.6 ( 0.7 ) & 25.8 ( 1.7 ) & 0.35 ( 0.02 ) + [email protected] h2 & 12.2 ( 2.8 ) & 108.2 ( 1.3 ) & 29.4 ( 3.0 ) & 0.38 ( 0.03 ) + [email protected] h3 & 9.1 ( 2.8 ) & 69.0 ( 1.5 ) & 24.4 ( 4.7 ) & 0.24 ( 0.04 ) + & & & & + [email protected] h1 & 16.2 ( 4.1 ) & 27.8 ( 0.8 ) & 25.6 ( 2.0 ) & 0.44 ( 0.03 ) + [email protected] h2 & 12.4 ( 4.1 ) & 93.7 ( 1.1 ) & 31.9 ( 3.2 ) & 0.42 ( 0.03 ) + & & & & + [email protected] h1 & 32.0 ( 4.8 ) & 100.3 ( 0.7 ) & 22.1 ( 1.7 ) & 0.75 ( 0.05 ) + [email protected] h2 & 12.8 ( 4.8 ) & 77.0 ( 1.3 ) & 14.6 ( 3.1 ) & 0.20 ( 0.05 ) + & & & & + [email protected] h1 & 28.8 ( 2.8 ) & 110.3 ( 0.8 ) & 27.4 ( 1.8 ) & 0.84 ( 0.05 ) + [email protected] h2 & 15.3 ( 2.8 ) & 80.0 ( 1.3 ) & 23.0 ( 3.0 ) & 0.37 ( 0.05 ) + & & & & + [email protected] h1 & 86.5 ( 5.3 ) & 96.5 ( 0.2 ) & 22.5 ( 0.4 ) & 2.07 ( 0.03 ) + [email protected] h2 & 19.3 ( 5.3 ) & 41.5 ( 0.8 ) & 23.1 ( 2.1 ) & 0.47 ( 0.04 ) + & & & & + [email protected] h1 & 25.5 ( 5.6 ) & 37.8 ( 0.6 ) & 21.4 ( 1.4 ) & 0.58 ( 0.03 ) + [email protected] h2 & & 106.0 ( 2.4 ) & & + & & & & + [email protected] h1 & 131.9 ( 7.3 ) & 101.2 ( 0.2 ) & 28.2 ( 0.4 ) & 3.96 ( 0.05 ) + [email protected] h2 & & 47.6 ( 1.6 ) & & + & & & & + [email protected] h1 & 42.4 ( 6.4 ) & 36.3 ( 0.5 ) & 28.7 ( 1.2 ) & 1.30 ( 0.05 ) + [email protected] h2 & 36.9 ( 6.4 ) & 102.9 ( 0.6 ) & 35.9 ( 1.5 ) & 1.41 ( 0.05 ) + & & & & + [email protected] h1 & 53.0 ( 4.5 ) & 33.6 ( 0.2 ) & 32.4 ( 0.5 ) & 1.82 ( 0.03 ) + [email protected] h2 & 41.8 ( 4.5 ) & 98.4 ( 0.3 ) & 32.4 ( 0.6 ) & 1.44 ( 0.03 ) + [email protected] mc 1 & @xmath467.6 ( 11.2 ) & @xmath443.0 ( 0.2 ) & 2.5 ( 0.4 ) & @xmath40.18 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4571.0 ( 35.4 ) & @xmath427.7 ( @xmath410.1 ) & 5.9 ( 0.1 ) & @xmath43.56 ( 0.04 ) + [email protected] mc 2 & @xmath4238.4 ( 15.1 ) & @xmath485.5 ( @xmath410.1 ) & 3.0 ( 0.1 ) & @xmath40.75 ( 0.01 ) + [email protected] mc 3 & @xmath4234.4 ( 16.2 ) & @xmath417.0 ( @xmath410.1 ) & 2.2 ( 0.1 ) & @xmath40.56 ( 0.02 ) + [email protected] mc 4 & @xmath428.8 ( 9.1 ) & @xmath495.5 ( 0.2 ) & 2.6 ( 0.3 ) & @xmath40.08 ( 0.01 ) + & & & & + [email protected] mc 1 & & & & + & & & & + [email protected] mc 1 & @xmath4823.8 ( 38.4 ) & @xmath411.7 ( 0.2 ) & 4.7 ( 0.5 ) & @xmath40.46 ( 0.04 ) + [email protected] mc 2 & @xmath4143.6 ( 16.4 ) & 6.1 ( @xmath410.1 ) & 0.6 ( 0.1 ) & @xmath40.09 ( 0.01 ) + & & & & + [email protected] mc 1 & & & & + & & & & + [email protected] mc 1 & @xmath4131.1 ( 14.5 ) & @xmath414.2 ( 0.1 ) & 6.0 ( 0.3 ) & @xmath40.83 ( 0.03 ) + [email protected] mc 2 & @xmath485.6 ( 14.5 ) & @xmath429.7 ( 0.2 ) & 4.2 ( 0.4 ) & @xmath40.38 ( 0.03 ) + & & & & + [email protected] mc 1 & @xmath4126.1 ( 15.5 ) & @xmath426.6 ( 0.1 ) & 3.1 ( 0.2 ) & @xmath40.42 ( 0.02 ) + [email protected] mc 2 & @xmath488.3 ( 18.7 ) & @xmath415.4 ( 0.1 ) & 2.1 ( 0.5 ) & @xmath40.20 ( 0.03 ) + & & & & + [email protected] mc 1 & @xmath4765.0 ( 97.1 ) & 36.2 ( @xmath410.1 ) & 1.8 ( 0.2 ) & @xmath41.46 ( 0.10 ) + [email protected] mc 2 & @xmath4658.2 ( 97.1 ) & 26.5 ( 0.1 ) & 3.8 ( 0.3 ) & @xmath42.66 ( 0.15 ) + [email protected] mc 3 & @xmath4640.9 ( 97.1 ) & 31.4 ( 0.1 ) & 2.2 ( 0.2 ) & @xmath41.49 ( 0.12 ) + [email protected] mc 4 & @xmath42623.0 ( 190.0 ) & 9.0 ( 0.3 ) & 7.1 ( 0.6 ) & @xmath419.86 ( 1.96 ) + & @xmath42261.7 ( 190.0 ) & 1.1 ( 0.2 ) & 6.4 ( 0.4 ) & @xmath415.45 ( 1.17 ) + & @xmath41405.5 ( 190.0 ) & 16.2 ( 0.4 ) & 6.8 ( 0.7 ) & @xmath410.24 ( 1.50 ) + & & & & + [email protected] mc 1 & @xmath4480.0 ( 39.5 ) & 11.1 ( @xmath410.1 ) & 5.3 ( 0.1 ) & @xmath42.70 ( 0.03 ) + & @xmath4120.0 ( 39.5 ) & 3.9 ( 0.1 ) & 4.1 ( 0.2 ) & @xmath40.52 ( 0.03 ) + [email protected] mc 2 & @xmath4298.3 ( 23.2 ) & 22.3 ( 0.1 ) & 6.2 ( 0.3 ) & @xmath41.98 ( 0.08 ) + & @xmath4130.2 ( 23.2 ) & 28.3 ( 0.2 ) & 3.5 ( 0.4 ) & @xmath40.48 ( 0.06 ) + [email protected] mc 3 & @xmath4428.2 ( 49.2 ) & 38.4 ( @xmath410.1 ) & 1.8 ( 0.1 ) & @xmath40.82 ( 0.04 ) + [email protected] mc 4 & @xmath450.2 ( 14.6 ) & 47.3 ( 0.4 ) & 7.8 ( 0.8 ) & @xmath40.42 ( 0.04 ) + & & & & + [email protected] mc 1 & @xmath4238.3 ( 10.4 ) & 68.9 ( 0.1 ) & 8.3 ( 0.1 ) & @xmath42.11 ( 0.03 ) + [email protected] mc 2 & @xmath4123.7 ( 10.4 ) & 25.8 ( 0.1 ) & 2.3 ( 0.2 ) & @xmath40.31 ( 0.02 ) + [email protected] mc 3 & @xmath428.7 ( 8.7 ) & 155.5 ( 0.3 ) & 3.5 ( 0.6 ) & @xmath40.11 ( 0.02 ) + [email protected] mc 4 & @xmath433.7 ( 6.0 ) & 9.4 ( 0.2 ) & 2.3 ( 0.7 ) & @xmath40.08 ( 0.02 ) + & @xmath428.5 ( 6.0 ) & 7.4 ( 0.2 ) & 1.3 ( 0.3 ) & @xmath40.04 ( 0.02 ) + [email protected] mc 5 & @xmath468.1 ( 7.1 ) & 34.6 ( 0.1 ) & 1.8 ( 0.2 ) & @xmath40.13 ( 0.01 ) + & @xmath443.6 ( 7.1 ) & 30.1 ( 0.1 ) & 1.0 ( 0.2 ) & @xmath40.047 ( 0.006 ) + & @xmath439.8 ( 7.1 ) & 32.0 ( 0.1 ) & 1.5 ( 0.3 ) & @xmath40.06 ( 0.01 ) + & @xmath432.0 ( 7.1 ) & 33.0 ( @xmath410.1 ) & 0.4 ( 0.1 ) & @xmath40.013 ( 0.005 ) + & & & & + [email protected] mc 1 & @xmath4123.4 ( 6.8 ) & 29.0 ( @xmath410.1 ) & 3.9 ( 0.1 ) & @xmath40.51 ( 0.01 ) + [email protected] mc 2 & @xmath480.9 ( 6.8 ) & 20.8 ( 0.1 ) & 4.6 ( 0.2 ) & @xmath40.40 ( 0.01 ) + [email protected] mc 3 & @xmath466.0 ( 4.6 ) & 150.9 ( 0.1 ) & 3.5 ( 0.1 ) & @xmath40.25 ( 0.01 ) + [email protected] mc 4 & @xmath456.5 ( 6.8 ) & 15.2 ( 0.1 ) & 2.7 ( 0.2 ) & @xmath40.16 ( 0.01 ) + [email protected] mc 5 & @xmath449.1 ( 6.8 ) & 10.1 ( 0.1 ) & 4.4 ( 0.4 ) & @xmath40.23 ( 0.01 ) + [email protected] mc 6 & @xmath426.9 ( 4.6 ) & 124.7 ( 0.1 ) & 1.9 ( 0.3 ) & @xmath40.05 ( 0.01 ) + [email protected] mc 7 & @xmath412.9 ( 4.0 ) & 73.8 ( 0.3 ) & 2.7 ( 0.7 ) & @xmath40.04 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4422.4 ( 19.1 ) & 0.5 ( @xmath410.1 ) & 5.3 ( 0.1 ) & @xmath42.39 ( 0.03 ) + [email protected] mc 2 & @xmath466.0 ( 7.1 ) & 30.2 ( 0.1 ) & 3.1 ( 0.3 ) & @xmath40.22 ( 0.02 ) + [email protected] mc 3 & @xmath453.5 ( 7.1 ) & 35.3 ( 0.1 ) & 3.1 ( 0.6 ) & @xmath40.18 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath41146.2 ( 56.3 ) & 37.8 ( @xmath410.1 ) & 3.7 ( @xmath410.1 ) & @xmath44.50 ( 0.03 ) + & & & & + [email protected] mc 1 & @xmath454.5 ( 4.1 ) & 110.6 ( 0.1 ) & 4.9 ( 0.3 ) & @xmath40.28 ( 0.01 ) + [email protected] mc 2 & @xmath435.9 ( 4.1 ) & 97.7 ( 0.1 ) & 3.1 ( 0.3 ) & @xmath40.12 ( 0.01 ) + [email protected] mc 3 & @xmath411.6 ( 4.0 ) & 14.2 ( 0.5 ) & 4.1 ( 1.4 ) & @xmath40.05 ( 0.02 ) + & @xmath49.1 ( 3.0 ) & 21.2 ( 0.9 ) & 5.7 ( 1.7 ) & @xmath40.05 ( 0.02 ) + [email protected] mc 4 & @xmath48.4 ( 3.0 ) & 143.6 ( 0.6 ) & 2.7 ( 1.1 ) & @xmath40.02 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4199.5 ( 14.0 ) & 18.1 ( @xmath410.1 ) & 3.4 ( 0.1 ) & @xmath40.72 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4325 ( 11.4 ) & 21.6 ( @xmath410.1 ) & 3.6 ( 0.1 ) & @xmath41.25 ( 0.02 ) + [email protected] mc 2 & @xmath473 ( 11.4 ) & 47.2 ( 0.1 ) & 4.8 ( 0.3 ) & @xmath40.37 ( 0.02 ) + [email protected] mc 3 & @xmath465 ( 11.4 ) & 55.3 ( 0.1 ) & 3.0 ( 0.3 ) & @xmath40.21 ( 0.01 ) + [email protected] mc 4 & @xmath463 ( 11.4 ) & 34.5 ( 0.1 ) & 5.7 ( 0.4 ) & @xmath40.38 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4107.6 ( 8.2 ) & 45.6 ( 0.1 ) & 3.66 ( 0.2 ) & @xmath40.42 ( 0.02 ) + [email protected] mc 2 & @xmath457.3 ( 8.2 ) & 110.7 ( 0.1 ) & 2.09 ( 0.2 ) & @xmath40.13 ( 0.01 ) + [email protected] mc 3 & @xmath449.4 ( 8.2 ) & 21.6 ( 0.1 ) & 2.19 ( 0.3 ) & @xmath40.12 ( 0.01 ) + [email protected] mc 4 & @xmath440.3 ( 8.2 ) & 52.8 ( 0.2 ) & 5.00 ( 0.5 ) & @xmath40.21 ( 0.02 ) + [email protected] mc 5 & @xmath424.5 ( 3.3 ) & 37.8 ( 0.1 ) & 2.80 ( 0.4 ) & @xmath40.07 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4260.6 ( 5.7 ) & 18.1 ( @xmath410.1 ) & 3.0 ( 0.1 ) & @xmath40.84 ( 0.01 ) + [email protected] mc 2 & @xmath424.2 ( 5.7 ) & 30.4 ( 0.3 ) & 2.8 ( 0.4 ) & @xmath40.07 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4242.5 ( 9.0 ) & 32.7 ( @xmath410.1 ) & 3.8 ( 0.1 ) & @xmath40.99 ( 0.02 ) + [email protected] mc 2 & @xmath487.1 ( 9.0 ) & 18.5 ( 0.1 ) & 2.8 ( 0.2 ) & @xmath40.26 ( 0.01 ) + [email protected] mc 3 & @xmath429.5 ( 6.0 ) & 23.6 ( 0.1 ) & 0.8 ( 0.2 ) & @xmath40.02 ( 0.01 ) + & @xmath425.9 ( 6.0 ) & 25.0 ( 0.1 ) & 0.8 ( 0.3 ) & @xmath40.02 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath41373.0 ( 55.2 ) & 35.1 ( @xmath410.1 ) & 5.7 ( @xmath410.1 ) & @xmath48.33 ( 0.02 ) + [email protected] mc 2 & @xmath4122.9 ( 6.9 ) & 51.5 ( @xmath410.1 ) & 3.0 ( 0.1 ) & @xmath40.39 ( 0.01 ) + [email protected] mc 3 & @xmath436.5 ( 4.8 ) & 11.8 ( 0.1 ) & 5.3 ( 0.4 ) & @xmath40.21 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4200.0 ( 12.1 ) & 45.0 ( 0.1 ) & 3.2 ( 0.1 ) & @xmath40.68 ( 0.02 ) + [email protected] mc 2 & @xmath4102.9 ( 8.2 ) & 14.6 ( 0.1 ) & 5.1 ( 0.2 ) & @xmath40.56 ( 0.02 ) + [email protected] mc 3 & @xmath440.0 ( 3.9 ) & 23.0 ( 0.1 ) & 3.1 ( 0.2 ) & @xmath40.13 ( 0.01 ) + [email protected] mc 4 & @xmath4141.7 ( 8.3 ) & 34.9 ( 0.2 ) & 2.9 ( 0.3 ) & @xmath40.43 ( 0.07 ) + & @xmath4137.9 ( 8.3 ) & 37.5 ( 0.2 ) & 2.6 ( 0.3 ) & @xmath40.38 ( 0.07 ) + [email protected] mc 5 & @xmath459.6 ( 5.2 ) & 56.6 ( 0.1 ) & 3.2 ( 0.2 ) & @xmath40.20 ( 0.02 ) + & @xmath448.5 ( 5.2 ) & 51.5 ( 0.2 ) & 4.8 ( 0.6 ) & @xmath40.25 ( 0.03 ) + & & & & + [email protected] mc 1 & @xmath41341.5 ( 16.0 ) & 48.6 ( @xmath410.1 ) & 3.6 ( @xmath410.1 ) & @xmath45.09 ( 0.03 ) + & & & & + [email protected] mc 1 & @xmath4477.3 ( 23.8 ) & 20.3 ( @xmath410.1 ) & 3.1 ( 0.1 ) & @xmath41.57 ( 0.02 ) + [email protected] mc 2 & @xmath428.5 ( 1.9 ) & 37.3 ( 0.1 ) & 2.3 ( 0.2 ) & @xmath40.07 ( 0.01 ) + & @xmath425.0 ( 1.9 ) & 34.3 ( 0.1 ) & 2.5 ( 0.2 ) & @xmath40.07 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4670.6 ( 20.6 ) & 21.5 ( @xmath410.1 ) & 3.6 ( @xmath410.1 ) & @xmath42.59 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath497.3 ( 3.8 ) & 27.7 ( 0.1 ) & 4.0 ( 0.1 ) & @xmath40.42 ( 0.01 ) + [email protected] mc 2 & @xmath444.7 ( 3.8 ) & 19.3 ( 0.1 ) & 4.0 ( 0.3 ) & @xmath40.19 ( 0.01 ) + [email protected] mc 3 & @xmath424.8 ( 3.8 ) & 38.0 ( 0.1 ) & 1.7 ( 0.3 ) & @xmath40.05 ( 0.01 ) + [email protected] mc 4 & @xmath421.0 ( 3.8 ) & 60.0 ( 0.1 ) & 1.3 ( 0.3 ) & @xmath40.03 ( 0.01 ) + [email protected] mc 5 & @xmath420.8 ( 4.0 ) & 44.5 ( 0.1 ) & 0.7 ( 0.3 ) & @xmath40.02 ( 0.01 ) + [email protected] mc 6 & @xmath412.7 ( 4.0 ) & 71.7 ( 0.2 ) & 1.1 ( 0.5 ) & @xmath40.02 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4210.7 ( 6.6 ) & 59.4 ( @xmath410.1 ) & 2.5 ( @xmath410.1 ) & @xmath40.56 ( 0.01 ) + [email protected] mc 2 & @xmath424.5 ( 6.6 ) & 103.2 ( 0.1 ) & 1.9 ( 0.3 ) & @xmath40.05 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4113.6 ( 6.7 ) & 22.4 ( @xmath410.1 ) & 4.2 ( 0.1 ) & @xmath40.51 ( 0.01 ) + [email protected] mc 2 & @xmath416.1 ( 6.7 ) & 46.9 ( 0.2 ) & 1.4 ( 0.3 ) & @xmath40.02 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4395.9 ( 12.1 ) & 53.8 ( 0.2 ) & 3.8 ( 0.1 ) & @xmath41.61 ( 0.14 ) + & @xmath4161.3 ( 12.1 ) & 50.0 ( 0.4 ) & 6.6 ( 0.5 ) & @xmath41.14 ( 0.14 ) + [email protected] mc 2 & @xmath435.8 ( 6.6 ) & 125.7 ( 0.1 ) & 1.5 ( 0.2 ) & @xmath41.02 ( 0.07 ) + [email protected] mc 3 & @xmath424.5 ( 6.6 ) & 132.5 ( 0.1 ) & 0.9 ( 0.2 ) & @xmath41.72 ( 0.07 ) + & & & & + [email protected] mc 1 & @xmath4163.1 ( 4.8 ) & 45.5 ( @xmath410.1 ) & 2.80 ( 0.1 ) & @xmath40.49 ( 0.01 ) + & @xmath425.2 ( 4.8 ) & 50.2 ( 0.2 ) & 3.46 ( 0.4 ) & @xmath40.09 ( 0.01 ) + [email protected] mc 2 & @xmath427.1 ( 4.8 ) & 54.7 ( 0.1 ) & 2.05 ( 0.4 ) & @xmath40.06 ( 0.01 ) + [email protected] mc 3 & @xmath418.5 ( 4.8 ) & 130.0 ( 0.1 ) & 1.07 ( 0.3 ) & @xmath40.021 ( 0.005 ) + [email protected] mc 4 & @xmath416.2 ( 4.8 ) & 35.7 ( 0.2 ) & 1.34 ( 0.3 ) & @xmath40.023 ( 0.005 ) + & & & & + [email protected] mc 1 & @xmath4811.7 ( 25.4 ) & 65.5 ( @xmath410.1 ) & 5.7 ( 0.1 ) & @xmath44.95 ( 0.05 ) + & @xmath4221.1 ( 25.4 ) & 58.8 ( 0.1 ) & 3.3 ( 0.2 ) & @xmath40.79 ( 0.04 ) + [email protected] mc 2 & @xmath452.5 ( 9.1 ) & 6.4 ( 0.1 ) & 1.9 ( 0.2 ) & @xmath40.11 ( 0.01 ) + [email protected] mc 3 & @xmath431.2 ( 9.1 ) & 36.2 ( 0.2 ) & 3.5 ( 0.8 ) & @xmath40.11 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath476.4 ( 5.4 ) & 63.4 ( 0.1 ) & 7.7 ( 0.3 ) & @xmath40.63 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4379.8 ( 6.3 ) & 26.9 ( @xmath410.1 ) & 3.2 ( @xmath410.1 ) & @xmath41.30 ( 0.01 ) + & @xmath467.3 ( 6.3 ) & 23.1 ( 0.1 ) & 1.8 ( 0.2 ) & @xmath40.13 ( 0.01 ) + [email protected] mc 2 & @xmath431.8 ( 5.4 ) & 67.9 ( 0.2 ) & 2.9 ( 0.5 ) & @xmath40.10 ( 0.01 ) + [email protected] mc 3 & @xmath419.2 ( 5.4 ) & 39.6 ( 0.2 ) & 3.2 ( 0.6 ) & @xmath40.07 ( 0.01 ) + [email protected] mc 4 & @xmath419.4 ( 5.4 ) & 71.6 ( 0.3 ) & 1.6 ( 0.8 ) & @xmath40.03 ( 0.01 ) + [email protected] mc 4 & @xmath424.4 ( 3.0 ) & 117.0 ( 0.1 ) & 1.8 ( 0.4 ) & @xmath40.05 ( 0.01 ) + & @xmath412.4 ( 3.0 ) & 120.6 ( 0.5 ) & 4.5 ( 1.0 ) & @xmath40.06 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath473.3 ( 6.6 ) & 34.9 ( 0.1 ) & 2.6 ( 0.1 ) & @xmath40.20 ( 0.01 ) + [email protected] mc 2 & @xmath418.6 ( 5.6 ) & 58.3 ( 0.3 ) & 5.9 ( 0.9 ) & @xmath40.12 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4139.1 ( 4.4 ) & 43.9 ( @xmath410.1 ) & 3.5 ( 0.1 ) & @xmath40.51 ( 0.01 ) + [email protected] mc 2 & @xmath443.7 ( 4.4 ) & 24.3 ( @xmath410.1 ) & 1.2 ( 0.1 ) & @xmath40.056 ( 0.004 ) + [email protected] mc 3 & @xmath417.1 ( 4.4 ) & 69.0 ( 0.1 ) & 2.0 ( 0.3 ) & @xmath40.036 ( 0.005 ) + & & & & + [email protected] mc 1 & @xmath464.2 ( 3.6 ) & 54.1 ( @xmath410.1 ) & 3.3 ( 0.1 ) & @xmath40.226 ( 0.006 ) + [email protected] mc 2 & @xmath426.9 ( 3.6 ) & 5.6 ( 0.1 ) & 1.3 ( 0.2 ) & @xmath40.038 ( 0.005 ) + & @xmath416.6 ( 3.6 ) & 8.1 ( 0.2 ) & 2.1 ( 0.4 ) & @xmath40.037 ( 0.006 ) + [email protected] mc 3 & @xmath411.8 ( 3.6 ) & 74.2 ( 0.2 ) & 3.2 ( 0.5 ) & @xmath40.040 ( 0.006 ) + & & & & + [email protected] mc 1 & @xmath4224.7 ( 20.1 ) & 23.0 ( 0.1 ) & 3.4 ( 0.1 ) & @xmath40.81 ( 0.03 ) + [email protected] mc 2 & @xmath432.2 ( 4.8 ) & 49.6 ( 0.1 ) & 4.7 ( 0.4 ) & @xmath40.16 ( 0.01 ) + [email protected] mc 3 & @xmath428.5 ( 6.4 ) & 102.5 ( 0.1 ) & 1.0 ( 0.3 ) & @xmath40.03 ( 0.01 ) + [email protected] mc 4 & @xmath420.9 ( 5.1 ) & 6.7 ( 0.3 ) & 5.3 ( 0.6 ) & @xmath40.12 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath430.7 ( 4.2 ) & 50.2 ( 0.1 ) & 4.2 ( 0.3 ) & @xmath40.14 ( 0.01 ) + [email protected] mc 2 & @xmath430.4 ( 4.2 ) & 14.0 ( 0.1 ) & 2.8 ( 0.3 ) & @xmath40.09 ( 0.01 ) + [email protected] mc 3 & @xmath475.4 ( 4.3 ) & 84.5 ( 0.1 ) & 3.7 ( 0.1 ) & @xmath40.30 ( 0.02 ) + & @xmath445.7 ( 4.3 ) & 80.9 ( 0.1 ) & 2.5 ( 0.2 ) & @xmath40.123 ( 0.005 ) + & @xmath427.3 ( 4.3 ) & 89.3 ( 0.4 ) & 7.1 ( 0.7 ) & @xmath40.21 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4261.9 ( 22.1 ) & 94.6 ( 0.1 ) & 5.2 ( 0.1 ) & @xmath41.46 ( 0.03 ) + [email protected] mc 2 & @xmath4115.1 ( 7.2 ) & 76.4 ( @xmath410.1 ) & 4.2 ( 0.1 ) & @xmath40.52 ( 0.01 ) + [email protected] mc 3 & @xmath429.5 ( 5.3 ) & 4.7 ( 0.1 ) & 1.3 ( 0.2 ) & @xmath40.042 ( 0.005 ) + & @xmath415.7 ( 5.3 ) & 48.6 ( 0.1 ) & 1.8 ( 0.3 ) & @xmath40.030 ( 0.005 ) + & & & & + [email protected] mc 1 & @xmath4293.1 ( 21.4 ) & 62.9 ( 0.1 ) & 6.7 ( 0.2 ) & @xmath42.07 ( 0.04 ) + [email protected] mc 2 & @xmath4204.1 ( 16.2 ) & 80.2 ( 0.1 ) & 6.5 ( 0.2 ) & @xmath41.41 ( 0.03 ) + [email protected] mc 3 & @xmath4120.4 ( 13.0 ) & 95.6 ( 0.1 ) & 4.7 ( 0.2 ) & @xmath40.61 ( 0.02 ) + [email protected] mc 4 & @xmath465.1 ( 13.1 ) & 4.4 ( 0.1 ) & 3.8 ( 0.3 ) & @xmath40.26 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4134.1 ( 7.1 ) & 77.5 ( @xmath410.1 ) & 3.3 ( 0.1 ) & @xmath40.47 ( 0.01 ) + [email protected] mc 2 & @xmath429.1 ( 7.3 ) & 7.7 ( 0.1 ) & 1.0 ( 0.2 ) & @xmath40.030 ( 0.005 ) + [email protected] mc 3 & @xmath426.7 ( 7.3 ) & 19.8 ( 0.1 ) & 2.8 ( 0.4 ) & @xmath40.08 ( 0.01 ) + [email protected] mc 4 & @xmath424.7 ( 7.1 ) & 93.0 ( 0.5 ) & 13.4 ( 1.3 ) & @xmath40.35 ( 0.03 ) + [email protected] mc 5 & @xmath419.2 ( 5.5 ) & 27.7 ( 0.2 ) & 3.7 ( 0.5 ) & @xmath40.08 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath41138.4 ( 37.2 ) & 100.6 ( @xmath410.1 ) & 3.3 ( 0.1 ) & @xmath43.94 ( 0.07 ) + & @xmath4986.0 ( 37.2 ) & 96.8 ( @xmath410.1 ) & 2.4 ( 0.1 ) & @xmath42.49 ( 0.06 ) + [email protected] mc 2 & @xmath4499.8 ( 23.6 ) & 80.5 ( @xmath410.1 ) & 5.0 ( 0.1 ) & @xmath42.67 ( 0.04 ) + [email protected] mc 3 & @xmath477.0 ( 9.5 ) & 64.3 ( 0.2 ) & 9.5 ( 0.7 ) & @xmath40.78 ( 0.04 ) + [email protected] mc 4 & @xmath475.7 ( 15.2 ) & 3.2 ( 0.1 ) & 2.2 ( 0.3 ) & @xmath40.18 ( 0.02 ) + & @xmath475.9 ( 15.2 ) & 6.9 ( 0.1 ) & 3.0 ( 0.4 ) & @xmath40.24 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4139.1 ( 8.3 ) & 86.0 ( 0.1 ) & 2.7 ( 0.2 ) & @xmath40.40 ( 0.02 ) + & @xmath477.0 ( 8.3 ) & 80.8 ( 0.1 ) & 3.9 ( 0.3 ) & @xmath40.32 ( 0.02 ) + [email protected] mc 2 & @xmath479.4 ( 8.3 ) & 68.9 ( 0.1 ) & 4.9 ( 0.3 ) & @xmath40.42 ( 0.02 ) + [email protected] mc 3 & @xmath419.7 ( 6.2 ) & 6.2 ( 0.2 ) & 1.1 ( 0.4 ) & @xmath40.02 ( 0.01 ) + & @xmath418.7 ( 6.2 ) & 4.1 ( 0.2 ) & 1.9 ( 0.9 ) & @xmath40.04 ( 0.01 ) + [email protected] mc 4 & @xmath428.8 ( 5.1 ) & 95.2 ( 0.2 ) & 3.4 ( 0.5 ) & @xmath40.10 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4244.5 ( 9.6 ) & 57.8 ( @xmath410.1 ) & 3.8 ( 0.1 ) & @xmath40.99 ( 0.02 ) + [email protected] mc 2 & @xmath480.3 ( 4.8 ) & 45.9 ( 0.1 ) & 3.4 ( 0.1 ) & @xmath40.29 ( 0.01 ) + [email protected] mc 3 & @xmath432.6 ( 4.8 ) & 6.5 ( 0.1 ) & 3.1 ( 0.3 ) & @xmath40.11 ( 0.01 ) + [email protected] mc 4 & @xmath417.6 ( 4.6 ) & 72.7 ( 0.1 ) & 1.2 ( 0.2 ) & @xmath40.023 ( 0.003 ) + & & & & + [email protected] mc 1 & @xmath4213.3 ( 9.7 ) & 49.0 ( @xmath410.1 ) & 3.1 ( 0.1 ) & @xmath40.70 ( 0.02 ) + & @xmath478.0 ( 9.7 ) & 44.1 ( 0.1 ) & 2.7 ( 0.2 ) & @xmath40.22 ( 0.02 ) + & @xmath460.0 ( 9.7 ) & 53.1 ( 0.1 ) & 1.6 ( 0.3 ) & @xmath40.10 ( 0.01 ) + [email protected] mc 2 & @xmath478.4 ( 7.1 ) & 110.2 ( 0.1 ) & 4.8 ( 0.3 ) & @xmath40.40 ( 0.02 ) + [email protected] mc 3 & @xmath462.5 ( 9.7 ) & 5.6 ( 0.1 ) & 2.9 ( 0.4 ) & @xmath40.19 ( 0.02 ) + [email protected] mc 4 & @xmath448.5 ( 7.1 ) & 83.4 ( 0.2 ) & 3.7 ( 0.5 ) & @xmath40.19 ( 0.02 ) + [email protected] mc 5 & @xmath424.7 ( 7.1 ) & 95.6 ( 0.3 ) & 2.9 ( 0.8 ) & @xmath40.08 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4156.4 ( 11.9 ) & 45.2 ( 0.1 ) & 3.7 ( 0.2 ) & @xmath40.61 ( 0.03 ) + & @xmath4117.0 ( 11.9 ) & 41.5 ( 0.1 ) & 2.1 ( 0.1 ) & @xmath40.26 ( 0.01 ) + & @xmath447.0 ( 11.9 ) & 43.3 ( 0.1 ) & 0.7 ( 0.2 ) & @xmath40.04 ( 0.02 ) + [email protected] mc 2 & @xmath462.1 ( 5.1 ) & 54.8 ( 0.1 ) & 2.1 ( 0.3 ) & @xmath40.14 ( 0.02 ) + & @xmath457.6 ( 5.1 ) & 52.6 ( 0.1 ) & 1.7 ( 0.2 ) & @xmath40.10 ( 0.03 ) + [email protected] mc 3 & @xmath447.8 ( 8.6 ) & 99.9 ( 0.2 ) & 4.8 ( 0.5 ) & @xmath40.25 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4282.1 ( 8.8 ) & 51.9 ( @xmath410.1 ) & 2.5 ( 0.1 ) & @xmath40.74 ( 0.03 ) + & @xmath4138.6 ( 8.8 ) & 55.1 ( 0.1 ) & 2.4 ( 0.2 ) & @xmath40.35 ( 0.02 ) + & @xmath497.6 ( 8.8 ) & 46.0 ( 0.1 ) & 3.2 ( 0.3 ) & @xmath40.34 ( 0.03 ) + & @xmath493.3 ( 8.8 ) & 49.0 ( 0.1 ) & 1.7 ( 0.2 ) & @xmath40.17 ( 0.03 ) + [email protected] mc 2 & @xmath4228.0 ( 8.9 ) & 112.1 ( @xmath410.1 ) & 4.3 ( 0.1 ) & @xmath41.03 ( 0.02 ) + [email protected] mc 3 & @xmath4110.5 ( 8.9 ) & 93.8 ( 0.1 ) & 6.0 ( 0.2 ) & @xmath40.70 ( 0.02 ) + [email protected] mc 4 & @xmath482.2 ( 8.1 ) & 79.6 ( 0.2 ) & 2.0 ( 0.3 ) & @xmath40.18 ( 0.06 ) + & @xmath463.5 ( 8.1 ) & 84.5 ( 0.6 ) & 5.0 ( 1.0 ) & @xmath40.34 ( 0.08 ) + & @xmath445.1 ( 8.1 ) & 81.5 ( 0.7 ) & 2.8 ( 0.4 ) & @xmath40.13 ( 0.02 ) + [email protected] mc 5 & @xmath447.0 ( 8.2 ) & 6.5 ( 0.1 ) & 2.8 ( 0.4 ) & @xmath40.14 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4197.2 ( 7.5 ) & 48.8 ( @xmath410.1 ) & 1.9 ( 0.1 ) & @xmath40.39 ( 0.01 ) + & @xmath4161.2 ( 7.5 ) & 51.1 ( 0.1 ) & 1.5 ( 0.2 ) & @xmath40.25 ( 0.03 ) + & @xmath477.7 ( 7.5 ) & 50.1 ( 0.1 ) & 0.7 ( 0.1 ) & @xmath40.06 ( 0.02 ) + [email protected] mc 2 & @xmath444.0 ( 11.1 ) & 85.2 ( 0.4 ) & 35.8 ( 0.8 ) & @xmath40.69 ( 0.07 ) + [email protected] mc 3 & @xmath426.0 ( 8.1 ) & 6.7 ( 0.1 ) & 2.62 ( 0.2 ) & @xmath40.071 ( 0.005 ) + & & & & + [email protected] mc 1 & @xmath450.5 ( 4.1 ) & 45.2 ( 0.1 ) & 3.9 ( 0.2 ) & @xmath40.21 ( 0.01 ) + [email protected] mc 2 & @xmath434.5 ( 4.5 ) & 8.6 ( 0.1 ) & 0.8 ( 0.1 ) & @xmath40.031 ( 0.004 ) + & @xmath429.6 ( 4.5 ) & 6.1 ( 0.1 ) & 2.0 ( 0.3 ) & @xmath40.064 ( 0.006 ) + [email protected] mc 3 & @xmath420.5 ( 4.1 ) & 52.8 ( 0.2 ) & 3.4 ( 0.5 ) & @xmath40.07 ( 0.01 ) + [email protected] mc 4 & @xmath419.3 ( 4.1 ) & 35.2 ( 0.2 ) & 2.5 ( 0.4 ) & @xmath40.052 ( 0.006 ) + [email protected] mc 5 & @xmath417.8 ( 4.5 ) & 13.4 ( 0.2 ) & 1.9 ( 0.5 ) & @xmath40.035 ( 0.006 ) + [email protected] mc 6 & @xmath412.4 ( 3.3 ) & 103.4 ( 0.2 ) & 1.7 ( 0.5 ) & @xmath40.022 ( 0.006 ) + & & & & + [email protected] mc 1 & @xmath4173.7 ( 8.4 ) & 118.7 ( @xmath410.1 ) & 2.5 ( 0.1 ) & @xmath40.47 ( 0.01 ) + [email protected] mc 2 & @xmath435.8 ( 8.4 ) & 64.8 ( 0.1 ) & 2.7 ( 0.4 ) & @xmath40.10 ( 0.01 ) + [email protected] mc 3 & @xmath439.3 ( 2.7 ) & 93.6 ( 0.1 ) & 1.9 ( 0.2 ) & @xmath40.08 ( 0.01 ) + & @xmath420.2 ( 2.7 ) & 95.8 ( 0.1 ) & 1.7 ( 0.3 ) & @xmath40.037 ( 0.006 ) + & & & & + [email protected] mc 1 & @xmath4105.9 ( 7.5 ) & 94.1 ( 0.1 ) & 3.2 ( 0.1 ) & @xmath40.36 ( 0.01 ) + [email protected] mc 2 & @xmath420.2 ( 5.1 ) & 112.9 ( 0.2 ) & 2.3 ( 0.4 ) & @xmath40.05 ( 0.01 ) + [email protected] mc 3 & @xmath451.9 ( 6.6 ) & 107.1 ( 0.4 ) & 3.3 ( 0.3 ) & @xmath40.18 ( 0.05 ) + & @xmath440.5 ( 6.6 ) & 108.8 ( 0.2 ) & 1.8 ( 0.5 ) & @xmath40.08 ( 0.05 ) + [email protected] mc 4 & @xmath422.5 ( 1.5 ) & 149.0 ( @xmath410.1 ) & 0.7 ( 0.1 ) & @xmath40.017 ( 0.001 ) + & & & & + [email protected] mc 1 & @xmath4140.5 ( 6.5 ) & 109.7 ( @xmath410.1 ) & 3.8 ( 0.1 ) & @xmath40.57 ( 0.01 ) + [email protected] mc 2 & @xmath443.8 ( 6.5 ) & 5.0 ( 0.1 ) & 1.7 ( 0.4 ) & @xmath40.08 ( 0.01 ) + [email protected] mc 3 & @xmath423.8 ( 6.5 ) & 18.3 ( 0.1 ) & 1.1 ( 0.3 ) & @xmath40.028 ( 0.006 ) + & & & & + [email protected] mc 1 & @xmath4199.0 ( 17.9 ) & 100.3 ( @xmath410.1 ) & 3.6 ( 0.1 ) & @xmath40.77 ( 0.02 ) + [email protected] mc 2 & @xmath443.1 ( 7.6 ) & 110.6 ( 0.1 ) & 3.7 ( 0.3 ) & @xmath40.17 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath459.8 ( 5.8 ) & 24.3 ( 0.1 ) & 4.4 ( 0.3 ) & @xmath40.28 ( 0.02 ) + [email protected] mc 2 & @xmath431.2 ( 5.8 ) & 100.0 ( 0.2 ) & 3.3 ( 0.5 ) & @xmath40.11 ( 0.01 ) + [email protected] mc 3 & @xmath426.2 ( 4.7 ) & 66.2 ( 0.1 ) & 1.2 ( 0.2 ) & @xmath40.03 ( 0.01 ) + & @xmath425.1 ( 4.7 ) & 64.6 ( 0.2 ) & 1.8 ( 0.5 ) & @xmath40.05 ( 0.01 ) + & @xmath416.2 ( 4.7 ) & 61.8 ( 0.1 ) & 0.8 ( 0.3 ) & @xmath40.014 ( 0.003 ) + [email protected] mc 4 & @xmath440.0 ( 6.0 ) & 31.4 ( 0.4 ) & 7.3 ( 0.8 ) & @xmath40.24 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath482.5 ( 6.4 ) & 47.1 ( 0.1 ) & 2.2 ( 0.2 ) & @xmath40.19 ( 0.01 ) + [email protected] mc 2 & @xmath467.2 ( 6.4 ) & 26.2 ( 0.1 ) & 4.3 ( 0.2 ) & @xmath40.30 ( 0.01 ) + [email protected] mc 3 & @xmath460.0 ( 6.4 ) & 93.2 ( 0.1 ) & 2.9 ( 0.2 ) & @xmath40.19 ( 0.01 ) + [email protected] mc 4 & @xmath419.4 ( 6.4 ) & 70.7 ( 0.2 ) & 2.4 ( 0.5 ) & @xmath40.05 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4354.9 ( 11.8 ) & 75.1 ( @xmath410.1 ) & 4.0 ( 0.1 ) & @xmath41.52 ( 0.01 ) + [email protected] mc 2 & @xmath438.0 ( 11.8 ) & 58.1 ( 0.1 ) & 2.0 ( 0.3 ) & @xmath40.08 ( 0.01 ) + [email protected] mc 3 & @xmath427.4 ( 8.8 ) & 97.4 ( 0.3 ) & 4.7 ( 1.0 ) & @xmath40.14 ( 0.02 ) + [email protected] mc 4 & @xmath421.2 ( 4.4 ) & 12.2 ( 0.1 ) & 0.9 ( 0.2 ) & @xmath40.021 ( 0.004 ) + & @xmath418.5 ( 4.4 ) & 10.1 ( 0.1 ) & 1.8 ( 0.4 ) & @xmath40.035 ( 0.006 ) + [email protected] mc 5 & @xmath419.7 ( 6.5 ) & 23.0 ( 0.5 ) & 3.8 ( 1.0 ) & @xmath40.08 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4350.7 ( 15.8 ) & 97.1 ( @xmath410.1 ) & 4.4 ( 0.1 ) & @xmath41.64 ( 0.03 ) + & @xmath4118.1 ( 15.8 ) & 90.9 ( 0.1 ) & 4.6 ( 0.3 ) & @xmath40.56 ( 0.03 ) + [email protected] mc 2 & @xmath4206.6 ( 15.8 ) & 77.8 ( 0.1 ) & 4.1 ( 0.1 ) & @xmath40.91 ( 0.02 ) + [email protected] mc 3 & @xmath480.3 ( 13.2 ) & 105.8 ( 0.1 ) & 3.2 ( 0.4 ) & @xmath40.27 ( 0.02 ) + [email protected] mc 4 & @xmath421.0 ( 6.3 ) & 45.8 ( 0.3 ) & 6.3 ( 1.0 ) & @xmath40.14 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4493.1 ( 23.7 ) & 77.1 ( @xmath410.1 ) & 3.4 ( 0.1 ) & @xmath41.80 ( 0.02 ) + [email protected] mc 2 & @xmath490.1 ( 9.5 ) & 107.3 ( 0.1 ) & 3.6 ( 0.2 ) & @xmath40.34 ( 0.02 ) + [email protected] mc 3 & @xmath431.3 ( 5.7 ) & 47.7 ( 0.2 ) & 5.7 ( 0.5 ) & @xmath40.19 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4176.1 ( 5.0 ) & 89.4 ( @xmath410.1 ) & 4.0 ( 0.1 ) & @xmath40.76 ( 0.01 ) + & @xmath425.9 ( 5.0 ) & 84.3 ( 0.1 ) & 2.4 ( 0.4 ) & @xmath40.07 ( 0.01 ) + [email protected] mc 2 & @xmath443.3 ( 5.0 ) & 73.7 ( 0.1 ) & 2.7 ( 0.2 ) & @xmath40.12 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4302.4 ( 17.7 ) & 80.1 ( @xmath410.1 ) & 2.7 ( 0.1 ) & @xmath40.85 ( 0.02 ) + & @xmath4249.5 ( 17.7 ) & 75.5 ( @xmath410.1 ) & 3.9 ( 0.1 ) & @xmath41.04 ( 0.02 ) + [email protected] mc 2 & @xmath4222.6 ( 13.7 ) & 101.0 ( @xmath410.1 ) & 3.8 ( 0.1 ) & @xmath40.89 ( 0.02 ) + & @xmath4154.2 ( 13.7 ) & 106.9 ( 0.1 ) & 3.3 ( 0.1 ) & @xmath40.54 ( 0.02 ) + [email protected] mc 3 & @xmath472.5 ( 9.5 ) & 8.2 ( 0.1 ) & 1.4 ( 0.1 ) & @xmath40.11 ( 0.01 ) + [email protected] mc 4 & @xmath442.4 ( 9.5 ) & 19.1 ( 0.1 ) & 1.1 ( 0.2 ) & @xmath40.05 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4305.4 ( 14.0 ) & 100.3 ( @xmath410.1 ) & 3.6 ( 0.1 ) & @xmath41.18 ( 0.02 ) + [email protected] mc 2 & @xmath4169.1 ( 5.0 ) & 80.5 ( @xmath410.1 ) & 3.0 ( 0.1 ) & @xmath40.55 ( 0.01 ) + [email protected] mc 3 & @xmath494.6 ( 10.0 ) & 52.4 ( 0.1 ) & 2.6 ( 0.2 ) & @xmath40.26 ( 0.02 ) + [email protected] mc 4 & @xmath449.9 ( 14.0 ) & 94.2 ( 0.1 ) & 2.3 ( 0.3 ) & @xmath40.12 ( 0.01 ) + [email protected] mc 5 & @xmath424.4 ( 5.0 ) & 72.0 ( 0.1 ) & 2.2 ( 0.3 ) & @xmath40.06 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4551.6 ( 19.4 ) & 98.2 ( 0.1 ) & 4.4 ( 0.1 ) & @xmath42.60 ( 0.07 ) + & @xmath4265.0 ( 19.4 ) & 102.5 ( 0.1 ) & 3.4 ( 0.2 ) & @xmath40.95 ( 0.06 ) + [email protected] mc 2 & @xmath4483.8 ( 54.3 ) & 7.0 ( @xmath410.1 ) & 1.8 ( 0.1 ) & @xmath40.95 ( 0.02 ) + [email protected] mc 3 & @xmath4152.2 ( 15.8 ) & 67.1 ( 0.1 ) & 2.9 ( 0.2 ) & @xmath40.46 ( 0.03 ) + [email protected] mc 4 & @xmath465.5 ( 10.2 ) & 49.9 ( 0.1 ) & 1.3 ( 0.2 ) & @xmath40.09 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4173.4 ( 8.4 ) & 112.2 ( @xmath410.1 ) & 3.0 ( 0.1 ) & @xmath40.55 ( 0.01 ) + [email protected] mc 2 & @xmath4134.9 ( 8.4 ) & 87.1 ( @xmath410.1 ) & 2.6 ( 0.1 ) & @xmath40.37 ( 0.01 ) + [email protected] mc 3 & @xmath459.6 ( 10.3 ) & 99.1 ( 0.2 ) & 4.9 ( 0.5 ) & @xmath40.31 ( 0.02 ) + [email protected] mc 4 & @xmath448.2 ( 6.6 ) & 51.1 ( 0.2 ) & 2.9 ( 0.4 ) & @xmath40.15 ( 0.02 ) + & @xmath429.9 ( 6.6 ) & 48.2 ( 0.2 ) & 1.6 ( 0.5 ) & @xmath40.05 ( 0.02 ) + [email protected] mc 5 & @xmath429.7 ( 4.5 ) & 42.0 ( 0.3 ) & 5.0 ( 0.7 ) & @xmath40.16 ( 0.02 ) + & @xmath419.4 ( 4.5 ) & 38.8 ( 0.1 ) & 1.5 ( 0.4 ) & @xmath40.03 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath480.6 ( 5.9 ) & 15.9 ( 0.1 ) & 3.8 ( 0.1 ) & @xmath40.32 ( 0.01 ) + [email protected] mc 2 & @xmath479.5 ( 5.9 ) & 44.8 ( 0.1 ) & 2.7 ( 0.2 ) & @xmath40.23 ( 0.01 ) + [email protected] mc 3 & @xmath420.8 ( 5.5 ) & 91.9 ( 0.2 ) & 2.8 ( 0.6 ) & @xmath40.06 ( 0.01 ) + & @xmath420.8 ( 5.5 ) & 94.8 ( 0.2 ) & 1.4 ( 0.3 ) & @xmath40.03 ( 0.01 ) + [email protected] mc 4 & @xmath418.3 ( 5.9 ) & 71.5 ( 0.3 ) & 3.8 ( 0.5 ) & @xmath40.07 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4169.2 ( 9.3 ) & 97.6 ( 0.1 ) & 3.7 ( 0.1 ) & @xmath40.66 ( 0.02 ) + [email protected] mc 2 & @xmath424.9 ( 8.1 ) & 82.8 ( 0.4 ) & 4.4 ( 0.8 ) & @xmath40.12 ( 0.02 ) + [email protected] mc 3 & @xmath440.1 ( 9.0 ) & 60.6 ( 0.4 ) & 8.6 ( 0.8 ) & @xmath40.11 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4323.3 ( 20.3 ) & 100.0 ( 0.1 ) & 4.4 ( 0.3 ) & @xmath41.52 ( 0.07 ) + & @xmath4311.2 ( 15.4 ) & 94.9 ( 0.1 ) & 6.8 ( 0.3 ) & @xmath42.27 ( 0.07 ) + [email protected] mc 2 & @xmath472.2 ( 15.4 ) & 52.0 ( 0.1 ) & 3.6 ( 0.4 ) & @xmath40.28 ( 0.02 ) + [email protected] mc 3 & @xmath455.3 ( 8.6 ) & 7.5 ( 0.1 ) & 2.2 ( 0.3 ) & @xmath40.13 ( 0.01 ) + & @xmath443.8 ( 8.6 ) & 13.7 ( 0.1 ) & 1.5 ( 0.3 ) & @xmath40.07 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4765.5 ( 49.5 ) & 96.3 ( 0.1 ) & 4.4 ( 0.1 ) & @xmath43.55 ( 0.08 ) + [email protected] mc 2 & @xmath4139.6 ( 13.6 ) & 74.2 ( 0.1 ) & 4.3 ( 0.1 ) & @xmath40.63 ( 0.02 ) + [email protected] mc 3 & @xmath4242.1 ( 31.8 ) & 82.8 ( 0.1 ) & 1.6 ( 0.3 ) & @xmath40.40 ( 0.10 ) + & @xmath4144.3 ( 31.8 ) & 86.7 ( 0.4 ) & 5.8 ( 1.2 ) & @xmath40.89 ( 0.15 ) + [email protected] mc 4 & @xmath494.8 ( 6.3 ) & 106.3 ( 0.1 ) & 4.4 ( 0.1 ) & @xmath40.44 ( 0.01 ) + & @xmath483.6 ( 6.3 ) & 107.8 ( @xmath410.1 ) & 0.9 ( 0.1 ) & @xmath40.08 ( 0.01 ) + [email protected] mc 5 & @xmath459.5 ( 13.6 ) & 38.8 ( 0.2 ) & 6.5 ( 0.5 ) & @xmath40.41 ( 0.03 ) + [email protected] mc 6 & @xmath444.0 ( 8.9 ) & 5.4 ( 0.1 ) & 1.2 ( 0.2 ) & @xmath40.06 ( 0.01 ) + [email protected] mc 7 & @xmath437.6 ( 10.5 ) & 11.8 ( 0.1 ) & 1.3 ( 0.3 ) & @xmath40.05 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4187.9 ( 11.1 ) & 66.8 ( @xmath410.1 ) & 1.6 ( 0.1 ) & @xmath40.32 ( 0.01 ) + [email protected] mc 2 & @xmath468.0 ( 11.1 ) & 38.3 ( 0.2 ) & 5.0 ( 0.6 ) & @xmath40.36 ( 0.03 ) + [email protected] mc 3 & @xmath442.9 ( 11.1 ) & 11.7 ( 0.2 ) & 2.4 ( 0.6 ) & @xmath40.11 ( 0.02 ) + [email protected] mc 4 & @xmath437.9 ( 11.1 ) & 7.8 ( 0.2 ) & 2.4 ( 0.6 ) & @xmath40.10 ( 0.02 ) + [email protected] mc 5 & @xmath437.8 ( 3.2 ) & 45.9 ( @xmath410.1 ) & 1.6 ( 0.1 ) & @xmath40.062 ( 0.003 ) + [email protected] mc 6 & @xmath431.1 ( 5.3 ) & 104.1 ( 0.3 ) & 8.5 ( 0.8 ) & @xmath40.28 ( 0.02 ) + & @xmath418.8 ( 5.3 ) & 109.8 ( 0.2 ) & 1.8 ( 0.5 ) & @xmath40.03 ( 0.01 ) + [email protected] mc 7 & @xmath416.2 ( 5.3 ) & 84.6 ( 0.3 ) & 2.7 ( 0.6 ) & @xmath40.05 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath41357.8 ( 47.1 ) & 97.8 ( 0.1 ) & 6.2 ( 0.1 ) & @xmath48.91 ( 0.11 ) + [email protected] mc 2 & @xmath437.0 ( 12.0 ) & 12.7 ( 0.2 ) & 2.7 ( 0.7 ) & @xmath40.11 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4901.3 ( 26.0 ) & 63.5 ( @xmath410.1 ) & 5.5 ( 0.1 ) & @xmath45.29 ( 0.04 ) + [email protected] mc 2 & @xmath4243.2 ( 26.0 ) & 19.9 ( 0.1 ) & 2.2 ( 0.2 ) & @xmath40.57 ( 0.03 ) + [email protected] mc 3 & @xmath4204.0 ( 26.0 ) & 13.6 ( 0.1 ) & 2.6 ( 0.2 ) & @xmath40.56 ( 0.03 ) + [email protected] mc 4 & @xmath4115.6 ( 26.0 ) & 83.3 ( 0.2 ) & 4.7 ( 0.4 ) & @xmath40.58 ( 0.04 ) + [email protected] mc 5 & @xmath488.8 ( 26.0 ) & 90.9 ( 0.2 ) & 3.5 ( 0.5 ) & @xmath40.33 ( 0.03 ) + [email protected] mc 6 & @xmath456.6 ( 11.3 ) & 38.1 ( 0.1 ) & 2.3 ( 0.3 ) & @xmath40.14 ( 0.01 ) + & & & & + [email protected] mc 1 & @xmath4605.0 ( 11.0 ) & 64.8 ( 0.4 ) & 20.3 ( 0.8 ) & @xmath46.83 ( 0.06 ) + [email protected] mc 2 & @xmath4445.9 ( 13.2 ) & 74.3 ( 0.1 ) & 2.3 ( 0.2 ) & @xmath41.08 ( 0.09 ) + & @xmath4414.0 ( 13.2 ) & 72.4 ( 0.1 ) & 1.7 ( 0.1 ) & @xmath40.76 ( 0.08 ) + [email protected] mc 3 & @xmath4206.0 ( 15.9 ) & 14.0 ( 0.1 ) & 3.2 ( 0.1 ) & @xmath40.70 ( 0.02 ) + [email protected] mc 4 & @xmath4146.6 ( 28.7 ) & 80.1 ( 0.2 ) & 3.0 ( 0.5 ) & @xmath40.46 ( 0.05 ) + [email protected] mc 5 & @xmath499.2 ( 14.3 ) & 20.5 ( 0.1 ) & 2.2 ( 0.3 ) & @xmath40.23 ( 0.02 ) + [email protected] mc 6 & @xmath488.6 ( 29.4 ) & 91.9 ( 0.3 ) & 1.2 ( 0.6 ) & @xmath40.11 ( 0.05 ) + & & & & + [email protected] mc 1 & @xmath42383.8 ( 93.8 ) & 65.8 ( @xmath410.1 ) & 2.9 ( 0.1 ) & @xmath47.43 ( 0.14 ) + & @xmath4531.4 ( 93.8 ) & 61.8 ( 0.1 ) & 2.9 ( 0.3 ) & @xmath41.61 ( 0.14 ) + [email protected] mc 2 & @xmath4208.4 ( 21.5 ) & 6.1 ( 0.1 ) & 2.2 ( 0.1 ) & @xmath40.48 ( 0.02 ) + & & & & + [email protected] mc 1 & @xmath4417.2 ( 11.4 ) & 21.6 ( @xmath410.1 ) & 2.9 ( 0.1 ) & @xmath41.30 ( 0.02 ) + [email protected] mc 2 & @xmath433.2 ( 11.4 ) & 5.6 ( 0.3 ) & 2.7 ( 0.7 ) & @xmath40.10 ( 0.02 ) + + far & @xmath42 + & and ( @xmath2v@xmath43 @xmath7 10 km s@xmath23 ) @xmath41 @xmath2v@xmath44 + near & @xmath45 + & and ( @xmath2v@xmath43 @xmath7 10 km s@xmath23 ) @xmath41 @xmath2v@xmath44 + no classification & ( @xmath2v@xmath35@xmath2 @xmath4 10 km s@xmath23 ) @xmath41 @xmath2v@xmath46@xmath2 @xmath41 @xmath2v@xmath35@xmath2 + tangent point & @xmath2v@xmath46@xmath2 @xmath47 @xmath2v@xmath44 + & + & + + near & @xmath2v@xmath48 @xmath41 ( @xmath2v@xmath43 @xmath4 10 km s@xmath23 ) + associated with an hii region & ( @xmath2v@xmath43 @xmath4 10 km s@xmath23 ) @xmath41 @xmath2v@xmath48 @xmath41 ( @xmath2v@xmath43 @xmath7 10 km s@xmath23 ) + ambiguous & ( @xmath2v@xmath43 @xmath7 10 km s@xmath23 ) @xmath41 @xmath2v@xmath48 @xmath41 ( @xmath2v@xmath44 @xmath4 10 km s@xmath23 ) + no classification & ( @xmath2v@xmath44 @xmath4 10 km s@xmath23 ) @xmath41 @xmath2v@xmath48 @xmath41 @xmath2v@xmath44 + tangent point & @xmath2v@xmath48 @xmath47 @xmath2v@xmath44 + [email protected] & @xmath422.3 ( 0.5 ) & 2.7 & 13.8 & @xmath4153.5 & 8.2 & 2.7@xmath49 & 14.4 & 6.0 & n + mc 1 & @xmath411.7 ( 0.2 ) & 1.6 & 14.9 & @xmath4153.5 & 8.2 & 1.6@xmath50 & 8.7 & 7.0 & n + mc 2 & 6.10 ( 0.03 ) & & & @xmath4153.5 & 8.2 & 17.6@xmath51 & 95.2 & 9.6 & f + & & & & & & & & & + [email protected] & @xmath417.9 ( 0.1 ) & 2.6 & 14.1 & @xmath4162.1 & 8.3 & 14.1@xmath52 & @xmath4238.0 & 6.0 & f + mc 1 & @xmath414.2 ( 0.1 ) & 2.2 & 14.5 & @xmath4162.1 & 8.3 & 14.5@xmath53 & @xmath4244.9 & 6.4 & fs + & & & & & & & & & + [email protected] & @xmath425.7 ( 0.1 ) & 3.6 & 13.1 & @xmath4167.0 & 8.4 & 3.6@xmath54 & @xmath433.1 & 5.0 & n + mc 1 & @xmath426.6 ( 0.1 ) & 3.7 & 13.1 & @xmath4167.0 & 8.4 & 3.7@xmath54 & @xmath433.9 & 4.9 & ns + mc 2 & @xmath415.4 ( 0.1 ) & 2.6 & 14.2 & @xmath4167.0 & 8.4 & 2.6@xmath55 & @xmath423.6 & 6.0 & n + & & & & & & & & & + [email protected] & 12.5 ( 0.1 ) & 2.2 & 14.5 & 167.1 & 8.4 & 14.6@xmath56 & @xmath488.9 & 6.4 & f + mc 4a & 9.0 ( 0.3 ) & 1.7 & 15.0 & 167.1 & 8.4 & 15.1@xmath57 & @xmath491.9 & 6.9 & fs + mc 4b & 1.1 ( 0.2 ) & 0.3 & 16.5 & 167.1 & 8.4 & 0.3@xmath58 & @xmath41.6 & 8.3 & n + mc 4c & 16.2 ( 0.4 ) & 2.6 & 14.1 & 167.1 & 8.4 & 14.1@xmath59 & @xmath486.1 & 5.9 & fs + & & & & & & & & & + [email protected] & 9.4 ( 0.1 ) & 1.7 & 15.0 & 166.6 & 8.4 & 15.0@xmath60 & @xmath439.2 & 6.8 & f + mc 1a & 11.11(0.03 ) & 2.0 & 14.7 & 166.6 & 8.4 & 14.8@xmath61 & @xmath438.6 & 6.6 & fs + mc 1b & 3.9 ( 0.1 ) & 0.8 & 15.9 & 166.6 & 8.4 & 15.9@xmath62 & @xmath441.6 & 7.7 & fs + & & & & & & & & & + [email protected] & 70.2 ( 0.3 ) & 5.8 & 10.9 & 166.2 & 8.4 & 10.9@xmath63 & 5.7 & 3.0 & f + mc 1 & 68.9 ( 0.1 ) & 5.8 & 10.9 & 166.2 & 8.4 & 10.9@xmath63 & 5.7 & 3.0 & fs + mc 2 & 25.8 ( 0.1 ) & 3.5 & 13.2 & 166.2 & 8.4 & 3.5@xmath64 & 1.9 & 5.1 & n + mc 4a & 9.4 ( 0.2 ) & 1.7 & 15.0 & 166.2 & 8.4 & 1.7@xmath65 & 0.9 & 6.8 & n + mc 4b & 7.4 ( 0.2 ) & 1.4 & 15.3 & 166.2 & 8.4 & 1.4@xmath66 & 0.7 & 7.1 & n + mc 5a & 34.6 ( 0.1 ) & 4.2 & 12.5 & 166.2 & 8.4 & 4.2@xmath67 & 2.2 & 4.4 & n + mc 5b & 30.1 ( 0.1 ) & 3.9 & 12.8 & 166.2 & 8.4 & 3.9@xmath68 & 2.0 & 4.7 & n + mc 5c & 32.0 ( 0.1 ) & 4.0 & 12.7 & 166.2 & 8.4 & 4.0@xmath67 & 2.1 & 4.6 & n + mc 5d & 33.03 ( 0.04 ) & 4.1 & 12.6 & 166.2 & 8.4 & 4.1@xmath67 & 2.1 & 4.5 & n + & & & & & & & & & + [email protected] & 17.7 ( 0.3 ) & 2.7 & 14.0 & 164.6 & 8.3 & 14.0@xmath69 & 4.9 & 5.9 & f + mc 2 & 20.8 ( 0.1 ) & 3.0 & 13.7 & 164.6 & 8.3 & 13.7@xmath70 & 4.8 & 5.6 & fs + mc 4 & 15.2 ( 0.1 ) & 2.4 & 14.3 & 164.6 & 8.3 & 14.3@xmath59 & 5.0 & 6.2 & fs + mc 5 & 10.1 ( 0.1 ) & 1.7 & 14.9 & 164.6 & 8.3 & 14.9@xmath61 & 5.2 & 6.8 & fs + & & & & & & & & & + [email protected] & 2.1 ( 0.3 ) & 0.4 & 16.3 & 164.2 & 8.3 & 16.3@xmath71 & @xmath4113.5 & 8.1 & f + mc 1 & 0.48 ( 0.03 ) & 0.1 & 16.6 & 164.2 & 8.3 & 16.6@xmath72 & @xmath4115.7 & 8.4 & fs + & & & & & & & & & + [email protected] & 40.3 ( 0.4 ) & 4.3 & 12.3 & 161.6 & 8.3 & 4.3@xmath73 & @xmath445.7 & 4.4 & n + mc 1 & 37.80 ( 0.01 ) & 4.1 & 12.5 & 161.6 & 8.3 & 4.1@xmath74 & @xmath444.1 & 4.5 & ns + & & & & & & & & & + [email protected] & 34.1 ( 1.7 ) & 3.8 & 12.8 & 160.2 & 8.3 & 3.8@xmath67 & 34.1 & 4.8 & n + mc 1 & 18.10 ( 0.02 ) & 2.5 & 14.1 & 160.2 & 8.3 & 2.5@xmath52 & 22.4 & 6.1 & n + & & & & & & & & & + [email protected] & 33.5 ( 0.5 ) & 3.8 & 12.8 & 159.5 & 8.3 & 12.8@xmath75 & @xmath44.5 & 4.9 & f + mc 1 & 21.55 ( 0.02 ) & 2.8 & 13.8 & 159.5 & 8.3 & 2.8@xmath76 & @xmath41.0 & 5.8 & n + mc 4 & 34.5 ( 0.1 ) & 3.8 & 12.8 & 159.5 & 8.3 & 12.8@xmath75 & @xmath44.5 & 4.8 & fs + & & & & & & & & & + [email protected] & 15.8 ( 1.3 ) & 2.2 & 14.3 & 159.0 & 8.3 & 14.4@xmath69 & 82.6 & 6.3 & f + mc 1 & 18.10 ( 0.02 ) & 2.5 & 14.1 & 159.0 & 8.3 & 14.1@xmath77 & 81.3 & 6.1 & fs + & & & & & & & & & + [email protected] & 31.7 ( 0.5 ) & 3.6 & 13.0 & 158.6 & 8.3 & 13.0@xmath75 & @xmath458.9 & 5.1 & f + mc 1 & 35.07 ( 0.01 ) & 3.8 & 12.8 & 158.6 & 8.3 & 12.8@xmath78 & @xmath457.9 & 4.9 & fs + mc 3 & 11.8 ( 0.1 ) & 1.8 & 14.8 & 158.6 & 8.3 & 1.8@xmath53 & @xmath48.0 & 6.8 & n + & & & & & & & & & + [email protected] & 40.6 ( 3.8 ) & 4.1 & 12.4 & 157.9 & 8.3 & 12.4@xmath79 & @xmath432.5 & 4.6 & f + mc 1 & 45.0 ( 0.1 ) & 4.4 & 12.2 & 157.9 & 8.3 & 12.2@xmath80 & @xmath431.9 & 4.4 & fs + mc 2 & 14.6 ( 0.1 ) & 2.1 & 14.5 & 157.9 & 8.3 & 2.1@xmath81 & @xmath45.4 & 6.5 & n + mc 3 & 23.0 ( 0.1 ) & 2.9 & 13.7 & 157.9 & 8.3 & 2.9@xmath82 & @xmath47.5 & 5.7 & n + mc 4a & 34.9 ( 0.2 ) & 3.8 & 12.8 & 157.9 & 8.3 & 12.8@xmath78 & @xmath433.5 & 4.9 & fs + mc 4b & 37.5 ( 0.2 ) & 3.9 & 12.6 & 157.9 & 8.3 & 12.2@xmath78 & @xmath433.0 & 4.8 & fs + & & & & & & & & & + [email protected] & 51.0 ( 0.1 ) & 4.6 & 11.9 & 155.5 & 8.3 & 4.6@xmath83 & 22.3 & 4.2 & n + mc 1 & 48.60 ( 0.01 ) & 4.5 & 12.1 & 155.5 & 8.3 & 4.5@xmath84 & 21.8 & 4.3 & ns + & & & & & & & & & + [email protected] & 27.6 ( 0.9 ) & 3.1 & 13.4 & 154.1 & 8.2 & 3.1@xmath68 & @xmath434.7 & 5.6 & n + mc 1 & 21.51 ( 0.01 ) & 2.6 & 13.9 & 154.1 & 8.2 & 2.6@xmath82 & @xmath429.1 & 6.0 & ns + & & & & & & & & & + [email protected] & 27.5 ( 0.5 ) & 3.0 & 13.4 & 153.0 & 8.2 & 13.4@xmath85 & @xmath4117.0 & 5.6 & f + mc 1 & 27.7 ( 0.1 ) & 3.1 & 13.4 & 153.0 & 8.2 & 13.4@xmath85 & @xmath4116.7 & 5.6 & fs + mc 2 & 19.3 ( 0.1 ) & 2.4 & 14.1 & 153.0 & 8.2 & 14.1@xmath86 & @xmath4122.8 & 6.2 & fs + & & & & & & & & & + [email protected] & 34.7 ( 1.2 ) & 3.3 & 12.9 & 146.9 & 8.1 & 13.0@xmath78 & @xmath411.3 & 5.4 & f + & & & & & & & & & + [email protected] & 25.2 ( 1.2 ) & 2.6 & 13.6 & 143.5 & 8.1 & 13.6@xmath85 & 35.6 & 6.1 & f + mc 1 & 22.36 ( 0.04 ) & 2.4 & 13.8 & 143.5 & 8.1 & 13.8@xmath85 & 36.2 & 6.3 & fs + & & & & & & & & & + [email protected] & 56.5 ( 0.4 ) & 4.4 & 11.8 & 140.9 & 8.1 & 11.8@xmath87 & @xmath40.8 & 4.6 & f + mc 1a & 53.8 ( 0.1 ) & 4.2 & 11.9 & 140.9 & 8.1 & 11.9@xmath83 & @xmath40.8 & 4.7 & fs + mc 1b & 50.0 ( 0.4 ) & 4.0 & 12.1 & 140.9 & 8.1 & 12.1@xmath83 & @xmath40.8 & 4.8 & fs + mc 3 & 132.5 ( 0.1 ) & 7.3 & 8.9 & 140.9 & 8.1 & & & 2.8 & + & & & & & & & & & + [email protected] & 44.1 ( 0.9 ) & 3.7 & 12.4 & 140.2 & 8.1 & 12.4@xmath80 & @xmath413.0 & 5.1 & f + mc 1a & 45.51 ( 0.02 ) & 3.8 & 12.3 & 140.2 & 8.1 & 12.3@xmath80 & @xmath413.0 & 5.1 & fs + mc 1b & 50.2 ( 0.2 ) & 4.0 & 12.1 & 140.2 & 8.1 & 12.1@xmath83 & @xmath412.6 & 4.9 & fs + mc 4 & 35.7 ( 0.2 ) & 3.2 & 12.9 & 140.2 & 8.1 & 12.9@xmath78 & @xmath413.5 & 5.6 & fs + & & & & & & & & & + [email protected] & 64.2 ( 0.1 ) & 4.7 & 11.4 & 138.9 & 8.0 & 4.7@xmath88 & @xmath422.7 & 4.4 & n + mc 1a & 65.47 ( 0.02 ) & 4.7 & 11.4 & 138.9 & 8.0 & 4.7@xmath88 & @xmath423.0 & 4.3 & ns + mc 1b & 58.8 ( 0.1 ) & 4.4 & 11.7 & 138.9 & 8.0 & 4.4@xmath88 & @xmath421.6 & 4.6 & ns + mc 2 & 6.4 ( 0.1 ) & 0.8 & 15.3 & 138.9 & 8.0 & 0.8@xmath81 & @xmath43.8 & 7.8 & n + mc 3 & 36.2 ( 0.2 ) & 3.2 & 12.9 & 138.9 & 8.0 & 3.2@xmath74 & @xmath415.7 & 5.6 & n + & & & & & & & & & + [email protected] & 63.5 ( 0.2 ) & 4.6 & 11.4 & 138.7 & 8.0 & 4.6@xmath88 & @xmath427.4 & 4.4 & n + mc 1 & 63.4 ( 0.1 ) & 4.6 & 11.4 & 138.7 & 8.0 & 4.6@xmath88 & @xmath427.4 & 4.4 & ns + & & & & & & & & & + [email protected] & 51.1 ( 0.8 ) & 4.0 & 12.0 & 138.0 & 8.0 & 12.0@xmath83 & @xmath44.2 & 4.9 & f + mc 1a & 26.89 ( 0.01 ) & 2.6 & 13.5 & 138.0 & 8.0 & 2.6@xmath89 & @xmath40.9 & 6.1 & n + mc 1b & 23.1 ( 0.1 ) & 2.3 & 13.8 & 138.0 & 8.0 & 2.3@xmath68 & @xmath40.8 & 6.4 & n + mc 3 & 39.6 ( 0.2 ) & 3.4 & 12.6 & 138.0 & 8.0 & 3.4@xmath90 & @xmath41.2 & 5.4 & n + & & & & & & & & & + [email protected] & 38.7 ( 1.2 ) & 3.3 & 12.7 & 137.2 & 8.0 & 12.7@xmath90 & @xmath4199.4 & 5.5 & f + mc 1 & 34.9 ( 0.1 ) & 3.1 & 12.9 & 137.2 & 8.0 & 12.9@xmath78 & @xmath4202.9 & 5.7 & fs + & & & & & & & & & + [email protected] & 50.9 ( 2.3 ) & 4.0 & 12.0 & 136.3 & 8.0 & 12.0@xmath83 & @xmath4111.2 & 5.0 & f + mc 1 & 43.85 ( 0.02 ) & 3.6 & 12.4 & 136.3 & 8.0 & 12.4@xmath80 & @xmath4114.6 & 5.3 & fs + mc 2 & 24.29 ( 0.04 ) & 2.3 & 13.6 & 136.3 & 8.0 & 2.4@xmath89 & @xmath421.7 & 6.3 & n + & & & & & & & & & + [email protected] & 70.2 ( 0.6 ) & 4.8 & 11.1 & 131.3 & 7.9 & 4.8@xmath88 & @xmath446.5 & 4.4 & n + mc 1 & 54.12 ( 0.04 ) & 4.0 & 11.8 & 131.3 & 7.9 & 4.0@xmath83 & @xmath439.3 & 5.0 & n + mc 2a & 5.6 ( 0.1 ) & 0.6 & 15.2 & 131.3 & 7.9 & 0.6@xmath52 & @xmath46.1 & 7.9 & n + mc 2b & 8.1 ( 0.2 ) & 0.9 & 15.0 & 131.3 & 7.9 & 0.9@xmath91 & @xmath48.5 & 7.7 & n + mc 3 & 74.2 ( 0.2 ) & 4.9 & 10.9 & 131.3 & 7.9 & 4.9@xmath92 & @xmath448.2 & 4.3 & ns + & & & & & & & & & + [email protected] & 26.9 ( 0.4 ) & 2.4 & 13.4 & 129.9 & 7.9 & 13.4@xmath93 & 2.3 & 6.3 & f + mc 1 & 23.0 ( 0.1 ) & 2.1 & 13.6 & 129.9 & 7.9 & 13.4@xmath93 & 2.4 & 6.6 & fs + mc 4 & 6.7 ( 0.3 ) & 0.7 & 15.1 & 129.9 & 7.9 & 0.7@xmath91 & 0.1 & 7.8 & n + & & & & & & & & & + [email protected] & 82.5 ( 0.3 ) & 5.2 & 10.5 & 128.3 & 7.9 & 5.2@xmath92 & 5.5 & 4.2 & n + mc 1 & 50.2 ( 0.1 ) & 3.8 & 12.0 & 128.3 & 7.9 & 3.8@xmath84 & 3.9 & 5.2 & n + mc 2 & 14.0 ( 0.1 ) & 1.4 & 14.3 & 128.3 & 7.9 & 1.4@xmath82 & 1.5 & 7.2 & n + mc 3a & 84.5 ( 0.1 ) & 5.3 & 10.4 & 128.3 & 7.9 & 5.3@xmath92 & 5.6 & 4.1 & ns + mc 3b & 80.9 ( 0.1 ) & 5.2 & 10.6 & 128.3 & 7.9 & 5.2@xmath92 & 5.4 & 4.2 & ns + mc 3c & 89.3 ( 0.4 ) & 5.5 & 10.2 & 128.3 & 7.9 & 5.5@xmath92 & 5.8 & 4.0 & ns + & & & & & & & & & + [email protected] & 76.3 ( 0.3 ) & 4.9 & 10.7 & 125.5 & 7.8 & 10.7@xmath92 & @xmath444.7 & 4.4 & f + mc 1 & 62.9 ( 0.1 ) & 4.3 & 11.3 & 125.5 & 7.8 & 4.3@xmath83 & @xmath418.2 & 4.8 & n + mc 2 & 80.2 ( 0.1 ) & 5.1 & 10.5 & 125.5 & 7.8 & 10.5@xmath92 & @xmath444.0 & 4.3 & fs + mc 4 & 4.4 ( 0.1 ) & 0.5 & 15.2 & 125.5 & 7.8 & 0.5@xmath91 & @xmath42.0 & 8.1 & n + & & & & & & & & & + [email protected] & 78.2 ( 1.7 ) & 5.0 & 10.6 & 125.4 & 7.8 & 10.6@xmath92 & 14.8 & 4.4 & f + mc 1 & 77.52 ( 0.04 ) & 5.0 & 10.6 & 125.4 & 7.8 & 10.6@xmath92 & 14.8 & 4.4 & fs + mc 2 & 7.7 ( 0.1 ) & 0.8 & 14.8 & 125.4 & 7.8 & 0.8@xmath91 & 1.1 & 7.8 & n + mc 3 & 19.8 ( 0.1 ) & 1.8 & 13.8 & 125.4 & 7.8 & 1.8@xmath89 & 2.5 & 6.9 & n + mc 5 & 27.7 ( 0.3 ) & 2.4 & 13.2 & 125.4 & 7.8 & 2.4@xmath74 & 3.3 & 6.4 & n + & & & & & & & & & + [email protected] & 101.1 ( 0.1 ) & 6.0 & 9.6 & 124.9 & 7.8 & 6.0@xmath87 & @xmath422.0 & 3.8 & n + mc 1a & 100.56 ( 0.02 ) & 6.0 & 9.6 & 124.9 & 7.8 & 6.0@xmath87 & @xmath421.9 & 3.8 & ns + mc 1b & 96.79 ( 0.02 ) & 5.8 & 9.8 & 124.9 & 7.8 & 5.8@xmath87 & @xmath421.3 & 3.9 & ns + mc 2 & 80.46 ( 0.03 ) & 5.1 & 10.5 & 124.9 & 7.8 & 5.1@xmath92 & @xmath418.8 & 4.3 & n + mc 3 & 64.3 ( 0.2 ) & 4.4 & 11.2 & 124.9 & 7.8 & 4.4@xmath83 & @xmath416.1 & 4.8 & n + mc 4a & 3.2 ( 0.1 ) & 0.3 & 15.3 & 124.9 & 7.8 & 0.3@xmath52 & @xmath41.2 & 8.2 & n + mc 4b & 6.9 ( 0.1 ) & 0.7 & 14.9 & 124.9 & 7.8 & 0.7@xmath91 & @xmath42.6 & 7.9 & n + & & & & & & & & & + [email protected] & 76.5 ( 0.4 ) & 4.9 & 10.6 & 124.0 & 7.8 & 10.6@xmath87 & @xmath437.1 & 4.5 & f + mc 1a & 86.0 ( 0.1 ) & 5.4 & 10.2 & 124.0 & 7.8 & 10.2@xmath92 & @xmath435.7 & 4.2 & fs + mc 1b & 80.8 ( 0.1 ) & 5.1 & 10.4 & 124.0 & 7.8 & 10.4@xmath92 & @xmath436.4 & 4.3 & fs + mc 2 & 68.9 ( 0.1 ) & 4.6 & 11.0 & 124.0 & 7.8 & 11.0@xmath87 & @xmath438.3 & 4.7 & fs + mc 3a & 6.2 ( 0.2 ) & 0.6 & 14.9 & 124.0 & 7.8 & 0.6@xmath91 & @xmath42.2 & 7.9 & n + mc 3b & 4.1 ( 0.2 ) & 0.4 & 15.1 & 124.0 & 7.8 & 0.4@xmath91 & @xmath41.5 & 8.1 & n + & & & & & & & & & + [email protected] & 108.1 ( 0.5 ) & 6.4 & 9.1 & 121.5 & 7.7 & 6.4@xmath79 & @xmath46.7 & 3.8 & n + mc 1a & 48.97 ( 0.04 ) & 3.6 & 11.9 & 121.5 & 7.7 & 3.6@xmath84 & @xmath43.8 & 5.4 & n + mc 1b & 44.1 ( 0.1 ) & 3.3 & 12.1 & 121.5 & 7.7 & 3.3@xmath84 & @xmath43.5 & 5.6 & n + mc 1c & 53.1 ( 0.1 ) & 3.8 & 11.7 & 121.5 & 7.7 & 3.8@xmath83 & @xmath44.0 & 5.3 & n + mc 2 & 110.2 ( 0.1 ) & 6.5 & 9.0 & 121.5 & 7.7 & 6.5@xmath94 & @xmath46.8 & 3.7 & ns + mc 3 & 5.6 ( 0.1 ) & 0.6 & 14.9 & 121.5 & 7.7 & 0.6@xmath91 & @xmath40.6 & 8.0 & n + mc 4 & 83.4 ( 0.2 ) & 5.2 & 10.2 & 121.5 & 7.7 & 5.2@xmath83 & @xmath45.5 & 4.3 & n + mc 5 & 95.6 ( 0.3 ) & 5.8 & 9.7 & 121.5 & 7.7 & 5.8@xmath83 & @xmath46.0 & 4.0 & n + & & & & & & & & & + [email protected] & 119.5 ( 1.4 ) & & & 117.4 & 7.7 & 7.7 @xmath22 1.2 & @xmath418.7 & 3.7 & + mc 1 & 94.1 ( 0.1 ) & 5.7 & 9.6 & 117.4 & 7.7 & 5.7@xmath83 & @xmath414.0 & 4.2 & n + mc 2 & 112.9 ( 0.2 ) & 6.9 & 8.4 & 117.4 & 7.7 & 7.7 @xmath22 1.20 & @xmath418.7 & 3.8 & s + mc 3a & 107.1 ( 0.4 ) & 6.4 & 8.9 & 117.4 & 7.7 & 6.4@xmath95 & @xmath415.7 & 3.9 & n + mc 3b & 108.8 ( 0.2 ) & 6.6 & 8.8 & 117.4 & 7.7 & & & 3.9 & + mc 4 & 149.02 ( 0.02 ) & & & 117.4 & 7.7 & 7.7 @xmath22 1.2 & @xmath418.7 & 3.2 & + & & & & & & & & & + [email protected] & 112.1 ( 0.1 ) & 6.8 & 8.5 & 117.3 & 7.7 & & & 3.8 & + mc 1 & 109.67 ( 0.04 ) & 6.6 & 8.7 & 117.3 & 7.7 & & & 3.8 & s + mc 2 & 5.0 ( 0.1 ) & 0.5 & 14.8 & 117.3 & 7.7 & 0.5@xmath96 & 2.0 & 8.1 & n + mc 3 & 18.3 ( 0.1 ) & 1.6 & 13.7 & 117.3 & 7.7 & 1.6@xmath89 & 6.7 & 7.1 & n + & & & & & & & & & + [email protected] & 107.0 ( 1.1 ) & 6.5 & 8.7 & 115.1 & 7.6 & & & 4.0 & + mc 1 & 100.29 ( 0.04 ) & 6.1 & 9.1 & 115.1 & 7.6 & & & 4.1 & s + mc 2 & 110.6 ( 0.1 ) & 6.8 & 8.4 & 115.1 & 7.6 & & & 3.9 & s + & & & & & & & & & + [email protected] & 76.6 ( 0.9 ) & 4.9 & 10.1 & 110.5 & 7.5 & 10.1@xmath83 & 14.1 & 4.8 & f + mc 1 & 75.11 ( 0.02 ) & 4.8 & 10.2 & 110.5 & 7.5 & 10.2@xmath83 & 14.2 & 4.8 & fs + mc 2 & 58.1 ( 0.1 ) & 4.0 & 11.1 & 110.5 & 7.5 & 4.0@xmath83 & 5.5 & 5.3 & n + mc 4a & 12.2 ( 0.1 ) & 1.1 & 13.9 & 110.5 & 7.5 & 1.1@xmath97 & 1.5 & 7.6 & n + mc 4b & 10.1 ( 0.1 ) & 0.9 & 14.1 & 110.5 & 7.5 & 0.9@xmath97 & 1.3 & 7.7 & n + mc 5 & 23.0 ( 0.5 ) & 1.9 & 13.1 & 110.5 & 7.5 & 1.9@xmath98 & 2.6 & 6.9 & n + & & & & & & & & & + [email protected] & 89.0 ( 0.2 ) & 5.5 & 9.4 & 108.6 & 7.5 & 5.6@xmath80 & @xmath434.9 & 4.5 & n + mc 1a & 89.43 ( 0.03 ) & 5.6 & 9.4 & 108.6 & 7.5 & 5.6@xmath80 & @xmath435.0 & 4.5 & ns + mc 1b & 84.3 ( 0.1 ) & 5.3 & 9.6 & 108.6 & 7.5 & 5.3@xmath80 & @xmath433.2 & 4.6 & ns + mc 2 & 73.7 ( 0.1 ) & 4.7 & 10.2 & 108.6 & 7.5 & 4.7@xmath83 & @xmath429.8 & 4.9 & n + & & & & & & & & & + [email protected] & 101.7 ( 0.4 ) & 6.4 & 8.5 & 107.8 & 7.4 & & & 4.2 & + mc 1 & 100.28 ( 0.02 ) & 6.3 & 8.6 & 107.8 & 7.4 & & & 4.3 & s + mc 2 & 80.52 ( 0.02 ) & 5.1 & 9.8 & 107.8 & 7.4 & 5.1@xmath80 & 5.3 & 4.7 & n + mc 3 & 52.4 ( 0.1 ) & 3.6 & 11.3 & 107.8 & 7.4 & 3.6@xmath84 & 3.8 & 5.6 & n + mc 4 & 94.2 ( 0.1 ) & 5.9 & 9.0 & 107.8 & 7.4 & & & 4.4 & s + mc 5 & 72.0 ( 0.1 ) & 4.7 & 10.2 & 107.8 & 7.4 & 4.7@xmath83 & 4.9 & 5.0 & n + & & & & & & & & & + [email protected] & 91.3 ( 0.1 ) & 5.8 & 9.0 & 104.4 & 7.4 & 8.9@xmath99 & @xmath43.1 & 4.5 & f + mc 1a & 98.2 ( 0.1 ) & 6.3 & 8.4 & 104.4 & 7.4 & 8.4@xmath68 & @xmath42.9 & 4.4 & fs + mc 1b & 102.5 ( 0.1 ) & 6.8 & 7.9 & 104.4 & 7.4 & & & 4.3 & + mc 2 & 6.99 ( 0.02 ) & 0.6 & 14.1 & 104.4 & 7.4 & 0.6@xmath97 & @xmath40.2 & 8.0 & n + mc 3 & 67.1 ( 0.1 ) & 4.4 & 10.3 & 104.4 & 7.4 & 4.4@xmath83 & @xmath41.5 & 5.2 & n + mc 4 & 49.9 ( 0.1 ) & 3.5 & 11.3 & 104.4 & 7.4 & 3.5@xmath84 & @xmath41.2 & 5.8 & n + & & & & & & & & & + [email protected] & 99.8 ( 0.7 ) & 6.6 & 8.1 & 103.1 & 7.3 & & & 4.37 & + mc 1 & 112.22 ( 0.04 ) & & & 103.1 & 7.3 & 7.3 @xmath22 1.4 & @xmath414.1 & 4.1 & + mc 2 & 87.14 ( 0.04 ) & 5.5 & 9.1 & 103.1 & 7.3 & 5.5@xmath78 & @xmath410.6 & 4.7 & n + mc 3 & 99.1 ( 0.2 ) & 6.5 & 8.2 & 103.1 & 7.3 & & & 4.38 & s + mc 4a & 51.1 ( 0.2 ) & 3.5 & 11.1 & 103.1 & 7.3 & 3.5@xmath90 & @xmath46.8 & 5.7 & n + mc 4b & 48.2 ( 0.2 ) & 3.4 & 11.3 & 103.1 & 7.3 & 3.4@xmath90 & @xmath46.5 & 5.8 & n + mc 5a & 42.0 ( 0.3 ) & 3.0 & 11.7 & 103.1 & 7.3 & 3.0@xmath90 & @xmath45.8 & 6.1 & n + mc 5b & 38.8 ( 0.1 ) & 2.8 & 11.8 & 103.1 & 7.3 & 2.8@xmath90 & @xmath45.4 & 6.2 & n + & & & & & & & & & + [email protected] & 22.5 ( 0.7 ) & 1.8 & 12.9 & 102.9 & 7.3 & 12.9@xmath98 & 103.4 & 7.0 & f + mc 1 & 15.9 ( 0.1 ) & 1.3 & 13.3 & 102.9 & 7.3 & 13.4@xmath93 & 107.2 & 7.4 & fs + mc 3b & 94.8 ( 0.2 ) & 6.1 & 8.6 & 102.9 & 7.3 & & & 4.5 & + & & & & & & & & & + [email protected] & 100.9 ( 0.3 ) & & & 99.9 & 7.3 & 7.3 @xmath22 1.4 & 39.2 & 4.4 & + mc 1 & 97.83 ( 0.04 ) & 6.6 & 7.9 & 99.9 & 7.3 & 7.3 @xmath22 1.4 & 39.2 & 4.5 & s + mc 2 & 12.7 ( 0.2 ) & 1.1 & 13.5 & 99.9 & 7.3 & 1.1@xmath89 & 5.7 & 7.6 & n + & & & & & & & & & + [email protected] & 65.8 ( 0.1 ) & 4.6 & 8.8 & 81.2 & 6.7 & 8.8@xmath68 & @xmath430.9 & 5.6 & f + mc 1 & 63.53 ( 0.02 ) & 4.4 & 9.0 & 81.2 & 6.7 & 9.0@xmath89 & @xmath431.5 & 5.7 & fs + mc 2 & 19.9 ( 0.1 ) & 1.5 & 11.9 & 81.2 & 6.7 & 1.5@xmath98 & @xmath45.3 & 7.4 & n + mc 3 & 13.6 ( 0.1 ) & 1.1 & 12.4 & 81.2 & 6.7 & 1.1@xmath89 & @xmath43.7 & 7.7 & n + mc 4 & 83.3 ( 0.2 ) & & & 81.2 & 6.7 & 6.7 @xmath22 1.7 & @xmath423.5 & 5.2 & + mc 5 & 90.9 ( 0.2 ) & & & 81.2 & 6.7 & 6.7 @xmath22 1.7 & @xmath423.5 & 5.0 & + mc 6 & 38.1 ( 0.1 ) & 2.7 & 10.7 & 81.2 & 6.7 & 2.7@xmath98 & @xmath49.5 & 6.6 & n + & & & & & & & & & + [email protected] & 59.2 ( 0.2 ) & 4.1 & 9.3 & 80.9 & 6.7 & 9.3@xmath89 & @xmath464.9 & 5.8 & f + mc 1 & 64.8 ( 0.4 ) & 4.5 & 8.9 & 80.9 & 6.7 & 8.9@xmath68 & @xmath462.0 & 5.7 & fs + mc 2a & 74.3 ( 0.1 ) & 5.4 & 8.1 & 80.9 & 6.7 & & & 5.4 & + mc 2b & 72.4 ( 0.1 ) & 5.2 & 8.3 & 80.9 & 6.7 & & & 5.4 & + mc 3 & 14.0 ( 0.1 ) & 1.1 & 12.3 & 80.9 & 6.7 & 1.1@xmath89 & @xmath47.6 & 7.7 & n + mc 4 & 80.1 ( 0.2 ) & 6.3 & 7.2 & 80.9 & 6.7 & & & 5.2 & + mc 5 & 20.5 ( 0.1 ) & 1.6 & 11.9 & 80.9 & 6.7 & 1.6@xmath98 & @xmath410.8 & 7.3 & n + mc 6 & 91.9 ( 0.3 ) & & & 80.9 & 6.7 & 6.7 @xmath22 1.7 & @xmath446.8 & 5.0 & + & & & & & & & & & + [email protected] & 67.9 ( 0.1 ) & & & 51.1 & 5.6 & 5.6 @xmath22 2.1 & @xmath433.9 & 6.0 & + mc 1a & 65.76 ( 0.02 ) & & & 51.1 & 5.6 & 5.6 @xmath22 2.1 & @xmath433.9 & 6.0 & s + mc 1b & 61.8 ( 0.1 ) & & & 51.1 & 5.6 & 5.6 @xmath22 2.1 & @xmath433.9 & 6.1 & s + mc 2 & 6.1 ( 0.1 ) & 0.5 & 10.6 & 51.1 & 5.6 & 0.5@xmath100 & @xmath43.0 & 8.2 & n + & & & & & & & & & + [email protected] & 26.5 ( 0.1 ) & & & 25.6 & 4.1 & 4.1 @xmath22 2.4 & 6.4 & 7.4 & + mc 1 & 21.55 ( 0.02 ) & 2.6 & 5.5 & 25.6 & 4.1 & 4.1 @xmath22 2.4 & 6.4 & 7.6 & s + mc 2 & 5.6 ( 0.3 ) & 0.5 & 7.6 & 25.6 & 4.1 & 0.6@xmath101 & 0.9 & 8.3 & n +	we report simultaneous h110@xmath0 and h@xmath1co line observations with the nrao green bank telescope toward 72 h ii regions in the spitzer / glimpse survey area ( @xmath2l@xmath2 = 10@xmath3 @xmath4 65@xmath3 and @xmath2b@xmath2 @xmath5 1@xmath3 ) . we used the h110@xmath0 line to establish the velocity of the h ii regions and h@xmath1co absorption lines to distinguish between near and far distances . accurate distances are crucial for the determination of physical properties of massive star formation regions . we resolved the distance ambiguity of 44 h ii regions . we detected multiple h ii regions along 18 lines of sight located in the longitude interval 12@xmath3 to 31@xmath3 , primarily a result of the relatively large telescope beam width . we could not resolve distance ambiguities for lines of sight with multiple h ii regions , since we could not determine which h@xmath1co lines were being absorbed against which h ii region . we examined the projected location of h ii regions whose distance ambiguities have been resolved ( in this work and other similar studies ) in the galactic plane and in a longitude - velocity diagram for a recognizable spiral arm pattern . although the highest density of points in the position - position plot approximately follows the spiral arms proposed by taylor and cordes ( 1993 ) , the dispersion is still about as large as the separation between their proposed arms . the longitude - velocity plot shows an increase in the density of sources at the points where the spiral arm loci proposed by taylor and cordes ( 1993 ) are approaching the locus of tangent point velocities and a lower density between the arm loci . however , it is not possible to trace spiral arms over significant segments of galactic longitude in the longitude - velocity plot . we conclude that a very large number of h ii regions in combination with more sophisticated galactic rotation models will be required to obtain a more continuous spiral pattern from kinematic studies of h ii regions than from fully sampled surveys of h i or co.
we want to thank g.rybnikov for his contribution and his support . his work has been a challenge for us for years . in this section , some standard facts about line combinatorics and ordered line combinatorics will be described . special attention will be given to maclane and rybnikov s line combinatorics . a _ combinatorial type _ ( or simply a _ ( line ) combinatorics _ ) is a couple @xmath24 , where @xmath25 is a finite set and @xmath26 , satisfying that : 1 . for all @xmath27 , @xmath28 ; 2 . for any @xmath29 , @xmath30 , @xmath31 such that @xmath32 . an _ ordered combinatorial type _ @xmath33 is a combinatorial type where @xmath25 is an ordered set . [ not - comb - type ] given a combinatorial type @xmath34 , the multiplicity @xmath35 of @xmath27 is the number of elements @xmath36 such that @xmath37 ; note that @xmath38 . the _ multiplicity of _ a combinatorial type is the number @xmath39 . ] ( 0,0 ) ( -90,80)@xmath40 ( -90,43)@xmath41 ( -55,80)@xmath42 ( -55,43)@xmath43 ( -55,5)@xmath44 ( 0,80)@xmath45 ( 0,43)@xmath46 ( 0,5)@xmath47 [ exam - maclane ] let us consider the 2-dimensional vector space on the field @xmath48 of three elements . such a plane contains 9 points and @xmath49 lines , @xmath40 of which pass through the origin . consider @xmath50 and @xmath51 , the set of lines in @xmath52 ( as a subset of @xmath53 ) . this provides a combinatorial type structure @xmath54 that we will refer to as _ maclane s combinatorial type_. figure [ mclane ] represents an ordered maclane s combinatorial type . let @xmath24 be a combinatorial type . we say a complex line arrangement @xmath55 is a _ realization _ of @xmath34 if and only if there are bijections @xmath56 and @xmath57 such that @xmath58 , one has @xmath59 . if @xmath33 is an ordered combinatorial type and the irreducible components of @xmath3 are also ordered , we say @xmath3 is an ordered realization if @xmath60 respects orders . the space of all complex realizations of a line combinatorics @xmath34 is denoted by @xmath61 . this is a quasiprojective subvariety of @xmath62 , where @xmath63 . if @xmath33 is ordered , we denote by @xmath64 the space of all ordered complex realizations of @xmath33 . there is a natural action of @xmath65 on such spaces . this justifies the following definition . the _ moduli space _ of a combinatorics @xmath34 is the quotient @xmath66 . the _ ordered moduli space _ @xmath67 of an ordered combinatorics @xmath33 is defined accordingly . [ mclane - rlz]let us consider the maclane line combinatorics @xmath54 . it is well known that such combinatorics has no real realization and that @xmath68 , however @xmath69 . the following are representatives for @xmath70 : @xmath71 where @xmath72 . we will refer to such ordered realizations as @xmath73 and @xmath74 given a line combinatorics @xmath75 , the automorphism group @xmath76 is the subgroup of the permutation group of @xmath25 preserving @xmath51 . let us consider an ordered line combinatorics @xmath33 . it is easily seen that @xmath77 acts on both @xmath78 and @xmath79 . note also that @xmath80 . the action of @xmath81 on the moduli spaces is as follows : matrices of determinant @xmath82 ( resp . @xmath83 ) fix ( resp . exchange ) the two elements of @xmath70 . of course complex conjugation also acts on @xmath70 exchanging the two elements . from the topological point of view one has that : * there exists a homeomorphism @xmath84 preserving orientations on both @xmath6 and the lines . such a homeomorphism does not respect the ordering . * there exists a homeomorphism @xmath84 preserving orientations on @xmath6 , but not on the lines . such a homeomorphism respects the ordering . also note that the subgroup of automorphisms that preserve the set @xmath85 is isomorphic to @xmath86 , since the vectors @xmath87 , @xmath88 and @xmath89 generate @xmath52 . we will denote by @xmath90 and @xmath91 the sets of @xmath41 lines such that @xmath92 and @xmath93 . since any transposition of @xmath94 in @xmath54 produces a determinant @xmath83 matrix in @xmath95 , one concludes from the previous paragraph that any transposition of @xmath94 induces a homeomorphism @xmath84 that exchanges @xmath96 and @xmath97 as representatives of elements of @xmath70 and globally fixes @xmath98 . [ exam - ryb - comb ] let @xmath14 and @xmath99 be ordered maclane realizations as above , where @xmath85 . let us consider a projective transformation @xmath100 ( resp . @xmath101 ) fixing the initial ordered set @xmath98 ( that is , @xmath102 @xmath103 ) and such that @xmath104 ( resp . @xmath105 ) and @xmath14 intersect each other only in double points outside the three common lines . note that @xmath106 can be chosen with real coefficients . let us consider the following ordered arrangements of thirteen lines : @xmath107 , where @xmath108 and @xmath109 ( resp @xmath110 ) if @xmath111 ( resp . @xmath112 ) . they produce the following combinatorics @xmath113 given by : @xmath114 the following combinatorial properties hold : 1 . the different arrangements @xmath115 have the same combinatorial type @xmath116 . the set of lines @xmath98 has the following distinctive combinatorial property : every line in @xmath98 contains exactly 5 triple points of the arrangement ; the remaining lines only contain 3 triple points . 3 . for the other @xmath117 lines we consider the equivalence relation generated by the relation of _ sharing a triple point_. there are two equivalence classes which correspond to @xmath118 and @xmath119 , @xmath120 . by the previous remarks one can group the set @xmath121 together in three subsets . one is associated with the set of lines @xmath98 ( referred to as @xmath122 ) , and the other two are combinatorially indistinguishable sets ( @xmath123 and @xmath124 ) such that @xmath125 and @xmath126 are maclane s combinatorial types . note that any automorphism of @xmath116 must preserve @xmath122 and either preserve or exchange @xmath123 and @xmath124 . therefore , @xmath127 . the following results are immediate consequences of the aforementioned remarks . the following are ( or induce ) homeomorphisms between the pairs @xmath128 and @xmath129 ( resp . @xmath130 and @xmath131 ) preserving the orientation of @xmath6 : 1 . complex conjugation , which reverses orientations of the lines . 2 . a transposition in @xmath122 , which preserves orientations of the lines we will refer to @xmath132 and @xmath133 ( resp . @xmath134 and @xmath135 ) as a type @xmath136 ( resp . type @xmath137 ) arrangements . any homeomorphism of pairs between a type @xmath136 and a type @xmath137 arrangement should lead ( maybe after composing with complex conjugation ) to an orientation - preserving homeomorphism of pairs between a type @xmath136 and a type @xmath137 arrangement . if such a homeomorphism existed , there should be an orientation - preserving homeomorphism of ordered maclane arrangements of type @xmath138 and @xmath99 . the purpose of the next section will be to prove that there is no orientation - preserving homeomorphism of ordered maclane arrangements of type @xmath138 and @xmath99 . even though the alexander invariant can be developed for general projective plane curves , we will concentrate on the case of line arrangements . let @xmath139 be a projective line arrangement where @xmath140 . let us denote its complement @xmath141 and @xmath17 its fundamental group . the derived series associated with this group is recursively defined as follows : @xmath142 , @xmath143 $ ] , @xmath144 , where @xmath145 is the derived subgroup of @xmath17 , i.e. the subgroup generated by @xmath146:=a b a^{-1 } b^{-1}$ ] , @xmath147 . note that the consecutive quotients are abelian . this property also holds for the lower central series defined as @xmath148 , @xmath149 $ ] , @xmath144 . it is clear that @xmath150 and @xmath151 . since @xmath152 , one can consider the inclusion @xmath153 as representing the universal abelian cover @xmath154 of @xmath155 , where @xmath156 , and therefore @xmath157 . the group of transformations @xmath158 of the cover acts on @xmath145 . this results in an action by conjugation on @xmath159 , @xmath160 : @xmath161 ) & \mapsto & g[a , b ] \ \mod g '' & = [ g,[a , b ] ] + [ a , b ] , \endarray\ ] ] where @xmath162 . this action is well defined since @xmath163 implies @xmath164\equiv [ a , b]$ ] mod @xmath165 . additive notation will be used for the operation in @xmath19 . this action endows the abelian group @xmath19 with a @xmath166-module structure , that is , a module on the group ring @xmath167 $ ] . if @xmath168 , then @xmath169 denotes its class in @xmath170 . for @xmath171 , we choose @xmath172 a meridian of @xmath173 in @xmath17 ; the class @xmath174 does not depend on the particular choice of the meridian in @xmath173 . note that @xmath175 is a basis of @xmath176 and therefore one can identify @xmath177=\zz[t_1^{\pm 1}, ... ,t_r^{\pm 1}].\ ] ] this module is denoted by @xmath178 and is referred to as the _ alexander invariant _ of @xmath179 . since we are interested in oriented topological properties of @xmath180 , the coordinates @xmath175 are well defined . the module structure of @xmath178 is in general complicated . one of its invariants is the zero set of the fitting ideals of the complexified alexander invariant of @xmath179 , that is , @xmath181 . this sequence of invariants is called the sequence of characteristic varieties of @xmath179 introduced by a. libgober @xcite . these are subvarieties of the torus @xmath182 ; in fact , irreducible components of characteristic varieties are translated subtori @xcite . our approach in studying the structure of the @xmath183-module @xmath178 is via the associated graded module by the augmentation ideal @xmath184 in order to do so , and to be able to do calculations , we need some formul on this module relating operations in @xmath19 . for the sake of completeness , these formul are listed below . however , since they are straightforward consequences of the definitions , their proof will be omitted . the symbol `` @xmath185 '' means that the equality is considered in @xmath19 : [ prop - alex - mod ] 1 . [ prop - comm ] @xmath186 \eqmap{2 } ( t_x-1)p$ ] @xmath187 , 2 . [ prop - prod2 ] @xmath188 \eqmap{2 } \sum_{i=1}^n \sum_{j=1}^m t_{ij } [ x_i , y_j]$ ] , where @xmath189 . 3 . [ prop - comm2 ] @xmath190 \eqmap{2 } -(t_x-1)(p_1+ ... +p_n)$ ] @xmath191 , 4 . @xmath192 \eqmap{2 } [ x , y]+(t_x-1)p_y-(t_y-1)p_x$ ] @xmath193 . 5 . [ prop - cnj ] @xmath194 \eqmap{2 } \sum_{i=1}^n \sum_{j=1}^m t_{ij } \big ( [ x_i , y_j]+\delta(i , j ) \big),$ ] where @xmath195-(t_{x_i}-1)[\beta_j , y_j].\ ] ] 6 . [ jacobi]_jacobi relations : _ @xmath196+(t_y-1)[z , x]+(t_z-1)[x , y ] \eqmap{2 } 0.\ ] ] let us recall a well - known result on presentations of fundamental groups of line arrangements , an immediate consequence of the zariski - van kampen method : [ propos - pres ] the group @xmath17 admits a presentation of the form @xmath197 , where @xmath198 , and @xmath199 @xmath200 , such that @xmath201 is a meridian of @xmath173 , and @xmath202 @xmath203 . any presentation @xmath204 of @xmath17 as in proposition [ propos - pres ] will be called a _ zariski presentation _ of @xmath17 . the free group @xmath205 will be referred to as the _ free group associated with _ the given presentation . for technical reasons it is important to consider the alexander invariant corresponding to the free group associated with a given presentation . such a module will be denoted by @xmath206 . the following is a standard presentation of the modules @xmath206 and @xmath206 in terms of a zariski presentation of @xmath17 . [ propos - pres - alex ] let @xmath197 be a zariski presentation of @xmath17 and let @xmath207 be its associated free group , then the module @xmath206 admits a presentation @xmath208 , where @xmath209 \gl\ ] ] and @xmath210 is the submodule of @xmath211 generated by the jacobi relations ( property [ prop - alex - mod]([jacobi ] ) ) @xmath212 moreover , the module @xmath178 can be obtained as a quotient of @xmath206 as @xmath213 , where @xmath214 is the submodule of @xmath215 generated by the relations @xmath216 . [ exam - dt ] as an example of how to obtain @xmath214 note that , if the lines @xmath173 and @xmath217 in @xmath179 intersect in a double point , then there is a relation in @xmath216 of type @xmath218 $ ] , where @xmath219 . using property [ prop - alex - mod]([prop - cnj ] ) , this relation can be written in @xmath206 as @xmath220-(t_i-1)[\ga_j , x_j]\in \cw$ ] . analogously , if the lines @xmath173 , @xmath217 and @xmath221 in @xmath179 intersect at a triple point , one obtains relations in @xmath17 of type @xmath222 $ ] , where @xmath223 , which can be rewritten in @xmath224 as @xmath225-(t_i-1)[\ga_j , x_j]+ t_j \big(x_{i , k}+(t_k-1)[\ga_i , x_i]-(t_i-1)[\ga_k , x_k]\big).\ ] ] let @xmath226 be the augmentation ideal in @xmath183 associated with the origin , that is , the kernel of homomorphism of @xmath183-modules , @xmath227 , @xmath228 , where @xmath229 has the trivial module structure . one can consider the filtration on @xmath178 associated with @xmath226 , that is , @xmath230 . the associated graded module @xmath231 , where @xmath232 is a graded module over @xmath233 . consider the ring @xmath234 , tensoring @xmath183 by successive powers of the ideal @xmath226 . this allows one to define truncations of the alexander invariant . [ def - trunc - alex - mod ] the @xmath235-module , @xmath236 will be called the _ @xmath237-th truncated alexander invariant of @xmath179_. the induced filtration is finite and will be denoted in the same way . [ exam - dt - trunc ] from example [ exam - dt ] , it is easy seen that the relations in @xmath238 coming from double and triple points are as follows . 1 . [ d ] if @xmath173 and @xmath217 intersect at a double point one has : @xmath239-(t_i-1)[\ga_j , x_j]=0,\ ] ] @xmath240 2 . [ t ] if @xmath173 , @xmath217 and @xmath221 intersect at a triple point one has : @xmath241-(t_i-1)[\ga_j , x_j]+ ( t_k-1)[\ga_i , x_i]-(t_i-1)[\ga_k , x_k]=0,\ ] ] @xmath242 for any @xmath243 , there is a natural morphism @xmath244 . we will sometimes refer to @xmath245 as @xmath246 and equalities in @xmath247 will be denoted by @xmath248 . a zariski presentation on @xmath17 induces a ( set - theoretical ) section @xmath249^{k_1} ... [x_r,[x_i , x_j]]^{k_r},\ ] ] which , accordingly , induces a section of @xmath250 on each @xmath247 denoted by @xmath251 . from property [ prop - alex - mod ] , we deduce that the kernel of @xmath252 equals @xmath253 . moreover @xmath254 equals @xmath255 . [ propos - modk ] let @xmath256 be a word on the letters @xmath257 . if @xmath258 and @xmath259 @xmath260 , then @xmath261\equivmap{k+1}[g,\psi(\bar q)]$ ] , @xmath262 . in particular , if @xmath263 then @xmath264 $ ] is a well - defined element of @xmath265 ; if @xmath266 this element can be written @xmath267 . the ring @xmath268 is not local , but note that an element @xmath269 is a unit if an only if @xmath270 . to see this note that @xmath271 and the kernel of the evaluation map @xmath272 is exactly @xmath273 . [ free - tilde ] note that everything in this section can also be reproduced by using the free group @xmath274 associated with a zariski presentation of @xmath17 and will be denoted by adding a tilde . for instance , @xmath275 is the alexander invariant associated with @xmath276 , and @xmath277 is the filtration associated with @xmath278 . note that the module @xmath279 is isomorphic ( as an abelian group ) to the graduate @xmath280 by means of the morphism @xmath281 where @xmath282 is a laurent polynomial in @xmath283 and @xmath284 is the @xmath285-truncated taylor polynomial of @xmath282 around the origin @xmath286 . note that this isomorphism depends strongly on the zariski presentation of @xmath17 . more specifically , it depends on changes in the set of generators . for instance , any automorphism of @xmath17 that sends @xmath201 to @xmath287 , ( with @xmath288 ) induces both an automorphism of @xmath279 and a filtered automorphism of @xmath289 which is the identity on @xmath290 , but not on @xmath279 : @xmath291\mapsto [ x_i { \alpha_i},x_j { \alpha_j}]\equivmap{k } [ x_i , x_j]+t_j(t_i-1)\alpha_j - t_i(t_j-1)\alpha_i.\ ] ] nevertheless , the following result is true for these graduate groups . [ prop - gr - comb ] the graduate groups @xmath292 ( if @xmath293 ) depend only on the combinatorics @xmath34 of @xmath179 and will be denoted by @xmath294 . the following result is an immediate consequence of proposition [ propos - modk ] and it explains why @xmath279 is a more manageable object . [ lineal ] the formula ( [ eq - key ] ) only depends on @xmath295 . let @xmath4 and @xmath5 be two ordered line arrangements sharing the ordered combinatorics @xmath34 . consider two zariski presentations @xmath296 and @xmath297 of the fundamental groups of @xmath298 and @xmath299 , where the subscripts of the generators @xmath198 respect the ordering of the irreducible components . the abelian groups @xmath300 and @xmath301 can be canonically identified with @xmath302 so that @xmath303 . hence @xmath304 we will study the existence of isomorphisms @xmath305 such that @xmath306 is the identity . let @xmath307 be the free group associated with the zariski presentation of @xmath15 , @xmath8 . a morphism @xmath308 is called a _ homologically trivial morphism _ if @xmath309 satisfies @xmath310 . a morphism @xmath305 is called a _ homologically trivial isomorphism _ if it is induced by a homologically trivial morphism @xmath311 , i.e. , if @xmath306 is the identity . note that @xmath311 might not be unique . [ rem - htm ] 1 . [ rem - htm - expl ] in other words , a morphism @xmath305 is homologically trivial if there exists @xmath312 such that @xmath313 . 2 . any homologically trivial isomorphism @xmath314 induces a @xmath183-module morphism @xmath315 . [ rem - htm3 ] any homologically trivial isomorphism @xmath314 respects the filtrations @xmath316 and produces isomorphisms @xmath317 . by identifying @xmath318 , @xmath319 is the _ identity_. in order to state some properties of homologically trivial isomorphisms , we need to introduce some notation . note that the homologically trivial morphism @xmath311 also induces morphisms on the alexander invariants of the associated free groups @xmath320 ( @xmath8 ) and on their truncations @xmath321 . let us denote by @xmath322 the induced homologically trivial morphisms of the truncated modules . a straightforward computation proves that @xmath323 homologically trivial isomorphisms induce a particular kind of isomorphisms of the @xmath183-modules @xmath324 which are worth studying . [ rem - gr - combinat ] a direct attempt to prove that two modules are homologically trivial isomorphic is almost intractable . one would have to check if , for some choice @xmath325 , such a @xmath183-module isomorphism exists . the lack of linearity in this approach is the reason why we consider the truncated modules @xmath326 . applying corollary [ lineal ] , we are faced with simply solving a linear system as follows . let @xmath305 be a homologically trivial morphism , then there exists @xmath325 such that @xmath327 . therefore there exist @xmath328-morphisms @xmath329 induced by @xmath314 for any @xmath243 . note that @xmath330 where @xmath331 $ ] , @xmath332 and @xmath333 , since @xmath334\equivmap{2}[x_i\alpha_i , x_j\alpha_j]\ ] ] only depends on @xmath335 , the class of @xmath336 , by proposition [ lineal ] . in order to prove that @xmath337 is well defined , one must solve a linear system of equations on the variables @xmath338 in the abelian group @xmath339 . if an integer solution exists , one can repeat the procedure on @xmath340 , obtaining again a linear system of equations in the abelian group @xmath341 , and so on . in this work , we only need to consider @xmath337 . let us consider an ordered line arrangement @xmath179 with a fixed zariski presentation @xmath342 . let us denote @xmath34 its ordered combinatorics . [ free ] the @xmath343-module @xmath344 is free of rank @xmath345 , where @xmath346 is the multiplicity of the combinatorial type of @xmath179 ( notation [ not - comb - type ] ) . it is straightforward by taking suitable orderings on the relations . [ jac - homo ] let @xmath347 . then @xmath348 . a consequence of the fact that @xmath349 . [ rem - gen - gr1 ] note that @xmath350 is an abelian group with _ trivial _ action of @xmath351 . a system of generators is given by @xmath352 ; the relations as an abelian group are exactly those in @xmath353 and the relations and in example [ exam - dt - trunc ] , which depend only on the combinatorial type by . in our particular case we can effectively compute the 2-nd truncated alexander invariant . the following result is an easy computation . [ lem - mcl - rank ] for any maclane arrangement @xmath179 , the abelian group @xmath354 is free of rank @xmath355 , and its subgroup @xmath356 is free of rank @xmath357 . [ thm - main ] there is no homologically trivial isomorphism between @xmath358 and @xmath359 . consider suitable zariski presentations @xmath360 and @xmath361 of the ordered arrangements @xmath14 and @xmath99 ( for instance , we have used the suitable zariski presentations provided in @xcite and other presentations obtained using the software in @xcite ) . we identify the corresponding free groups @xmath362 and @xmath363 with a free group @xmath364 . recall that their combinatorial type has multiplicity @xmath365 and we suppose that the relations are ordered upon the following condition : @xmath366 ( in particular @xmath367 ) . let us suppose that a homologically trivial homomorphism @xmath368 exists . consider the corresponding elements @xmath369 that induce such a morphism ( remark [ rem - htm ] ) . consider @xmath370 and @xmath371 , the alexander invariants of @xmath372 and @xmath99 . let @xmath373 $ ] be the ground ring of both alexander invariants , where @xmath374 as usual . this mapping induces a @xmath375-isomorphism @xmath376 . by corollary [ lineal ] , @xmath337 only depends on the class @xmath377 . one has @xmath378 by jacobi relations play no role here . let us fix @xmath379 . since @xmath380 vanishes in @xmath381 , one deduces that @xmath382 should vanish in @xmath383 . equivalently @xmath384 should also vanish in @xmath385 . the vanishing of these terms , considered in the free group @xmath386 , produces a system of linear equations in the variables @xmath387 ( actually , even though there are 147 variables , only 126 appear in the equations ) . solving a system with 137 equations and 126 variables is not an easy task , but any computer will help . using maple8 , it takes 85 seconds of cpu time running on an athlon at 1.4mhz and 256 kb ram memory to obtain the linear set of solutions . it is an affine variety of dimension 98 of the form @xmath388 where @xmath389 @xmath390 , and @xmath391 . since @xmath392 one concludes that there is no integer solution . there is no orientation preserving homeomorphism between the pairs of ordered arrangements @xmath393 and @xmath394 . for rybnikov s arrangements , one obtains similar results . for any rybnikov s arrangement @xmath395 , the abelian group @xmath396 is free of rank @xmath397 , and its subgroup @xmath398 is free of rank @xmath399 . [ thm - main - ryb ] there is no homologically trivial isomorphism between @xmath400 and @xmath401 . one way to prove this statement is to follow the computational strategy proposed for maclane arrangements . first one needs zariski presentations of @xmath402 . this was done by means of the software in @xcite . in this case the linear system obtained consists of 531 equations and 420 variables ( again , out of the 792 variables @xmath387 , only 420 appear in the equations ) and it took the same processor a total of 23,853 seconds of cpu time to compute the solutions . the space of solutions has dimension 252 , that is , it can be written as @xmath403 , where @xmath404 @xmath390 , and @xmath391 . since @xmath405 , one again concludes that there is no integer solution . another proof that does nt depend as strongly on computations can be obtained from theorem [ thm - main ] as follows . let us assume that a homologically trivial isomorphism exists between @xmath406 and @xmath407 . such an isomorphism induces an @xmath375-isomorphism between @xmath408 and @xmath409 . let @xmath410 , where @xmath411 is the ideal generated by @xmath412 , and let @xmath413 denote @xmath414 . note that @xmath415 ( resp . @xmath416 ) can be considered as the @xmath417-module obtained from the inclusion of the complements @xmath418 ( resp . @xmath419 ) . moreover , these inclusions define epimorphisms of @xmath417-modules @xmath420 and @xmath421 . proving the existence of a homologically trivial isomorphism @xmath422 that matches in the commutative diagram , and using theorem [ thm - main ] one obtains a contradiction . @xmath423 consider @xmath424 the @xmath417-submodule of @xmath425 generated by the elements @xmath426 , @xmath427 and consider the commutative diagram . since @xmath428 and @xmath429 . @xmath430 since @xmath415 and @xmath431 are free abelian of the same rank , can be obtained from , by proving that @xmath432 is an isomorphism , which is the statement of lemma [ lema - comm - diag ] . [ lema - comm - diag ] the epimorphism @xmath433 in is injective we break the proof in several steps . 1 . [ ceros ] @xmath434 in @xmath435 if @xmath436 . + this can be proved case by case ( all the equalities are considered in @xmath435 ) : 1 . [ cerosa ] if @xmath437 and @xmath438 ( or vice versa , since @xmath439 ) : this is a consequence of example [ exam - dt - trunc ] since the lines @xmath173 and @xmath217 intersect transversally . [ cerosb ] if @xmath440 : this is a consequence of ( [ cerosa ] ) and the jacobi relations ( property [ prop - alex - mod]([jacobi ] ) ) . 3 . if @xmath441 : using the jacobi relations ( property [ prop - alex - mod]([jacobi ] ) ) and ( [ cerosb ] ) it is enough to check that @xmath434 , @xmath442 , @xmath438 . if @xmath173 and @xmath217 intersect at a double point , then ( [ cerosa ] ) proves the result . otherwise , there exists a line @xmath443 ( @xmath444 ) such that @xmath173 , @xmath217 and @xmath443 intersect at a triple point . by example [ exam - dt - trunc]([t ] ) one has @xmath445 , but @xmath446 by ( [ cerosb ] ) , thus we are done . 2 . @xmath447 . it is a direct consequence of ( o[ceros ] ) . [ otres ] @xmath448 , i.e. , @xmath449 . + since @xmath450 is generated by @xmath451 , it is clear that @xmath452 decomposes in the required sum . it remains to prove that it is a direct sum . + one can consider @xmath452 as a quotient @xmath453 , hence it is enough to check that there is a system of generators @xmath454 of @xmath224 such that : @xmath455 note that a system of relators can be obtained combinatorially as @xmath456 ( if @xmath457 is a double point ) or @xmath458 ( if @xmath459 is a triple point ) . relations coming from double points satisfy . for the triple point relations note that any triple point @xmath459 such that @xmath460 , verifies that @xmath461 ; therefore , condition is also satisfied . 4 . @xmath462 . + by ( o[ceros ] ) , the abelian group @xmath425 is generated by @xmath426 , @xmath463 , and @xmath464 , @xmath465 ; the relators are obtained from the singular points ( see example [ exam - dt - trunc ] ) and the jacobi relations @xmath353 . + by ( o[ceros ] ) and remark [ rem - gen - gr1 ] , we find that @xmath466 is generated by the elements @xmath464 , @xmath465 and the relations are exactly those in @xmath353 and the relations and in example [ exam - dt - trunc ] . the arguments used in ( o[otres ] ) also show that only double and triple points in @xmath54 provide non - trivial relations and thus one obtains the same system of generators and relations of @xmath467 . the proof of lemma [ lema - comm - diag ] is combinatorial and depends strongly on the properties of @xmath116 . this lemma corresponds to a key statement of the proof of ( ? ? ? * lemma 4.3 ) which is worth mentioning . this last section will be devoted to proving that the fundamental groups of @xmath468 and @xmath469 are not isomorphic . [ rem - m - combinat ] associated with a combinatorial type @xmath24 , there is a family of groups , where @xmath470 and @xmath471 is given by generators and relations as a quotient of @xmath472 as described in remark [ rem - gr - combinat ] . note that , if @xmath34 has a realization @xmath179 , then one has identifications @xmath473 and @xmath474 . there is a natural injective map @xmath475 given by the permutation of the generators of @xmath476 ; we identify @xmath477 with its image in @xmath478 . another subgroup of @xmath479 , denoted by @xmath480 , is defined as those automorphisms of @xmath476 that induce an automorphism of @xmath481 . it is easily seen that @xmath482 . the first goal of this section is to prove that rybnikov s combinatorial type @xmath113 ( described in ( [ eq - ryb - ct-2 ] ) ) is homologically rigid ; we will follow the ordering ( [ eq - ryb - eqs ] ) . in order to do so , we are going to study @xmath480 for an ordered combinatorics @xmath75 having at most triple points . we will denote @xmath484 , @xmath485 , and @xmath486 . let us first describe the groups @xmath476 and @xmath487 : @xmath488 where @xmath489 is the subgroup generated by @xmath426 ( @xmath490 ) and @xmath491 is the subgroup generated by @xmath492 and @xmath493 ( @xmath494 ) . any isomorphism @xmath495 induces a map @xmath496 . let us represent @xmath495 by means of a matrix @xmath497 such that @xmath498 ( note that such a matrix is not uniquely determined : each column is only well defined modulo the vector @xmath499 ) . the conditions required for this map to define a morphism on the quotient @xmath487 are called _ admissibility conditions _ and can be expressed as follows : we are going to express the admissibility conditions in a more useful way . let @xmath505 and let @xmath503 be a matrix representing @xmath506 . fix @xmath507 and consider the submatrix @xmath508 of @xmath503 which contains the rows associated with @xmath509 . let @xmath510 , @xmath511 . we denote by @xmath512 , the column vectors @xmath513 of @xmath514 . 1 . [ lem - adm - cond1 ] the vectors @xmath515 span @xmath516 . 2 . [ lem - adm - cond2]@xmath517 . [ lem - adm - cond3]for any @xmath518 and for any @xmath519 , the vectors @xmath520 and @xmath521 are linearly dependent ( i.e. , span a sublattice of @xmath516 of rank less than two ) . in particular , if @xmath522 , then @xmath523 spans a rank - one sublattice of @xmath516 . [ lem - adm - cond4 ] there exists @xmath524 such that @xmath523 spans a rank - two sublattice of @xmath516 and @xmath525 . * @xmath506 is an automorphism . * the sum of the columns of @xmath503 is a multiple of @xmath526 . * it is an immediate consequence of the admissibility conditions . * if no such @xmath527 exists , then all the vectors @xmath528 are linearly dependent , which contradicts . the last part is a consequence of . 1 . [ def - adm-1 ] there exists @xmath507 , such that @xmath537 spans a rank - two sublattice . [ def - adm-1b ] for every @xmath27 and for every @xmath538 , @xmath539 and @xmath540 are linearly dependent . [ def - adm-3 ] @xmath541 . 1 . [ exam - adm - adm ] with the above notation , @xmath547 is 3-admissible by lemma [ lem - adm - cond ] . [ exam - adm - triple ] the combinatorics @xmath548 of a triple point ( that is , @xmath549 , @xmath550 ) is 3-admissible , simply using @xmath551 , @xmath552 , @xmath553 . [ exam - adm - general ] let @xmath24 be a combinatorics such that * @xmath554 ; * @xmath555 and @xmath556 define non - empty subcombinatorics in general position w.r.t . @xmath557 ( that is , @xmath558 and @xmath559 implies that @xmath560 ) ; * at most one line of @xmath557 _ intersects _ @xmath561 in a non - multiple point of @xmath557 . + then @xmath34 is not 3-admissible . it is enough to see that if @xmath562 is a set of non - zero vectors satisfying conditions and of definition [ def - adm ] , one has that @xmath544 . * @xmath558 and @xmath563 ; since @xmath560 then , by remark [ ref - def - adm ] , @xmath544 . * @xmath564 ; considering any @xmath565 and using the previous case , one has @xmath566 . the same argument works for @xmath567 . in particular , @xmath544 if @xmath568 . * @xmath569 and @xmath27 such that @xmath570 . since all the vectors associated with @xmath509 but one are proportional , then this must also be the case for @xmath539 . + all the vectors ( but at most one ) are proportional . to conclude we apply condition of definition [ def - adm ] . [ exam - adm - ceva ] ceva s line combinatorics is 3-admissible . + ceva s line combinatorics is given by the following realization : + + ( 0,0 ) ( -60,20)@xmath571 ( -20,25)@xmath572 ( -25,55)@xmath573 ( -25,95)@xmath574 ( -40,125)@xmath575 ( -100,20)@xmath576 + @xmath577 @xmath578 for example , the following is a 3-admissible set of vectors for ceva : @xmath579 5 . [ exam - adm - mcl ] maclane s line combinatorics @xmath54 is not 3-admissible . + we will use the combinatorics given in figure [ mclane ] . let us assume that maclane is 3-admissible , then one has a list of non - zero vectors @xmath580 associated with each line . we will first see that @xmath581 and @xmath582 can not be proportional . if they were ( @xmath583 ) , using @xmath94 , @xmath584 and @xmath585 one would have that @xmath586 and finally , using @xmath587 , one obtains @xmath588 , and hence , by @xmath589 and @xmath590 , all vectors are proportional , which contradicts condition of definition [ def - adm ] . + therefore , @xmath581 and @xmath582 are linearly independent . after a change of basis , one can assume that @xmath591 , @xmath592 , @xmath593 , and therefore @xmath594 . we will briefly describe the conditions and how they affect the vectors until a contradiction is reached : @xmath595 + hence @xmath596 does not satisfy the condition of definition [ def - adm ] with @xmath597 on the double point @xmath585 . let @xmath601 denote the set of lines of @xmath34 and let @xmath602 denote the set of triple points of @xmath34 . we will first prove that any isomorphism @xmath598 induces a map @xmath603 . consider a triple point @xmath604 ; then @xmath547 is an admissible subcombinatorics of @xmath34 ( example [ exam - adm ] ) and defines a triple point . the map @xmath603 given by @xmath605 is defined . we will next prove that such a map is indeed injective , and hence a permutation ( in order for this to make sense , one has to assume that @xmath606 and hence @xmath607 ) . assume @xmath599 is not injective , and let @xmath608 . one has to consider two different cases depending on whether or not @xmath609 and @xmath610 share a line : 1 . if @xmath611 do not share a line , i.e , @xmath612 and @xmath613 , where all the subscripts are pairwise different . by reordering the columns , let us write @xmath614 . let @xmath615 be the submatrix of @xmath503 corresponding to the rows @xmath616 . using : @xmath617 where @xmath618 , @xmath619 . the sublattice @xmath16 of @xmath620 generated by the columns @xmath621 should have maximal rank equal to @xmath41 . note that @xmath622 equals the rank of the matrix @xmath623 obtained by subtracting the last row from the first ones , forgetting the last row and replacing the first column by the sum of the first three : @xmath624 which does not have rank @xmath41 . if @xmath611 share a line , say @xmath625 and @xmath626 . then , analogously to the previous case , one obtains a similar matrix to ( [ eq - matrix1 ] ) but where the rows @xmath627 and @xmath628 are identified , and we proceed in a similar way . [ propos - c3 ] for any @xmath598 , @xmath34 pointwise @xmath47-admissible , @xmath599 satisfies the following triangle property : @xmath603 preserves triangles , that is , if @xmath634 are in a triangle , then @xmath635 are also in a triangle . let @xmath634 be three triple points in a triangle , say @xmath636 , @xmath637 , @xmath638 . let us assume that @xmath635 are not in a triangle . one has two possibilities , either two of them do not share a line or three of them share a line . 1 . two of them , say @xmath639 do not share a line . after reordering , we can suppose that @xmath640 and @xmath641 . for @xmath642 there are several possibilities but we may assume @xmath643 . consider the submatrix @xmath644 of @xmath503 given by the rows of @xmath645 . applying successively to the rows defined by @xmath645 one has : @xmath646 moreover , @xmath647 , @xmath648 , @xmath649 and @xmath650 as in the proof of proposition [ propos - permut ] we need @xmath651 , where @xmath652 is the matrix obtained from @xmath644 by subtracting the last row from the first ones and forgetting the last row . we obtain : @xmath653 which can not have rank @xmath41 . if @xmath639 and @xmath642 have a common line , say @xmath654 , we follow the same strategy and obtain the desired result . note that proposition [ propos - permut ] does not automatically ensure that in general an automorphism of the combinatorics is produced . for instance , consider the combinatorics @xmath34 given by the lines @xmath655 , and the following triple points @xmath94 , @xmath656 , and @xmath657 ( the remaining intersections are double points ) . it is easy to see that such a combinatorics is pointwise 3-admissible . let @xmath495 be given by the following matrix : @xmath658 satisfying the admissibility relations . this induces the following maps : @xmath659 where @xmath660 . it is not difficult to prove that any combinatorics ( with only double and triple points ) of less than 6 lines ( other than @xmath548 ) is not @xmath47-admissible . in example [ exam - adm ] it is shown for the whole combinatorics of eight lines . let us check the remaining cases , that is , six and seven lines . note that , up to a combinatorics automorphism , there is a unique way to remove one line . there are , however , two possible ways to remove two lines , depending on whether they intersect or not at a triple point . therefore we only have to check the following cases : 1 . for 7 lines + @xmath663 : @xmath664 . 2 . for 6 lines , removing two lines intersecting at a triple point + @xmath665 : @xmath666 . 3 . for 6 lines , removing two lines intersecting at a double point + @xmath667 : @xmath668 . 1 . [ rem - ryb - triang36 ] the triple point @xmath671 in @xmath116 is the only one that belongs to 36 triangles . [ rem - ryb - triang - triple ] any triple point @xmath507 in @xmath116 except for @xmath587 and @xmath672 satisfies that @xmath94 and @xmath509 are in a triangle . [ rem - ryb - triang-2triple ] any two triple points in @xmath54 sharing a line belong to a triangle . [ rem - ryb - triang-3triple ] for any three triple points @xmath645 in @xmath54 , there exists another triple point @xmath527 such that @xmath673 belong to a triangle ( @xmath674 ) . 1 . [ cor - mclane1 ] @xmath676 2 . [ cor - mclane2 ] @xmath599 either preserves ( resp . exchanges ) the triple points of @xmath123 and @xmath124 in @xmath116 inducing an automorphism 3 . [ cor - mclane3 ] the action of @xmath599 on @xmath123 and @xmath124 comes from an automorphism ( resp . isomorphism ) of their combinatorics . part is true by propositions [ propos - permut]-[propos - c3 ] and remark [ rem - ryb - triang]([rem - ryb - triang36 ] ) . by remark [ rem - ryb - triang]([rem - ryb - triang - triple ] ) , the points @xmath587 and @xmath672 are either preserved or exchanged . in order to prove and , we may suppose that @xmath677 . recall that the subcombinatorics defined by @xmath678 is isomorphic to @xmath54 . since triangles are preserved by @xmath599 ( proposition [ propos - c3 ] ) , according to remark [ rem - ryb - triang]([rem - ryb - triang-2triple ] ) the images of any two triple points in @xmath678 sharing a line also share a line . this implies , using remark [ rem - ryb - triang]([rem - ryb - triang-3triple ] ) again , that the image of any three triple points on @xmath678 sharing a line are also three points sharing a line . since any line in @xmath123 or @xmath124 has at least three triple points , we conclude and . we will use proposition [ cor - mclane ] repeatedly . we can compose @xmath506 with an element of @xmath670 in order to have @xmath599 preserve the triple points in @xmath680 . recall that @xmath681 comes from an automorphism @xmath682 of @xmath54 which respects @xmath94 . composing again with an element of @xmath670 we may suppose that @xmath683 is the identity on @xmath94 . it is enough to prove that it is also the case for @xmath252 . if it is not the case , we may assume ( by conjugation with an element of @xmath670 ) that @xmath684 . there are two possibilities to be checked , depending on whether @xmath685 and @xmath117 are fixed or permuted . in both cases , the triple points @xmath94 , @xmath587 and @xmath672 are fixed . using the arguments in the proofs of propositions [ propos - permut ] and [ propos - c3 ] , one can obtain the induced matrices @xmath503 with all the admissibility relations , which , modulo @xmath502 do not have a maximal rank in either case . let @xmath679 ; by propositions [ propos - combinat ] and [ propos - c3 ] , @xmath599 comes from an automorphism of the combinatorics . composing with the inverse of such an automorphism , we may suppose that @xmath686 . it is hence enough to prove that any isomorphism @xmath679 that induces the identity on @xmath599 is just @xmath687 . from the definition of @xmath547 , @xmath545 , we deduce the following . let us fix the @xmath688-column ; all the entries in this column corresponding to @xmath545 such that @xmath689 are equal . we deduce from this that we can choose @xmath503 such that all the elements outside the diagonal are constant in their column . adding multiples of @xmath502 , we obtain that @xmath503 can be chosen to be diagonal . we also know that for each @xmath545 , the diagonal terms corresponding to @xmath51 are equal and since any two elements can be joined by a _ chain _ of triple points , we deduce that all the diagonal terms are equal . since @xmath506 is an automorphism , they are equal to @xmath690 and then @xmath691 . let @xmath406 and @xmath407 be the fundamental groups of @xmath468 and @xmath692 respectively . any isomorphism @xmath693 will produce an automorphism @xmath694 , that is , we can consider @xmath695 . by theorem [ thm - main - comb ] , @xmath506 induces an automorphism of @xmath116 . since the identifications @xmath696 are made up to the action of @xmath670 , we may assume that @xmath506 induces @xmath687 . moreover , eventually exchanging @xmath468 ( resp . @xmath692 ) for @xmath697 ( resp . @xmath698 ) , see example [ exam - ryb - comb ] by means of the automorphism given by complex conjugation , we may assume that @xmath699 ( proposition [ thm - main - comb ] ) . therefore @xmath700 is a homologically trivial isomorphism between @xmath406 and @xmath407 , something which is ruled out by theorem [ thm - main - ryb ] .	following the general strategy proposed by g.rybnikov , we present a proof of his well - known result , that is , the existence of two arrangements of lines having the same combinatorial type , but non - isomorphic fundamental groups . to do so , the alexander invariant and certain invariants of combinatorial line arrangements are presented and developed for combinatorics with only double and triple points . this is part of a more general project to better understand the relationship between topology and combinatorics of line arrangements . one of the main subjects in the theory of hyperplane arrangements is the relationship between combinatorics and topological properties . to be precise , one has to make the following distinction : for a given hyperplane arrangement @xmath0 , one can study the topological type of the pair @xmath1 or the topological type of the complement @xmath2 . for the first concept we will use the term _ relative topology _ of @xmath3 , whereas for the second one we will simply say _ topology _ of @xmath3 . it is clear that if two hyperplane arrangements have the same relative topology , then they have the same topology , but the converse is not known . in a well - known and very cited work @xcite , g. rybnikov found an example of two line arrangements @xmath4 and @xmath5 in the complex projective plane @xmath6 having the same combinatorics but different topology . the most common way to prove that two topologies of line arrangements are different is to check that the fundamental groups of their complements are not isomorphic . recently , the authors of this work have provided an example of two line arrangements with different relative topologies ( see @xcite ) . the contribution of @xcite is that it refers to real arrangements , that is , arrangements that admit real equations for each line ( note that rybnikov s example can not have real equations ) . the proof proposed by rybnikov has two steps . let @xmath7 , @xmath8 . 1 . [ r1 ] recall that the homology of the complement of a hyperplane arrangement depends only on combinatorics . this way , one can identify the abelianization of @xmath9 and @xmath10 with an abelian group @xmath11 combinatorially determined . rybnikov proves that no isomorphisms exist between @xmath9 and @xmath10 that induce the identity on @xmath11 . in particular , this result proves that both arrangements have different relative topologies . the reason can be outlined as follows : any automorphism of the combinatorics of rybnikov s arrangement can be obtained from a diffeomorphism of @xmath6 , and hence it produces an automorphism of fundamental groups . since any homeomorphism of pairs @xmath12 induces an automorphism of the combinatorics of @xmath13 , after composition one can assume that any homeomorphism of pairs induces the identity on @xmath11 . + the strategy rests on the study of the lower central series ( lcs ) . since @xmath4 and @xmath5 are constructed using the maclane arrangement @xmath14 ( see example [ mclane - rlz ] ) , it is enough to study , by some combinatorial arguments , the lcs of @xmath14 with an extra structure ( referred to as an ordered arrangement ) . although this part is explained in ( * ? ? ? * section 3 ) , computations are hard to verify . 2 . [ r2 ] the second step is essentially combinatorial . the main point is to truncate the lcs of @xmath15 such that the quotient @xmath16 depends only on the combinatorics . rybnikov proposes to prove that an automorphism of @xmath16 induces the identity on @xmath11 ( up to sign and automorphisms of the combinatorics ) . this proof is only outlined in ( * ? ? ? * proposition 4.2 ) . it is worth pointing out that such a result can not be expected for any arrangement . also ( * ? ? ? * proposition 4.3 ) needs some explanation of its own . the main difference between relative topology and topology of the complement in terms of isomorphisms of the fundamental group is that homeomorphisms of pairs induce isomorphisms that send meridians to meridians , whereas homeomorphisms of the complement can induce any kind of isomorphism , and even if we know that the isomorphism induces the identity on homology , this is not enough to claim that meridians are sent to meridians . the aim of our work is to follow the idea behind rybnikov s work and , using slightly different techniques , provide detailed proofs of his result . this is part of a more general project by the authors that aims to better understand the relationship between topology and combinatorics of line arrangements . the following is a more detailed description of the layout of this paper . in section [ sec - defs ] , the more relevant definitions are set , as well as a description of rybnikov s and maclane s combinatorics . sections [ sec - alex ] and [ sec - method ] provide a proof of step ( r[r1 ] ) . in order to do so , we propose a new approach related to derived series , which is also useful in the study of characteristic varieties and the alexander invariant . the alexander invariant of a group @xmath17 , with a fixed isomorphism @xmath18 , is the quotient @xmath19 considered as a module over the ring @xmath20 $ ] , which is the group algebra of @xmath21 . by fixing an ideal @xmath22 , the truncated module by @xmath23 is studied . the problem is reduced to solving a system of linear equations . section [ sec - combinat ] is devoted to the study of combinatorial properties of a line arrangement which ensure that any automorphism of the fundamental group of the complement _ essentially _ induces the identity on homology ( that is , the analogous of ( * ? ? ? * proposition 4.2 ) ) . this is an interesting question that can be applied to general line arrangements . for the sake of simplicity , we only present our progress on line arrangements with double and triple points . this provides a proof for the second step ( r[r2 ] ) .
we use a pre - computed database of 46800 synthetic flux and polarization spectra for field strengths of 1400 mg and a wide range of effective temperatures to generate synthetic zeeman spectra , which we subject to an automatic optimization scheme with the aim to recover the parameters describing the original magnetic field configuration . we adopt input field geometries which involve the sum of non - aligned dipole and quadrupole components . we also provide for off - centre shifts of the configuration . by adding gaussian noise we create input spectra with and 20 , respectively . = 0.80 off - centred dipole - quadrupole combinations are reliably recovered even for by our code , which can handle up to 12 free parameters but becomes inefficient if higher multipoles are included . the poor convergence is caused by the complexity of the @xmath0-landscape , which develops an increasing number of secondary minima for more complex field geometries ( for a detailed investigation , see euchner et al . , 2002 ) . we find that a given set of phase - resolved zeeman spectra can be reproduced within the noise by quite different formal representations of the field , which , however , all seem to describe rather similar actual field geometries . we are confident , therefore , that observed phase - resolved zeeman spectra can provide quite definite information on the field structure . = 0.85 the magnetic field structure over the surface of accreting white dwarfs in cataclysmic binaries becomes accessible to observation only in states of low or switched - off accretion . following the first detection of zeeman lines in the polar eferidani ( @xmath1 = 81 min ) by wheatley & ramsay ( 1998 ) , we obtained phase - resolved flux and circular polarization spectra of this object with fors1 at the eso vlt on 22 nov 2000 . fig.1 shows the spectra collected in four almost equal phase bins after removal of a slight rotational temperature variation . there is no obvious variation of the zeeman structures with phase . fits of inclined multipole field models with the angle @xmath2 between magnetic and rotational axis left free yielded . the lack of phase dependence of the zeeman features further suggests that the tesseral / sectoral multipole components are weak . we fitted the mean flux and polarization spectra , therefore , with an expansion including only the zonal components up to a maximum degree . fig.2 shows the dipole+quadrupole fit , which is substantially better than the pure dipole ( not shown ) . fig.3 depicts the result for the expansion up to . the dipole is replaced by large coefficients for the higher - order multipole components which substantially improve the fit . while the higher - order multipole fit accounts for many of the observed details all the wiggles in the blue are genuine zeeman features there are still noticeable discrepancies : ( i ) the observed lack of circular polarization around 5700 requires that a larger fraction of the field occurs at negative @xmath3 ( angle between line of sight and field direction ) , and ( ii ) the observed narrow spectral feature at 5120 can be fitted by additional field contributions with @xmath4 between 60 and 100 mg pointing away from the observer , possibly generated by a steeper rise of @xmath4 towards the lower ( unseen ) pole than provided by the model .	we have developed a new method to derive the magnetic field distribution on the surfaces of rotating magnetic white dwarfs from phase - resolved flux and circular polarization spectra . an optimization code based on an evolutionary strategy is used to fit synthetic zeeman spectra for a variety of model geometries described in the framework of a truncated multipole expansion . we demonstrate that the code allows the reconstruction of relatively complex fields using noise - added synthetic input spectra . as a first application , we analyze flux and circular polarization spectra of the polar eferi in a low state of accretion taken with fors1 at the eso vlt .
in the framework of the semiclassical theory of waves [ 1 - 5 ] , i.e. , short wavelength asymptotics , the uniform ( global ) description of the wavefield is complicated by the formation of caustic singularities [ 1 , 2 , 6 - 8 ] . although a complete and deep understanding of the wavefield structure near caustic regions is obtained on the basis of catastrophe theory @xcite and the unfolding of the corresponding singularities can be treated by means of symplectic techniques @xcite , the application of such methods to realistic cases , e.g. , to waves in magnetically confined plasmas @xcite , appears rather difficult . therefore , with specific regard to physical applications , several asymptotic methods have been developed which yield numerically tractable equations , though being limited concerning the global properties of the asymptotic solutions . such asymptotic techniques can be classified into two different families , depending on whether the relevant wave equation is described in the phase space , _ microlocal techniques _ , or directly in the configuration space where the wavefield is defined , _ quasi - optical methods_. this work aims to give a detailed comparative analysis of two such techniques , namely , the wigner - weyl kinetic formalism [ 11 - 13 ] and the complex geometrical optics ( cgo ) method [ 14 - 17 ] which can be considered as benchmarks for microlocal and quasi - optical methods , respectively . specifically , in sec.2 , the wigner - weyl formalism and the complex geometrical optics method are reviewed and compared . in particular , it is pointed out that , within the wigner - weyl formalism , physically meaningful solutions should have a specific form , referred to as momentum distribution , which is characterized in sec.3 . on the basis of the mathematical properties of momentum distributions , our main result is obtained in sec.4 . in particular , it is shown that , in correspondence to appropriate boundary conditions , there exists a specific asymptotic solution of the wave kinetic equation relevant to the wigner - weyl formalism that can be written in terms of the corresponding solution of the complex geometrical optics equations . this allows us to relate the two considered methods as well as to determine the specific class of boundary conditions for which they are equivalent . in sec.5 , this general result is illustrated by means of an analytically tractable example , i.e. , the propagation of a gaussian beam of electromagnetic waves in an isotropic `` lens - like '' medium . in conclusion , a summary of the main results is given in sec.6 . in order to set up the framework , let us discuss the relevant boundary value problem for a generic _ scalar _ pseudodifferential wave equation , together with the required mathematical definitions . specifically , we will consider the case of a scalar ( real or complex ) wavefield @xmath0 propagating in the @xmath1-dimensional linear space @xmath2 with @xmath3 a generic set of cartesian coordinates and denote by @xmath4 the corresponding coordinates in the dual space @xmath5 . time - dependent wavefields are included as one of the coordinates can play the role of time , e.g. , @xmath6 , @xmath7 being a reference speed , and the corresponding dual coordinate is related to frequency , e.g. , @xmath8 . to some extent , the results for a scalar wave equation are valid also for a multi - component wave equation as the latter can be reduced to a set of _ independent _ scalar equations far from mode conversion regions @xcite . thereafter , the wigner - weyl formalism will be formulated entirely in the space @xmath9 , with coordinates @xmath10 , which is viewed as the trivial cotangent bundle @xcite over the configuration space @xmath11 where the wavefield is defined . no explicit reference to the propagation direction is made , differently from the classical derivations @xcite . as for the cgo method , it has been originally developed for solving the second - order partial differential equation relevant to the propagation of electromagnetic wave beams in stationary spatially nondispersive media . hence , we will need to discuss its application to generic pseudodifferential equations . in particular , it is shown that the cgo method yields an approximation of the wavefield directly in the configuration space , provided that the wave equation satisfies an appropriate condition . first , let us define the class of wave equations undergone to solution . a pseudodifferential wave equation is an equation of the form @xmath12 which admits propagating wave solutions . the operator @xmath13 is a pseudodifferential operator [ 3,18 - 20 ] acting on the wavefield as a fourier integral operator @xcite characterized by the bilinear phase function @xmath14 . here , @xmath15 belongs to a particular class of smooth functions , referred to as _ symbols _ , which , roughly speaking , behave like a polynomial in @xmath16 for @xmath17 large enough . specifically , a smooth function @xmath18 with @xmath19 and @xmath20 is a symbol of order @xmath21 if for every multi - indices @xmath22 , @xmath23 there is a constant @xmath24 such that @xmath25 . in virtue of the symbol estimate ( [ 2n ] ) the integral in ( [ 1n ] ) makes sense for @xmath26 , the space of tempered distribution @xcite . moreover , the scale length @xmath27 characterizes the variations of symbols with respect to the spatial coordinate @xmath28 and it can be eliminated by the rescaling @xmath29 and @xmath30 . it is worth noting that any linear differential operator with smooth and bounded coefficients is a pseudodifferential operator @xcite . boundary conditions of cauchy type are given on an @xmath31-dimensional hypersurface @xmath32 : for simplicity , one can assume @xmath32 to be the hyperplane @xmath33 where the wavefield @xmath34 , @xmath35 , is assigned together with as many derivatives @xmath36 as appropriate . we are interested in semiclassical solutions for which only the covectors @xmath16 with @xmath37 are significant , in the integral in ( [ 1n ] ) . as a consequence , the wavefield should be a highly oscillating function on the ( large ) scale length @xmath27 and it should correspond to a specific set of highly oscillating boundary conditions of the form @xmath38 the amplitude @xmath39 being slowly varying , that is , @xmath40 . in the wigner - weyl formalism , the weyl - symbol map @xmath41 is applied in order to represent the wave equation ( [ 1n ] ) in @xmath42 which is naturally endowed with a phase - space structure . the weyl - symbol map transforms an operator @xmath43 , @xmath44 being the space of schwartz s functions @xcite , into a tempered distribution by acting on the schwartz kernel @xmath45 of the operator @xmath46 according to @xcite @xmath47 it is worth noting that for the pseudodifferential operator in ( [ 1n ] ) the schwartz kernel is @xmath48 thus , the image of @xmath13 under the weyl - symbol map amounts to the formal series of decreasing order symbols @xmath49 where taylor expansion has been used and @xmath50 . series of that kind admit always an asymptotic resummation @xcite to a symbol of order @xmath51 , @xmath52 which is referred to as _ weyl symbol _ of @xmath13 . in ( [ weyl ] ) the @xmath53 denotes the asymptotic equivalence of symbols @xcite . on the other hand , one can consider the correlation operator @xcite @xmath54 whose schwartz kernel is given by the tensor product @xmath55 , then the weyl - symbol map yields the _ wigner function _ @xmath56 the wave equation ( [ 1n ] ) can be written in the equivalent form @xmath57 and , on applying the weyl - symbol map , one gets @xcite @xmath58 in the semiclassical limit one has @xmath59 so that , on assuming the same ordering for the wigner function @xcite , the foregoing equation separates into [ 1.1 ] @xmath60 where @xmath61 and @xmath62 are the real and imaginary parts of the weyl symbol ; in particular , one has @xmath63 which is the condition for weak absorption and/or instabilities @xcite . equation ( [ 1.1a ] ) is a constraint to eq.([1.1b ] ) which , on the other hand , has the form of a kinetic equation in the @xmath64-@xmath16 phase space , @xmath65 being the corresponding poisson brackets . we will refer to the whole system ( [ 1.1 ] ) as the _ wave kinetic equation_. in general , a solution of the wave kinetic equation is a tempered distribution , however , one usually restricts the class of solutions to semiclassical measures . this allows us to make sense of the integrals of the form @xcite @xmath66 which expresses the expectation value of a physical quantity represented by the pseudodifferential operator @xmath46 with weyl symbol @xmath67 . in the following we will assume further regularity with respect to @xmath64 so that the expectation values of physical quantities can be defined _ locally _ , that is , [ physics ] @xmath68 makes sense as a smooth function in @xmath69 . in particular , the wavefield intensity amounts to @xmath70 ) . on the hyperplane @xmath71 the wavefield @xmath34 has been assigned and one can compute the corresponding wigner function @xmath72 , @xmath73 being the correlation operator associated to @xmath74 and @xmath75 the coordinates dual to @xmath76 ; then a solution @xmath77 should match @xmath78 in some appropriate sense . specifically , one should impose that the local value @xmath79 of any physical quantity evaluated on @xmath32 is the same whether it is evaluated by @xmath77 or by @xmath78 . within this formulation , the weyl symbol @xmath67 should be restricted to @xmath80 where @xmath81 is defined . one can note that the suitable embeddings @xmath82 of @xmath80 into @xmath83 such that @xmath82 lies over @xmath32 , i.e. , @xmath84 with @xmath85 the canonical projection @xcite , are of the form @xmath86 with @xmath87 a generic smooth function ; correspondingly , the restriction of a symbol is readily defined as @xmath88 . then the boundary value conditions read [ 6n ] @xmath89 which is equivalent to @xmath90 the function @xmath91 can not be arbitrary as ( [ 6nb ] ) should satisfy the constraint ( [ 1.1a ] ) which reads @xmath92 hence , the appropriate functions @xmath91 are obtained on solving the so - called _ local dispersion relation _ evaluated on @xmath32 . since @xmath32 has been assumed to be noncharacteristic , i.e. , @xmath93 on @xmath82 , the function @xmath91 is well defined and smooth at least locally , in view of the implicit function theorem . on the other hand , it is not unique since ( [ 7n ] ) may have multiple solutions , each one corresponding to a specific branch of the dispersion relation . in virtue of the superposition principle for linear wave equations , the total wavefield is a linear superposition of the contributions from each branch of the dispersion relation , the coefficients being determined by the cauchy boundary values of the normal derivatives @xmath94 . therefore , one has a specific cauchy boundary value problem for the wigner function of each branch and the sum over all branches yields the total wigner function . since one has @xmath95 where the indices @xmath96 and @xmath97 run over all branches , and , on noting that the average @xmath98 over short scale oscillations cancels out the mixed terms , @xmath99 for @xmath100 , whereas @xmath101 , one gets @xmath102 that is , the projection of the total wigner function yields the _ averaged wavefield intensity _ and , thus , it does not account for , e.g. , the formation of short - scale diffraction patterns . let us now turn to the complex geometrical optics ( cgo ) method and , in particular , let us discuss its application to pseudodifferential wave equations . this is based on approximating the solution of ( [ 1n ] ) by a smooth wave function of the form @xmath103 where , according to the semiclassical limit , @xmath104 and @xmath105 . in addition to the standard oscillating exponential @xmath106 , the wave object ( [ 1.3 ] ) exhibits a novel scale length @xmath107 which accounts for intermediate - scales variations of the amplitude profile @xmath108 with @xmath109 . in general , such an intermediate scalelength @xmath110 can be determined by both ( strong ) absorption @xcite and diffraction @xcite ; however , in this paper , it is assumed that the medium is weakly nondissipative [ cf . comments after equations ( [ 1.1 ] ) ] so that only diffraction effects are significant . the total short- and intermediate - scale variations of the wavefield are accounted for by the _ complex eikonal function _ @xmath111 . the relevant equations for the three unknown functions @xmath112 , @xmath113 and @xmath114 are determined on substituting the ansatz ( [ 1.3 ] ) into the wave equation ( [ 1n ] ) . for the specific case for which @xmath13 is a differential operator this is straightforward . on the other hand , for the general case , one should deal with the nonlocal response of the operator @xcite . with this aim it is convenient writing ( [ 1n ] ) in the configuration space in terms of the schwartz kernel , namely , @xmath115 where the exact kernel has been replaced by the @xmath116 where @xmath117 is the weyl symbol . actually , one could make use of other symbol maps @xcite , yielding asymptotically equivalent results ; here the weyl - symbol maps has been chosen for direct comparison with the wigner - weyl approach . let us further assume that the kernel @xmath118 amounts to a distribution smoothly dependent on @xmath119 and with compact support in @xmath120 . in virtue of the paley - wiener - schwartz theorem @xcite , this is equivalent to assume that the corresponding weyl symbol @xmath121 extends to an _ entire _ function of the complex - valued dual vector @xmath122 , smoothly dependent on @xmath123 . from a physical standpoint , the foregoing assumption implies that nonlocal effects have a finite range : the response @xmath124 of the operator @xmath13 depends only on the value of the wavefield @xmath125 in a compact set . within this condition , one can substitute the complex eikonal ansatz ( [ 1.3 ] ) into ( [ 1n ] ) and expand in taylor series with respect to @xmath126 . as a result one has @xcite @xmath127 where @xmath128 and , for the paley - wiener - schwartz theorem , the estimate @xmath129 is still valid for complex - extended symbols . to leading orders in @xmath130 , and for a weakly dissipative media , i.e. , @xmath63 , one gets [ 11n ] @xmath131 \right ] u(x),\end{aligned}\ ] ] @xmath132 and @xmath133 being the real and imaginary parts of @xmath117 extended in the complex @xmath134 space and , in general , are complex valued . it is worth noting that the foregoing equations can be formally obtained from the standard geometrical optics equations , e.g. , in the form given by littlejohn and flynn @xcite , by replacing @xmath135 with @xmath136 . the cgo equations ( [ 11n ] ) have been dealt with both by means of the characteristics method in the complex domain @xcite and on expanding the equations with respect to @xmath137 @xcite . in particular , on referring to the latter approach , in the weak - diffraction regime @xmath138 , terms up to order @xmath139 should be considered in the cgo equation for the complex eikonal @xmath140 , which , after separating the real and imaginary parts , amounts to [ 1.5 ] @xmath141 equations ( [ 1.5a ] ) and ( [ 1.5b ] ) constitute a set of coupled first - order partial differential equations for @xmath142 and @xmath143 with @xmath144 and @xmath145 . as for the complex amplitude @xmath146 , only the lowest order approximation with respect to @xmath147 is significant , so that the real amplitude @xmath148 is decoupled from the phase @xmath149 $ ] ( not considered hereafter ) and determined by means of the transport equation @xmath150 = 2d '' \big(x , k(x)\big)\ \|u(x)\|^{2}.\ ] ] this is formally the same equation as the geometrical optics transport equation @xcite , but diffraction effects are accounted for through the wavevector - field @xmath135 which differs from that obtained in the geometrical optics . the approximated form ( [ 1.5 ] ) of the cgo equations is the one used in physical applications . moreover , in the zero - diffraction regime ( @xmath151 ) , one has @xmath152 , thus terms up to first order only should be considered , with the result that equations ( [ 1.5a ] ) and ( [ 1.5b ] ) are decoupled and the whole set of cgo equations ( [ 1.5 ] ) reduces to the standard geometrical optics equations , @xmath113 being effectively zero . equations ( [ 1.5a ] ) and ( [ 1.5b ] ) are usually solved by computing the characteristic curves @xcite of ( [ 1.5a ] ) with ( [ 1.5b ] ) regarded as a constraint with the result that the characteristics curves thus obtained resemble the geometrical optics rays @xcite . therefore , the appropriate boundary conditions should be enough to determine the initial values of the complex vector @xmath153 evaluated on the boundary surface @xmath32 . such conditions are obtained from the cauchy data ( [ 1nbound ] ) on writing @xmath154 for some functions @xmath155 and @xmath156 such that @xmath157 and @xmath158 . from ( [ 12n ] ) one readily gets the value of the component of the complex vector @xmath159 tangent to @xmath32 , namely , @xmath160 and @xmath161 . the remaining normal component is obtained on imposing that the cgo equations ( [ 1.5 ] ) are satisfied on @xmath32 ; this yields two equation for @xmath162 and @xmath163 , _ viz . _ , [ 13n ] @xmath164 where @xmath165 and @xmath166 are obtained in terms of the first- and second - order @xmath16 derivatives of @xmath167 and evaluated at @xmath168 . equation ( [ 13na ] ) is an @xmath169 perturbation of the local dispersion relation ( [ 7n ] ) , hence , it can be solved by @xmath170 and , in correspondence of ( [ 14n ] ) , eq.([13nb ] ) yields @xmath171 . as in the wigner - weyl formalism , if multiple solutions are found , one should write the wavefield as a sum of contributions from each branch of the local dispersion relation . from the foregoing discussion , one should note that the cgo method yields the solution directly in the configuration space , but one should deal with the set of partial differential equations ( [ 1.5 ] ) , the numerical solution of which can be rather cumbersome . although the characteristics technique can be used for eq.([1.5a ] ) , the constraint ( [ 1.5b ] ) should be solved in parallel , thus increasing the computational complexity of the problem . as for the global properties of the cgo solution , to our knowledge no general result is still available , though numerical solutions @xcite show that the cgo solution is regular even near focal points where the standard geometrical optics solution exhibits a caustic singularity . in contrast , the wigner - weyl formalism appears better suited for numerical solutions . in particular , the wave kinetic equation can be solved along the corresponding hamiltonian orbits in the phase space so that it is reduced to a set of ordinary differential equations that require limited computational efforts and the solution thus obtained has a global validity in the phase space since the hamiltonian orbits do not cross each other . in this respect , the constraint ( [ 1.1a ] ) does not constitute a limitation as @xmath167 is a constant of motion . moreover , there is no limitation on the nonlocal response of pseudodifferential operators to which the wigner - weyl formalism applies . on the other hand , the solution in the phase space should be projected into the configuration space and , thus , an integral with respect to the momentum @xmath16 should be carried out numerically . notwithstanding these differences , the wigner - weyl kinetic formalism and the complex geometrical optics method share a number of features , e.g. , the solution of the local dispersion relation ( [ 14n ] ) relevant to the cgo method is obtained , to the lowest significant order in @xmath147 , on evaluating the corresponding solution ( [ 7n ] ) for @xmath172 . in the following sections , it will be proved that one can project the wave kinetic equation from the phase space into the configuration space in such a way that the cgo eq.([1.5 ] ) are recovered . as discussed in sec.2 , the solutions of the wave kinetic equation are usually sought in the space @xmath173 of tempered distributions @xcite , or in the space of semiclassical measures @xcite . the first formulation is the more general , whereas the second follows from the physical requirement that expectation values ( [ star ] ) are well defined . moreover , we have pointed out that , for general wave propagation problems , physics requires a stronger condition on the wigner function , namely , the expectation values of physical quantities should be locally defined according to ( [ starstar ] ) . for instance , if one deals with a time - dependent wavefield for which @xmath174 and @xmath175 , the integral @xmath176 yields the wave action density @xmath177 in the space - time @xcite . in this section , a mathematical characterization of such novel solutions is given and the corresponding differential calculus is put forward . first , let us note that for any schwartz function @xmath178 and for any tempered distribution @xmath179 one can define a tempered distribution @xmath180 over the configuration space only , given by @xmath181 where , in general , angle brackets and the integral are alternative ways to denote the action of a distribution on the corresponding test function . the distribution @xmath182 is smooth with respect to @xmath69 if and only if @xmath183 amounts to a smooth function @xmath184 . in this case the map @xmath185 for @xmath186 defines at every point location @xmath123 a tempered distribution @xmath187 with @xmath188 this is a consequence of the completeness of @xmath189 along with the identity @xmath190 where @xmath191 is a compact - supported function that approximates the dirac s @xmath192-function for @xmath193 . let us now consider a symbol @xmath194 , that is , @xmath39 fulfills the symbol estimate ( [ 2n ] ) for every order @xmath51 . it follows that @xmath195 and for any @xmath196 which is smooth with respect to @xmath64 one can define @xmath197 and this is a smooth function on @xmath11 as required in ( [ starstar ] ) . the definition ( [ 1 m ] ) should be extended for any symbol @xmath198 with arbitrary order . with this aim , the class of physically admissible solutions [ in the sense of ( [ starstar ] ) ] is restricted . in particular , it is appropriate considering the tempered distributions @xmath199 that satisfy the following conditions : 1 . @xmath182 is smooth with respect to @xmath64 , 2 . the restriction @xmath200 amounts to a distribution with compact support , i.e. , @xmath201 , where @xmath202 is continuously embedded in @xmath189 in the weak topology . such a distribution @xmath182 will be called _ momentum distribution _ since for every @xmath64 its restriction @xmath200 represents the distribution of momentum @xmath16 over @xmath64 . for short let us write @xmath203 for the space of momentum distributions . within this formulation for every @xmath204 and for every @xmath198 eq.([1 m ] ) is well posed and defines a smooth function on @xmath11 . let us note that the foregoing definition of the space @xmath205 is not the optimal one as functions rapidly decreasing in @xmath16 are also admissible momentum distributions . however , in the semiclassical limit these functions can be ignored and only compact - supported distributions are significant . let us now address the derivatives of a momentum distribution @xmath204 . first , the derivatives with respect to the momentum @xmath16 are defined throughout every order . specifically , since @xmath206 @xmath207 is smooth with respect to @xmath64 ; moreover , for every @xmath186 , @xmath208 hence , @xmath209 is compactly supported , and , thus , @xmath210 . in terms of the integral notation the latter result evaluated for symbols reads @xmath211 which is the `` integration - by - parts '' formula . on the other hand , the derivatives with respect to @xmath64 should be dealt with more carefully . for simplicity , let us consider first - order derivative @xmath212 for @xmath203 . one has that @xmath213 in virtue of ( [ 1mprimo ] ) , so that @xmath212 is smooth with respect to @xmath64 . furthermore , @xmath214 hence @xmath215 is compactly supported and @xmath216 . the explicit formula for the derivative is obtained on noting that for every symbol @xmath198 , @xmath217 and @xmath218 which in the integral notation takes the form @xmath219 the same result would be obtained from the definition @xmath220 through straightforward but longer calculations . higher - order derivatives can be defined by recurrence , but they are not explicitly needed in the following . searching for solutions of the wave kinetic equations ( [ 1.1 ] ) in the space of momentum distributions leads to the weak formulation [ 2.1bar ] @xmath221a(x , k)\ d^{n}k=0,\end{gathered}\ ] ] with @xmath222 . furthermore , we are interested in semiclassical solutions for which only large - enough momenta are significant . on recalling that @xmath223 , we will search for solution of ( [ 2.1bar ] ) in the form @xmath224 where @xmath142 and @xmath225 are smooth functions to be determined ; in particular , @xmath142 defines a lagrangian manifold @xmath226 in the @xmath64-@xmath16 phase space . moreover , @xmath227 is normalized , i.e. , @xmath228 , and such that @xmath229 for any multi - index @xmath22 . the integrals in ( [ 2.3 ] ) are well posed since @xmath230 are symbols of order @xmath231 , and the corresponding quantities @xmath232 express the statistical moments of the distribution @xmath182 , i.e. , they expresses how important are the _ deviations _ @xmath233 of the momentum from the lagrangian manifold @xmath226 . in particular , @xmath234 in view of the normalization condition . in the semiclassical limit it is assumed that the scale length @xmath235 characterizing the range of the momentum deviations is large enough as compared to @xmath236 , namely , @xmath237 . in appendix , it is proved that , within the weak formulation ( [ 2.1bar ] ) , the momentum distribution @xmath182 , satisfying the foregoing conditions , can be represented by the asymptotic series , cf . eq.([a.5 ] ) , @xmath238 controlled by the small parameter @xmath239 . it is worth noting that @xmath240 is thus represented by a distribution which is point supported on the lagrangian manifold @xmath241 and completely determined by its statistical moments @xmath242 . in correspondence of the asymptotic expansion ( [ 2.5 ] ) , eq.([2.1abar ] ) reduces to @xmath243 as shown in details in appendix , cf . , in particular , eq.([a.6 ] ) . formally , eq.([2.6 ] ) constitutes an infinite set of algebraic equations for the statistical moments @xmath244 characterizing the momentum distribution , where each equation , labelled by @xmath23 , is expressed as an asymptotic series in @xmath245 ; the function @xmath142 is determined by imposing that the system ( [ 2.6 ] ) admits nontrivial solutions . equations ( [ 2.6 ] ) are valid for a general momentum distribution @xmath182 which satisfies ( [ 2.1abar ] ) . in particular , on setting @xmath246 for @xmath247 , eq.([2.5 ] ) reduces to @xmath248 which is the geometrical - optics - like solution obtained by bornatici and kravtsov @xcite and by sparber , markowich and mauser @xcite , whereas in ( [ 2.6 ] ) the only nontrivial equation reduces to the geometrical optics eikonal equation @xcite for @xmath142 , namely , @xmath249 . on the basis of the asymptotic expansion ( [ 2.5 ] ) for a momentum distribution , one can prove the main result of this paper , that is , relating the wave kinetic equation to the cgo equations for suitable boundary conditions . first let us consider the specific momentum distribution for which @xmath250 that is , odd - order moments have been set to zero , whereas even - order moments have been related to a single vector field @xmath145 , with @xmath113 an unknown smooth function . correspondingly , the momentum distribution ( [ 2.5 ] ) takes the form @xmath251^{2n } \delta\big(k-\partial_{x}s(x)\big),\end{aligned}\ ] ] the second identity being obtained by means of the multinomial formula @xmath252 with @xmath253 ( no sum over @xmath254 ) . let us note that the momentum distribution ( [ 2.8bar ] ) is symmetric with respect to the lagrangian manifold @xmath241 , as even - order moments only appear ; in particular , the second order moment @xmath255 for @xmath256 is negative , so that such a distribution can not be interpreted as a probability measure . the momentum distribution ( [ 2.8bar ] ) should be multiplied by @xmath225 to get the whole wigner function ( [ 2.1 ] ) . let us consider the case for which @xmath257 ) and @xmath258 is ordered according to @xmath259 with @xmath260 . as a consequence , @xmath235 is the shortest scale length characterizing the wavefield intensity @xmath225 , hence it can be identified with the scale length @xmath110 defined after eq.([1.3 ] ) , namely , @xmath261 and @xmath262 . then one has the following : first , let us prove the statement @xmath263 . in view of the ansatz ( [ 2.7 ] ) , all the equations obtained from ( [ 2.6 ] ) with @xmath23 such that @xmath264 is an even integer reduce to the same equation which reads [ 2.10 ] @xmath265^{2n } d'\big(x , k(x)\big ) = 0,\ ] ] and , analogously , all the equations obtained from ( [ 2.6 ] ) with @xmath23 such that @xmath264 is an odd integer reduce to @xmath266^{2n+1 } d'\big(x , k(x)\big)=0,\ ] ] where the second identity in both equations ( [ 2.10 ] ) follows on using the multinomial formula ( [ 2.9bar ] ) . equations ( [ 2.10 ] ) constitute a set of two coupled equations for the real functions @xmath142 and @xmath143 , and , to lowest significant orders in @xmath147 , they are the same as the cgo equations ( [ 1.5a ] ) and ( [ 1.5b ] ) ; this completes the proof of @xmath263 . as for @xmath267 , on account of the differential calculus for momentum distributions put forward in sec.3 , the term connected with the poisson brackets in eq.([2.1bbar ] ) should be written in the form , cf . eq.([5 m ] ) , @xmath268 using the specific momentum distribution ( [ 2.8bar ] ) for which @xmath269 yields @xmath270 - \left[\frac{\partial d'\big(x , k(x)\big)}{\partial k_i}\frac{\partial a\big(x , k(x)\big)}{\partial x^i}\right . \\ \nonumber & \qquad\qquad \left . -\frac{\partial d'\big(x , k(x)\big)}{\partial x^i } \frac{\partial a\big(x , k(x)\big)}{\partial k_i}\right ] \|\psi(x)\|^2 + o(\epsilon^2)\\ \label{17n } & = \frac{\partial}{\partial x^i } \left[\frac{\partial d'\big(x , k(x)\big)}{\partial k_i } \|\psi(x)\|^2 \right ] a\big(x , k(x)\big ) + o(\epsilon).\end{aligned}\ ] ] the last identity follows on noting that taking the derivative of ( [ 2.10a ] ) with respect to @xmath271 yields @xmath272 with @xmath273 . equation ( [ 17n ] ) implies that @xmath274 \delta\big(k-\partial_{x}s(x)\big ) + o(\epsilon)\ ] ] in the weak sense . hence , from the wave kinetic equation , to lowest order in @xmath147 , one gets the transport equation @xmath275 = 2 d''\big(x , k(x)\big)\ \|\psi(x)\|^{2},\ ] ] which reduces to the cgo transport eq.([1.5c ] ) for @xmath276 , cf . equation ( [ 2.10bar ] ) , on noting that , to lowest significant order , @xmath277 = e^{-2\phi(x)}\ \frac{\partial}{\partial x^{i}}\left[\frac{\partial d'\big(x , k(x)\big)}{\partial k_{i}}\ \|u(x)\|^{2}\right],\ ] ] in view of eq.([2.10b ] ) . this concludes the proof . the foregoing result shows that there exists a specific form of the wigner function for which the wave kinetic equation is reduced to the cgo equations . in order to compare the wavefield intensities predicted by the wave kinetic description with that obtained on solving the cgo equations , one should complete the foregoing argument by discussing cauchy boundary conditions . with reference to ( [ 6n ] ) and ( [ 7n ] ) one has the following : _ let @xmath77 be the weak solution of the wave kinetic equation ( [ 1.1 ] ) corresponding to the cauchy boundary conditions @xmath278 for some smooth functions @xmath279 , @xmath280 and @xmath281 with @xmath282 satisfying the cgo ordering defined after ( [ 12n ] ) , and let @xmath142 , @xmath143 and @xmath258 be solution of the cgo equations ( [ 1.5 ] ) with cauchy boundary conditions given by the same function @xmath283 , @xmath156 and @xmath284 . then @xmath77 can be approximated according to @xmath285 in the weak sense of sec.3 . _ first , let us note that the cauchy data ( [ 20n ] ) is a particular case of ( [ 6nb ] ) which corresponds to @xmath286 in particular , the wigner function corresponding to the complex - eikonal wave object ( [ 12n ] ) can be written in the form ( [ 22n ] ) . in order to prove the foregoing statement , we will make use of the previous result of this section . specifically , we have proved that the wigner function @xmath287 @xmath182 being the momentum distribution ( [ 2.8bar ] ) , solves asymptotically the wave kinetic equation in the weak sense within an @xmath288 accuracy . moreover , @xmath289 in view of ( [ 16bar ] ) . as for the boundary condition ( [ 20n ] ) one gets @xmath290 where @xmath291 . according to ( [ 14n ] ) , @xmath292 , and , thus , @xmath293 in the weak sense , so that @xmath294 matches the boundary conditions ( [ 20n ] ) . since the solution of the wave kinetic equation along with the cauchy boundary condition ( [ 20n ] ) is unique and since @xmath295 is an @xmath288 solution , it follows that @xmath296 which concludes the proof of ( [ 21n ] ) . this implies that , whenever the solutions of both the wave kinetic equation and the cgo equations exist , thus , in particular , the cauchy boundary conditions are of the form ( [ 22n ] ) , the wigner - weyl formalism and the complex geometrical optics method are equivalent within an @xmath288 accuracy . in particular , the wavefield intensity predicted by the wigner - weyl kinetic formalism is the same as that predicted by the cgo method , namely , @xmath297 ) . in the next section , an analytically tractable case is considered as an example . specifically , the solution of the wave kinetic equation relevant to the paraxial propagation of a gaussian wave beam in a `` lens - like '' medium is obtained and shown to be the same as the corresponding cgo solution . let us address the case of a monochromatic @xmath298 beam of electromagnetic waves propagating in a _ loss less _ `` lens - like '' medium @xcite with _ real _ refractive index @xmath299^{\frac{1}{2}}$ ] . it is assumed that the wavefield is localized near the axis @xmath300 of the medium , that is , @xmath301 ; moreover , the wave electric field is written in the form @xmath302 , i.e. , it is polarized along the @xmath303 axis and propagates in the @xmath64-@xmath28 plane . the relevant wave equation for the wavefield real amplitude @xmath304 is thus the helmholtz equation . the corresponding weyl symbol is real valued and given by @xmath305,\ ] ] thus , the dispersion relation @xmath306 yields two branches , to be referred to as the progressive and the regressive waves . as for the cauchy boundary conditions , let us assume that the wavefield is purely gaussian at @xmath307 , i.e. , @xmath308 $ ] , @xmath309 being the initial width , and the propagation occurs along the @xmath28 axis , so that one should solve the dispersion relation @xmath306 for @xmath310 . on assuming that each branch of the dispersion relation carries half of the wavefield intensity , one can consider the progressive wave only which reads @xmath311 where @xmath312 is the wavevector at @xmath300 and the paraxial approximation @xmath313 has been exploited as relevant to the weak - diffraction regime @xcite . it is convenient noting that the dispersion relation corresponding to the second form of ( [ 3.1 ] ) can be written as @xmath314 which is formally analogous to the dispersion relation relevant to a quantum harmonic oscillator @xcite with unit mass and @xmath315 , @xmath316 , @xmath317 being the characteristic frequency of the oscillator , @xmath318 and @xmath319 . in particular , the frequency @xmath320 corresponds to the shifted wavevector @xmath321 along the propagation direction @xmath28 . the shift occurs because of the oscillations of the wavefield along the propagation direction @xmath28 . this analogy allows to make use of the well - known solution of the wave kinetic equation for the quantum harmonic oscillator @xcite to describe the paraxial propagation of a gaussian beam in the lens - like " medium . more specifically , the solution of the wave kinetic equation for the harmonic oscillator corresponding to an initially gaussian wave packet @xmath322 $ ] is @xcite [ 3.2pre ] @xmath323\ w_{0}^{2 } , \end{gathered}\ ] ] where @xmath324 is the width of the wave packet as a function of time and @xmath325 , @xmath51 being the mass of the oscillator and @xmath326 the initial displacement of the gaussian from the centre of the elastic force acting on the oscillator . correspondingly , the solution for the wave electric field intensity in the `` lens - like '' medium , with the considered launching conditions , is [ 3.2 ] @xmath327w_{0}^{2 } = \left[1+\big(\big(\frac{l}{z_{r}}\big)^{2}-1\big)\sin^{2}(z / l)\right ] w_{0}^{2},\end{gathered}\ ] ] with @xmath328 the rayleigh range in the medium . in eq.([3.2b ] ) , it has been explicitly indicated that the solution obtained from the wave kinetic equation amounts to the averaged intensity @xmath329 , rather than to the exact value @xmath330 since two branches of the dispersion relation exist each one carrying half of the wavefield intensity , cf . comments after eqs.([7nprime ] ) . the intensity ( [ 3.2a ] ) and the beam width ( [ 3.2b ] ) are the same as the corresponding quantities obtained from the cgo solution @xcite . as a consequence the intensity profile ( [ 3.2 ] ) accounts for diffraction effects as shown in fig.1 . one can conclude that , according to results of sec.4 , the _ kinetic formalism can be used to describe the effects of diffraction on the propagation of wave beams _ , and , for the case under consideration , it yields the same result as the cgo method . nevertheless , the detailed structure of the wavefield , i.e. , the oscillations along the propagation direction @xmath28 , the effects of the curvature of phase fronts and the gouy shift , which are available from the cgo solution @xcite , _ can not be resolved by means of the wave kinetic equation _ , which instead gives the averaged intensity distribution . within the framework of semiclassical wave propagation , two specific asymptotic techniques have been considered , namely , the wigner - weyl kinetic formalism and the complex geometrical optics ( cgo ) method . a detailed comparative analysis of these techniques has been given in sec.2 , for the case of scalar pseudodifferential wave equations , with cauchy boundary conditions . in particular , in the wigner - weyl formalism , the wavefield is represented in the phase space by the wigner function which is a solution to the wave kinetic equation . in the most general case the wigner function amounts to a tempered distribution . however , physical considerations lead to the definition of a novel class of weak solutions which have been characterized in sec.3 . such specific weak solutions are referred to as _ momentum distributions _ since , for every point location @xmath64 in the configuration space , they give the distribution of momentum @xmath16 over @xmath64 , in the @xmath64-@xmath16 phase space . on the other hand , the cgo method yields an asymptotics solution of a pseudodifferential wave equation directly in the configuration space , in terms of three smoothly varying functions representing , the phase , the wavefield envelope and the amplitude , respectively . in sec.4 , on the basis of the mathematical framework developed in sec.3 , we have proved that , whenever both the solutions of the wave kinetic equation and of the cgo equations exist , thus , in particular , the cauchy data are of the form ( [ 22n ] ) , the former can be approximated by a momentum distribution , cf . , equation ( [ 21n ] ) , written in terms of the three smoothly varying functions that solve the cgo equations ( [ 1.5 ] ) . as a consequence , the two considered techniques are asymptotically equivalent and , in particular , to lowest significant order , the wavefield intensity predicted by the wigner - weyl formalism is the same as that predicted by the cgo method , cf . eq.([intensity ] ) . in addition , one can conclude that the wigner - weyl kinetic formalism properly describes the wavefield near focal points . this is also shown by comparing the solution of the kinetic equation to that of the cgo equations for a specific case , namely , the propagation of electromagnetic gaussian wave beams in an isotropic `` lens - like '' medium , cf . in particular , the relevant solution of the wave kinetic equation has been obtained on the basis of the analogy between the `` lens - like '' medium and the quantum harmonic oscillator . the author is grateful to m. bornatici for many useful discussions , suggestions and for carefully reading the manuscript . for the revised version of this paper , valuable discussions with c. dappiaggi and a. orlandi are gratefully acknowledged . this work was supported by the italian ministry of university scientific research and technology ( murst ) and the italian institute for the physics of matter ( infm ) . in this appendix the asymptotic series expansion ( [ 2.5 ] ) of the momentum distribution is proved and the corresponding equations ( [ 2.6 ] ) are derived from the weak form ( [ 2.1abar ] ) of the dispersion relationship ( [ 1.1a ] ) . since both @xmath331 and @xmath67 are symbols , they are , in particular , smooth functions , and one can apply the taylor s formula @xmath332 @xmath333 being an @xmath1-dimensional multi - index and @xmath334 the remainder of order @xmath335 relevant to the expansions of @xmath167 and @xmath39 , respectively . more specifically , on making use of ( [ a.2 ] ) to evaluate the left - hand side of ( [ 2.1abar ] ) , one gets @xmath336 where @xmath337 are the statistical moments of the momentum distribution @xmath338 , c.f . eq.([2.3 ] ) . in virtue of the symbol estimate ( [ 2n ] ) , symbols are such that , e.g. , @xmath339 in the semiclassical limit @xmath340 uniformly in @xmath64 , hence , the asymptotic series expansion ( [ a.3 ] ) is controlled by the ( small ) parameter @xmath341 . moreover , on noting that @xmath342 eq.([a.3 ] ) takes the form @xmath343 a(x , k)\ d^{n}k\end{gathered}\ ] ] and , in view of the arbitrariness of @xmath67 , one gets @xmath344 in the weak sense . it is worth noting that the derivation of ( [ a.4 ] ) does not depend on the explicit form of the symbol @xmath331 , thus , on setting @xmath345 , eq.([a.4 ] ) reduces to @xmath346 which is just the general asymptotic expansion ( [ 2.5 ] ) of the momentum distribution . going back to eq.([2.1abar ] ) , its solution is obtained on setting the expansion ( [ a.4 ] ) to zero and exploiting the linear independence of the derivatives of the dirac s @xmath192-function , thus yielding a set of equations for the statistical moments , namely , @xmath347 which is just eq.([2.6 ] ) . it is worth noting that eq.([a.6 ] ) can be also obtained on substituting ( [ a.5 ] ) into ( [ 2.1abar ] ) and exploiting the leibniz s formula @xmath348 which expresses the derivative of a product to any orders . v. guillemin and s. stenberg , _ geometric asymptotics _ ( amer . math survey , amer . soc . , new york 1976 ) . j. j. duistermaat , comm . pure appl . vol.xxvii , 207 ( 1974 ) . a. martinez , _ an introduction to semiclassical and microlocal analysis _ ( springer - verlag , new york , 2002 ) . a. kravtsov and yu . i. orlov , _ geometrical optics of inhomogeneous media _ ( springer - verlag , berlin , 1990 ) . r. g. littlejohn and w. g. flynn , phys . a * 44 * , 5239 ( 1991 ) . a. kravtsov and yu . i. orlov , _ caustics , catastrophes and wave fields _ ( springer - verlag , berlin , 1999 ) . j. ehlers and e. t. newman , _ j. math . phys . _ * 41 * , 3344 ( 2000 ) , and references therein . j. f. nye , _ natural focusing and fine structure of light _ ( institute of physics publishing , bristol , 1999 ) and references therein . v. guillemin and s. stenberg , _ symplectic techniques in physics _ ( cambridge university press , cambridge , 1984 ) . t. stix _ plasma waves _ ( american institute of physics , new york , 1992 ) . s. w. mcdonald and a. n. kaufman , phys . a * 32 * , 1708 ( 1985 ) . r. g. littlejohn , phys . rep . * 138 * , 193 ( 1986 ) . s. w. mcdonald , phys . * 158 * , 337 ( 1988 ) . s. choudhary and l. b. felsen , proc . of the ieee * 62 * , 1530 ( 1974 ) . a. kravtsov , g. w. forbes and a. a. asatryan , in progress in optics vol.xxxix ( 1999 ) , pp.1 - 61 . a. bravo - ortega and a. h. glasser , phys . fluids b * 3 * , 529 ( 1991 ) . e. mazzucato , phys . fluids b * 1 * , 1855 ( 1989 ) . m. e. taylor , _ partial differential equations ii : qualitative studies of linear equations _ , springer , new york ( 1996 ) . l. hrmander , _ lectures on nonlinear hyperbolic differential equations _ ( springer - verlag , berlin heidelberg , 1997 ) , section 8.4 . j. j. duistermaat , _ fourier integral operators _ , birkhuser , boston ( 1996 ) . l. hrmander , _ the analysis of linear partial differential operators i : distribution theory and fourier analysis _ , springer , berlin ( 2003 ) . o. maj and m. bornatici , _ proc . of the 30@xmath349 eps conf . on controlled fusion and plasma physics _ eca * 27a * , p-2.1 ( 2003 ) . m. bornatici and yu . a. kravtsov , plasma phys . controlled fusion * 42 * , 255 ( 2000 ) . c. sparber , p. a. markowich and n. j. mauser , asymptotic analysis * 33 * , 153 ( 2003 ) . m. bornatici and o. maj , plasma phys . controlled fusion * 45 * , 707 ( 2003 ) .	the relationship between two different asymptotic techniques , namely , the wigner - weyl kinetic formalism and the complex geometrical optics method , is addressed within the framework of semiclassical theory of wave propagation . more specifically , in correspondence to appropriate boundary conditions , the solution of the wave kinetic equation , relevant to the wigner - weyl formalism , is obtained in terms of the corresponding solution of the complex geometrical optics equations . in particular , this implies that the two considered techniques yield the same wavefield intensity . such a result is also discussed on the basis of the analytical solution of the wave kinetic equation specific to gaussian beams of electromagnetic waves propagating in a `` lens - like '' medium for which the complex geometrical optics solution is already available .
the temporal coherence of a source of radiation is a key quantity which discriminates a laser - like device from thermal emitters . while the first order coherence function , @xmath0 , which is the correlator of field amplitudes at different times , reflects the coherence of the emitted photons , its second order counterpart @xmath1 , that involves intensity correlation , gives insights into the emission statistics . in conventional semiconductor lasers which rely on stimulated emission and population inversion , the emission statistics transits from thermal below lasing threshold to coherent as revealed by @xmath2 and 1 respectively . even in the limiting case of a vanishing threshold occurring in devices where spontaneous emission is perfectly funneled into the lasing mode , laser oscillation can still be identified since @xmath3 approaches a value of 1 with increasing pump power @xcite . an example of a light emitting device which may operate as a coherent light source without population inversion , is a semiconductor microcavity in the strong light matter coupling regime @xcite . the strong coupling results in the emergence of hybrid eigen modes called exciton - polaritons ( later , _ polaritons _ ) @xcite . being very low - mass bosons , they can undergo a dynamic bose - einstein condensation ( bec ) process at elevated temperatures . analogous to a conventional laser operating in the weak coupling regime ( for instance the vertical cavity surface emitting laser , vcsel ) , the formation of a bec of polaritons is accompanied by a nonlinear increase of the intensity of the emitted light and a drop of the spectral linewidth @xcite.the latter is a typical , yet not unambiguous signature of first order temporal coherence of the emitted radiation . a more sophisticated approach relying on michelson interferometry was discussed in @xcite , where strongly enhanced coherence times in the regime of polariton lasing have been demonstrated by using low noise pump sources . in analogy to cold atom becs , the first order spatial coherence ( long range order ) , has been considered as a smoking gun criterion for the claim of a polariton bec @xcite . however , the second order coherence function which represents another key signature of coherent light , is less well understood . one difficulty arises from the lack of accurate theoretical quantum description of the behavior of @xmath4 in real conditions , such as finite temperatures . indeed , common methods are based on stochastic gross - pitaevskii @xcite equations or expectation values evolution @xcite that rely on severe assumptions regarding high order correlations such as coherence functions . additionally , the @xmath5 dependence could not be addressed by semiclassical models . as a consequence of these complications , reports on the second order coherence function in polariton systems have not been fully conclusive in unified experimental theoretical works . a second difficulty is based on the results of present experimental works : here , in most cases , a surprisingly slow drop down of @xmath3 above threshold is commonly observed in two - dimensional microcavities @xcite with a persisting value significantly above unity , thus implying a strong deviation from a poissonian photon statistics . even an increase of @xmath3 with increasing pump power has been reported , both for gaas and cdte based samples @xcite . it is reasonable to assume that these peculiarities are a result of a large number of states in the continuous dispersion of polaritons confined in planar microcavities , which can contribute to the condensation phenomena and unavoidably lead to mode competition effects . such effects further convolute the coherence phenomena inherent to the polariton condensation , and therefore are considered as another main obstacle towards a precise understanding of the evolution of second - order temporal coherence in solid state condensates beyond a phenomenological level . in this article we present a systematic combined experimental and theoretical study of the power and temperature dependence of the @xmath6 function both in confined and planar strongly coupled microcavity structures . spatial confinement has been used to enhance condensation processes in trap structures and micropillars@xcite with a good overview given in @xcite . in our case spatial confinement in etched micropillars allows us to form a single mode polariton condensate , thus @xmath3 converges towards unity for large pump powers . our results are supported by a stochastic quantum trajectory approach which provides unprecedented insight into the interplay between state occupation , particle fluctuations , and photon coherence . in particular , we show that accounting for dissipative polariton - phonon interaction and polariton self interactions in a fully quantum manner is sufficient to retrieve the ( resolution limited ) @xmath3 behavior in confined structures where a discrete dispersion assists the condensation . the sample which we study is a high - q algaas alloy - based @xmath7 planar microcavity with twelve gaas quantum wells ( qws ) of 13 nm width each located in the optical antinodes of the confined electromagnetic field . the design of the sample is similar to the one described in @xcite . in order to enhance the q - factor , we increased the number of algaas / alas mirrorpairs in the top(bottom ) distributed bragg reflector ( dbr ) segment to 23(27 ) . the q - factor was experimentally estimated to exceed 10000 for highly photonic structures . to provide lateral confinement of the polariton modes , we fabricated micropillars with the diameters 6 @xmath8 m . electron beam lithography and electron - cyclotron - resonance reactive - ion - etching were used for deeply etching the cavity . a scanning electron microscope ( sem ) image of the sample is shown in fig . [ fig1](a ) . we also extracted the rabi - splitting of the device via white light reflection and found a value of 10.1 mev for the heavy hole - based exciton @xcite . the sample is pumped with a pulsed ti : sa laser tuned to the first bragg minimum of the stop band located approximately 100 mev above the free exciton resonance . the pulse width was 50 ps . it is worth noting , that in most of the previous studies devoted to the second order correlation function in polariton systems , very short pumping pulses of 1 - 3 ps have been used . on one hand , short pulsing is advantageous over long - pulse excitation schemes since conventional single photon detectors with time resolution of the order of hundreds ps can be used to extract the second order correlation function at zero time delay @xmath3 @xcite . the finite length of the emission pulse effectively acts as a time filter in this configuration @xcite as only photons which are emitted within the lifetime of the condensate and reservoir are correlated . on the other hand , it has been realized by now that fundamental behavior of a polariton laser ( in terms of its linewidth and emission frequency ) is drastically modified by the ultra short excitation . it results in significant line broadenings and energy shifts @xcite , which strongly affects the coherence properties . as our pulse width is far below the resolution of our detectors of 400 ps and still much shorter than the pulse repetition period of around 12 ns , our excitation scheme can be considered as a compromise between reasonable time resolution and keeping equilibrium conditions in the experiment . for excitation beam diameters smaller than the pillar width , we gain high sensitivity of the output to the spatial position of the laser leading to excitement of desired optical modes @xcite . however , using narrow beams breaks the homogeneity of the pumping scheme . therefore small pumping spots lead to a repulsive potential driving the polaritons away from the center , i.e . out of the ground state , causing condensation at high energies and @xmath9 vectors @xcite . to suppress this effect and ensure a homogeneous excitation of the sample , we expanded the beam diameter to @xmath10 m . this value is more than six times bigger than the pillar width . moreover , to further improve the homogeneity of the excitation , a closed optical aperture and a lense were used for beamshaping . the dispersions were recorded using a 4f fourier setup @xcite . tight optical confinement provided by the micropillar landscape gives rise to a characteristic set of discrete optical modes in the lower polariton dispersion , see fig . [ fig1](b ) . on top of the mode spectrum , we detect an emission band around 1.545 ev which we attribute to radiation of free excitons which scatters through the sidewalls of the pillars . in contrast , for the case of the planar ( not etched ) structure , the dispersion has a continuous parabolic shape , see fig . [ fig1](d ) , and the free exciton radiation is completely suppressed . we also note , that the energy of free excitons is somewhat blue shifted compared to the planar sample as a result of the modified strained environment . above a distinct threshold occuring at pump powers around 45 w/@xmath11 , both for the planar and the structured sample , we observe a massive occupation of the polariton ground state at the bottom of the lower polariton branch , indicating the formation of a polariton condensate . the spectra in fig . [ fig1](c , e ) were recorded at injection powers of ca . 50 w/@xmath11 . additionally the dispersion for @xmath12 ( lower row ) and @xmath13k ( upper row ) for different excitation powers are shown in fig . the transition from the mode structure emission for low powers ( ( a ) and ( d ) ) to the slight broadening of the fundamental mode shortly before threshold ( ( b ) and ( e ) ) and the ground state condensation above threshold ( ( c ) and ( f ) ) can be seen for both temperatures . in order to give unambiguous evidence for polariton lasing in our quasi - zero - dimensional system , we carefully investigate the luminescence from the ground state of a pillar with a detuning @xmath14 mev at a sample temperature of 40 k. the dispersion is integrated from @xmath15 m to @xmath16 m and fitted by a lorentzian . we observe a distinct threshold behaviour in the input - output - characteristic with a non - linearity occuring at 45 w/@xmath11 , see fig . [ fig3](a ) . while the energy of the ground state features a linear blueshift with increasing pump power below threshold ( caused by interaction with the exciton reservoir ) , the blueshift becomes logarithmic as soon as the threshold is reached . this behaviour displayed in fig . [ fig3](c ) is in agreement with data reported in @xcite and indicates that repulsive polariton - polariton interactions play a dominant role . furthermore , we clearly observe narrowing of the emission linewidth ( fwhm ) at the threshold , also shown in fig . [ fig3](c ) indicating the transition from thermal to coherent emission of light . the behaviour for the pillar at @xmath12 k is similar and therefore not shown . [ fig3](b ) and ( d ) show analogous data for the planar measurement . the distinct threshold in fig . [ fig3](b ) , accompanied by the continuous blueshift above and the drop of the linewidth at the threshold in fig . [ fig3](d ) prove polaritonic condensation also in the planar regime . further , we study the temporal second order correlation function , @xmath17 where @xmath18 is the emission intensity at time t and @xmath19 means time averaging . for the photon statistics measurements , we use a hanbury - brown and twiss configuration with two avalanche photodiodes having a time resolution of 400 ps each @xcite . the light emitted from the sample is filtered by a monochromator with a resolution of 0.2 mev to avoid errors caused by interaction with higher - energy states . figure [ fig4](a ) shows a zoom onto the @xmath20 peak for the planar sample ( black ) and the pillar sample ( red ) far above ( @xmath21 ) threshold . the difference in height of the @xmath20 peak visualizes the difference in the @xmath6 value . the inset shows the corresponding normalized @xmath6 measurements . the data has been normalized to the average of the side peaks . figure [ fig4](b ) analogously shows the same behaviour for @xmath12k ( red ) and @xmath13k ( black ) measured on the pillar sample . the inset clearly shows a @xmath6 value bigger than @xmath22 for both measurements . peak for the planar sample ( black ) and pillar sample ( red ) taken at @xmath21 . inset shows corresponding normalized @xmath6 measurements ( b ) zoom onto the @xmath20 peak for @xmath13k ( black ) and @xmath12k ( red ) taken at @xmath23 on the pillar sample . inset shows corresponding normalized @xmath6 measurements ( a ) and ( b ) visualize the difference in the @xmath6 value . ( c ) comparison between pillar ( red circles ) and planar ( black squares ) emission : for large pump powers , @xmath3 drops significantly faster in the micropillar as a result of spatial confinement . ( d ) @xmath3 dependence on the pumping power for the pillar sample at 5k ( red circles ) and 40k ( black squares ) with fit ( line ) . the enhanced phonon effects for higher temperatures cause a slower drop of the @xmath3-value . ] figure [ fig4](c ) shows the power dependence of @xmath3 in the planar microcavity and the pillar structure for a detuning of @xmath24 . below threshold one would expect @xmath25 for thermal emission . since the temporal resolution of the measurements exceeds the temporal correlation time of the emitter this can not be resolved leading to @xmath26 @xcite . the increase of the central peak around @xmath27 at the threshold can be attributed to an enhanced two - photon emission probability and indicates that a fully coherent state is not yet formed . far above threshold the system has reached a coherent state represented by @xmath26 . to estimate the actual value of @xmath28 , the integrated intensity of the @xmath20 peak , @xmath29 , is divided by the average intensity of the @xmath30 peak , @xmath31 , where @xmath32 is the number of side peaks . in the planar case , @xmath3 is significantly increased at large excitation powers , and remains almost constant at a value around @xmath33 for increasing pump power . this is in qualitative agreement with previous reports @xcite . the effect has been assigned to a reservoir depletion triggered by polariton self - scattering from the ground state into the higher - energy states . an alternative interpretation of such behavior reveres to multimode polariton lasing from the ground state and existence of a finite set of emitting @xmath9-values , which leads to an increase of @xmath3 . in the structure with pillars , both of the above mentioned effects should be strongly suppressed due to the discrete mode spectrum and optical confinement . as shown in fig . [ fig3 ] , the pillar device exhibits single - mode lasing on the fundamental mode . the mode spacing from the ground state to the next higher state amounts to around @xmath34mev . this should drastically reduce the condensate depletion effect at cryogenic temperatures @xcite . we directly evidence this improvement by observing a significantly faster drop of the @xmath3 towards unity in the micropillar structure , see fig . [ fig4](c ) . we have investigated the power dependent behaviour of @xmath3 for temperatures of 5 and 40 k in fig . [ fig4](d ) in the same @xmath35 m pillar device . with increasing temperature , the drop of @xmath3 towards 1 successively becomes slower , which suggests a phonon induced condensate depletion effect , leading to enhanced number of fluctuations even in the presence of optical confinement . the system is described by the hamiltonian @xmath36 the first two terms : free propagation term , @xmath37 , and polariton - polariton elastic scattering term , @xmath38 , encompass the coherent processes . the last two terms namely the coupling to a thermal bath of accustic phonons , @xmath39 , and the nonresonant excitation , @xmath40 , are written using the rotating wave approximation and correspond to incoherent processes . the incoherent processes are treated through a monte carlo wave function approach @xcite given the corresponding quantum jump operators @xmath41 } \hat a_{\mathbf{k}},\\ \label{j3 } \hat { \cal j}_{{{\mathbf{k}}_1}{{\mathbf{k}}_2}}^{+ } & = & { \sqrt { \gamma _ { { { \mathbf{k}}_1}{{\mathbf{k}}_2}}^\textrm{ph}{{\bar n}_\textrm{ph } } ( { { { e_{{{\mathbf{k}}_1 } } } - { e_{{{\mathbf{k}}_2 } } } } } ) } } \hat a_{{{\mathbf{k}}_1}}^\dag { \hat a_{{{\mathbf{k}}_2}}},\\ \label{j4 } \hat { \cal j}_{{{\mathbf{k}}_1}{{\mathbf{k}}_2}}^{- } & = & { \sqrt { \gamma _ { { { \mathbf{k}}_1}{{\mathbf{k}}_2}}^\textrm{ph}\left [ { { { \bar n}_\textrm{ph } } ( { { { e_{{{\mathbf{k}}_1 } } } - { e_{{{\mathbf{k}}_2 } } } } } ) + 1 } \right ] } } { \hat a_{{{\mathbf{k}}_1}}}\hat a_{{{\mathbf{k}}_2}}^\dag,\end{aligned}\ ] ] where @xmath42 are annihilation operators of the polariton modes , @xmath43 is a wave vector - dependent lifetime of polaritons , @xmath44 is a pumping source destribution , @xmath45 is a polariton dispersion , @xmath46 , @xmath47 is a polariton - phonon scattering rate , and @xmath48 is a phonon bath destribution . using the experimental parameters , we simulate the quantum dynamics of our system initially prepared in a vacuum state . in brief , the procedure consists in evolution of the system wave function through the schrdinger equation , @xmath49 with an effective hamiltonian @xmath50 where the non - hermitian parts results in a decay of the norm @xmath51 from unity . further , a random number , @xmath52 , is generated . if during the evolution by eq . ( [ sch ] ) , the condition @xmath53 is met , a jump occurs . after each jump , the state @xmath54 should be renormalized and a new number @xmath52 generated . each realization , @xmath55 , of this protocol yields a quantum trajectory , @xmath56 , where @xmath57 , @xmath58 . later , we approximate the density matrix of the system as @xmath59 where @xmath60 is the density matrix of the system . the approximate expectation value , @xmath61 , of any system observable @xmath62 can be found from @xmath63 \mathop{\rightarrow}\limits_{n \to \infty } { \rm{tr}}\left [ { \hat o\hat { { \rho}}(t ) } \right ] = \langle { \hat { o}\left ( t \right ) } \rangle .\ ] ] in our computations we account for 2 excited states in addition to the ground state in a truncated hilbert space @xcite of sufficient dimension to avoid particle number saturation for the pump power we consider . to mimic the finite resolution @xmath64 of the hbt setup , we reconstruct the @xmath1 function from the full ground state decay statistics imposed by the action of the operator ( [ j1 ] ) on the ground state via @xmath65 . then we reconstruct the resolution limited @xmath3 as @xmath66 as shown in fig.[fig4](d ) that compares theory to experiment for 2 different temperature , we were able to reproduce the drop of @xmath3 towards 1 quantitatively for the entire investigated temperature range . the discrete dispersion strongly assists the onset of coherence in the system for the reason that a polariton goes down the dispersion ( via phonon emission ) at a rate proportional to @xmath67 while the probablity of going up ( via phonon absorption ) is proportional to @xmath68 . indeed , given the relatively large gaps of about 0.8 mev imposed by the lateral confinement ( see fig.[fig1](b ) ) the average number of phonons available for such transitions remains much lower than unity and a polariton reaching the ground state can hardly thermalize especially at large pump powers . the saturation of the @xmath69 to a value of about 1.4 results from the averaging imposed by the hbt resolution . in our simulation before averaging we systematically obtain @xmath70 or even slightly above at the peak around threshold . noteworthy , our model is additionally able to reproduce @xmath71 far below threshold which results from the absence of up stimulation process given the low polariton occupation involved . finally we underline that the quantitative level discrepancies between theory and experiment could be wiped out adding a new set of quantum jump operators accounting for a weak pure dephasing in the system @xcite and therefore slightly reducing the coherence . these results underline the power of the quantum jump approach to our discrete mode polariton scheme and support our conclusion that the presence of optical confinement strongly improves the coherence of our polariton laser and partly protects the device from phonon induced condensate depletion effects for higher temperatures . we have evaluated the effect of optical confinement and the lattice temperature on the temporal coherence function behavior of a polariton condensate . experimentally , we observe a significantly faster drop of @xmath72 towards unity above the polariton laser threshold , compared to planar polariton microcavity reference stuctures . by combining our experimental data with a microscopic model based on the quantum jump approach , we were able to successfully describe the influence of phonons on the coherence evolution in our polariton devices , which is of paramount important for the design of next generation coherent polariton light sources . the authors would like to thank the state of bavaria for financial support . we thank monika emmerling and adriana wolf for expert sample processing , marco de gregorio for experimental support , and m. mttnen for help with the theory . i.g.s . acknowledges financial support from the academy of finland through its centre of excellence programs ( projects no . 250280 and no . 251748 ) ; the government of russian federation , grant 074-u01 ; and the dynasty foundation . this publication was funded by the german research foundation ( dfg ) and the university of wrzburg in the funding programme open access publishing .	we study the second order temporal coherence of the emission from a semiconductor microcavity in the strong coupling regime . we evidence the favorable influence of spatial confinement , realized by etching micropillar structures , on the temporal coherence of solid state quasi - condensates which evolve in our device above threshold . by fitting the experimental data with a microscopic quantum theory based on a monte carlo wavefunction approach , we scrutinize the influence of the crystal lattice temperature ( interaction with acoustic phonons ) and pump power on the condensate s temporal coherence . phonon - mediated transitions in the optical mode structure are observed , which in the case of a cofined structure splits into a set of discrete resonances . by increasing the pump power beyond the condensation threshold , the temporal coherence significantly improves in the pillar devices , as revealed in the transition from thermal to coherent statistics of the emitted light . 10 s. strauf and f. jahnke , `` single quantum dot nanolaser , '' laser & photonics reviews * 5 * , 607633 ( 2011 ) . a. imamoglu , r. j. ram , s. pau , and y. yamamoto , `` nonequilibrium condensates and lasers without inversion : exciton - polariton lasers , '' phys . rev . a * 53 * , 42504253 ( 1996 ) . c. weisbuch , m. nishioka , a. ishikawa , and y. arakawa , `` observation of the coupled exciton - photon mode splitting in a semiconductor quantum microcavity , '' phys . rev . lett . * 69 * , 33143317 ( 1992 ) . a. v. kavokin , j. j. baumberg , g. malpuech , and f. p. laussy , _ microcavities _ ( oxford university press , 2006 ) . a. p. d. love , d. n. krizhanovskii , d. m. whittaker , r. bouchekioua , d. sanvitto , s. a. rizeiqi , r. bradley , m. s. skolnick , p. r. eastham , r. andreacute , and l. s. dang , `` intrinsic decoherence mechanisms in the microcavity polariton condensate , '' phys . rev . lett . * 101 * , 067404 ( 2008 ) . b. deveaud - pldran , `` on the condensation of polaritons , '' j. opt . soc . am . b * 29 * , a138a145 ( 2012 ) . j. kasprzak , m. richard , s. kundermann , a. baas , p. jeambrun , j. m. j. keeling , f. m. marchetti , m. h. szymaska , r. andr , j. l. staehli , v. savona , p. b. littlewood , b. deveaud , and l. s. dang , `` bose - einstein condensation of exciton polaritons , '' nature * 443 * , 409414 ( 2006 ) . h. deng , d. press , s. gtzinger , g. s. solomon , r. hey , k. h. ploog , and y. yamamoto , `` quantum degenerate exciton - polaritons in thermal equilibrium , '' phys . rev . lett . * 97 * , 146402 ( 2006 ) . j. fischer , i. g. savenko , m. d. fraser , s. holzinger , s. brodbeck , m. kamp , i. a. shelykh , c. schneider , and s. hfling , `` spatial coherence properties of one dimensional exciton - polariton condensates , '' phys . rev . lett . * 113 * , 203902 ( 2014 ) . m. wouters and v. savona , `` stochastic classical field model for polariton condensates , '' phys . rev . b * 79 * , 165302 ( 2009 ) . h. haug , t. doan , h. t. cao , and d. t. thoai , `` temporal first - and second - order correlations in a polariton condensate , '' phys . rev . b * 85 * , 205310 ( 2012 ) . i. g. savenko , t. c. h. liew , and i. a. shelykh , `` stochastic gross - pitaevskii equation for the dynamical thermalization of bose - einstein condensates , '' phys . rev . lett . * 110 * , 127402 ( 2013 ) . i. g. savenko , e. b. magnusson , and i. a. shelykh , `` density - matrix approach for an interacting polariton system , '' phys . rev . b * 83 * , 165316 ( 2011 ) . h. deng , g. weihs , d. snoke , j. bloch , and y. yamamoto , `` polariton lasing vs. photon lasing in a semiconductor microcavity , '' proc . natl . acad . sci . usa . * 100 * , 1531815323 ( 2003 ) . j .- s . tempel , f. veit , m. amann , l. e. kreilkamp , a. rahimi - iman , a. lffler , s. hfling , s. reitzenstein , l. worschech , a. forchel , and m. bayer , `` characterization of two - threshold behavior of the emission from a gaas microcavity , '' phys . rev . b * 85 * , 075318 ( 2012 ) . a. rahimi - iman , a. v. chernenko , j. fischer , s. brodbeck , m. amthor , c. schneider , a. forchel , s. hfling , s. reitzenstein , and m. kamp , `` coherence signatures and density - dependent interaction in a dynamical exciton - polariton condensate , '' phys . rev . b * 86 * , 155308 ( 2012 ) . j. kasprzak , m. richard , a. baas , b. deveaud , r. andre , j .- p . poizat , and l. s. dang , `` second - order time correlations within a polariton bose - einstein condensate in a cdte microcavity , '' phys . rev . lett . * 100 * , 067402 ( 2008 ) . t. horikiri , p. schwendimann , a. quattropani , s. hfling , a. forchel , and y. yamamoto , `` higher order coherence of exciton - polariton condensates , '' phys . rev . b * 81 * , 033307 ( 2010 ) . g. dasbach , m. schwab , m. bayer , d. krizhanovskii , and a. forchel , `` tailoring the polariton dispersion by optical confinement : access to a manifold of elastic polariton pair scattering channels , '' phys . rev . b * 66 * , 201201 ( 2002 ) . d. bajoni , p. senellart , e. wertz , i. sagnes , a. miard , a. lemaitre , and j. bloch , `` polariton laser using single micropillar gaas - gaalas semiconductor cavities , '' phys . rev . lett . * 100 * , 047401 ( 2008 ) . t. klein , s. klembt , e. durupt , c. kruse , d. hommel , and m. richard , `` polariton lasing in high - quality selenide - based micropillars in the strong coupling regime , '' appl . . lett . * 107 * , 071101 ( 2015 ) . c. schneider , k. winkler , m. d. fraser , m. kamp , y. yamamoto , e. a. ostrovskaya , and s. hfling , `` exciton - polariton trapping and potential landscape engineering , '' arxiv:1510.07540 ( 2015 ) . h. deng , g. weihs , c. santori , j. bloch , and y. yamamoto , `` condensation of semiconductor microcavity exciton polaritons , '' science * 298 * , 199202 ( 2002 ) . j .- s . tempel , f. veit , m. amann , l. kreilkamp , m. bayer , a. rahimi - iman , a. lffler , s. hfling , s. reitzenstein , l. worschech , and a. forchel , `` distinguishing photon and polariton lasing from gaas microcavities by spectral and temporal analysis of the two - threshold behavior , '' arxiv:1101.5316 ( 2011 ) . v. kalevich , m. afanasiev , v. lukoshkin , d. solnyshkov , g. malpuech , k. kavokin , s. tsintzos , z. hatzopoulos , p. savvidis , and a. kavokin , `` controllable structuring of exciton - polariton condensates in cylindrical pillar microcavities , '' arxiv:1411.3222 ( 2014 ) . l. ferrier , e. wertz , r. johne , d. d. solnyshkov , p. senellart , i. sagnes , a. lemaitre , g. malpuech , and j. bloch , `` interactions in confined polariton condensates , '' phys . rev . lett . * 106 * , 126401 ( 2011 ) . c. w. lai , n. y. kim , s. utsunomiya , g. roumpos , h. deng , m. d. fraser , t. byrnes , p. recher , n. kumada , t. fujisawa , and y. yamamoto , `` coherent zero - state and pi - state in an exciton - polariton condensate array , '' nature * 450 * , 529532 ( 2007 ) . g. roumpos , m. lohse , w. h. nitsche , j. keeling , m. h. szymaska , p. b. littlewood , a. lffler , s. hfling , l. worschech , a. forchel , and y. yamamoto , `` power - law decay of the spatial correlation function in exciton - polariton condensates , '' proc . natl . acad . sci . usa * 109 * , 64676472 ( 2012 ) . s. m. ulrich , c. gies , s. ates , j. wiersig , s. reitzenstein , c. hofmann , a. lffler , a. forchel , f. jahnke , and p. michler , `` photon statistics of semiconductor microcavity lasers , '' phys . rev . lett . * 98 * , 043906 ( 2007 ) . j. wiersig , c. gies , f. jahnke , m. amann , t. berstermann , m. bayer , c. kistner , s. reitzenstein , c. schneider , s. hfling , a. forchel , c. kruse , j. kalden , d. hommel , `` direct observation of correlations between individual photon emission events of a microcavity laser , '' nature * 460 * , 245249 ( 2009 ) . m. amthor , s. weienseel , j. fischer , m. kamp , c. schneider , and s. hfling , `` electro - optical switching between polariton and cavity lasing in an ingaas quantum well microcavity , '' opt . express * 22 * , 3114631153 ( 2014 ) . k. mlmer , y. castin , and j. dalibard , `` monte carlo wave - function method in quantum optics , '' j. opt . soc . am . b * 10 * , 524538 ( 1993 ) . h. flayac , i. savenko , m. mttnen , and t. ala - nissila , `` quantum treatment for bose - einstein condensation in non - equilibrium systems , '' phys . rev . b * 92 * , 115117 ( 2015 ) . h. flayac and v. savona , `` heralded preparation and readout of entangled phonons in a photonic crystal cavity , '' phys . rev . lett . * 113 * , 143603 ( 2014 ) .
low to intermediate mass stars ( 0.8 to 8 @xmath2 ) undergo spectacular structural changes during the last phases of their evolution . according to the interacting stellar wind model ( isw ; kwok purton and fitzgerald 1978 ) , the spherically symmetric pne are formed by the interaction of two isotropic stellar winds , a slow and dense one from the asymptotic giant branch ( agb ) phase and a fast and tenuous one during the pn phase . the generalized isw model considers in addition the contribution of an equatorial density enhancement at the exit of the agb phase that produces a density contrast leading to the formation of axisymmetric shapes ( e.g. balick 1987 ) that may range from mildly elliptical to bipolar . in fact , the majority of planetary nebulae ( pne ) and proto pne ( ppne ) show axisymmetric morphologies . in some cases highly collimated , high speed bipolar outflows are also found . the causes of the equatorial density enhancement and the jet - like outflows are still under debate ( e.g. balick & frank 2002 ) and the two most likely being the presence of magnetic fields ( e.g. garcia segura & lpez 2000 , frank & blackman 2004 ) and post - common envelope , close binary nuclei ( e.g. soker & livio 1994 , de marco 2009 ) . sahai and trauger ( 1998 ) proposed as a shaping mechanism for the bipolar and multi polar pne , the presence of highly collimated outflows developed during the post agb or ppne phase . all these elements represent the main considerations in recent morphological classification studies of pne ( e.g. parker et al . 2006 , miszalski et al . 2008 , sahai et al . 2011 , lagadec et al . however , imaging alone can be in some cases deceiving in describing the real shape of a pn due to the inherent uncertainty introduced by the effects of the projection on the plane of the sky for a three dimensional nebula . the simplest example is that of an axisymmetric nebula , such as a bipolar , with a thick waist observed pole - on , in which case the nebula appears as a round doughnut projected on the sky . in these cases spatially resolved , high spectral resolution spectroscopy becomes an ideal tool to explore the three dimensional structure of the nebula by examining the doppler shifts in the emission line profile and assuming in a first approximation a homologous expansion for the nebula . most of these morpho - kinematic studies have been performed on relatively large , spatially resolved pne ( e.g. lpez et al . 2012 , clark et al . 2010 , garca daz et al . 2009 ) but they can also be very revealing when studying spatially unresolved , compact pne , as we show here . [ cols="<,<,<,<,<,<,<,<,<,<,<,<",options="header " , ] @xmath3 stanghellini & haywood 2010 + @xmath4 acker & neiner 2003 in this work , we perform a morpho kinematic study of five , relatively bright , compact pne with no discernable structure and with seeing limited angular sizes ranging from 5 to 10 arcsec . no high spatial resolution images for these objects were found in the literature or the usual repositories of images of pne . these objects were chosen from the the san pedro martir kinematic catalogue of galactic planetary nebulae ( lpez et al . 2012 ) on the basis of their line emission spectra that show the presence of fast , collimated bipolar outflows . the objects selected are : m 132 , m 220 , m 231 and m 242 and m 315 . based on their galactic coordinates , distances and systemic velocities they seem located in the galactic bulge or close to it , see table 1 . the central stars for four of them have been classified as wolf - rayet type ( acker & neiner 2003 ) and the fifth one as a weak emission line star or wels ( tylenda , acker & stenholm 1993 ) . as mentioned above , the long - slit , spectroscopic observations reveal the presence of highly collimated , fast , bipolar outflows surrounded by a thick equatorial enhancement , as a torus or a ring . we combine these data with the 3d morpho kinematic code shape ( steffen & lopez 2006 , steffen et al . 2011 ) to analyze the 3d structure of these outflows and the relation of their appearance with different projection on the sky . in section 2 , the observation and data reduction are presented . in section 3 , we describe the parameters used in the morpho kinematic code shape as well as the modelling results . we finish by summing up the results of this work in section 4 . high resolution , long - slit spectra of the pne m 132 , m 220 , m 231 , m242 and m 315 were obtained using the manchester echelle spectrometer , mes spm ( meaburn et al . 2003 ) on the 2.1 m telescope at the observatorio astronmico nacional at san pedro martir observatory in baja california , mexico , in its @xmath5/7.5 configuration . the observing runs for m 132 , m242 , m 220 , m 231 and m 315 took place in 2009 july , 2007 june , 2006 july , 2004 june and 2004 june , respectively . in all cases mes - spm was equipped with a site3 ccd detector with 1024@xmath61024 square pixels , each 24 @xmath7 m on a side ( @xmath80.312 arcsec pixel@xmath9 ) . a 90 bandwidth filter was used to isolate the 87@xmath10 order containing the h@xmath11 and [ n ii ] @xmath12@xmath136548 , 6584 , nebular emission lines . two - by - two binning was employed in both the spatial and spectral directions . consequently , 512 increments , each 0624 long gave a projected slit length of 532 on the sky . the slit was 150 @xmath7 m wide ( @xmath8 11 km @xmath1 and 19 ) . the slit was oriented north south ( p.a . @xmath14 ) for m 132 , m 220 and 315 . for m 231 and m 242 two slit positions are available , p.a . @xmath15 and p.a . @xmath14 and p.a . = @xmath16 and p.a . = @xmath17 , respectively . all the spectroscopic integrations were of 1800 s duration . the wavelength calibration was performed using a th / ar calibration lamp to an accuracy of @xmath18 1 km @xmath1 when converted to radial velocity . the stellar continua were not subtracted . only in two cases , m 242 and m 231 , a very faint continuum could be discerned within the very short spectra range of our single order echelle spectra . the data reduction was performed using the standard iraf routines . individual images were bias subtracted and cosmic rays removed . all spectra were also calibrated to heliocentric velocity . these spectra are part of the spm kinematic catalogue of galactic planetary nebulae ; lpez et al . 2012 ) and are available at http://kincatpn.astrosen.unam.mx where images from the digitized sky survey are also shown with the location of the slits overlaid for each object . modelling with shape follows three main steps . first , defining the geometrical forms to use ; shape has a variety of objects such as a sphere , torus , cone , cube , etc . whose basic forms can be modified by the user ( e.g. squeeze , twist , boolean , etc ) . second , an emissivity distribution is assigned to each individual object or structure , and third , a velocity law is chosen as a function of position . shape gives as result a two dimensional image and synthetic p v arrays , which are rendered from the 3d model to be compared with the observed data . the parameters of the model are then iteratively adjusted until a satisfactory solution is obtained . figure 1 presents the expected shapes of bi dimensional line profiles or position velocity diagrams ( pv ) from a spatially resolved , long slit , spectrum for some representative cases , such as an expanding bubble , a face on torus , a bipolar nebula with its lower part tilted into the plane of the sky and finally , the same bipolar with an equatorial torus . figure 2 presents the observed [ n ii ] emission line spectra with the corresponding synthetic p v diagrams . dimensional line profiles clearly show the structure of fast , bipolar outflows and a central thick waist , presumably an equatorial density enhancement in the form of a torus or ring . this information is used in combination with the code shape to iteratively find a combined geometry and expanding velocity field solution that replicates the observed position velocity diagrams of the bi - dimensional line profiles . the resultant synthetic p v diagrams are shown side by side to the observed ones , to their right , in the panels in figure 2 . these were reproduced assuming an axisymmetric geometry and a hubble type velocity law increasing linearly with distance from the central star to reconstruct the outflows along the polar direction . a key characteristic of cylindrically symmetric and homologously expanding nebulae is that the p v diagrams can be stretched in velocity in such a way that the outline of the image and the p v diagram match . this match determines the factor of proportionality in the mapping between position and velocity . it must however be noted that the shape model can not constrain the extent of the nebula along the sightline and thus there can be more than one solution to the fit . moreover , the information in the present cases comes from the line profiles alone ( no morphology information ) from single slit cuts over the nebula and without a previous knowledge of the position angle of the main axis of the nebula . to our only advantage is the slit width of two arcsec that includes a significant portion of the core of these compact nebulae , providing sufficient information to investigate their inner structure . considering these limitations we have therefore assumed reasonable sizes for all objects and considered the observed expansion velocities as lower limits . under these conditions we have selected the simplest solution that best fit the data . the equatorial thick torus / ring component was assumed to expand radially from the central star . similar solutions have been used for ngc 6337 ( garca daz et al . 2009 ) and and more recently for bd + 30 3639 ( akras & steffen 2012 ) . in order to reproduce the line profiles shown in figure 2 , the inclination of the symmetry axes of m 132 , m 220 , m 231 , m 242 and m 315 were found to be @xmath19 , @xmath20 , @xmath19 , @xmath21 and @xmath19 respectively , with respect to the line of sight , with an uncertainty in this case of the order of 20% , beyond this figure the line profiles start to show significant tilt differences . in figure 3 , we present the shape mesh models that best fit each of the pn under study , displaying the geometry of the system at four different orientations given by position angle and inclination : @xmath22 , before rendering . all models share as common elements the bipolar outflows , the torus or thick ring and an outer shell . the unusual , wide line profile of m 132 has been noted before by medina et al . the p v diagram shown in figure 2 ( top row , left panels ) for this object reveals fast collimated bipolar outflows reaching @xmath23 km @xmath1 . the orientation of m 132 is nearly pole - on , @xmath19 with respect to the line of sight as derived from the model , and the torus intersected by the slit is clearly revealed by the bright knots at the centre of the profile . the torus component is expanding radially with a velocity of @xmath2415 km @xmath1 . faint extensions beyond the bright equatorial knots disclose the presence of an outer , thin shell . the p v diagram for m 220 ( figure 2 , top row , right panels ) indicates a tilted axisymmetric structure , either an elliptical or a mild bipolar nebula . the central region is dominated by two bright equatorial knots from a toroidal structure . the knots are displaced one respect to the other due to the inclination of the torus that shows the front and back sides approaching and receding due to its radial expansion . these knots have expansion velocities with respect to the core of @xmath2415 km @xmath1 . the observed expansion velocity of the lobes with respect to the core is @xmath18 65 km @xmath1 and for an inclination with respect to the line of sight of @xmath20 the expansion velocity for the bipolar outflow transforms into @xmath18 125 km @xmath1 and 25 km @xmath1 for the equatorial toroid . m 231 shows for the slit oriented at p.a . = @xmath25 , a p v diagram ( figure 2 , middle row , left panels ) dominated by two bright knots located at very nearly the same position along the slit and separated by 30 km @xmath1 , fainter extensions to these knots reveal a collimated bipolar outflow reaching @xmath18 90 km @xmath1 . there is also a faint and seemingly , slightly tilted shell component that surrounds the bright knots or blobs . this tilt indicates an ellipsoidal shell expanding at a similar velocity to the bright knots . the slit oriented at p.a . = @xmath26 ( figure 2 , middle row , right panels ) shows the same components except for the tilt in the faint shell which suggests that the slit at p.a . = @xmath25 lies close to its major axis whereas the slit at p.a . = @xmath26 cuts across it . contrary to the previous cases where the bright knots have been identified with emission from a thick toroidal structure , in this case the characteristics of the bi - dimensional line profile indicate that the bright knots are part of an expanding bipolar system , probably a pair of compact , bright lobes , perhaps with a structure similar to the pn mz 3 ( or m 192 ) ( redman et al . 2000 ) but seen nearly pole - on and with material beyond the knots that expands at significantly larger speed . for m 242 its p v diagram reveals clearly a collimated bipolar structure with both lobes showing knots or condensations at their tips ( figure 2 , bottom row , left panels ) . the lobes have an extent of about 10 arcsec from the centre where again the presence of a conspicuous equatorial torus is indicated by the line profile . the knots are observed to be expanding with a velocity of @xmath27@xmath1815 km @xmath1 with respect to the core . considering the @xmath21 angle of inclination derived from the model for this nebula , the polar knots expand at a deprojected velocity of @xmath18 70 km @xmath1 from the core . the torus expands radially with a velocity of @xmath2415 km @xmath1 , very similar to the case of m 1 - 32 . m 315 looks nearly as a twin brother to m 242 but observed close to the line of sight ( figure 2 , bottom row , left panels ) . the bi - dimensional line profile in this case is dominated by a bright , compact , core containing a central toroid and a nearly pole - on , collimated outflow that extends to @xmath18 100 km @xmath1 . an inclination of @xmath19 with respect to the line of sight is obtained here from the model . in this work , we carried out a morpho kinematic study of five compact pne . the data were drawn from the kinematic catalogue of galactic planetary nebulae ( lpez et al . the sample was selected on the basis of compact pne with no discernible or apparent structure from their ground based images , no high spatial resolution imaging available ( e.g. _ hst _ ) and showing the presence of collimated , high - speed , bipolar outflows in their emission line spectra . fortuitously , the five pne selected under the criteria described above turned out to have wr type central stars . it is worth pointing out that the spm catalogue lists at present 84 pne with wr type nuclei , 92 spatially resolved bipolar out of which 43 are labeled as showing fast projected outflows and 152 pne from the galactic bugle region . it is our intention to conduct next a thorough search in the catalogue to select from the lists mentioned above additional pne that match similar criteria adopted here . in this work , we demonstrate that with practically no morphological information and limited spectral information consisting of long slit , spatially resolved , echelle spectra combined with the reconstruction code shape , a fair morpho - kinematic representation of the nebulae can be achieved . the process yields a first order description of the main morphological elements of the nebula , the inclination of the main outflow symmetry axis with respect to the line of sight and with it an estimate of the true expansion velocity for the bipolar outflow . we find that these planetary nebulae have very similar structural components and the differences in their emission line spectra derive mostly from their different projection on the sky and degree of collimation of the bipolar outflows . all cases studied here are characterized by the presence of an equatorial density enhancement or toroid , a collimated bipolar outflow and an outer shell . the bipolar expansion velocities range from 70 to 200 km @xmath1 whereas the tori are found to expand rather slowly , in the order of 15 to 25 km @xmath1 . the results of this work strongly indicate the importance of evaluating the true structure of compact pne for statistical studies that consider the different morphological classes for evolution and population synthesis studies . given the distance to the galactic bulge , there is a large population of unresolved pne there that need to be investigated in detail . whether the fact that all the compact pne studied here turned out to have a wr - type nucleus has an statistical significance for the bulge pne with collimated bipolar outflows or not deserves further investigation . likewise a detailed reinvestigation of the kinematics of pn with wr type nucleus such as the work of medina et al . ( 2006 ) with spatially resolved , long slit spectra of higher spectral resolution should turn fruitful results and possibly establish a stronger link between pne with wr type nuclei and fast , collimated , bipolar outflows . it is also interesting to note that lopez et al . ( 2011 ) associate toroidal structures with close binary central stars and miszalski ( 2011 ) also associates bipolar outflows with close binary nuclei in pne . given the characteristics of the present sample , it would be interesting to search for the presence of binary nuclei in these objects . + + + we acknowledge the financial support of dgapa - unam through grants in110011 and in100410 . s. a. gratefully acknowledges a postdoctoral scholarship from dgapa - unam . the authors are grateful to the anonymous referee for his valuable comments and suggestions .	we present high resolution , long slit spectroscopic observations of five compact ( @xmath0 10 arcsec ) planetary nebulae located close to the galactic bulge region and for which no high spatial resolution images are available . the data have been drawn from the san pedro mrtir kinematic catalogue of galactic planetary nebulae ( lpez et al . 2012 ) . the central star in four of these objects ( m 132 , m 220 , m 231 and m 315 ) is of wr type and the fifth object ( m 242 ) has a wels type nucleus . these observations reveal the presence in all of them of a dense and thick equatorial torus - like component and high speed , collimated , bipolar outflows . the code shape is used to investigate the main morpho kinematic characteristics and reproduce the 3d structure of these objects assuming a cylindrical velocity field for the bipolar outflows and a homologous expansion law for the torus / ring component . the deprojected expansion velocities of the bipolar outflows are found to be in the range of 65 to 200 km @xmath1 , whereas the torus / ring component shows much slower expansion velocities , in the range of 15 to 25 km @xmath1 . it is found that these planetary nebulae have very similar structural components and the differences in their emission line spectra derive mostly from their different projections on the sky . the relation of their morpho kinematic characteristics with the wr type nuclei deserves further investigation . [ firstpage ] ism : jets and outflows , ism : kinematics and dynamics , planetary nebula : individual ( m 132 , m 220 , m 231 , m 242 , m 315 )
modelling the flow of complex fluids is a very intricate problem which is far from being solved up to now . besides studies which aim at improving phenomenological rheological models ( purely macroscopic constitutive laws ) , only a few attempts are made to recover the rheological behavior of a complex fluid from elementary physical processes arising in its microstructure . + the mesoscopic model which has been proposed by hbraud and lequeux in @xcite deals with simple shear flows of concentrated suspensions . it is obtained by dividing the material in a large number of mesoscopic elements ( `` blocks '' ) with a given shear stress @xmath0 ( @xmath0 is a real number ; it is in fact an extra - diagonal term of the stress tensor in convenient coordinates ) and by considering the evolution of the probability density @xmath1 which represents the distribution of stress in the assembly of blocks . under various assumptions on the evolution of the stresses of the blocks which will be described below , the equation for the probability density @xmath1 for a block to be under stress @xmath0 at time @xmath2 may be written as : [ eq : syst - p ] _ tp =- b(t ) _ p+d(p(t ) ) ^2_p - p+_0 ( ) ( 0;t);[eq : p ] + p0 ; + p(0,)=p_0 ( ) , [ eq : p0er ] where for @xmath3 , we denote by @xmath4 in equation ( [ eq : p ] ) , @xmath5}$ ] denotes the characteristic function of the open set @xmath6 $ ] and @xmath7 the dirac delta function on @xmath8 . each term arising in the above equation ( hl equation in short ) has a clear physical interpretation . when a block is sheared , the stress of this block evolves with a variation rate @xmath9 proportional to the shear rate @xmath10 ( @xmath11 is an elasticity constant ) ; in this study , the shear rate @xmath10 , and therefore the function @xmath12 , are assumed to be in @xmath13 . when the modulus of the stress overcomes a critical value @xmath14 , the block becomes unstable and may relax into a state with zero stress after a characteristic relaxation time @xmath15 . this phenomenon induces a rearrangement of the blocks and is modelled through the diffusion term @xmath16 . the diffusion coefficient @xmath17 is assumed to be proportional to the amount of stress which has to be redistributed by time unit and the positive parameter @xmath18 is supposed to represent the mechanical fragility " of the material . in all that follows , the parameters @xmath18 , @xmath15 and @xmath14 are positive , and the initial data @xmath19 in is a given probability density ; that is @xmath20 we will be looking for solutions @xmath21 in @xmath22 such that @xmath23 belongs to @xmath24 to the nonlinear parabolic partial differential equation . the subscript @xmath0 refers to integration over @xmath8 with respect to @xmath0 , whereas the subscript @xmath2 refers to time integration on @xmath25 $ ] for any @xmath26 . note that the average stress in the material is given by @xmath27 and therefore the above condition on @xmath28 ensures that the average stress is an essentially bounded function of time . + actually in practice , the shear rate is not uniform in the flow and in order to better describe the coupling between the macroscopic flow and the evolution of the microstructure we introduce and study in a second paper @xcite a micro - macro model . in this model @xmath29 is also a function of the macroscopic space variables and the average stress defined by is inserted into the macroscopic equation governing the velocity of the macroscopic flow . in order to lighten the notation and without loss of generality we assume from now on that @xmath30 and @xmath31 this amounts to changing the time and stress scales . the main difficulties one encounters in the mathematical analysis come from the nonlinearity in the diffusion term and also and even more from the fact that the parabolic equation may degenerate when the viscosity coefficient @xmath32 vanishes , and this will be shown to may appear only when @xmath33 . this difficulty is illustrated on a simplified example just below and also in section [ sec : stat ] where we discuss the existence of stationary solutions in the case when the shear rate @xmath34 is a constant . let us first of all look at the following simplified model which already includes the difficulties we are going to face to in the study of equation . we consider the equation : @xmath35 - 1,1[}(\sigma)\ ; , \end{array}\right.\ ] ] where @xmath36 - 1,1[}$ ] is the characteristic function of the interval @xmath37 - 1,1[$ ] . the initial condition is on purpose chosen in such a way that @xmath38 . the function @xmath39 - 1,1[}(\sigma)$ ] is a stationary solution to this equation and for this solution @xmath40 is identically zero . but it is not the unique solution to in @xmath41 . it is indeed possible to construct a so - called _ vanishing viscosity solution _ for which @xmath42 for all @xmath43 , and there are actually infinitely many solutions to this equation . ( this statement is obtained as a corollary of lemma [ lem:2 ] in section [ sec : deg ] below . ) as far as equation ( [ eq : syst - p ] ) is concerned , we show that , in the case when @xmath33 and @xmath44 , we may have either a unique or infinitely many solutions , depending on the initial data ( see proposition [ prop : deg ] in section [ sec : deg ] ) . on the other hand , we are able to prove the following existence and uniqueness result in the non - degenerate case when @xmath45 : [ th : main1 ] let the initial data @xmath19 satisfy the conditions @xmath46 and assume that @xmath47 then , for every @xmath48 , there exists a unique solution @xmath29 to the system in @xmath49 . moreover , @xmath50 , @xmath51 for all @xmath52 , @xmath53 and for every @xmath48 there exists a positive constant @xmath54 such that @xmath55 besides @xmath56 so that the average stress @xmath57 is well - defined by ( [ eq : def - tau ] ) in @xmath58 . the first step toward the existence proof of solutions to will consist in the study of so - called vanishing viscosity approximations , which are the unique solutions to the family of equations [ syst : p - eps ] _ tp_=-b(t)_p_+(d(p_(t))+)^2_p _ -_p_+ _ 0 ( ) ; [ eq : p - eps ] + p_0 ; + p_(0,)=p_0 ( recall that we have rescaled the time and stress units to get @xmath31 and @xmath30 ) . section [ sec : visc ] below is devoted to the proof of the following [ prop : visc ] let @xmath48 be given . we assume that the initial data satisfies the same conditions as in the statement of the theorem . then , for every @xmath48 and @xmath59 , there exists a unique solution @xmath60 to in @xmath61 . moreover , @xmath62 , @xmath63 , @xmath64 @xmath65 and for every @xmath48 , there exist positive constants @xmath66 , @xmath67 and @xmath68 which are independent of @xmath69 such that @xmath70 @xmath71 and @xmath72 theorem ( [ th : main1 ] ) is then proved in section [ sec : nondeg ] while the degenerate case is investigated in section [ sec : deg ] . lastly , the description of stationary solutions in the constant shear rate case is carried out in section [ sec : stat ] . this section is devoted to the proof of proposition [ prop : visc ] . + we begin with the following : [ lem : unique ] let @xmath19 satisfy . then for every @xmath48 and @xmath73 , there exists at most one solution @xmath60 to in @xmath74 . moreover , @xmath75 ( thus , the initial condition makes sense ) and @xmath76 for almost every @xmath2 in @xmath25 $ ] . * proof of lemma [ lem : unique ] : * we begin with proving that every solution to in @xmath74 satisfies . we fix @xmath77 and we consider a cut - off @xmath78 function @xmath79 with compact support which is equal to 1 when @xmath80 and to 0 when @xmath81 and such that @xmath82 where here and below @xmath83 denotes a positive constant that is independent of @xmath84 . notice that @xmath85 is equal to 0 on @xmath37-\infty,\,-2r]$ ] , on @xmath86 $ ] and on @xmath87 . now , we multiply ( [ eq : p - eps ] ) by @xmath88 and integrate over @xmath89\times \rr$ ] to obtain @xmath90 we bound from above the terms on the right - hand side as follows . first , we have @xmath91 thanks to ( [ c.1 ] ) and using that @xmath92 and @xmath93 . next , @xmath94 thanks again to ( [ c.1 ] ) , cauchy - schwarz inequality and since @xmath95 is in @xmath96 . finally , @xmath97 and the right - hand side goes to @xmath98 as @xmath84 goes to infinity since @xmath99 is in @xmath100 . all this together yields @xmath101 for almost every @xmath2 in @xmath25 $ ] . in particular , this implies that @xmath102 . + let us now argue by contradiction by assuming that there exist two solutions @xmath103 and @xmath104 to ( [ eq : p - eps ] ) corresponding to the same initial data @xmath19 . by subtracting the equations satisfied by @xmath103 and @xmath104 respectively , we obtain ; [ eq : q ] + where @xmath105 . we multiply ( [ eq : q ] ) by @xmath106 and integrate over @xmath8 with respect to @xmath0 to obtain , after integrations by parts , @xmath107 we first remark that since @xmath108 thanks to , we get @xmath109 with the help of cauchy - schwarz inequality . next , using the sobolev embedding of @xmath110 into @xmath111 , we bound from above the terms on the right - hand side in the following way : @xmath112 therefore , comparing with we deduce @xmath113 finally , by applying the gronwall lemma , we prove that @xmath114 , thus @xmath115 . the uniqueness of the solution follows . @xmath116 + [ rem : unic]the same proof shows that if there exists a solution to in @xmath117 such that @xmath118 , then it is unique in this space . we now turn to the existence part in the statement of proposition [ prop : visc ] . from now on we fix a positive constant @xmath119 . the proof of proposition [ prop : visc ] will be carried out by the schauder fixed point theorem . for given positive constants @xmath120 and @xmath84 , we introduce @xmath121 and @xmath122 two closed convex subsets of respectively @xmath123 and @xmath124 as follows : @xmath125 to simplify notation we denote @xmath126 we first prove the following [ prop:1 ] let @xmath48 and let @xmath127 such that @xmath128 . then , for every @xmath129 in @xmath130 and @xmath106 in @xmath122 , there exists a unique solution @xmath29 in @xmath131 to [ eq : syst - lin ] _ tp(t,)=-b(t)_p(t,)+a(t)^2_p(t,)- _ [ -1,1 ] ( ) p(t , ) + _ 0 ( ) ; [ eq : plin ] + p(0,)=p_0 ( ) . moreover , @xmath132 , @xmath29 is non - negative and @xmath133 with @xmath134 and @xmath135 where @xmath136 . in addition , 1 . if @xmath137 , then @xmath29 is in @xmath138 and @xmath139 2 . if @xmath140 ( thus @xmath141 ) , then @xmath142 . more precisely , we have @xmath143 moreover @xmath144 and @xmath145 . * proof of proposition [ prop:1 ] : * let us first observe that for every @xmath106 in @xmath122 , @xmath146 since @xmath147 for almost every @xmath2 in @xmath25 $ ] . therefore the source term @xmath148 in is in @xmath149 and the existence and the uniqueness of a solution @xmath150 to the system is well - known ( see for example @xcite ) . in particular , the initial condition makes sense . owing to the fact that the source term is non -negative , the proof that @xmath151 is also standard ( see again @xcite ) . we now check the pointwise inequality . + this is ensured by the maximum principle with observing that @xmath152 and @xmath153 given respectively by and are the unique solutions to the systems @xmath154 and @xmath155 respectively . we now turn to the proof of statement _ i. _ and assume that @xmath19 belongs to @xmath111 . then , using the two facts that for every @xmath156 , @xmath157 and @xmath158 , is easily deduced from @xmath159 with the help of and since @xmath160 . + suppose now that @xmath161 . this together with the assumption @xmath162 , guarantees that @xmath141 ( see also below ) . using again , we now have @xmath163 since @xmath164 . with the help of and observing that @xmath165 , we then deduce . + we now use this bound to check that @xmath144 and @xmath53 . indeed , for any @xmath2 , any sequence @xmath166 in @xmath25 $ ] which converges to @xmath2 and @xmath167 , we have @xmath168 for any fixed @xmath169 the first term in the right - hand side goes to @xmath98 as @xmath170 goes to infinity since @xmath132 and then the second term is arbitrarily small as @xmath169 goes to infinity . the same argument yields the continuity of @xmath17 with respect to @xmath2 . @xmath116 + the following proposition aims at checking the required assumptions to apply the schauder fixed point theorem . [ prop:2 ] let @xmath171 be given . we assume that @xmath172 let @xmath59 , @xmath173^{-\,2}\;.\ ] ] then , for every @xmath174 , the function @xmath175 , with @xmath29 being the solution to the system , maps @xmath176 into itself . moreover @xmath177 is continuous and @xmath178 is relatively compact in @xmath179 . * proof of proposition [ prop:2 ] : * * step 1 : * _ @xmath177 is well - defined . _ according to proposition [ prop:1 ] , @xmath29 is in @xmath180 and @xmath53 . we now prove that with our choice for @xmath181 ( which ensures that @xmath182 ) , @xmath183 for this , we again use the inequality @xmath159 , the definition of @xmath153 , the rough estimate @xmath184 and to obtain @xmath185 for @xmath186 . it only remains now to check that @xmath187 . we thus go back to and observe that this condition holds provided @xmath188 since we already have demanded that @xmath189 a sufficient condition is then @xmath190\,\leq 1\;,\ ] ] which reduces to @xmath191 with @xmath192 given by . + our next step will consist in establishing _ a priori _ bounds on @xmath29 in @xmath193 . + * step 2 * : _ _ a priori _ bounds . _ if we multiply equation ( [ eq : plin ] ) by @xmath29 and integrate by parts over @xmath8 with respect to @xmath0 we easily obtain @xmath194 since from the sobolev embedding of @xmath110 into @xmath111 and the bound on @xmath195 we get @xmath196 we may write @xmath197 we recall that @xmath160 and we apply the gronwall lemma to obtain @xmath198 we now return to ( [ eq : endsigmap ] ) and integrate it over @xmath199 $ ] to obtain @xmath200 * step 3 * : _ the function @xmath177 is continuous . _ we consider a sequence @xmath201 in @xmath202 such that @xmath203 converges to @xmath129 strongly in @xmath204 and @xmath205 converges to @xmath106 strongly in @xmath96 , and we denote @xmath206 . we have to prove that @xmath207 converges strongly to @xmath29 in @xmath96 and @xmath208 converges to @xmath32 strongly in @xmath209 , with @xmath210 . in virtue of ( [ eq : bd - l2 ] ) and ( [ in : endsigmap ] ) , the sequence @xmath211 is bounded in @xmath212 then , @xmath213 is bounded in @xmath214 and @xmath215 is bounded in @xmath216 . since @xmath217 is bounded in @xmath216 , @xmath218 and @xmath219 is bounded in @xmath216 , @xmath220 is bounded in @xmath216 . this together with the fact that @xmath221 is bounded in @xmath222 implies that , up to a subsequence , @xmath211 converges strongly towards @xmath223 in @xmath224 ( the convergence being weak in @xmath225 ) thanks to a well - known compactness result @xcite . in particular , @xmath221 converges to @xmath29 almost everywhere . thus @xmath151 and by the fatou s lemma , @xmath226 almost everywhere on @xmath199 $ ] . hence @xmath29 belongs to @xmath122 . we are going to show that the convergence is actually strong in @xmath96 . in virtue of in proposition [ prop:1 ] , we dispose of a uniform _ a priori _ bound on @xmath211 in @xmath138 ( hence also on @xmath29 ) . for the strong convergence in @xmath96 we then argue as follows . for any fixed positive real number @xmath227 , we have @xmath228 owing to the fact that @xmath211 and @xmath29 belong to a bounded subset of @xmath229 . we then conclude by letting @xmath170 next @xmath227 go to infinity . + we now prove that @xmath208 converges to @xmath32 strongly in @xmath204 . we shall actually prove that @xmath208 converges to @xmath32 strongly in @xmath230 and then use the fact that @xmath231 is bounded in @xmath58 , in virtue of and because @xmath221 lies in @xmath122 . let us fix @xmath232 . then , we have @xmath233 because @xmath29 and @xmath221 belong to @xmath122 . since @xmath211 converges to @xmath29 strongly in @xmath234 , we conclude that @xmath208 converges to @xmath32 in @xmath235 by letting @xmath170 next @xmath227 go to infinity in ( [ in : convged ] ) . + in order to pass to the limit in the equation satisfied by @xmath211 ( thereby proving that @xmath210 ) , we now observe that the strong convergence of @xmath205 to @xmath106 in @xmath96 , together with the argument in above shows that @xmath236 converges to @xmath195 strongly in @xmath237 . it is then easily proved that @xmath29 is a weak solution to and since @xmath29 is in @xmath222 it is the unique solution to corresponding to @xmath129 and @xmath106 . in particular , the whole sequence @xmath221 converges and not only a subsequence . * step 4 * : _ @xmath238 is relatively compact . _ + let @xmath239 be a sequence in @xmath178 . we have to prove that we may extract a subsequence which converges strongly in @xmath240 . exactly as for the proof of the continuity , the _ a priori _ estimates ( [ eq : bd - l2 ] ) and ( [ in : endsigmap ] ) ensure that the sequence @xmath211 is bounded in @xmath212 . since @xmath241 is bounded @xmath242 , we can mimic the argument in step 3 above to deduce that up to a subsequence the sequence @xmath221 converges to some @xmath29 in @xmath122 strongly in @xmath243 and that @xmath208 converges to @xmath244 strongly in @xmath204 . @xmath116 we are now in position to conclude the proof of proposition [ prop : visc ] . + let @xmath171 and @xmath59 being given . we are going to prove the existence of a unique solution on @xmath245 $ ] . + being given an initial data @xmath19 which satisfies , existence of a solution @xmath60 is ensured from proposition [ prop:2 ] by applying the schauder fixed point theorem on `` short '' time interval @xmath246 $ ] with @xmath247 and where @xmath248 . this solution is uniquely defined in virtue of lemma [ lem : unique ] and we know from that @xmath249 . moreover from proposition [ prop:1 ] @xmath250 and by construction @xmath251 . therefore @xmath252 satisfies the same conditions as @xmath19 . then , repeating the same argument we may build a solution to with initial data @xmath252 on @xmath253 $ ] with @xmath254 , where latexmath:[$r_2=r_1 + 1=\int_\rr if we now glue this solution to @xmath60 at @xmath256 we obtain the unique solution to on @xmath257 $ ] . it is now clearly seen that for any integer @xmath258 we may build a solution to on @xmath259 $ ] with @xmath260 obviously goes to @xmath261 together with @xmath170 , existence ( and uniqueness ) of the solution @xmath60 to is obtained on every time interval . + for the proof of we argue as for the proof of in proposition [ prop:1 ] . defining @xmath262 as in with @xmath129 replaced by @xmath263 and @xmath195 by @xmath264 we obtain @xmath265\\ & \leq & \vert p_0\vert_{l^\infty}+\frac{1}{\alpha\,\sqrt\pi}\;\sqrt{\int_0^t d(p_\varepsilon)}\\ & \leq&\vert p_0\vert_{l^\infty}+ \frac{\sqrt\alpha\,\sqrt t}{\sqrt{\pi}}\;.\end{aligned}\ ] ] then @xmath266 from which follows gathering together and and , with the notation of the proposition , @xmath267 + the proof of follows the same lines as the proof of . indeed , we again use the pointwise inequality @xmath268 and replace @xmath195 by @xmath269 and @xmath129 by @xmath270 in and use to deduce @xmath271 whence with @xmath66 being the quantity in the right - hand side of . in order to prove , we multiply ( [ eq : p - eps ] ) by @xmath272 and we integrate by parts over @xmath8 with respect to @xmath0 to obtain @xmath273 we use the @xmath274 bound to bound the right - hand side and we integrate with respect to @xmath275 over @xmath199 $ ] to deduce with @xmath276 using that @xmath277 . @xmath116 the main result of this section corresponds to the statement of theorem [ th : main1 ] and fully describes the issue of existence and uniqueness of solutions to the hl equation in the non - degenerate case . it is summarized in the following : [ prop : non - deg ] let @xmath279 satisfy . we assume that @xmath278 . then , the hl equation has a unique solution @xmath29 in @xmath280 and @xmath29 is the limit ( in @xmath281 ) of @xmath282 when @xmath283 goes to @xmath98 where @xmath60 is the vanishing viscosity solution whose existence and uniqueness is ensured by proposition [ prop : visc ] . moreover , @xmath284 , @xmath56 and @xmath285 . furthermore , @xmath53 and for every @xmath48 there exists a positive constant @xmath54 such that @xmath286 we begin with proving the following : [ lem : non - degeneracy ] we assume that @xmath19 satisfies . then , if @xmath278 , @xmath287 for every @xmath288 $ ] , with @xmath60 being the unique solution to provided by proposition [ prop : visc ] and , actually , for every @xmath48 there exists a positive constant @xmath54 such that @xmath289 for every @xmath290 . note that this bound from below is independent of @xmath69 , but it comes out from the proof that it depends on @xmath19 and on the shear @xmath34 . * proof of lemma [ lem : non - degeneracy ] : * the proof relies on the bound from below in that we integrate over @xmath291 to obtain @xmath292 let us define @xmath293}$ ] . the function @xmath294 is a gaussian probability density with mean @xmath295 and squared width @xmath296 . therefore , for every @xmath297 , we have @xmath298 which implies @xmath299 in the zero shear case ( @xmath300 , thus @xmath301 ) the proof is over and @xmath302 in the general case , a strictly positive bound from below is available as long as the support of @xmath19 is not contained in @xmath303 . we thus define @xmath304 then @xmath305 ( @xmath306 possibly even infinite ) , the support of @xmath19 is contained in @xmath307 , and for every @xmath308 , holds for some positive constant @xmath309 defined by @xmath310 it is worth emphasizing that this quantity is independent of @xmath69 . if @xmath311 , the proof is over and @xmath312 fits . let us now examine the case when @xmath313 and @xmath314 . + we go back to , take @xmath2 in @xmath315 $ ] and denote @xmath316 for shortness . then @xmath317 since @xmath318 and @xmath319 thanks to . the proof of lemma [ lem : non - degeneracy ] then follows by defining @xmath320 with @xmath309 given by and @xmath321 being the positive quantity in the right - hand side of , that is @xmath322 @xmath116 * proof of proposition [ prop : non - deg ] : * + we first go back to the proof of the bound on @xmath95 and more precisely we look at and observe that in virtue of @xmath323 let now @xmath324 denote any sequence in @xmath325 $ ] which goes to @xmath98 as @xmath170 goes to infinity . to shorten the notation we denote by @xmath211 instead of @xmath326 the corresponding sequence of solutions to . with the above bound on @xmath211 and , we know that @xmath211 is bounded in @xmath222 independently of @xmath170 . moreover thanks to and @xmath211 is bounded in @xmath327 and we also dispose of a uniform bound on @xmath328 in virtue of . therefore arguing exactly as in the proof of proposition [ prop:2 ] ( step 4 ) where we have proved that the mapping @xmath329 is relatively compact in @xmath330 we show that @xmath211 converges to some @xmath29 strongly in @xmath96 and @xmath208 converges to @xmath32 in @xmath209 . then @xmath29 is a solution to the initial problem in @xmath331 , @xmath285 and @xmath332 . moreover , @xmath333 this non - degeneracy condition on the viscosity coefficient ensures that there is at most one solution to in @xmath334 ( this follows by an obvious adaptation of the proof of lemma [ lem : unique ] to this case ) . therefore the limiting function @xmath29 is uniquely defined and does not depend on the sequence @xmath324 . moreover the whole sequence @xmath211 converges to this unique limit and not only a subsequence . @xmath116 + as a conclusion of this subsection let us make the following comment which is a byproduct of proposition [ prop : non - deg ] . let @xmath29 be a solution to in @xmath335 , then as soon as @xmath17 is positive for some time @xmath2 it remains so afterwards since the solution can be continued in a unique way beginning from time @xmath2 . throughout this section we assume that @xmath33 and therefore the support of @xmath19 is included in @xmath336 $ ] . assume that we dispose of a solution to in @xmath335 . we may define @xmath337 by @xmath338 according to the comment at the end of the previous section for every @xmath339 , @xmath340 while @xmath341 for all @xmath2 in @xmath342 $ ] . on @xmath343 , the hl equation reads @xmath344 the above system reduces to @xmath345 $ } \;. \end{array } \right.\ ] ] the second equation in is compatible with the first one as long as @xmath346 $ } \;.\ ] ] therefore there exists a maximal time interval @xmath347 $ ] on which the hl equation may reduce to a mere transport equation and this is for an intrinsic time @xmath192 ( possibly infinite ) defined by @xmath348 note that @xmath192 is completely determined by the data @xmath19 and @xmath34 . if @xmath349 , the steady state @xmath350 is a solution of the hl equation for all time . we shall now exhibit circumstances under which it is not the unique solution . for convenience , we restrict ourselves to the case when @xmath44 ( we then have obviously @xmath349 ) . for @xmath351 such that @xmath352 , let us denote by @xmath353 the function from @xmath354 to @xmath354 defined by @xmath355 and by @xmath356 [ prop : deg ] let @xmath279 satisfy and be such that @xmath357 , then 1 . if @xmath353 satisfies @xmath358 then @xmath359 is the unique solution to in @xmath360 ; 2 . otherwise , has an infinite number of solutions in @xmath360 . the set of solutions to is made of the steady state @xmath359 and of the functions @xmath361 defined by @xmath362 where @xmath106 is the unique solution to in @xmath360 such that @xmath363 on @xmath370,+\infty[$ ] . besides , @xmath364 [ lem:1 ] let @xmath351 such that @xmath365 the function @xmath366 is in @xmath3670,+\infty[)$ ] , and is positive on @xmath370,+\infty[$ ] . in addition , @xmath368 on @xmath370,+\infty[$ ] . * proof of lemma [ lem:1 ] : * it is easy to check that @xmath3690,+\infty[)$ ] , and that @xmath370 on @xmath370,+\infty[$ ] . since @xmath33 , the function @xmath19 is supported in @xmath371 $ ] . thus , for any @xmath372 @xmath373 it follows that for any @xmath372 , @xmath374 @xmath116 [ lem:2 ] let @xmath375 and @xmath376 such that @xmath377 let us consider the problem @xmath378 1 . if @xmath353 satisfies ( [ eq : condfp0 ] ) then @xmath359 is the unique solution to ( [ eq : ii ] ) in @xmath360 ; 2 . otherwise , ( [ eq : ii ] ) has an infinite number of solutions in @xmath360 . the set of solutions to ( [ eq : ii ] ) is made of the steady state @xmath379 and of the functions @xmath380 defined by @xmath381 where @xmath382 is the unique solution to in @xmath360 such that @xmath383 on @xmath370,+\infty[$ ] . [ cor:1 ] the initial data @xmath384 - 1,1[}$ ] fulfills the assumptions of the above lemma and @xmath385 . therefore there are infinitely many solutions to the equation in the introduction . * proof of corollary [ cor:1 ] : * the only point to be checked is that @xmath385 . with the standard notation @xmath386 , and by using and symmetry considerations , simple calculations yield @xmath387.\end{aligned}\ ] ] since @xmath388 for @xmath389 going to @xmath261 , @xmath390 near @xmath98 and the integrability of @xmath391 on @xmath392 $ ] follows . @xmath116 * proof of lemma [ lem:2 ] * let us consider a non - negative function @xmath393 . the unique solution in @xmath360 of the problem @xmath394 is given by @xmath395 where @xmath396 . any solution to ( [ eq : ii ] ) thus satisfies @xmath397 and therefore @xmath398 it follows that the function @xmath399 is solution in @xmath400 to the nonlinear integral equation @xmath401 on the other hand , if @xmath402 is solution to ( [ eq : inteq ] ) it is easy to check that the function @xmath403 defined by ( [ eq : wd ] ) is solution to ( [ eq : iii ] ) . if condition ( [ eq : condfp0 ] ) is fulfilled , equation ( [ eq : inteq ] ) has a unique solution in @xmath400 ( the constant function equal to zero ) and the steady state @xmath404 thus is the unique solution to ( [ eq : ii ] ) in @xmath360 ; otherwise , the set of solutions to ( [ eq : inteq ] ) is made of the steady state @xmath404 and of the family @xmath405 with @xmath406 where the function @xmath389 is defined on @xmath407 by @xmath408 statement _ ii . _ is obtained by denoting by @xmath382 the solution to ( [ eq : iii ] ) associated with the function @xmath409 . @xmath116 * proof of proposition [ prop : deg ] : * the solution @xmath410 to equation satisfies the inequalities @xmath411 where @xmath412 and @xmath413 are defined in @xmath360 by @xmath414 therefore on the one hand @xmath415 and on the other hand @xmath416 if ( [ eq : condfp0 ] ) is not fulfilled , using ( [ eq : binf ] ) and the property that @xmath353 is strictly increasing on @xmath407 , we obtain that @xmath417 where @xmath409 is the function defined in the proof of lemma [ lem:2 ] . as for any @xmath418 , there exists @xmath419 such that @xmath420 on @xmath421 $ ] the same reasoning as in the non - degenerate case leads to the conclusion that @xmath282 converges up to an extraction to @xmath29 in @xmath4220,+\infty[\times \rr)$ ] and in @xmath423,l^2(\rr))$ ] for any @xmath424 , @xmath29 being a solution to in @xmath4250,+\infty[,l^2_\sigma)$ ] such that @xmath426 on @xmath370,+\infty[$ ] . throughout this section the shear rate @xmath34 is assumed to be a given constant and we are looking for solutions in @xmath427 to the following system : * if @xmath300 , any probability density which is compactly supported in @xmath336 $ ] is a solution to which satisfies @xmath428 . if @xmath429 , these are the only stationary solutions ( and there are infinitely many ) , whereas when @xmath430 there exists a unique stationary solution corresponding to a positive value of @xmath431 , which is explicitly given by and below . this solution is even and with exponential decay at infinity . * if @xmath432 , for any @xmath433 , there exists a unique stationary solution to , and it corresponds to a positive value for @xmath431 , which is implicitly given by and below . this solution has exponential decay at infinity . * proof of proposition [ prop : solstat ] : * + * the case when @xmath300 * + we first observe that any non - negative function @xmath29 which is normalized in @xmath427 and with support in @xmath336 $ ] is a solution to the system since in that case all terms in equation cancel . we now examine the issue of existence of solutions of such that @xmath434 . for simplicity we denote @xmath435 . for given constant @xmath436 , it is very easy to calculate explicitly the solutions of on each of the three regions @xmath437 , @xmath438 $ ] and @xmath439 . using compatibility conditions on @xmath8 and the fact that @xmath29 has to be in @xmath427 one obtains : @xmath440 the compatibility condition @xmath435 happens to be then automatically satisfied and the normalization constraint @xmath441 imposes that @xmath431 solves @xmath442 since @xmath443 , we immediately reach a contradiction when @xmath444 , whereas when @xmath445 equation admits a unique positive solution ; namely @xmath446 * the case when @xmath432 * + first of all , we observe that if @xmath447 every term in equation but @xmath448 vanish . thus @xmath29 has to be a non - zero constant which is in contradiction with @xmath449 . so necessarily @xmath436 . for given positive constant @xmath431 , we then solve as above and obtain @xmath450 with @xmath451 @xmath452 and @xmath453 it is tedious but easy to check that this function always fulfills the self - consistency condition @xmath435 and that the normalization condition @xmath285 reads @xmath454 for any @xmath455 ( the negative values of @xmath34 are dealt with by replacing @xmath0 by @xmath456 ) , the left - hand side of ( [ eq : norm - new ] ) is a continuous function which goes to @xmath261 when @xmath431 goes to infinity and goes to zero when @xmath431 goes to @xmath98 . this already ensures the existence of at least one steady state for any @xmath433 . moreover , setting @xmath457 ( for example ) we may rewrite the left - hand side of ( [ eq : norm - new ] ) as f(z ) = + . next we check that the function @xmath458 is monotone decreasing ( thus , the left - hand side of is increasing with respect to @xmath431 ) , whence the uniqueness result .	the mathematical properties of a nonlinear parabolic equation arising in the modelling of non - newtonian flows are investigated . the peculiarity of this equation is that it may degenerate into a hyperbolic equation ( in fact a linear advection equation ) . depending on the initial data , at least two situations can be encountered : the equation may have a unique solution in a convenient class , or it may have infinitely many solutions . [ section ] [ section ] [ section ] [ section ] [ section ] [ section ] [ section ]
communication networks are increasingly being taxed by the enormous demand for instantly available , streaming multimedia . ideally , we would like to maximize the reliability and data rate of a system while simultaneously minimizing the delay . yet , in the classical fixed blocklength setting , the reliability function of a code goes to zero as the rate approaches capacity even in the presence of feedback . this seems to imply that , close to capacity , it is impossible to keep delay low and reliability high . however , this lesson is partially an artifact of the block coding framework . the achievable tradeoff changes in a streaming setting where all bits do not need to be decoded by a fixed deadline , but rather , each individual bit must be recovered after a certain delay . in this setting , the reliability function measures how quickly the error probability on each bit estimate decays as a function of the delay . surprisingly , the achievable error exponent can be quite large at capacity if a noiseless feedback link is available and cleverly exploited @xcite . the distinguishing feature of these streaming architectures with feedback is the use of an ultra - reliable special codeword that is transmitted to notify the decoder when it is about to make an error . while this `` red alert '' codeword requires a significant fraction of the decoding space to attain its very large error exponent , the remaining `` standard '' codewords merely need their error probability to vanish in the blocklength . one question that seems intimately connected to the streaming delay - reliability tradeoff is how large the red alert error exponent can be made for a fixed blocklength codebook of a given rate . beyond this streaming motivation , the red alert problem is also connected to certain sensor network scenarios . for example , consider a setting where sensors must send regular updates to the basestation using as little power as possible , i.e. , using the standard codewords . if an anomaly is detected , the sensors are permitted to transmit at higher power in order to alert the basestation with high reliability , which corresponds to our red alert problem . prior work has characterized the red alert exponent for discrete memoryless channels ( dmcs ) @xcite . in this paper , we determine the red alert exponent for point - to - point additive white gaussian noise ( awgn ) channels that operate under block power constraints on both the regular and red alert messages . we derive matching upper and lower bounds on the red alert exponent with a focus on the resulting high - dimensional geometry of the decoding regions . our code construction can be viewed as a generalization of that used in the discrete case . previous studies on protecting a special message over a dmc have relied on some variant of the following code construction . first , designate the special codeword to be the repetition of a particular input symbol . then , generate a fixed composition codebook at the desired rate . this composition is chosen to place the `` standard '' codewords as far as possible from the special codeword ( as measured by the kullback - leibler ( kl ) divergence between induced output distributions ) while still allocating each codeword a decoding region large enough to ensure a vanishing probability of error . by construction , the rest of the space is given to the special codeword . early work by kudryashov used this strategy to achieve very high error exponents in the bit error setting under an expected delay constraint @xcite . in @xcite , borade , nakibolu , and zheng study `` bit''-wise and `` message''-wise unequal error protection ( uep ) problems and error exponents . the red alert problem is a message - wise uep problem in which one message is special and the remaining messages are standard . while @xcite focuses on general dmcs near capacity , lemma 1 of that paper develops a general sharp bound on the red alert exponent for dmcs at any rate below capacity ( both with and without feedback ) . specializing to the exponent achieved at capacity , let @xmath0 denote the input alphabet , @xmath1 the channel transition matrix , and @xmath2 the capacity - achieving output distribution of the dmc . then , the optimal red alert exponent at capacity is @xmath3 where @xmath4 is the kl divergence . we also mention recent work by nakibolu _ _ @xcite that considers the generalization where a strictly positive error exponent is required of the standard messages . for the binary symmetric channel ( bsc ) , the optimal red alert exponent has a simple and illustrative form . this exponent can be inferred from the general expression in ( * ? ? ? * lemma 1 ) or via a direct proof due to sahai and draper @xcite ( which appeared concurrently with the conference version @xcite of @xcite ) . let @xmath5 denote the crossover probability of the bsc and @xmath6 the probability that a symbol in the codebook is a one . then , the optimal red alert exponent as a function of rate @xmath7 for the bsc is @xmath8 where @xmath9 , @xmath10 , and @xmath11 . csiszr studied a related problem where multiple special messages require higher reliability in @xcite . upper bounds for multiple special messages with different priority levels were also developed in @xcite . in @xcite , borade and sanghavi examined the red alert problem from a coding theoretic perspective . as shown by wang @xcite , similar issues arise in certain sparse communication problems where the receiver must determine whether a codeword was sent or the transmitter was silent . the fundamental mechanism through which high red alert exponents are achieved is a binary hypothesis test . by designing the induced distributions at the output of the channel to be far apart as measured by kl divergence , we can distinguish whether the red alert or some standard codeword was sent . the test threshold is biased to minimize the probability of missed detection and is analyzed via an application of stein s lemma . this sort of biased hypothesis test occurs in numerous other communication settings with feedback , such as @xcite and , as mentioned earlier , these codes are also used as a component in streaming data systems ( see , for instance , @xcite ) . there is also a rich literature on the interplay between hypothesis testing and information theory , which we can not do justice to here ( see , for instance , @xcite ) . first , we mention some of our notational choices . we will use boldface lowercase letters to denote column vectors , @xmath12 to denote the all zeros vector , and @xmath13 to denote the all ones vector . throughout the paper , the @xmath14 function is taken to be the natural logarithm and rate is measured in nats instead of bits . we use @xmath15 to denote the euclidean norm of the vector @xmath16 . the transmitter has a _ message _ @xmath17 that it wants to convey to the receiver . one of the messages , @xmath18 , is a red alert message that will be afforded extra error protection . we assume the red alert message is chosen with some probability greater than @xmath19 and the remaining messages are chosen with equal probability . the _ encoder _ @xmath20 maps the message @xmath21 into a length-@xmath22 real - valued codeword @xmath16 for transmission over the channel , @xmath23 . let @xmath24 denote the codeword used for message @xmath21 and let @xmath25 denote the entire codebook , @xmath26 . the codebook must satisfy both an average block power constraint across codewords , @xmath27 in addition , the red alert codeword must satisfy a less stringent power constraint , @xmath28 for some @xmath29 . the _ rate _ of the codebook is @xmath30 nats per channel use . note that our codebook average power constraint ( [ e : avgpower ] ) is less strict that the usual block power constraint @xmath31 . our achievable scheme can be easily modified to meet this constraint using expurgation . furthermore , our red alert power constraint ( [ e : alertpower ] ) is less strict than a peak power constraint @xmath32 @xmath33 , where @xmath34 denotes the @xmath35th symbol of the red alert codeword . our scheme sets the red alert codeword to be @xmath36 , which naturally satisfies a peak power constraint . therefore , our main results hold under an average power constraint and peak power constraint as well . we omit the red alert codeword from the average block power constraint for the sake of simplicity . another possibility would be to consider only an average block power constraint over both the standard and red alert codeword . this would lead to two different tensions between maximizing the red alert exponent and maximizing the rate . the first would be the allocation of the decoding regions and the second would be the allocation of power based on the probability of a red alert message . by using two separate power constraints , we can state our results in a simpler form that does not depend on the red alert probability . the _ channel _ outputs the transmitted vector , corrupted by independent and identically distributed ( i.i.d . ) gaussian noise : @xmath37 where @xmath38 for some noise variance @xmath39 . the signal observed by the receiver is sent into a _ decoder _ which produces an estimate @xmath40 of the transmitted message @xmath21 , @xmath41 . we are concerned with three quantities , the _ probability of missed detection _ of the red alert message @xmath42 , the _ probability of false alarm _ @xmath43 , and the _ average probability of error _ of all other messages @xmath44 : @xmath45 we say that a _ red alert exponent _ of @xmath46 is achievable if for every @xmath47 and @xmath22 large enough , there exists a rate @xmath48 encoder and a decoder such that @xmath49 in other words , we would like the red alert codeword to have as large an error exponent as possible while keeping the other error probabilities small . the standard codewords do not need to have a positive error exponent of course , the rate must be lower than the awgn capacity , @xmath50 , where @xmath51 we now review some basic facts of high - dimensional geometry that will be useful in our analysis . let @xmath52 denote the @xmath22-dimensional ball centered at @xmath53 with radius @xmath54 . recall that the volume of @xmath52 is @xmath55 where @xmath56 is the gamma function ( * ? ? ? we define @xmath57 to be the surface of @xmath52 . its surface area ( or , more precisely , the @xmath58-dimensional volume of its surface ) is @xmath59 ( * ? ? ? 1 , eq . 19 ) . the dimension of the @xmath60 function will always be clear from the context . we also define @xmath61 to be the spherical shell centered at @xmath62 from radius @xmath63 to @xmath64 . the angle between two @xmath22-dimensional vectors @xmath62 and @xmath65 is @xmath66 where @xmath67 takes values between @xmath19 and @xmath68 . let @xmath69 denote the @xmath22-dimensional cone with its origin at @xmath62 , its center axis running from @xmath62 to @xmath65 , and of half - angle @xmath70 which takes values from @xmath19 to @xmath71 . the solid angle @xmath72 of an @xmath22-dimensional cone of half - angle @xmath70 is the fraction of surface area that it carves out of an @xmath22-dimensional sphere , @xmath73 note that the solid angle is the same for any sphere radius @xmath54 . [ l : surfaceratio]the solid angle of a cone with half - angle @xmath70 satisfies @xmath74 see the math leading up to equation 28 in @xcite for a proof . in the binary case , the simplest characterization of the optimal codebook is a statistical one : the red alert codeword is the zero vector and the remaining codewords are of a constant composition . from one perspective , this can be visualized as placing the red alert codeword in the `` center '' of the space with the other codewords encircling it ( see figure [ f : topdown ] ) . this corresponds to choosing the red alert codeword to be the all zeros ( or all ones ) vector . the standard codewords are generated using the distribution that maximizes the kl divergence between output distributions while still supporting a rate @xmath48 . while this two - dimensional illustration is quite useful for understanding the binary case , it can be misleading in the gaussian case . specifically , it suggests that we should place the red alert codeword at the origin which turns out to be suboptimal . ( -30,-30)(30,30 ) ( 0,0)30 ( 0,0)(-2,-2)(2,2 ) another way of looking at the binary construction is to visualize each fixed composition as a parallel ( or circle of constant latitude ) on a sphere ( see figure [ f : sideview ] ) . that is , the code lives on the hamming cube in @xmath22 dimensions , which can be imagined as a sphere by taking the all zeros and all ones vectors as the two poles and specifying the parallels by their hamming weight . from this viewpoint , the binary construction sets the red alert codeword to be one of the poles and chooses the remaining codewords on the furthest parallel that can support a codebook of rate @xmath48 . this perspective leads naturally to the right construction for the gaussian case . essentially , the standard codewords are placed uniformly along a constant parallel . this can be achieved by generating the standard codewords using a capacity - achieving code with a fraction @xmath75 of the total power . the red alert codeword is placed at the furthest limit of the red alert power constraint ( e.g. , at @xmath76 ) and the standard codewords are offset in the opposite direction ( e.g. , by @xmath77 ) . see figure [ f : offsetcodebook ] for an illustration . in the high - dimensional limit , most of the codewords will live on a parallel , thus mimicking the binary construction . this scheme leads us to the optimal red alert exponent . [ t : main ] for an awgn channel with red alert power constraint @xmath78 , average power constraint @xmath79 , and rate @xmath48 , the optimal red alert exponent is @xmath80 we prove achievability in lemma [ l : achievable ] and provide a matching upper bound in lemma [ l : converse ] . in the conference version of this paper @xcite , we used a different code construction that lead to a smaller achievable red alert exponent . the codewords were generated uniformly on the sphere of radius @xmath81 and we only kept those that fell within a cone of appropriate half - angle . this type of construction turns out not to achieve as dense a packing as the construction used in this paper . in appendix [ a : cone ] , we explore the reasons why this occurs in the binary case . in appendix [ a : coneexponent ] , we state the achievable red alert exponent for the conical construction . ( -50,-15)(55,15 ) ( -50,-2)(-46,2 ) ( 40,0)15 ( -48,0)(0,0 ) ( -24,-4 ) ( 0,0)(40,0 ) ( 12,-4 ) ( 40,0)(40,15 ) ( 44,3 ) ( 0,0)1.3 our codebook construction for @xmath25 consists of the following steps : 1 . choose @xmath82 so that @xmath83 . the red alert codeword is placed at the boundary of the red alert power constraint , @xmath84 . 3 . choose @xmath85 so that @xmath86 and choose @xmath87 so that @xmath88 4 . draw @xmath89 codewords @xmath90 i.i.d . according to a gaussian distribution with mean zero and variance @xmath91 . 5 . to each of these codewords , add an offset @xmath92 so that the transmitted codeword for each message ( other than @xmath93 ) is @xmath94 . we will show that this procedure yields a random codebook @xmath25 whose false alarm probability and average probability of error are both less than @xmath95 . afterwards , we will characterize the probability of missed detection for the red alert codeword . this will in turn imply the existence of a good fixed codebook . in this section , we will show that the red alert error exponent stated in theorem [ t : main ] is achievable . we begin by stating useful large deviations bounds that will play a role in both the proof of the achievability and of the converse . next , we show that any standard codeword plus noise lies at a certain distance from the red alert codeword with high probability . afterwards , we argue that , with high probability , any standard codeword plus noise is contained in a cone of a certain half - angle that is centered on the red alert codeword . by combining the distance and angle bounds , we can constrain the decoding region for the standard codewords to the intersection of a cone with a shell . the remainder of @xmath96 can thus be allocated to the decoding region for the red alert codeword , for which we will bound the resulting probability of a missed detection . our upper and lower bounds on the probability of error are proven by deriving bounds on the size and shape of the decoding regions and then applying cramr s theorem to get large deviations bounds . define @xmath97 to be the moment generating function of a random variable @xmath98 , @xmath99 \ , \end{aligned}\ ] ] and @xmath100 to be the fenchel - legendre transform ( * ? ? ? * definition 2.2.2 ) of @xmath101 , @xmath102 \ . \end{aligned}\ ] ] [ t : cramer ] let @xmath103 be the normalized sum of @xmath22 i.i.d . variables @xmath104 with finite mean and rate function @xmath100 . then , for every closed subset @xmath105 , @xmath106 and , for every open subset @xmath107 , @xmath108 see , for instance , ( * ? ? ? * theorem 2.2.3 ) for a proof . we will be particularly interested in how this bound applies to the length of i.i.d . gaussian vectors , which corresponds to setting the @xmath109 to be chi - square random variables ( with one degree of freedom ) . the moment generating function for such random variables is @xmath110 which yields a rate function of @xmath111 . the following lemma formalizes the notion that the squared @xmath112-norm of an i.i.d . gaussian vector concentrates sharply around its variance . thus , for large @xmath22 , the decoding region can be restricted to a thin spherical shell . [ l : gaussiannorm ] let @xmath113 be a length-@xmath22 vector with i.i.d . zero - mean gaussian entries of variance @xmath114 . then , for any @xmath115 , @xmath116 and , for any @xmath117 , @xmath118 see appendix [ a : distanceproofs ] for the proof . recall that the q - function returns the probability that a scalar gaussian random variable with mean zero and unit variance is greater than or equal to @xmath119 , @xmath120 and is upper bounded as @xmath121 the next lemma is about the well - known fact that an i.i.d . gaussian vector is approximately orthogonal to any fixed vector . -shell of power @xmath122 that is offset away from the origin with power @xmath123 . thus , with high probability , any random codeword meets the power constraint.,width=211 ] [ l : gaussianortho ] let @xmath113 be a length-@xmath22 vector with i.i.d . zero - mean gaussian entries with variance @xmath114 and let @xmath62 be a length-@xmath22 vector with @xmath124 for some fixed @xmath125 . then , for any @xmath126 and @xmath22 large enough , @xmath127 see appendix [ a : distanceproofs ] for the proof . in figure [ f : offsetcb ] , the codebook is illustrated from the perspective of the origin . using the above lemma , it can be shown that all but a vanishing fraction of codewords have power close to @xmath79 and are nearly orthogonal with respect to any fixed vector . we now characterize how far away a codeword plus noise is from the red alert codeword with high probability . [ l : distance ] for any @xmath126 and @xmath22 large enough , the distance from the red alert codeword to the codeword for a standard message , @xmath128 , plus noise is at least @xmath129 with high probability , @xmath130 see appendix [ a : distanceproofs ] for the proof . we now upper bound the @xmath22-dimensional angle between a fixed vector and the same vector plus i.i.d . gaussian noise . [ l : gaussianangle ] let @xmath113 be a length-@xmath22 vector with i.i.d . zero - mean gaussian entries with variance @xmath114 and let @xmath62 be a length-@xmath22 vector with @xmath124 for some fixed @xmath125 . for any @xmath126 and @xmath22 large enough , the probability that the angle between @xmath62 and @xmath131 exceeds @xmath132 is upper bounded by @xmath133 , @xmath134 see appendix [ a : angleproofs ] for the proof . and the angle @xmath135 between a codeword plus noise and the red alert codeword.,width=259 ] in figure [ f:3dangles4 ] , we have depicted the distance @xmath129 and the angle @xmath135 from the red alert codeword to a standard codeword plus noise . notice that both the noise and the codewords are ( nearly ) orthogonal to the axis along which the red alert codeword lies . now consider a cone centered on the red alert codeword that contains a standard codeword plus noise with high probability . the next lemma upper bounds the required half - angle for the cone . [ l : angle ] let @xmath136 denote the cone centered on the red alert codeword with axis running towards the origin and half - angle @xmath135 . for any @xmath126 , @xmath128 , and @xmath22 large enough , if the half - angle @xmath135 is greater than or equal to @xmath137 then the cone contains the codeword for message @xmath21 plus noise with high probability , i.e. , @xmath138 see appendix [ a : angleproofs ] for the proof . now that we know the decoding region can be confined to a conical shell , we can bound the probability of missed detection for the red alert codeword . [ l : achievable ] for any rate @xmath48 , the following red alert exponent is achievable @xmath80 choose @xmath126 . in lemma [ l : distance ] , @xmath129 is a lower bound on the distance between the red alert codeword and a standard codeword plus noise . from lemma [ l : angle ] , we have an upper bound on the half - angle needed to capture a standard codeword plus noise in the cone @xmath139 centered on the red alert codeword . if the received vector lies in the cone and is at least distance @xmath129 from the red alert codeword , then the decoder assumes the red alert message was not transmitted . otherwise , it declares that the red alert message was sent . for @xmath22 large enough , we know that the probability that a random codeword plus noise , @xmath140 , leaves this region is at most @xmath95 . therefore , the probability of false alarm ( averaged over the randomness in the codebook ) is upper bounded by @xmath95 . if the received vector falls in the decoding region for standard messages , we simply subtract the offset @xmath141 and apply a maximum likelihood decoder to make an estimate of the transmitted message . since the rate of the codebook is chosen to be slightly less than the capacity ( for the power level @xmath91 ) , it is straightforward to show that the average probability of error for a given message is at most @xmath95 . since the average false alarm probability and average error probability are small , it follows that there exists at least one fixed codebook with a small false alarm probability and average error probability . we now turn to upper bounding the probability of missed detection . assume the red alert codeword is transmitted . define @xmath142 where @xmath143 is specified by step 3 ) of the codebook construction in section [ s : codebook ] . using lemma [ l : gaussiannorm ] , the probability that the noise pushes the red alert codeword further than @xmath129 ( as specified in lemma [ l : distance ] ) can be upper bounded by @xmath144 the probability that the received vector falls into the cone of half - angle @xmath135 is given by the fraction of surface area of a sphere carved out by the cone . using lemma [ l : surfaceratio ] , this can be calculated as @xmath145 pulling terms into the exponent we get @xmath146 for @xmath22 large enough , we get that the probability is upper bounded by @xmath147 . since the noise is an i.i.d . gaussian vector , its magnitude and direction are independent . therefore , the probability of missed detection is upper bounded as @xmath148 for @xmath22 large enough . for @xmath143 and @xmath133 small enough and @xmath22 large enough , the exponent @xmath149 can be made equal to @xmath150 finally , we can solve for @xmath75 in terms of @xmath48 to get @xmath151 . substituting this into the expression above yields the desired result . note that at @xmath152 , the coherent gain @xmath153 , which is the largest benefit we could hope for . at @xmath154 , the coherent gain vanishes . we can interpret our achievability result from a hypothesis testing perspective . let @xmath155 denote the event that a standard codeword is transmitted and let @xmath156 denote the event that the red alert codeword is transmitted . under @xmath155 , the entries of @xmath157 are i.i.d . according to a gaussian distribution with mean @xmath158 and variance @xmath159 . under @xmath156 , the entries are i.i.d . gaussian with mean @xmath160 and variance @xmath114 . using the chernoff - stein lemma ( * ? ? ? * theorem 11.8.3 ) , we can bound the missed detection probability of the optimal hypothesis test via the kl divergence between the two distributions , @xmath161 . a bit of calculation will reveal that this kl divergence corresponds exactly to the red alert exponent . one can obtain the same exponent by plugging these distributions into the red alert exponent expression from ( * ? ? ? * lemma 1 ) . however , this does not in itself constitute a proof as the results of @xcite are for dmcs without cost constraints . we now develop an upper bound on the red alert exponent . our bound relies on the fact that , in order to recover the standard messages reliably , we must allocate a significant volume of the output space for decoding them , which contributes to the probability of missed detection . an overview of the main steps in the proof is provided below . * in lemma [ l : shell ] , we argue that a constant fraction of the codewords live in a thin shell and strictly satisfy the power and error constraints . * with high probability , the standard codewords plus noise are concentrated in a thin shell . lemma [ l : volume ] establishes this fact as well as the minimum volume required for the decoding region to attain a given probability of error . * to minimize the probability of missed detection , we should pack this volume into the thin shell to maximize the distance from the red alert codeword ( see figure [ f : converseregion ] for an illustration ) . lemma [ l : redalert ] bounds the distance and angle from the red alert codeword to the resulting decoding region ( see figure [ f : converse ] for an illustration ) . * finally , in lemma [ l : converse ] , we bound the probability that the noise carries the red alert codeword into the decoding region for the standard codewords . [ l : shell ] assume that a sequence of codebooks satisfies the average block power constraint @xmath79 and has average probability of error @xmath44 that tends to zero . then for any @xmath162 and @xmath22 large enough , there exists a shell of width @xmath163 that contains @xmath164 codewords , each with probability of error at most @xmath165 , and average power at most @xmath166 . see appendix [ a : converseproofs ] for the proof . [ l : volume ] assume that , for some @xmath167 , @xmath168 codewords , each with probability of error at most @xmath169 lie in the shell @xmath170 . then , for @xmath22 large enough , the decoding region for these codewords must include a subset of the noise - inflated shell @xmath171 with volume at least @xmath172 see appendix [ a : converseproofs ] for the proof . [ l : redalert ] assume that a sequence of codebooks has rate @xmath48 and an average probability of error @xmath44 that tends to zero as @xmath22 increases . then , for sufficiently small @xmath173 and @xmath22 large enough , the probability of missed detection @xmath42 is lower bounded by the probability that the noise vector has squared norm @xmath174 between @xmath175 and @xmath176 and lies at an angle @xmath177 between @xmath178 and @xmath179 where @xmath180 consider the standard codewords from a red alert codebook . from lemma [ l : shell ] , for any @xmath181 and @xmath22 large enough , at least @xmath164 codewords with power at most @xmath166 and probability of error at most @xmath169 must lie in a shell @xmath182 for some @xmath183 . from lemma [ l : volume ] , it follows that the decoding region for these codewords falls within the noise - inflated shell @xmath171 and has volume at least @xmath184 . ( -52,-44)(53,44 ) ( 25,-44)(55,44 ) ( -50,0)87 - 3232 ( -48,-2)(-52,2 ) ( -50,2)(-40,15 ) ( -40,23)red alert ( -40,18)codeword ( 15,0)38 - 67.967.9 ( 15,0)38 ( -50,0)87 - 23.7 - 15.5 ( -50,0)8715.523.7 ( 15,0)30 - 52.152.1 ( 15,0)3052.1 - 52.1 ( -50,0)87 - 3030 ( -50,0)(37,0 ) ( 0,3)@xmath185 ( 49,0 ) ( 40,-40 ) ( -28,-28)(-15,-15 ) ( -28,-32)noise - inflated shell to get our lower bound , we need to pack this volume in the noise - inflated shell such that it minimizes @xmath42 . since the noise vector is i.i.d . gaussian , the probability that the red alert codeword is pushed to a certain point is determined solely by a decreasing function of the distance . let @xmath186 denote the set of all points at distance @xmath185 or greater from the red alert codeword @xmath187 the optimal volume packing corresponds to the intersection of the set @xmath186 and the noise shell + @xmath188 with @xmath185 chosen such that the volume of the set is equal to @xmath184 . let @xmath189 denote the resulting region and see figure [ f : converseregion ] for an illustration . let @xmath190 denote the set of points in @xmath189 that sit at the minimum distance to the red alert codeword , @xmath191 and let @xmath192 be any of these points . let @xmath193 and @xmath194 denote the distance and angle from the red alert codeword @xmath195 to @xmath196 . we now seek to bound these quantities through a bound on the angle from the origin to @xmath196 . ( -52,-38)(53,38 ) ( -48,-2)(-52,2 ) ( 15,0)38 - 67.967.9 ( 15,0)38 ( 15,0)30 ( 15,0)38 - 5959 ( -50,0)87 - 23.7 - 15.5 ( -50,0)8715.523.7 ( 15,0)30 - 52.152.1 ( 15,0)3052.1 - 52.1 ( -50,0)92 - 2020 ( -50,0)89 - 2020 ( -50,0)(34.5,32.8 ) ( -50,0)(34.5,-32.8 ) ( 15,0)1.3 ( -50,0)(45,0 ) ( -50,0)19020.5 ( -28,4)@xmath135 ( 15,0)8059 ( 25,6)@xmath70 ( 0,22.5)@xmath129 ( 20,14 ) ( 29.5,35)(0,0)1.2 ( 30.5,39)@xmath196 ( 34.5,32.8)(0,0)1.2 ( 34.5,36.3)@xmath197 ( 35.7,23.9)(0,0)(3,7 ) ( 35.7,-23.9)(0,-7)(3,0 ) ( 49,0 ) let @xmath198 denote the half - angle of a cone , centered on the origin that contains the region @xmath189 ( and thus includes @xmath196 ) . the volume of this cone must be at least equal to that of @xmath189 since @xmath189 is a subset of the noise shell . therefore , @xmath198 is lower bounded by the half - angle @xmath70 of a cone whose volume is equal to the volume of @xmath189 ( see figure [ f : converse ] for an illustration ) . combining ( [ e : spherevol ] ) and lemma [ l : surfaceratio ] , for @xmath22 large enough , the volume of this cone is upper bounded by @xmath199 now , since we require this quantity to exceed @xmath184 , we can lower bound @xmath70 by @xmath200 we can further lower bound @xmath70 by setting @xmath201 to its maximum value @xmath202 , @xmath203 thus , for any @xmath126 , @xmath204 and @xmath205 small enough , and @xmath22 large enough , @xmath198 is lower bounded as follows @xmath206 the distance @xmath193 from @xmath195 to @xmath196 is upper bounded by the distance @xmath129 to a point @xmath197 that lies on the intersection of the outer shell ( at distance @xmath207 from the origin ) and the cone of half - angle @xmath70 . without loss of generality , assume that the red alert codeword is placed at @xmath208 for some @xmath209 . the direction of the red alert codeword is not important since we will always fill the noise shell relative to this direction . then @xmath210 is at least @xmath211 for any @xmath126 , @xmath205 and @xmath204 small enough , and @xmath22 large enough , this quantity is itself upper bounded by @xmath212 the half - angle @xmath135 of a cone , centered on the red alert codeword , that contains the point @xmath197 is lower bounded by @xmath213 which , for any @xmath214 , @xmath205 , @xmath204 , and @xmath133 small enough , and @xmath22 large enough is itself lower bounded by @xmath215 the probability of missed detection decreases if the distance @xmath193 from @xmath195 to @xmath196 is increased . the angle @xmath194 will simultaneously decrease . thus , by setting @xmath216 , we further lower bound the probability of missed detection . using the relation @xmath217 combined with ( [ e : theta ] ) , we find that @xmath218 . plugging in @xmath219 and @xmath70 , we obtain the following upper bound on @xmath220 : @xmath221 and the following lower bound on @xmath194 : @xmath222 finally , it follows that , for @xmath173 small enough ( but greater than @xmath19 for finite @xmath22 ) and @xmath22 large enough , the optimal packing contains all points from squared distance @xmath223 to @xmath224 from the red alert codeword and angle @xmath225 to @xmath226 where @xmath129 and @xmath135 are as in the statement of the theorem . thus , the probability of missed detection is lower bounded by the event that the noise falls into this region . [ l : converse ] for any rate @xmath48 , the red alert exponent is upper bounded by @xmath227 lemma [ l : redalert ] established that the probability of missed detection is lower bounded by the event that the noise has squared length between @xmath175 to @xmath176 and angle between @xmath178 and @xmath228 for some @xmath173 that tends to @xmath19 as @xmath22 tends to infinity . we now lower bound the probability of this event . define @xmath229 since the magnitude and angle of an i.i.d . gaussian vector are independent , the probability of missed detection is lower bounded as follows : @xmath230 by lemma [ l : surfaceratio ] , for @xmath22 large enough , the second term in the product can be lower bounded by @xmath231 which , for @xmath22 large enough , is itself lower bounded by @xmath232 now , substituting in the lower bound on @xmath135 from lemma [ l : redalert ] , we arrive at the following lower bound @xmath233 using the upper bound on @xmath129 from lemma [ l : redalert ] and applying theorem [ t : cramer ] for chi - square random variables ( and noting that @xmath173 and @xmath204 go to zero as @xmath22 goes to infinity ) , it follows that @xmath234 combining this with the lower bound on the angle event in ( [ e : angleexp ] ) , the exponent of the probability of missed detection is lower bounded by @xmath235as desired . db with @xmath236 . an upper bound on the point - to - point awgn error exponent is provided for comparison . , width=364 ] db with @xmath236 . an upper bound on the point - to - point awgn error exponent is provided for comparison . , width=364 ] in figures [ f : redalertplot1 ] and [ f : redalertplot2 ] , we have plotted the optimal red alert exponent for @xmath237db and @xmath238db , respectively , with red alert power constraints @xmath239 , and @xmath240 . for comparison , we have also plotted an upper bound on the awgn point - to - point error exponent from ( * ? ? ? * equation 4 ) . notice that the red alert exponent can be quite large at capacity , even when the red alert power constraint is equal to the average power constraint . we have developed sharp bounds on the error exponent for distinguishing a single special message from @xmath241 standard messages over an awgn channel . as discussed in the introduction , these bounds can be used to characterize the performance of certain data streaming architectures , where each bit must be decoded after a given delay . an interesting question for future study is how well a single special message can be protected at a given finite blocklength , i.e. , understanding the limits of unequal error protection in the non - asymptotic regime @xcite . the squared euclidean distance is the sum of @xmath22 i.i.d . squared gaussian random variables with variance @xmath114 . therefore , @xmath242 is the sum of @xmath22 i.i.d . chi - square random variables . applying theorem [ t : cramer ] and plugging in the chi - square rate function of @xmath243 , it follows that @xmath244 substituting in @xmath245 yields the first bound and @xmath246 yields the second . first , we write the probability that @xmath247 is greater than @xmath248 in terms of the q - function , @xmath249 substituting @xmath250 yields , @xmath251 which can be driven arbitrarily close to zero for @xmath22 large enough . we simply wish to bound the length of the vector from the special codeword to a standard codeword plus noise , @xmath252 . by expanding terms , we obtain : @xmath253 the first term is @xmath254 . the second term is the inner product of a fixed vector @xmath255 and an i.i.d . gaussian vector @xmath256 since @xmath257 is an element of a random gaussian codebook . thus , using lemma [ l : gaussianortho ] , it can be shown that the probability that this inner product is less than @xmath258 is at most @xmath259 for @xmath22 large enough . the third term is the squared norm of an i.i.d . gaussian vector with mean zero and variance @xmath260 . from lemma [ l : gaussiannorm ] , it follows that @xmath261 is less than @xmath262 with probability at most @xmath259 for @xmath22 large enough . combining these three bounds completes the proof . the angle between @xmath62 and @xmath263 is @xmath264 from lemma [ l : gaussianortho ] , for any @xmath265 and @xmath22 large enough , the probability that @xmath266 is at most @xmath204 . therefore , since @xmath267 we have that @xmath268 with probability at most @xmath204 . combining lemmas [ l : gaussiannorm ] and [ l : gaussianortho ] , we can also show that the probability that @xmath269 is at most @xmath204 for @xmath22 large enough . thus , the probability that @xmath270 is at most @xmath271 . choosing @xmath204 small enough yields the desired result . the angle between the axis of the cone and the standard codeword plus noise is @xmath272 since @xmath273 is a decreasing function of @xmath274 , an upper bound on the angle can be obtained by lower bounding @xmath274 . we will do this by lower bounding the numerator and upper bounding the denominator ( with high probability ) . expanding the numerator yields : @xmath275 the first term is simply @xmath276 . the second term is the inner product of a fixed vector and an i.i.d . gaussian vector . thus , using lemma [ l : gaussianortho ] , it can be shown that for any @xmath265 and @xmath22 large enough , the probability that the second term is less than @xmath277 is at most @xmath204 . the denominator is composed of two terms . the first is simply @xmath278 . following the proof of lemma [ l : distance ] , it can be shown that the second term @xmath279 is greater than @xmath280 with probability at most @xmath204 . combining these bounds , we get that the probability that @xmath274 is less than @xmath281 with probability at most @xmath282 . thus , so long as the half - angle @xmath135 is greater than or equal to @xmath283 the cone contains @xmath140 with probability at least @xmath284 for @xmath22 large enough . applying the trigonometric identity @xmath285 completes the proof . observe that at least one codeword has power at most @xmath79 , otherwise the average will be larger than @xmath79 . if we remove this codeword s contribution from the average , the remaining codewords have average power at most @xmath286 . now , we can find a codeword whose power must be at most @xmath286 . removing this codeword yields an average of @xmath287 . continuing this process , we can remove @xmath288 codewords that each have power at most @xmath289 . by the same argument , we can find @xmath290 codewords that each have probability of error at most @xmath169 . therefore , at least @xmath291 codewords must satisfy both these constraints simultaneously . the selected codewords live in the sphere of radius @xmath292 . we partition this sphere into shells of width @xmath205 each . it follows that at least one of these shells must contain @xmath293 codewords . finally , select @xmath22 large enough so that @xmath294 . [ proof of lemma [ l : volume ] ] assume that one of the codewords from the shell @xmath170 is transmitted . it follows from lemma [ l : gaussiannorm ] , that for any @xmath295 and @xmath22 large enough , the probability that @xmath296 is larger than @xmath297 or smaller than @xmath298 is upper bounded by @xmath299 . if the noise lands outside this `` noise shell , '' then we will assume that the transmitted codeword is decoded correctly . however , each codeword still needs to capture @xmath300 probability inside the shell to ensure the error probability does not exceed @xmath301 . now , consider the volume required for decoding a single codeword reliably . since the noise is i.i.d . gaussian , its probability distribution is rotationally invariant . this implies that the shape that uses the least volume to capture a given probability of error is a sphere centered on the codeword . let @xmath302 be the radius of this sphere . by lemma [ l : gaussiannorm ] , if @xmath303 , the probability that the noise falls inside this sphere goes to zero exponentially in @xmath22 which implies the probability of error goes to one . therefore , for @xmath22 large enough , the probability of error will always exceed the desired probability of error ( which is assumed to be bounded away from one ) . using ( [ e : spherevol ] ) , we get that the decoding region of each codeword must have volume at least @xmath304 for any @xmath295 . we find that we will need a volume of at least @xmath305 to reliably decode these codewords . we now develop some intuition for why the offset construction of section [ s : codebook ] is a better construction than the conical construction we used in our earlier work @xcite . the difference between these two constructions is easier to understand in a discrete setting so we will analyze the corresponding constructions for a bsc with crossover probability @xmath5 . for ease of analysis , we will calculate rate in bits per channel use ( rather than nats per channel use ) . first , recall that the bsc red alert exponent can be attained using a fixed composition codebook . specifically , each of the @xmath241 codewords is drawn independently and uniformly from the set of weight-@xmath306 binary sequences . if the rate is less than the induced mutual information , the average probability of error can be driven to zero @xmath307 the red alert codeword is taken to be all the all zeros vector . the decoder runs a hypothesis test between the two possible output distributions , bernoulli@xmath308 and bernoulli@xmath309 . the error exponent for the probability of missed detection is the kl divergence between the two distributions , @xmath310 as shown in @xcite , this is the optimal red alert exponent . we can construct a _ conical code _ of parameter @xmath311 by first drawing @xmath312 codewords i.i.d . according to a bernoulli@xmath313 distribution for some @xmath47 . let @xmath25 denote the resulting set of codewords . to guarantee the same red alert exponent , we only keep those codewords with hamming weight @xmath306 or greater and set the red alert codeword to be the all zeros vector . we now bound the rate of this construction . using theorem [ t : cramer ] , it can be shown that the probability that a bernoulli@xmath313 sequence @xmath16 has hamming weight at least @xmath306 is upper bounded as @xmath314 take @xmath315 to be the set of subset of codewords in @xmath25 with hamming weight @xmath306 or greater . using ( [ e : weightprob ] ) , the expected size of @xmath315 is upper bounded by @xmath316 & \leq 2^{n(c - d(q\\|0.5 ) -\epsilon ) } \\ & = 2^{n(1 - h_b(p ) - ( 1 - h_b(q ) ) -\epsilon ) } \\ & = 2^{n(h_b(q ) - h_b(p ) - \epsilon ) } \ .\end{aligned}\ ] ] it can be shown with a chernoff bound that the probability @xmath315 contains significantly more codewords vanishes doubly exponentially in @xmath22 . furthermore , it can be shown that the average probability of error of @xmath315 vanishes with @xmath22 . therefore , the rate of the conical codebook is @xmath317 for a red alert exponent of @xmath318 . now , observe that @xmath319 ( unless either @xmath6 or @xmath5 is equal to @xmath320 ) so @xmath321 , meaning that the rate of the conical construction is strictly less than the ( optimal ) constant composition construction . intuitively , this means that the usual i.i.d . bernoulli@xmath322 construction used to approach capacity does not pack codewords of higher ( or lower ) weights efficiently . constraining the weight of codewords is essential to the hypothesis test that leads to the red alert exponent . the constant composition ( or offset ) construction is successful since it optimizes the packing of codewords of a given weight . a similar phenomenon occurs in the awgn setting as shown below . for completeness , we review the awgn conical code that we proposed in @xcite and the resulting red alert exponent . the construction is comprised of three main steps : 1 . place the red alert codeword at the limit of the red alert power constraint , @xmath323 . 2 . draw @xmath324 codewords i.i.d . according to a gaussian distribution with mean @xmath19 and variance @xmath325 . 3 . of these codewords , only keep the first @xmath241 that lie in the cone @xmath326 where @xmath327 . ( if there are fewer than @xmath241 such codewords , declare an error . ) it can be shown that with high probability the resulting codebook contains @xmath241 codewords inside the cone of half - angle @xmath70 . we now turn to bounding the distance and angle from the standard decoding region to the red alert codeword . the distance can be bounded using the techniques used to prove lemma [ l : distance ] . it follows that for any @xmath126 and @xmath22 large enough , the squared distance from the red alert codeword to a standard codeword plus noise is at least @xmath129 with high probability , @xmath328 substituting in @xmath329 , we get that @xmath210 is equal to @xmath330 similarly , the techniques from lemma [ l : angle ] can be used to bound the angle . let @xmath136 denote the cone centered on the red alert codeword with axis running towards the origin and half - angle @xmath135 . for any @xmath126 and @xmath22 large enough , if the half - angle @xmath135 is larger than @xmath331 then the cone contains the codeword for message @xmath21 plus noise with high probability , i.e. , @xmath332 finally , these two bounds can be combined , as in the proof of lemma [ l : achievable ] , to get an an achievable red alert exponent of @xmath333 , @xmath334 , and @xmath335db with @xmath336 . , width=364 ] in figure [ f : conevsoffset ] , we have plotted this red alert exponent alongside the optimal one derived via the offset construction for average power constaints @xmath337 , @xmath338 and @xmath335db with @xmath336 . the authors would like to thank the anonymous reviewers whose suggestions improved the presentation of this work . they would also like to thank b. nakibolu for insightful discussions on the connections between this work and the discrete memoryless case . b. nakibolu , s. k. gorantla , l. zheng , and t. p. coleman , `` bit - wise unequal error protection for variable length block codes with feedback , '' _ ieee transactions on information theory _ , submitted january 2011 . available online : http://arxiv.org/abs/1101.1934 . s. k. gorantla , b. nakibolu , l. zheng , and t. p. coleman , `` bit - wise unequal error protection for variable length block codes with feedback , '' in _ proceedings of the ieee international symposium on information theory ( isit 2010 ) _ , ( austin , tx ) , june 2010 . available online : http://arxiv.org/abs/1101.1934 . s. borade , b. nakibolu , and l. zheng , `` some fundamental limits of unequal error protection , '' in _ proceedings of the ieee international symposium on information theory ( isit 2008 ) _ , ( toronto , canada ) , july 2008 . y. y. shkel and s. c. draper , `` cooperative reliability for streaming multiple access , '' in _ proceedings of the ieee international symposium on information theory ( isit 2010 ) _ , ( austin , tx ) , june 2010 . b. nazer and s. c. draper , `` gaussian red alert exponents : geometry and code constructions , '' in _ 48th annual allerton conference on communications , control , and computing _ , ( monticello , il ) , september 2010 .	consider the following unequal error protection scenario . one special message , dubbed the `` red alert '' message , is required to have an extremely small probability of missed detection . the remainder of the messages must keep their average probability of error and probability of false alarm below a certain threshold . the goal then is to design a codebook that maximizes the error exponent of the red alert message while ensuring that the average probability of error and probability of false alarm go to zero as the blocklength goes to infinity . this red alert exponent has previously been characterized for discrete memoryless channels . this paper completely characterizes the optimal red alert exponent for additive white gaussian noise channels with block power constraints .
rare - earth titanates @xmath11ti@xmath1o@xmath2 ( @xmath12 rare earth ) have attracted extensive research interests due to their exotic magnetism.@xcite these materials have a pyrochlore crystal structure with the space group @xmath13 , in which the magnetic rare - earth ions form a network of corner - sharing tetrahedra and are prone to a high degree of geometric frustration . however , these materials are greatly sensitive to weak perturbations ( e.g. single - ion anisotropy , dipolar interaction or quantum fluctuations ) beyond the nearest - neighboring exchange , which results in unconventional low - temperature magnetic and thermodynamic properties . one famous phenomenon is the classical spin - ice state in ho@xmath1ti@xmath1o@xmath2 and dy@xmath1ti@xmath1o@xmath2 ( with effective ferromagnetic exchange and ising anisotropy).@xcite at low temperatures , their moments have an ising anisotropy due to the strong crystal field with the local easy axis along the [ 111 ] axis . the ground states have a macroscopically degenerate `` 2-in , 2-out '' spin configuration in each tetrahedron . an interesting finding in the spin - ice materials is that the magnetic excitations could be emergent magnetic monopoles.@xcite once the flipping of a spin occurs , a local `` 3-in , 1-out '' or `` 3-out , 1-in '' spin configuration forms , which is equivalent to yielding two opposite magnetic monopoles in the adjacent tetrahedra.@xcite for dy@xmath1ti@xmath1o@xmath2 , the external magnetic field along different directions can easily break the degenerate spin - ice state and induce various magnetic states.@xcite for example , with increasing magnetic field along the [ 100 ] axis , the system enters a q = 0 state at @xmath14 0.5 t.@xcite this state has a single spin configuration chosen from the degenerate spin - ice ground state , with all the spins on each tetrahedron having a component along the field direction and actually forming a long - range order . when the field is along the [ 110 ] direction , the system enters a q = x state at @xmath14 0.4 t.@xcite the spin system is separated into two sets of chains parallel ( @xmath15 ) and perpendicular ( @xmath16 ) to the field , with long - range ferromagnetic order and short - range antiferromagnetic order , respectively.@xcite a more complicated case occurs when applying field along the [ 111 ] direction . there are two successive transitions from the spin - ice state to the kagom - ice state and then to the fully - polarized state at @xmath9 0.3 and 0.9 t , respectively.@xcite another notable characteristics of the spin - ice materials is the very slow spin dynamics at low temperatures , demonstrated by the dc magnetization , ac susceptibility , and specific - heat measurements , etc.@xcite due to the slow spin dynamics , these field - induced magnetic transitions show significant irreversibility , which has been probed by the magnetization and neutron scattering.@xcite yb@xmath1ti@xmath1o@xmath2 has been proposed to be a good candidate for the quantum version of spin ice.@xcite the crystal - field structure of the yb@xmath17 ion is a ground - state kramers doublet well - separated from the first excited doublet . the @xmath18-factor has large planar component , @xmath19 = 4.18 , compared to the [ 111 ] component @xmath20 = 1.77.@xcite the net interaction between the neighboring yb@xmath17 ions is ferromagnetic with a curie - weiss temperature of @xmath9 0.65 k.@xcite for these factors , this material can be described as an effective pseudospin-1/2 quantum - spin ice , with strong transverse quantum fluctuations of magnetic dipoles.@xcite the nature of ground state of yb@xmath1ti@xmath1o@xmath2 is still rather controversial . the specific - heat measurements have revealed a sharp peak at @xmath9 200 mk , signifying a first - order transition.@xcite however , the neutron scattering measurements have indicated that the low-@xmath3 state is weakly ferromagnetic , accompanied with short - range and dynamic spin correlations.@xcite furthermore , the neutron scattering above 200 mk has revealed the pinch - point structure , indicating the presence of a spin - liquid phase with spin - ice correlations.@xcite the excitations of magnetic monopoles are expected in this quantum spin - ice state.@xcite low - temperature heat transport is a powerful tool to probe the properties of elementary excitations@xcite and the field - induced magnetic transitions.@xcite in principle , the magnetic monopoles , as the elementary excitations in the spin - ice state , can also contribute to the heat transport by acting as either heat carriers or phonon scatterers . in a pioneer work by klemke _ et al._,@xcite it was concluded that the thermal conductivity ( @xmath0 ) of dy@xmath1ti@xmath1o@xmath2 is purely phononic and the field dependence of @xmath0 is attributed to phonon scattering by monopoles . in contrast , kolland _ et al . _ explained their @xmath0 data of dy@xmath1ti@xmath1o@xmath2 in terms of the magnetic monopoles making a large contribution to the @xmath0 in zero field.@xcite toews _ measured the low-@xmath3 @xmath0 of another spin - ice material , ho@xmath1ti@xmath1o@xmath2 , and analyzed the data with the considerations that the magnetic monopoles act as heat carriers as well as phonon scatterers.@xcite recently , we have also studied the low-@xmath3 thermal conductivity of dy@xmath1ti@xmath1o@xmath2 single crystals with @xmath8 [ 111 ] and the heat current parallel or perpendicular to the ( 111 ) plane.@xcite the zero - field @xmath10 was proved to be a purely phononic heat transport,@xcite consistent with the conclusion by klemke _ et al._@xcite furthermore , an irreversible @xmath4 behavior was observed in the field region where the magnetization shows a nearly reversible plateau . a picture of the pinning effect of magnetic monopoles by the weak disorders was proposed to explain this novel phenomenon.@xcite it seems that an accurate understanding on the low-@xmath3 heat transport of dy@xmath1ti@xmath1o@xmath2 has not been achieved . a very recent work on yb@xmath1ti@xmath1o@xmath2 has found a suppression of thermal conductivity in low fields.@xcite it was explained as a large monopole heat transport at @xmath21 200 mk . in physics , the quantum monopoles can be more promising than the classical monopoles for transporting heat because they are dispersive while the classical ones are dispersionless . a coherent propagating of monopoles was also indicated by the terahertz spectroscopy and microwave cavity techniques.@xcite however , more detailed studies on the temperature and field dependencies of thermal conductivity are desperately needed for establishing a complete understanding of the heat transport properties . in this work , we study in details the low-@xmath3 thermal conductivity of dy@xmath1ti@xmath1o@xmath2 and yb@xmath1ti@xmath1o@xmath2 single crystals in magnetic fields up to 14 t along the [ 111 ] , [ 100 ] and [ 110 ] directions . it is found that most results of dy@xmath1ti@xmath1o@xmath2 are not supportive to a sizeable heat transport of monopoles , while some results of yb@xmath1ti@xmath1o@xmath2 point to a considerable quantum monopole transport at @xmath21 200 mk . the @xmath0 at very low temperatures is mainly the phonon transport , but the spin fluctuations in yb@xmath1ti@xmath1o@xmath2 strongly scatter phonons . in both materials , the field - induced magnetic transitions can affect the phonon heat transport strongly . moreover , the magnetoelastic coupling are probably also playing a role . all these indicate rather complicated mechanisms of the heat transport in these materials . ti@xmath1o@xmath2 and yb@xmath1ti@xmath1o@xmath2 single crystals , which were orientated by using the x - ray laue photographs . the full width at half maximum ( fwhm ) of the rocking curves are shown in panels ( b ) and ( d).,width=321 ] high - quality dy@xmath1ti@xmath1o@xmath2 and yb@xmath1ti@xmath1o@xmath2 single crystals were grown using a floating - zone technique.@xcite the basic low - temperature properties of our crystals , including the magnetic susceptibility ( down to 2 k ) and specific heat ( down to 0.4 k ) , were checked and found to be in good consistency with most of the reported data from the literatures . some of these data were already shown in ref . . to exhibit the crystallinity of samples , some representative xrd data on the single crystals are shown in fig . the narrow widths of the rocking curve of the ( 440 ) bragg peak ( fwhm @xmath22 and @xmath23 for these two samples ) , demonstrates that the crystals have good crystallinity . for the @xmath0 measurements , the long - bar shaped samples were cut from the as - grown crystals along the [ 111 ] , [ 100 ] or [ 110 ] axes ( the uncertainty is smaller than 1@xmath24 ) after orientation by using back - reflection x - ray laue photographs . the @xmath0 was measured using a one heater , two thermometers " technique in a @xmath25he refrigerator at 300 mk @xmath26 30 k and a @xmath25he-@xmath27he dilution refrigerator at 50 mk @xmath26 1 k , equipped with a 14 t magnet.@xcite all the measurements were done with both the magnetic field and the heat current along the longest dimension of the samples , which can minimize the demagnetization effects.@xcite the thermal conductivities of dy@xmath1ti@xmath1o@xmath2 along the [ 111 ] , [ 100 ] , and [ 110 ] axes were measured at low temperatures down to 0.3 k and in magnetic fields up to 14 t on three samples with the sizes of [email protected]@xmath280.15 mm@xmath25 , [email protected]@xmath280.15 mm@xmath25 and [email protected]@xmath280.15 mm@xmath25 , respectively . in this work , the magnetic - field dependencies of @xmath0 for dy@xmath1ti@xmath1o@xmath2 were measured in two different processes . the first one is the standard steady - state technique with the field changing step by step.@xcite that is , the measurements were done by the following steps : ( i ) change field slowly to a particular value and keep it stable ; ( ii ) after the sample temperature is stabilized , apply a heat power at the free end of sample ; ( iii ) wait some time ( typically for several minutes at very low temperatures ) until the temperature gradient on the sample is nearly stabilized ( judged from the time dependencies of two ruo@xmath1 thermometers on the samples ; a relative change of less than 1 % within a time window of 2 or 3 minutes is the usual criterion ) ; ( iv ) record the temperatures of two thermometers and obtain the temperature gradient . using this process , the @xmath4 isotherms for three samples were measured at different temperatures and the irreversibilities were also probed with the measurements done in changing field ( from zero after zero - field cooling ) up and down in the step mode . the speed of changing field is slow enough to avoid observable heating effect from the eddy current . more exactly , the field ramping rate is not a constant , and it must be very slow at @xmath29 0.1 t ( 0.0010.01 t / min ) and could be quicker at @xmath14 0.1 t ( 0.010.05 t / min ) . note that this process is suitable for most of solid materials . however , it is known that dy@xmath1ti@xmath1o@xmath2 has very slow spin dynamics at low temperatures,@xcite which may cause some relaxation phenomenon in the temperature gradient measurements,@xcite if the magnetic excitations are involved in the heat transport behaviors . for this reason , the @xmath0 measurement done with the sweeping - field mode could not be very meaningful because the thermal relaxation and spin relaxation must be strongly mixed . to probe the time relaxation of the @xmath0 , the measurements were also done in a second process , which is different from the first one in the last step . namely , after the field and sample temperature are stabilized , apply a suitable heat power at the free end of sample and wait for about 30 minutes ; then , record the time dependence of temperature gradient on the sample . since the relaxation was found to be extremely slow at some particular fields , it is not possible to finish a relaxation measurement at different fields in a up and down loop . the relaxations were measured for three different samples at very low temperatures . in this case , the @xmath0 at a certain field was measured separately in the field - increasing and field - decreasing processes , respectively . in the former case , the sample was cooled to the measurement temperature ( for example , 0.36 k ) in zero field and then field was changed slowly to the particular values , and then the @xmath0 was measured in each fields . in the latter case , the sample was first cooled to 0.36 k in zero field ; after that the field was swept slowly to 12 t , which are well above the irreversible region as we show below , then decreased slowly to the target field . the thermal conductivities of yb@xmath1ti@xmath1o@xmath2 along the [ 100 ] , [ 110 ] , and [ 111 ] axes were measured at low temperatures down to 50 mk and in magnetic fields up to 14 t on three samples with dimensions of @xmath30 mm@xmath25 , @xmath31 mm@xmath25 and @xmath32 mm@xmath25 , respectively . the measurements of yb@xmath1ti@xmath1o@xmath2 single crystals are much simpler , since there is no observable relaxation effect . ti@xmath1o@xmath2 in zero field for heat current along the [ 111 ] , [ 100 ] or [ 110 ] axis . the dashed line shows the @xmath33 temperature dependence . the inset shows the temperature dependencies of the phonon mean free path @xmath34 divided by the averaged sample width @xmath35 for these samples.,width=283 ] figure 2 shows the temperature dependencies of @xmath0 for dy@xmath1ti@xmath1o@xmath2 along three directions and in zero field . apparently , the heat transport is nearly isotropic and shows a simple phonon transport . the phonon peaks are located at @xmath9 15 k and a rough @xmath33 dependence of @xmath0 is visible at subkelvin temperatures , which is however weaker than the standard @xmath36 behavior of phonon thermal conductivity at the boundary scattering limit.@xcite similar result has been obtained in the earlier works and the magnetic scattering on phonons was discussed to be important at temperatures below 10 k.@xcite the mean free path of phonons at low temperatures are calculated in the usual way.@xcite the phononic thermal conductivity can be expressed by the kinetic formula @xmath37,@xcite where @xmath38 is the phonon specific heat at low temperatures , @xmath39 is the average velocity and @xmath34 is the mean free path of phonons . the @xmath16 value of dy@xmath1ti@xmath1o@xmath2 crystals , obtained from our specific - heat data,@xcite is @xmath40 j / k@xmath27mol ( which is in good consistency with the result from other group@xcite ) . the inset to fig . 2 shows the ratios @xmath41 for three samples , where @xmath35 is the averaged sample width.@xcite it is clear that the ratios increase quickly at very low temperatures and are expected to approach 1 at @xmath42 0.3 k. this calculated result is compatible with the temperature dependence of @xmath0 , which is a bit weaker than @xmath36 . note that if there were other type of heat carriers ( e.g. magnetic monopoles ) that makes a large contribution to the @xmath0 , the actual phonon mean free path would be much smaller than those in fig . 2 , which is apparently not very reasonable . the low-@xmath3 thermal conductivities of dy@xmath1ti@xmath1o@xmath2 had been studied in magnetic fields up to only 7 t and the main finding was that the @xmath0 can be strongly suppressed by rather weak fields ( @xmath5 1 t).@xcite there are two different explanations for the field - induced suppression of @xmath0 in dy@xmath1ti@xmath1o@xmath2 . namely , it is due to either the suppression of a monopole heat transport or the field - enhanced magnetic scattering on phonons.@xcite note that the @xmath10 data shown in fig . 2 are not supportive for a sizeable monopole heat transport in zero field . in passing , a recent work on another spin - ice material , ho@xmath1ti@xmath1o@xmath2 , proposed that magnetic monopoles are playing a dual role in the heat transport ; that is , they can both transport heat and scatter phonons.@xcite in that work , the contribution of the monopole transport to the total thermal conductivity was estimated to be about 10 % , @xcite much smaller than that proposed for dy@xmath1ti@xmath1o@xmath2.@xcite ti@xmath1o@xmath2 : ( a , c , e ) data in fields up to 14 t and along three directions ; ( b , d , f ) the zoom - in of low - field data . in this figure , the magnetic field is increased step by step after cooling the samples in zero field . the directions of the magnetic field and heat current ( @xmath43 ) are the same.,width=321 ] in present work , the magnetic - field dependencies of @xmath0 are measured up to 14 t , as shown in fig . the low - field behaviors are almost the same as those results in earlier studies . for @xmath8 [ 111 ] , at very low temperatures , the @xmath0 display two anomalies in low fields : at 0.36 k , the @xmath4 curve shows a quick decrease and a peak - like transition at about 0.25 and 0.9 t , respectively , which are in good correspondence with the critical fields of the subsequential transitions from the low - field spin - ice state to the kagom - ice state then to the saturated state ( 3-in , 1-out " or 3-out , 1-in").@xcite at 0.5 k , the @xmath4 curve behaves like two step decreases at these two transitions . the sharpness of the transitions is quickly smeared out upon increasing temperature . for @xmath8 [ 100 ] , the 0.36-k @xmath4 curve shows a steep and broad step - like transitions at 00.2 t and 12 t , respectively . upon increasing temperature , the first transition gradually becomes weaker while the second one is enlarged and shifts to higher fields . at 5 k , the first step is completely gone and the second one evolves into a broad transition at 14 t. the first transition is in good correspondence with the critical field of a long - range - order transition,@xcite in which the external magnetic field chooses one of the degenerate ground states with all the spins having the components along the field . for @xmath8 [ 110 ] , the low - field @xmath4 behave rather similarly to the case of @xmath8 [ 100 ] . for example , at 0.36 k the @xmath4 curve also shows two step - like decreases at 00.25 and 0.81.5 t. the first step - like change of @xmath0 at 0.25 t is in good correspondence with a first - order transition from the low - field spin - ice state to the spin - chain state,@xcite in which the @xmath15-chains form a long - range ferromagnetic order and the @xmath16-chains form a short - range antiferromagnetic order . upon increasing temperature , the first transition gradually becomes weaker , while the second one becomes larger and shifts to higher fields . the drastic change of the heat transport at the magnetic phase transition has been widely observed in many magnetic materials and is usually related to the evolution of magnetic excitations.@xcite however , the particular features of @xmath4 at the magnetic transitions can be significantly different from each other , which depend on both the natures of magnetic transitions and the roles of magnetic excitations in the heat transport ( as heat carriers or phonon scatterers ) . in the spin - ice state of dy@xmath1ti@xmath1o@xmath2 , the monopoles are the only one type of magnetic excitations . in this regard , it was naively assumed that the quick decreases of @xmath0 at the low - field magnetic transitions are caused by the suppression of monopole transport.@xcite this scenario will be further analyzed in the following sections . a remarkable finding is that these @xmath4 curves still exhibit clear or strong field dependencies when the field is above 7 t , which have not been achieved in those earlier works.@xcite at subkelvin temperatures , the @xmath4 for @xmath8 [ 111 ] and [ 100 ] show a broad - valley - like feature ; that is , after arriving a minimum at several tesla , the @xmath0 gradually increases in high magnetic fields . in particular , at 0.36 k , the @xmath0 at 14 t @xmath44 [ 100 ] is already larger than that at 0.5 t , where the thermal conductivity was taken for the phononic background in a recent work.@xcite in addition , the @xmath4 for @xmath8 [ 111 ] and [ 100 ] are not yet saturated in 14 t. it is notable that the high - field @xmath4 behave rather differently for @xmath8 [ 110 ] . the @xmath4 show weak field dependence at several tesla , particularly at 0.36 k , which is also consistent with the earlier reported data in field up to 7 t.@xcite however , at low temperatures the @xmath0 show another strong decrease at @xmath14 7 t and there is no sign that the @xmath0 can be recovered even at 14 t. since this phenomenon does not appear for fields along other directions , a high - field polarization effect is not likely the reason . for now , there are no other experimental results for dy@xmath1ti@xmath1o@xmath2 in [ 110 ] field higher than 7 t , so it is unknown whether the strong suppression of @xmath0 at @xmath14 7 t ( @xmath44 [ 110 ] ) is related to some magnetic transitions . for clarifying , high - field investigations by using other measurements are necessary in the near future . the paramagnetic scattering effect related to the crystal - field levels of dy@xmath17 ions was proposed to play an important role in the field dependencies of @xmath0.@xcite it is known that the dy@xmath17 ions have a degenerate doublet of the lowest crystal - field level . in magnetic field , a zeeman splitting of the doublet can produce resonant scattering on phonons and give both a low - field suppression and a high - field recovery of @xmath0 , as some other magnetic materials have shown.@xcite however , since the paramagnetic scattering is a qualitatively isotropic effect , the obvious anisotropic and complicated high - field behaviors in fig . 3 should not be solely relevant to this mechanism . @xmath4 loops with magnetic field sweeping up and down along the [ 111 ] , [ 100 ] , or [ 110 ] directions . the data shown with solid symbols are measured in ascending field after the sample is cooled in zero field , while the open symbols show the data with descending field , as also indicated by the arrows . these data were measured in the normal way , that is , the waiting time for the temperature stabilization is in a scale of ten minutes . note that the irreversibilities are not visible at @xmath21 0.5 k ( not shown here).,width=321 ] it is found that the magnetic - field dependencies of @xmath0 display irreversible behavior at very low temperatures for all three field directions , as shown in fig . 4 . these data reproduce our earlier results for @xmath8 [ 111 ] and are similar to those reported by other group.@xcite the reversibility weakens quickly with increasing temperature and are hardly visible at @xmath45 0.5 k. note that this is a bit different from some earlier results that clear hysteresis was observed at 0.6 k.@xcite the reason is that those data were taken in a sweeping - field mode . there are significant differences in the irreversibility between the @xmath4 data with @xmath8 [ 111 ] and the data taken with @xmath8 [ 100 ] and [ 110 ] . first , the hysteresis of @xmath4 with @xmath8 [ 111 ] is located mainly in a field region between the two field - induced magnetic transitions , while those with @xmath8 [ 100 ] and [ 110 ] are located at the magnetic transitions . actually , the irreversibility for @xmath8 [ 100 ] and [ 110 ] seems to be a common supercooling " phenomenon of a first - order phase transition . note that the ultra - low-@xmath3 neutron scattering also demonstrated clear hysteresis of the field dependencies of bragg scattering intensity for @xmath8 [ 100 ] and [ 110].@xcite second , the irreversibility for @xmath8 [ 111 ] appears in the kagom - ice state , in which the magnetization curves do not show any hysteresis;@xcite whereas those for @xmath8 [ 100 ] and [ 110 ] have good correspondence with the magnetization hysteresis.@xcite third , the magnitudes of @xmath0 are larger for decreasing fields than those for increasing field with @xmath8 [ 111 ] , while the opposite behavior is observed with @xmath8 [ 100 ] and [ 110 ] . all these clearly indicate that the origin of the irreversibility with @xmath8 [ 111 ] is different from others . , [ 100 ] , or [ 110 ] directions and at 0.36 k. the details for these measurements are described in the experimental section . the open squares show the data measured in ascending field after the sample is cooled in zero field , and the open circles with descending field from high field , where no relaxation and irreversibility are observed . the horizontal lines are guides for eyes while the curves are fittings using formula ( [ e ] ) . all the @xmath46 are taken the values from the normal measurements , as shown in figs . 3 and 4.,width=321 ] it is known that the thermal conductivity of dy@xmath1ti@xmath1o@xmath2 shows not only irreversible behaviors but also a time relaxation.@xcite in present work , the relaxation effect is carefully studied and the representative data are shown in fig . 5 . note that at 0.36 k , the relaxation of @xmath0 can be very strong and very slow , which prevents us from getting the equilibrium state in the available time window of our @xmath25he refrigerator . however , the relaxation effect is not visible in high magnetic fields or at @xmath47 0.5 k , where the @xmath4 hysteresis in fig . 4 disappears . among three field directions , the relaxation of @xmath0 in @xmath8 [ 111 ] is the weakest . as shown in figs . 5(a ) and 5(b ) , in either zero field or 0.5 t ( the field showing largest hysteresis ) , the magnitude of relaxation is very small . typically , after waiting for about 30 min ( the starting point of the recorded data in fig . 5 ) , the measured @xmath0 change in a scale of only 12 % , which is nearly the same as the error of this measurement . this result confirms our earlier work that the relaxation effect in @xmath8 [ 111 ] is not significant.@xcite although one can not completely rule out the possibility of a relaxation with very long time scale ( e.g. much longer than the time window of this measurement ) , the magnitude of the @xmath0 changing with time for @xmath8 [ 111 ] is significantly smaller than those for other field directions . a bit larger relaxation of @xmath0 for @xmath8 [ 111 ] could be observed in a different measurement process , in which the magnetic field was continuously swept.@xcite the relaxation of @xmath0 is the most significant for @xmath8 [ 100 ] . as shown in fig . 5(c ) , although there is no sizeable relaxation effect after cooling in zero field , decreasing field to zero from high field ( 1 t or higher ) results in a slow relaxation . this is very different from the case of [ 111 ] field shown in fig . in addition , in non - zero fields , the relaxation could be very strong and is related to the history of applying field . for example , increasing field from 0 to 0.1 t does not cause significant relaxation , but decreasing field from high value to 0.1 t leads to a very large relaxation ; the @xmath0 can be changed about 10 % and is not saturated after a very long time of about 2400 min , as shown in fig . a contrary feature is observed at 0.15 t ; as shown in fig . 5(e ) , decreasing field from high value to 0.15 t does not cause significant relaxation , but increasing field from 0 to 0.15 t leads to a very slow relaxation . furthermore , the @xmath0 slowly relaxes towards larger values in the case of decreasing field to 0.1 t , while it slowly relaxes towards smaller values if increasing field to 0.15 t. such differences demonstrate that the hysteresis of @xmath4 is directly related to the relaxation effect , as discussed in the following text . the time dependence of @xmath0 can be fitted by a single exponential function , @xmath48 where @xmath49 is the characteristic time constant , and @xmath50 and @xmath51 are time independent parameters . the time constant @xmath49 could be as large as 11001600 min . the situation for @xmath8 [ 110 ] is very similar to that for @xmath8 [ 100 ] , but the relaxation is a bit weaker . for example , the time constants @xmath49 are about 900 min and the @xmath0 changes about 5 % with decreasing field to 0.1 t , as shown in fig . the slow spin dynamics of the spin ice has been probed by other experiments on dy@xmath1ti@xmath1o@xmath2 , such as the dc magnetization , ac susceptibility , and specific heat.@xcite it has been found that the spin relaxation time in the spin - ice state showed an exponential increase at low temperatures and is greater than 10@xmath27 s at @xmath42 0.45 k. the current explanations are mainly in terms of diffusive motion of monopoles . after cooling in zero field to a low but finite temperature , the density of monopoles decreases gradually with time . this could induce a slow relaxation with large time scale . however , it is notable in the present @xmath0 measurements that cooling in zero field does not lead to a significant relaxation of @xmath0 . this means that the monopole dynamics does not have close correlation with the heat transport of dy@xmath1ti@xmath1o@xmath2 . on the other hand , the magnetization measurements with the field - quench or pulse - field processes have revealed the monopole dynamics having a time scale of only several minutes.@xcite this is in contrast with the extremely slow relaxation of @xmath0 in magnetic field . more importantly , if the relaxation of @xmath0 in finite fields , as shown in figs . f ) , were mainly caused by the monopole dynamics ( either transporting heat or scattering phonons ) , one would expect opposite trends of @xmath52 for increasing or decreasing to the same field . again , none of the data supports this expectation . therefore , the relaxation of @xmath0 can not be well explained based on the assumption that the monopoles play an important role in the heat transport properties . furthermore , the anisotropic relaxation of @xmath0 for different field directions indicates that the slow spin dynamics associated with the field - induced transitions for @xmath8 [ 100 ] and [ 110 ] are strongly coupled with phonons , whereas that for @xmath8 [ 111 ] is not . loops with the magnetic field sweeping up and down along the [ 111 ] , [ 100 ] , or [ 110 ] directions . the data shown with squares are measured in the normal way , that is , the waiting time for the temperature stabilization is in a scale of several or ten minutes . the circles show the data extrapolated at @xmath53 from the time - relaxation measurements . the arrows indicate the direction of changing field.,width=226 ] the relaxation fittings using formula ( [ e ] ) can get not only the time constant but also the extrapolated values of @xmath54 at @xmath53 . then , the magnetic - field dependencies of @xmath0 in the expected equilibrium or completely relaxed state " can be obtained . figure 6 shows the comparison of the @xmath4 between the data of normal measurements and the data after complete relaxation for all three field directions . the main result is that the irreversibility in the [ 111 ] field changes only slightly and still clearly exists after long - time relaxation , but the irreversibilities in the [ 100 ] and [ 110 ] fields almost disappear for the data extrapolated at @xmath53 . this demonstrates that the irreversibilities of @xmath4 in the [ 100 ] and [ 110 ] fields are related to the extremely slow spin dynamics in dy@xmath1ti@xmath1o@xmath2 , while that in the [ 111 ] field has a different origin . as discussed in our previous work , a pinning effect of magnetic monopoles by crystal disorders can explain the particular irreversibility with @xmath8 [ 111].@xcite it is also notable that in this case the irreversibility of @xmath4 is actually not large , which is consistent with the previous conclusion on the minor role of monopoles in the heat transport . ( a , b ) , [ 110 ] ( c , d ) and [ 111 ] ( e h ) directions ( @xmath55 ) are shown . red and blue dots represent magnetic monopoles with opposite polarities . blue arrows represent the flipped spins , associated with a monopole movement from the dashed - dot position to the solid - dot position.,width=321 ] the present experimental results have already met with some difficulties in explaining the data when assuming large monopole heat transport . a quantitative analysis on the low - field decrease of @xmath0 is useful to make further clarification . figure 7 shows the schematic picture of monopole motions that can contribute to the heat transport along the magnetic - field directions . the changes of energy gap between the ground state and excited state and the density of monopoles are dependent on the field direction.@xcite when the field is applied along the [ 100 ] direction , the ground state and excited state split into three and two levels , respectively.@xcite because the split levels are rather complicated , a simplification should be carried out . according to the calculation of castelnovo _ al._,@xcite the density of monopoles decays in an approximately exponential speed because the energy gap increases . here , only the hopping from the lowest level of ground state to the lowest level of excited state is considered , whose gap is @xmath56 and @xmath57 4.35 k is the energy gap in zero field.@xcite figure 7(a ) presents this process , in which the spin 1 flips and a pair of monopoles @xmath58 and @xmath59 is created . @xmath59 has three choices of spin flip to recover to the ground state . if the spin 1 flips , the two monopoles annihilate , and there is no monopole movement . if one of the other two equivalent spins 2 or 4 flips , @xmath59 moves to the next tetrahedron and becomes @xmath60 , as shown by fig . this movement has a positive component along the heat current and makes contribution to thermal conductivity . after this process involving the heat transport , the energy of the two tetrahedra is lifted by @xmath61 . on one hand , increasing field leads to larger @xmath62 . as a result , this process becomes more and more difficult and the movement of monopoles are restricted . on the other hand , the energy gap @xmath63 for monopoles also increases and fewer monopoles can be excited . in fields along the [ 110 ] direction , both the ground state and the excited state split into three levels . with the same simplification , the energy gap between the lowest levels of ground state and excited state is @xmath64 ; that is , the gap for monopoles is field independent for @xmath8 [ 110 ] . this means the density of monopoles also does not vary with field . figure 7(c ) shows a pair of monopoles , @xmath58 and @xmath59 , created by this process . apart from the annihilation caused by the spin 3 , the inequivalent spins 2 and 4 provide other two recovery choices for @xmath59 . if the spin 4 flips , the monopole propagates along the direction perpendicular to heat current , and this process makes no contribution to heat transport . if the spin 2 flips , as shown by fig . 7(d ) , @xmath59 moves to the next tetrahedron and becomes @xmath60 , with a positive component along the heat current . this process accompanied by the heat transport leads to an energy lift of @xmath65 . this energy lift increases with field , and therefore the monopole movement is suppressed . in the case of @xmath8 [ 111 ] , the propagation of monopoles along the [ 111 ] direction is the most complicated . the ground state and excited state split into two and four levels in the [ 111 ] field , respectively . the excitation gap of monopoles @xmath58 and @xmath59 , as shown in fig . 7(e ) , is @xmath66 , which results in an increase of monopole density . for the movement of @xmath59 , two effective flips caused by the spins 2 and 4 are equivalent and lift the energy by @xmath67 . however , @xmath60 , which is generated in this process shown by fig . 7(f ) , does not leave the original ( 111 ) plane . further effective flips have to be considered . first , if the next flipping occurs with the spin 3 , as shown in fig . 7(g ) , @xmath60 moves within the ( 111 ) plane and becomes @xmath68 . though energy drops by @xmath69 in this process , no heat transport along the [ 111 ] direction occurs . second , if the spin 1 flips , it can be seen in fig . 7(h ) that @xmath60 goes across the ( 111 ) plane and contributes to heat transport . @xmath60 turns into @xmath70 and energy is lifted by @xmath71 . thus , two - step flips are needed to transporting heat , and the total energy lift @xmath72 also increases with field . ( b ) magnetic - field dependencies of the monopole density . ( c ) calculated monopole thermal conductivity using the kinetic formula after considering the field dependencies of @xmath63 , @xmath73 and velocity of monopole . the inset shows the calculation using the same formula , but the field dependencies of velocity are not considered . ( d ) the monopole thermal conductivity extracted from the experimental data in fig . 6 , assuming the decrease of raw - data @xmath0 from 0 to 0.5 t is just a decrease of @xmath74 . all the calculations and experimental data are taken at 0.36 k.,width=321 ] the above discussions indicate that magnetic field affects both the excitation and the propagation of monopoles . the former refers to the monopole energy @xmath63 and the density @xmath73 , while the latter refers to velocity @xmath75 . as discussed above , the monopole energy @xmath63 with @xmath8 [ 100 ] , [ 110 ] and [ 111 ] are @xmath76 , @xmath77 and @xmath78 , respectively , as shown in fig . for the monopole density , only free monopoles are concerned . according to the debye - hckle theory,@xcite the monopole density is determined by the energy gap , @xmath79 this formula gives the field dependence of the monopole density , as shown in fig . the specific heat of monopoles can be obtained by@xcite @xmath80 where @xmath81 represents the number of tetrahedra per unit volume . since the positive and negative monopoles would annihilate when they meet , the mean free path of monopoles is assumed to be the averaged distance of monopoles with the same polarity . thus , it can be written as @xmath82 if the velocity of monopoles is independent of field , according to the kinetic formula @xmath83 , @xmath84 at 0.36 k can be calculated and are shown in the inset to fig . this calculation shows that the @xmath74 increases strongly with @xmath8 [ 111 ] and is independent of field with @xmath8 [ 110 ] , which are completely different from the experimental observations . in fact , the velocity of monopole should be sensitive to the magnetic field and its direction , as discussed above . currently , the velocity of monopoles can not be calculated accurately . if only the relative change is concerned , based on the additional energy lift @xmath62 caused by the magnetic field , we simply assume that the monopole velocity decays exponentially with increasing @xmath62 ; that is , @xmath85 where @xmath86 is the monopole velocity at zero field and @xmath15 is an adjustable parameter . @xmath62 are given by @xmath61 , @xmath65 and @xmath87 for @xmath8 [ 100 ] , [ 110 ] and [ 111 ] , respectively . with this additional assumption , @xmath84 at 0.36 k are calculated and shown in fig . 8(c ) . for comparison , fig . 8(d ) shows the low - field @xmath74 data " at 0.36 k extracted from the experimental results for @xmath8 [ 100 ] , [ 110 ] and [ 111 ] . here , the relaxed " @xmath4 data from fig . 6 are used , since the above calculations do not take into account the relaxation effect . it is assumed that the decrease of @xmath0 from 0 to 0.5 t is simply due to the suppression of @xmath74 . the raw - data of @xmath0 at 0.5 t is taken as the background from the phonon transport . it seems that there is a rather good consistency between the calculations and the experimental data . furthermore , from the assumed @xmath74 data shown in fig . 8(d ) , the absolute values of @xmath88 along the [ 100 ] , [ 110 ] and [ 111 ] directions are @xmath89 , @xmath90 and @xmath91 w / km , respectively . this anisotropy can be explained as the difference in monopole ratio , @xmath92 , that conducts heat in different directions . based on the above discussion , only certain spin flips contribute to transporting heat . if we assume the probabilities of spin flips in one tetrahedron are equal and omit flips leading to all - in or all - out state , the probability for a monopole conducting heat is 2/3 , 1/3 and 2/9 in the [ 100 ] , [ 110 ] and [ 111 ] directions , respectively , which gives the monopole ratio , @xmath92 , participating in transporting heat . from eqs . ( 2 ) , ( 3 ) and ( 4 ) , the monopole density , specific heat and mean free path at 0.36 k are 1.13 @xmath28 10@xmath93 per tetrahedron , 176.88 j / km@xmath25 and 2.83 @xmath28 10@xmath94 m , respectively . using the formula @xmath95 , the @xmath86 in the [ 100 ] , [ 110 ] and [ 111 ] directions can be calculated to be 2.12 @xmath28 10@xmath27 , 1.59 @xmath28 10@xmath27 and 1.62 @xmath28 10@xmath27 m / s , respectively . these values are unreasonably too large since monopoles are known to be dispersionless.@xcite finally , we need to point out another experimental phenomenon that can not be explained using the assumption of monopole heat transport . as found in our former work , for @xmath8 [ 111 ] the field dependencies of @xmath0 along the field and perpendicular to it are nearly isotropic.@xcite as discussed above , these two perpendicular monopole transport are established with the processes shown by figs . 7(e)7(f)7(h ) and figs . 7(e)7(f)7(g ) , respectively . in the former case , the process in fig . 7(h ) is suppressed most strongly by magnetic field . according to magnetic susceptibility result , the spin 1 in fig . 7(e ) is fixed along field direction at about 0.3 t,@xcite which leads to the disappearance of @xmath96 . in the latter case , magnetic field only suppresses the process in fig . 7(f ) . at 0.3 t , although fixing spin 1 does not forbid the monopole transport directly , it reduces the degree of freedom of monopoles and forbids the subsequent monopole propagating after the flip of spin 1 . therefore , @xmath96 should have a kink - like decrease at 0.3 t instead of completely disappears . when field reaches 0.9 t , the spin 2 in fig . 7(f ) is also fixed@xcite and the @xmath74 is completely suppressed . therefore , the @xmath96 along the [ 111 ] field and perpendicular to it are expected to present obvious anisotropy , which is not supported by the experimental results.@xcite the analysis and calculations indicate that only some of the low - field results of @xmath4 can be qualitatively explained as the suppression of monopole thermal conductivity in magnetic fields . however , the quantitative analysis also points out that the monopoles are not likely making a large contribution to the heat transport . it is essentially coincided with some other experiments . for example , it has been found that the hop rate of monopoles is roughly proportional to the density of monopoles at low temperatures ( @xmath5 1.5 k).@xcite furthermore , a recent adiabatic susceptibility measurement revealed a brownian characteristic of the monopole movement in the spin - ice state,@xcite with a much smaller diffusion constant than that proposed by some heat transport work.@xcite there should be more important factors that are responsible for the field dependencies of @xmath0 in dy@xmath1ti@xmath1o@xmath2 . , [ 110 ] and [ 111 ] axes of yb@xmath1ti@xmath1o@xmath2 single crystals in zero field . the dashed line shows the @xmath97 dependence which the @xmath0 follow between 200 mk and 2 k. the inset shows the temperature dependencies of the ratio of the phonon mean free path @xmath34 to the averaged sample width @xmath35 . the @xmath34 is calculated assuming that the @xmath0 is purely phononic.,width=264 ] figure 9 shows the temperature dependence of @xmath0 of yb@xmath1ti@xmath1o@xmath2 single crystals in zero field . similar to the case of dy@xmath1ti@xmath1o@xmath2 , the thermal conductivity is nearly isotropy for the heat current along the [ 100 ] , [ 110 ] and [ 111 ] axes . the peak at 13 k is the typical feature of a phonon transport in insulators . moreover , there are two notable features of these data . first , the temperature dependencies display kink - like anomalies at @xmath9 200 mk where the slopes of @xmath10 changes strongly . this is likely related to the first - order transition of magnetism . second , at subkelvin temperatures , the @xmath0 do not show the @xmath36 dependence , which is expected for phonon thermal conductivity in the boundary scattering limit . actually , the @xmath0 nearly follow the @xmath97 dependence from about 2 k to 200 mk . this is consistent with a recent report.@xcite although the well - known @xmath36 ballistic behavior of phonons has been rarely observed in the transition - metal compounds , including the high-@xmath98 cuprates , the multiferroic manganites and the low - dimensional quantum magnets,@xcite the @xmath0 can usually exhibit the temperature dependence rather close to @xmath36 . in dy@xmath1ti@xmath1o@xmath2 , as shown in fig . 2 , the @xmath33 dependence of @xmath0 is observed at temperatures of several hundreds of millikelvins . in yb@xmath1ti@xmath1o@xmath2 , however , the @xmath97 dependence differs too much from the @xmath36 behavior . there are two possible reasons for this deviation from the boundary scattering limit . if there are sizeable scattering effects on phonons even at the lowest temperature , the phonon mean free path is significantly smaller that the sample size and @xmath36 behavior can not be observed . another possibility is the presence of the other term of heat conductivity , for example , from the magnetic excitations . in any case , the estimation of phonon mean free path can provide some useful information . with the @xmath16 value (= 1.61 @xmath99 j / k@xmath27mol ) from the specific - heat data,@xcite the phonon mean free path can be calculated assuming the @xmath0 is purely phononic . the inset to fig . 9 shows the temperature dependencies of the ratio @xmath41 . it is found that the @xmath41 ratios are only about 0.30.4 at the lowest temperatures , indicating that the boundary scattering limit of phonons is not established and there are still some microscopic scattering of phonons . it should be noted that the above calculation is based on the assumption of a purely phononic heat transport . if there were a magnetic term of heat transport , the phonon mean free path would be even smaller . nevertheless , it is clear that the phonon scattering is stronger in yb@xmath1ti@xmath1o@xmath2 , compared to dy@xmath1ti@xmath1o@xmath2 . it is understandable from the magnetic scattering since yb@xmath1ti@xmath1o@xmath2 is known to have strong spin fluctuations.@xcite all the zero - field @xmath10 curves show a kink - like decrease at 200 mk with lowering temperature . it might be that some term of thermal conductivity disappears when entering the low-@xmath3 state . this is likely a signature that the magnetic excitations ( monopoles ) in the quantum spin - ice state ( at @xmath21 200 mk ) can act as heat carriers.@xcite ti@xmath1o@xmath2 crystals at low temperatures . panels ( a , c , e ) show the data in fields up to 14 t and along three directions , while panels ( b , d , f ) show the zoom - in data of the low - field plots . since the magnetic fields are applied along the direction of the heat current ( @xmath43 ) , the demagnetization effect is negligibly small for these long - bar shaped samples.,width=321 ] figure 10 shows the magnetic - field dependencies of @xmath0 at low temperatures for these yb@xmath1ti@xmath1o@xmath2 single crystals . the data have two remarkable features . first , with increasing field , the @xmath0 exhibits strong changes in low fields , followed by a quick increase , and finally becomes field independent in high fields . it naively indicates that there are magnetic scattering effects on phonons that can be suppressed in high fields , where the spins are polarized . this is consistent with the zero - field @xmath10 data and the small phonon mean free path . the strong spin fluctuations revealed by the neutron scattering is most likely the source of phonon scattering in zero field . second , the low - field behaviors are somewhat different for different directions of magnetic field . for @xmath8 [ 100 ] , the @xmath0 increase monotonically with field when @xmath42 200 mk , but show a low - field valley when @xmath21 200 mk . the position of dip field shifts to higher field with increasing temperature . for @xmath8 [ 110 ] or [ 111 ] , the @xmath4 isotherms at the lowest temperatures show a dip " at 0.5 t. in particular , the dip is very sharp for @xmath8 [ 110 ] , reminiscent of a field - induced transition . in this regard , the absence the dip " for @xmath8 [ 100 ] indicates that either there is no field - induced transition or the transition field is very low ( smaller than 0.1 t , which is the step size of our @xmath4 measurements ) . as suggested in a recent work , the low - field decrease of @xmath0 at @xmath21 200 mk may be due to the suppression of monopole heat transport in magnetic field.@xcite however , it should be pointed out that the mechanism of monopole transport is far from well understood . the present @xmath4 data with field along three different directions show some peculiar phenomenon . to illustrate , at @xmath21 200 mk the low - field behavior of @xmath0 are qualitatively different for different field directions . the position of dip field shifts to higher field with increasing temperature for @xmath8 [ 100 ] , while it changes very weakly for @xmath8 [ 110 ] and [ 111 ] . this anisotropic behavior calls for a more precise theoretical description of the monopole transport . qualitatively , a low - field valley and a strong enhancement at high field in @xmath4 are also similar to the paramagnetic scattering effect.@xcite it is known that the yb@xmath17 ions have a kramers doublet of the lowest crystal - field level . in magnetic fields , a zeeman splitting of the doublet can produce resonant scattering on phonons and give both a low - field suppression and a high - field recovery of @xmath0.@xcite a simplified calculation has shown the qualitative behaviors of @xmath4 in the case of paramagnetic scattering ; one important feature is that the position of the @xmath4 minimum shifts to higher field with increasing temperature.@xcite since the changes of low - field valley with increasing temperature are not in good consistency with this calculation , the paramagnetic scattering may not be important in yb@xmath1ti@xmath1o@xmath2 . the low - field behavior of @xmath0 at @xmath42 200 mk displays different characteristics from the data at @xmath21 200 mk . the low-@xmath3 magnetization measurements have indicated that the ground state is ferromagnetic with the spin easy - axis in the [ 100 ] direction , but the residual spin fluctuations are evidenced.@xcite these results indicated that the spins are easily polarized in a rather weak field . in this regard , the quick increase of @xmath0 with field seems closely related to the spin polarization , which suppresses the spin fluctuations . in the field along the [ 110 ] direction , neutron scattering has revealed a field - induced transition at 0.5 t.@xcite with field lower than 0.5 t , there is significantly diffuse scattering intensity originated from the spin fluctuations . with increasing field , the diffuse scattering intensity is weakened and well - defined spin - wave excitations appear at 0.5 t , indicating a long - range ordered state ( spin polarization).@xcite since the spin waves are gapless at the critical field of 0.5 t , they can be well populated by the thermally excitations even at very low temperatures and strongly scatter phonons . this can explain the sharp 0.5t-dip " of @xmath4 curves . although there is no experimental investigations on this field - induced transition with field along other directions , the 0.5t-dip " of @xmath4 curves with @xmath8 [ 111 ] may have the same origin . ti@xmath1o@xmath2 crystals in zero field and in 8 t along different directions . the dashed lines indicate that the low - temperature data in 8 t follow the @xmath33 dependence . ( d ) temperature dependencies of the ratio of the phonon mean free path @xmath34 to the averaged sample width @xmath35 , calculated from the 8 t data.,width=321 ] in most cases , the magnetic scattering can be suppressed by strong magnetic field and the @xmath0 would be independent of field in high fields and display a @xmath36 dependence at low temperatures . however , yb@xmath1ti@xmath1o@xmath2 was found to be exceptional . figure 11 shows the @xmath10 with 8 t along three different directions . it can be seen that these data do not exhibit the expected @xmath36 behavior even at @xmath42 100 mk ; instead , the data follows a @xmath33 dependence at temperatures lower than 200 - 300 mk . the phonon mean free paths are calculated and are found to be smaller than the sample sizes , as shown in fig . 11(d ) , which means that there are still some kind of microscopic scattering on phonons and the boundary scattering limit is not achieved . this temperature dependence of @xmath0 in 8 t field is a peculiar phenomenon that is not easy to understand . since high field can effectively suppress spin fluctuations , there should be some other factor damping the phonon transport . firstly , there is no strong evidence for the large heat transport of monopoles in the spin - ice compound dy@xmath1ti@xmath1o@xmath2 . actually , most of the experimental data , including the zero - field @xmath10 , the @xmath4 with different field directions , the high - field behaviors , and even the relaxation phenomenon , are more or less incompatible with the supposition of large monopole transport in zero field . the quantitative analysis using some phenomenological theory gives a qualitative description of the low - field decrease of @xmath0 for @xmath100 , but it fails to explain the nearly isotropic behavior with @xmath8 [ 111 ] for @xmath0 along the field and perpendicular to it.@xcite furthermore , the calculations show that very high speeds ( @xmath6 10@xmath27 m / s ) of monopole propagating are required to describe the experimental data . this is unlikely since the monopole excitations are dispersionless and in principle they can only transport diffusively with a small averaged velocity . in the case of yb@xmath1ti@xmath1o@xmath2 , the kink - like feature of @xmath10 at 200 mk is a clear signature of considerable monopole transport at the quantum spin - ice state ( @xmath21 200 mk ) . since the quantum monopoles are dispersive , it is possible they have a highly mobile characteristic . secondly , judging from the @xmath10 data and the estimation of phonon mean free path , dy@xmath1ti@xmath1o@xmath2 does not exhibit strong phonon scattering in zero field , while yb@xmath1ti@xmath1o@xmath2 has rather strong scattering effect by the spin fluctuations . the high - field behaviors also demonstrate this difference . the main reason is that the monopole excitations are difficult to be thermally excited at very low temperatures due to the sizeable energy barrier ( @xmath9 4.35 k ) . thirdly , the field dependencies of @xmath0 are much more complicated in dy@xmath1ti@xmath1o@xmath2 . one phenomenon is that the low - field transitions have strongest effect and cause step - like decreases of @xmath0 . since both the qualitative and quantitative analysis based on the assumption of large monopole heat transport can not describe the experimental results , some unknown phonon scattering processes are relevant . in yb@xmath1ti@xmath1o@xmath2 , the field dependencies are mainly determined by the suppression of spin fluctuations when the spins are polarized , and the field - induced transitions can cause a sharp dip - like anomaly in @xmath4 . fourthly , for dy@xmath1ti@xmath1o@xmath2 , the @xmath0 in low fields exhibit a strong relaxation effect and irreversibility . these phenomenon are not observed in the @xmath0 measurements of yb@xmath1ti@xmath1o@xmath2 . this is mainly related to the extremely slow spin dynamics in the low - field states of dy@xmath1ti@xmath1o@xmath2 . fifthly , there are some peculiarity in the high - field heat transport of both materials . dy@xmath1ti@xmath1o@xmath2 displays strong field dependencies of @xmath0 in high field up to 14 t , which is quite anomalous considering the spins are polarized above @xmath9 2 t.@xcite yb@xmath1ti@xmath1o@xmath2 shows field independent @xmath0 in the polarized state . however , the @xmath33 behavior of high - field @xmath10 is not expectable if the magnetic scattering of phonons are completely suppressed . these indicate that besides the impacts associated with the low - field magnetic transitions and spin fluctuations , some other factors are also important . one factor can affect the phonon transport is the magnetostriction and magnetoelastic coupling in these materials . for example , at the transition from the kagom - ice phase to the saturated state with @xmath8 [ 111],@xcite the ultrasound measurement revealed sharp anomalies of the sound velocity and sound attenuation.@xcite furthermore , there is a low - field phenomenon of dy@xmath1ti@xmath1o@xmath2 that is not mentioned in the above discussions . for @xmath8 [ 100 ] or [ 110 ] and at very low temperatures , the weaker and broader step - like decreases of @xmath0 at 12 t could not be related to any magnetic transitions and should have a different origin . in fact , all the earlier experimental results of magnetization and neutron scattering have not found any magnetic transition or anomaly at 12 t for @xmath8 [ 100 ] or [ 110 ] . instead , this phenomenon may be related to the magnetostriction , which exhibited a small anomaly at 2 t along the [ 100 ] direction.@xcite in this regard , the lower - field ( @xmath5 1 t ) transitions can affect the @xmath0 in the same way since they show stronger effect on crystal lattice.@xcite the high - field behavior of yb@xmath1ti@xmath1o@xmath2 may be related to magnetoelastic coupling as well . in the spin - liquid tb@xmath1ti@xmath1o@xmath2 , a magnetoelastic mode is formed by the hybridization of the first excited crystal - field level and the transverse acoustic phonons.@xcite as a result , the phonon transport of tb@xmath1ti@xmath1o@xmath2 is so strongly damped that it behaves like a glassy state.@xcite in yb@xmath1ti@xmath1o@xmath2 , the coupling of the sound wave to quantum fluctuations has been investigated by the sound - velocity and sound - attenuation measurement.@xcite some anomalies in temperature and field - dependent sound velocity and attenuation were observed and were attributed to the first - order phase transition . moreover , these measurements ( down to 20 mk ) have found that both the sound velocity and sound attenuation change continuously with field up to 5 t , without showing any signature of saturation.@xcite this indicated that the magnetoelastic coupling is significant even in high fields . the structural distortions caused by the magnetoelastic coupling would prevent the phonons from transporting ballistically . in summary , the low-@xmath3 thermal conductivity of spin - ice dy@xmath1ti@xmath1o@xmath2 and quantum spin - ice yb@xmath1ti@xmath1o@xmath2 are studied . the detailed temperature and field dependencies of @xmath0 indicate that the mechanisms of heat transport in these materials are rather complicated . the phonons are the main heat carriers and several factors including the magnetic - monopole excitations , the field - induced magnetic transitions , the phonon scattering by spin fluctuations , and probably the structural distortions caused by the magnetoelastic coupling are involved . one conclusion is that the monopoles in dy@xmath1ti@xmath1o@xmath2 play a minor role , while the quantum monopoles in yb@xmath1ti@xmath1o@xmath2 may make a sizable contribution in carrying heat . the comprehensive theories for quantitatively describing the experimental results are called for and would be very useful for understanding the exotic magnetisms of these materials . further experimental investigations on the magnetoelastic coupling in high fields are also indispensable . we thank r. p. sinclair for the helpful assistant . this work was supported by the national natural science foundation of china , the national basic research program of china ( grant nos . 2015cb921201 and 2011cba00111 ) , and the fundamental research funds for the central universities ( program no . wk2030220014 ) . d. j. morris , d. a. tennant , s. a. grigera , b. klemke , c. castelnovo , r. moessner , c. czternasty , m. meissner , k. c. rule , j .- u . hoffmann , k. kiefer , s. gerischer , d. slobinsky , and r. s. perry , science * 326 * , 411 ( 2009 ) . j. a. hodges , p. bonville , a. forget , a. yaouanc , p. d. de rotier , g. andr , m. rams , k. krlas , c. ritter , p. c. m. gubbens , c. t. kaiser , p. j. c. king , and c. baines , phys . rev . lett . * 88 * , 077204 ( 2002 ) . r. m. dortenzio , h. a. dabkowska , s. r. dunsiger , b. d. gaulin , m. j. p. gingras , t. goko , j. b. kycia , l. liu , t. medina , t. j. munsie , d. pomaranski , k. a. ross , y. j. uemura , t. j. williams , and g. m. luke , phys . b * 88 * , 134428 ( 2013 ) .	we study the low - temperature thermal conductivity ( @xmath0 ) of dy@xmath1ti@xmath1o@xmath2 and yb@xmath1ti@xmath1o@xmath2 single crystals in magnetic fields up to 14 t along the [ 111 ] , [ 100 ] and [ 110 ] directions . the main experimental findings for dy@xmath1ti@xmath1o@xmath2 are : ( i ) the low-@xmath3 @xmath4 isotherms exhibit not only the step - like decreases at the low - field ( @xmath5 2 t ) magnetic transitions but also obvious field dependencies in high fields ( @xmath6 7 t ) ; ( ii ) at @xmath7 0.5 k , the @xmath4 curves show anisotropic irreversibility in low fields , that is , the @xmath4 hysteresis locates at the first - order transition with @xmath8 [ 100 ] and [ 110 ] , while it locates between two successive transitions with @xmath8 [ 111 ] ; ( iii ) the @xmath0 in the hysteresis loops for @xmath8 [ 100 ] and [ 110 ] show an extremely slow relaxation with the time constant of @xmath9 1000 min . the main experimental findings for yb@xmath1ti@xmath1o@xmath2 are : ( i ) the zero - field @xmath10 show a kink - like decrease at the first - order transition ( @xmath9 200 mk ) with decreasing temperature ; ( ii ) the low-@xmath3 @xmath4 isotherms show a decrease in low field and a large enhancement in high fields ; ( iii ) the low-@xmath3 @xmath4 curves show a sharp minimum at 0.5 t for @xmath8 [ 110 ] and [ 111 ] . the roles of monopole excitations , field - induced transitions , spin fluctuations and magnetoelastic coupling are discussed .
muon colliders _ _ have been proposed to provide lepton - lepton collisions while circumventing the energy limitations on electron - positron storage rings caused by synchrotron radiation . the larger muon mass suppresses synchrotron radiation energy losses by a factor @xmath0 relative to those of a circulating electron beam of the same energy and , incidentally , also opens up promising possibilities for @xmath1channel higgs boson production@xcite . recent feasibility and design studies for future muon colliders @xcite have begun to focus attention on the exciting physics possibilities for experiments using neutrino beams from the decays of the circulating high energy muons . this report explores the potential for a `` neutrino experiment at a muon collider '' , or @xmath2mc for short . a @xmath2mc program could operate either parasitically during a colliding beam experiment ; or it could be installed as part of a program in neutrino physics at a dedicated muon storage ring . amongst the potential physics topics for @xmath2mcs , neutrino oscillations have garnered the most intense experimental and theoretical activity , and particular possibilities for long baseline oscillation experiments exploiting a muon storage ring are covered elsewhere@xcite . in this report , we wish to also highlight the superb capabilities of neutrinos as probes of the strong and weak interaction dynamics of quarks and the parton structure of nucleons , as well as the power of a @xmath2mc in searches for evidence of new types of weak interactions . the remainder of this section lays out the expected experimental parameters and capabilities of a @xmath2mc and provides concise overviews for the more detailed physics discussions that follow . [ c]\|c\|ccc\| & @xmath3factory & higgs - factory & top - factory + @xmath4 & 20 gev & 50 gev & 175 gev + & 3.0 & 6.0 & 6.0 + & no dump & no dump & no dump + & 300 & 345 & 900 + & 90 & 40 & 110 + & 0.30 & 0.12 & 0.12 + & 0.90 & 0.72 & 0.72 + & 1.8 & 1.4 & 1.4 + & 1 & 1 & 1 + & 5.3 & 2.1 & 0.60 + & @xmath5 & @xmath6 & @xmath7 + & 2600 & 1500 & 370 + & @xmath8 & @xmath9 & @xmath10 + [ c]\|c\|cc\| & frontier & 2nd generation + & 500 gev & 5 tev + & 3.2 & 3.6 + & 0.5 & no dump + & 2000 & 15 000 + & 150 & 450 + & 0.12 & 0.03 + & 0.38 & 0.11 + & 0.30 & 0.22 + & 10 & 1 + & 2.1 & 0.021 + & @xmath11 & @xmath12 + & 430 & 100 + & @xmath13 & @xmath14 + [ c]\|r\|ccc\| & general & polarized & @xmath15 + & @xmath16 ccd & solid @xmath17 & liquid @xmath18 + mean density @xmath19 $ ] & 0.5 & 0.267 & 0.717 + length @xmath20 $ ] & 2 & 0.5 & 20 + thickness @xmath21 $ ] & 100 & 13.4 & 1430 + radius @xmath22 $ ] & 20 & 20 & 20 + mass @xmath23 $ ] & 126 & 16.8 & 1800 + integrated luminosity @xmath24 $ ] & @xmath25 & @xmath26 & @xmath27 + dis events / year at 50 gev & @xmath28 & @xmath12 & @xmath29 + dis events / year at 175 gev & @xmath30 & @xmath31 & @xmath32 + @xmath33 events / year at 50 gev & @xmath34 & na & @xmath35 + @xmath33 events / year at 175 gev & @xmath36 & na & @xmath37 + recent ideas for neutrino experiments at either muon colliders @xcite or dedicated neutrino factories@xcite represent reincarnations of earlier proposals for neutrino experiments at muon storage rings that date back at least to the 1960 s . the essential advantages of modern @xmath2mcs derive from the very large muon currents that might be produced and stored using the technologies developed for muon colliders . current design scenarios for muon colliders@xcite and neutrino factories @xcite envision of order @xmath38 positive and negative muons per year circulating and decaying in the storage ring . neutrinos from decays in the longest straight sections of the storage ring will emerge in intense collinear beams that are highly suitable for experiments . beams from such production straight sections should provide many orders of magnitude higher event rates than considered in the early versions of muon storage rings and , indeed , should be considerably more intense than today s `` conventional '' neutrino beams produced from @xmath39 decays . no need exists for a conventional beam s muon shielding berm , and detectors can be placed relatively close to the end of the production straight section . coupled with the relativistic kinematics of muon decay , this permits the possibility of detectors only tens of centimeters across and allows for the use of high precision vertex detectors as active neutrino targets . additional physics advantages over @xmath39 decay neutrino beams will result from the unique and precisely understood flux composition of the @xmath2mc beams . negative and positive muons decay according to @xmath40 producing pure 2-component neutrino beams but caution that the current experimental limit on the branching fraction for this decay is only @xmath41 . this limit will clearly be greatly improved upon from the consistency of the observed @xmath2mc spectra with predictions . ] via the perhaps best understood of all weak decay processes . these beams will be designated as @xmath42 or @xmath43 , respectively , in the rest of this report . experimental requirements for the two broad classes of neutrino physics at @xmath2mcs differ greatly , chiefly because the experiments would be conducted at very different baseline distances from beam production to the detector . experiments for neutrino interaction physics will be conducted as close to the muon ring as possible ( `` short baseline '' ) in order to maximize event rates and to subtend the neutrino beam with a target of small transverse cross section . on the other hand , the choice of baseline for neutrino oscillation studies will be dictated by the specific range of possible oscillation parameters under investigation , as discussed further in chapter [ ch : osc ] . oscillation parameters of current interest motivate the use of very long baselines , even extending to the possibility of transcontinental experiments @xcite . as an important caveat on the contents of this report , it should always be borne in mind that the ambitious technologies of these high current muon storage rings still only exist at the feasibility or early design study stage and it is by no means guaranteed that realizable devices will appear anytime soon . nevertheless , recent progress has been impressive , and the pace of r&d is accelerating . the reader is referred to the specialist literature for a more thorough overview of the technological challenges in building a muon collider@xcite or neutrino factory@xcite . event rates in all @xmath2mc experiments will be dominated by the charged current ( cc ) and neutral current ( nc ) deep inelastic scattering ( dis ) of neutrinos or antineutrinos with nucleons ( @xmath44 or @xmath45 ) : @xmath46 where @xmath47 or @xmath48 and @xmath49 represents a typically multi - particle hadronic final state . neutrino - nucleon dis cross sections scale with neutrino energy @xmath50 to a good approximation for neutrino energies above a few gev , with numerical values of @xcite : @xmath51{c}% \nu - cc\\ \nu - nc\\ \overline{\nu}-cc\\ \overline{\nu}-nc \end{array } \right ) \;\simeq\left ( \begin{array } [ c]{c}% 6.8\\ 2.1\\ 3.4\\ 1.3 \end{array } \right ) \times e_{\nu}[\text{gev}]\text { fb}. \label{eq : xsec}%\ ] ] at the many - gev energies of @xmath2mcs , @xmath52 dis is well described as the quasi - elastic scattering of neutrinos off one of the many quarks or antiquarks inside the nucleon through the exchange of a virtual @xmath53 or @xmath54 boson : @xmath55 all quarks @xmath56 participate in the nc process . the cc interactions change quark flavor , with neutrino interactions producing @xmath57-type and @xmath58-type final state quarks , @xmath59 , from @xmath60-type and @xmath61-type targets , @xmath62 and @xmath63 . antineutrinos participate in the charge - conjugate processes . much of the richness of neutrino interaction physics derives from the variety of processes contained in eqs . [ eq : ncnuq]-[eq : ccnubarq ] . neutrino flux spectra at @xmath2mcs will be precisely predictable since the decay of muons is a well - understood purely electroweak process . characteristics of the parent muon beam in the production straight section can be reliably calculated and modeled through a knowledge of the focusing magnet lattice and through beam monitoring . calibration of the muon energies in the storage ring might reach the level of a few parts per million fractional uncertainty @xcite . due to the differing angular coverages , the neutrino spectrum seen by an oscillation detector at a long baseline will differ from that seen by detectors placed at short baselines to study interaction physics . long baseline detectors will sample the very forward - going neutrinos , at angles in the muon rest frame@xmath64 and laboratory frame@xmath65 close to @xmath66 , while detectors close to the production straight section will instead accept a production solid angle bite that is comparable to the boosted forward hemisphere of the decaying neutrinos , @xmath67}. \label{eq : thetaforward}%\ ] ] figure [ fig : far_near ] gives an illustrative example of the neutrino spectra at @xmath2mcs for detectors at both short and long baselines@xcite , and table [ tab : beam_specs ] gives beam and event rate parameters for several other @xmath2mc scenarios . further explanation for the choices of storage ring parameters in table [ tab : beam_specs ] and a derivation for the following simple numerical expressions for event rates used to fill table [ tab : events ] are provided elsewhere@xcite . for short baseline detectors , @xmath68 = 2.1\times10^{-15}\times e_{\mu}[\text{gev}]\times n_{\mu}^{ss}[\text{yr}^{-1 } ] , \label{eq : beam_nsb}%\ ] ] where @xmath69 is the number of neutrino interactions per year of running per g@xmath70@xmath71 of a cylindrically symmetric target centered on the beam , @xmath72 is the muon beam energy and @xmath73 is the number of forward - going muons ( as opposed to muons circulating in the opposite direction in , e.g. , a collider ring ) decaying in the production straight section per year . equation [ eq : beam_nsb ] assumes the parent muon beam to have an angular divergence in the production straight section that is small compared to the @xmath74 natural divergence of the neutrino beam . this will normally be the case @xcite unless the choice of straight section is in the final focus region of a collider storage ring . table [ tab : events ] accounts simplistically for this exception ( in this case for the illustrative parameter set at 500 gev ) by increasing the angular divergence of the neutrino beam by a simple scale factor . in this circumstance , the angular coverage of the target would need to be increased by this same scale factor in order to retain the event rate predicted by the parameter @xmath69 . in contrast to short baseline detectors , the event rate in long baseline detectors is not sensitive to the geometry of the detector since the entire detector will always be bathed uniformly by the forward - going neutrino flux . the number of interactions in the detector will vary in proportion to the target mass @xmath75 and inversely as the square of the baseline length @xmath76 . this leads to a definition , analogous to eq . [ eq : beam_nsb ] , for the event rate benchmark @xmath77 : @xmath78 = \frac{1.6\times10^{-20}\times n_{\mu}^{ss}[yr^{-1}% ] \times(e_{\mu}[gev])^{3}}{(\gamma\cdot\delta\theta)^{2 } } , \label{eq : beam_nlb}%\ ] ] where @xmath77 is the number of neutrino interactions per kiloton@xmath70year of running with a target centered on the beam at a @xmath79 km distance from the production point . the previously discussed angular divergence scaling factor , @xmath80 , has been explicitly included . the event rates given in table [ tab : events ] are truly impressive . samples of thousands of events per kiloton might be recorded at oscillation experiments with baselines as long as thousands of kilometers . for neutrino interaction physics , samples as large as 10 billion events can be reasonably contemplated in compact targets close to the production straight section . equation [ eq : thetaforward ] shows that the radial extent of such targets can be as small 1020 cm . event rates for oscillation experiments will probably be less of an extrapolation from today s experiments than will be the case for interaction experiments due to the compensating rate decrease at the expected longer baselines . correspondingly , the innovations in neutrino detector design required to upgrade to the neutrino beams at @xmath2mcs are likely to be rather less substantial for oscillation experiments at long baselines than for interaction physics experiments . two significant changes expected for the design of oscillation detectors for @xmath2mcs are that ( i ) the 2-component beams provide strong motivation for a magnetic spectrometer to distinguish muon charge signs ; and ( ii ) larger detector masses might be financially justified in order to fully exploit the large financial investment in the muon storage ring . design considerations for detectors for oscillation @xmath2mcs are discussed in more detail in chapter [ ch : osc ] . in contrast to oscillation experiments , the increase in neutrino yield for @xmath2mcs relative to beams from pion decays as well as the collimation of the neutrino beams will allow the use of compact , specialized targets surrounded by high performance detectors . these detectors must operate at high rate in order to cope with the data sets implied in table [ tab : beam_specs ] . considerable thought must be given to triggering , data acquisition , event reconstruction and data handling considerations . mcs over that with collider experiment geometries . ( the figure is reproduced from @xcite , which used the terminology `` murine '' , for muon ring neutrino experiment , instead of @xmath2mc . ) neutrino targets could have a vertex plane of ccd pixel detectors spaced at intervals of approximately one millimeter . for comparison , the vxd3 vertexing detector at the sld experiment at slac , generally regarded as the best existing vertex detector in a collider experiment , has its two innermost ccd tracking planes at 2.8 cm and 3.8 cm from the interaction point ( ip ) . a schematic of a one - prong @xmath81 decay has been drawn to illustrate the advantages of closely spaced vertex detectors . for clarity of illustration , the kink deflection angle has been drawn much larger than would be typical . the 2 cm distance to decay for the @xmath81 charmed meson corresponds to the average boosted lifetime for a 120 gev @xmath81 . , width=480,height=384 ] figure [ hrdet ] provides an example@xcite of the sort of high rate general purpose neutrino detector that would be well matched to the intense neutrino beams at @xmath2mcs . the neutrino target is one meter long stack of ccd tracking planes represented by the small horizontal cylinder at mid - height on the right - hand side of the detector in fig . [ hrdet ] . its 10 cm radial extent could correspond to , e.g. , the 0.2 mrad divergence of the neutrino beam originating from a 500 gev muon beam 500 m upstream of the target . the scale of the entire detector is illustrated by a human figure in the lower left corner , emphasizing the striking contrast in target size with the kiloton - scale coarse - sampling calorimetric targets often used for past and present high rate neutrino experiments . the ccd target in fig . [ hrdet ] contains 750 planes of 300 micron thick silicon ccd s , corresponding to a mass per unit area of approximately 50 g@xmath70@xmath71 ; this translates to 2.5 radiation lengths or 0.5 interaction lengths . scaling to different target lengths and radii should be straightforward without altering the basic design of the surrounding detector . besides providing the mass for neutrino interactions , the tracking target allows for precise reconstruction of the event topologies from charged tracks , including event - by - event vertex tagging and reconstruction of those interactions containing heavy flavor final states . the fixed target geometry of @xmath2mc vertex detectors allows for much more frequent sampling than is possible in collider detectors : fig . [ vertexing ] gives a schematic comparison between the charm vertexing capabilities of the ccd detector of fig . [ hrdet ] and the current best vertexing detector in a collider experiment@xcite . the ccd target is backed by a hermetic detector reminiscent of many collider detector designs . an enveloping time projection chamber ( tpc ) provides track - following , momentum measurements , and particle identification for essentially all charged tracks emanating from the interactions . optionally , further particle identification might be available from a mirror that reflects cherenkov light to an instrumented back - plane directly upstream from the target . the mirror is backed by electromagnetic and hadronic calorimeters and , lastly , by iron - core toroidal magnets for muon identification . other possible specialized high rate neutrino target and detector possibilities include polarized solid protium - deuterium targets for spin physics ( sec . [ sec : qcd_pol ] ) and nuclear targets ( sec . [ sec : qcd_nucl ] ) for studies of @xmath82 dependence . a more massive tracking liquid target ( sec . [ ch : ew ] ) would be suitable for precision electroweak physics using neutrino - electron scattering . table [ tab : events ] provides a summary of some of the characteristics for examples of each of the three high rate target types discussed in this section and also gives plausible but very approximate integrated luminosities and event sample sizes for of the illustrative 50 gev and 175 gev beam parameters in table [ tab : beam_specs ] . this overview motivates and introduces the more detailed discussions that follow on : deep inelastic scattering and quantum chromodynamics ( sec . [ sec : qcd ] ) , quark mixing ( sec . [ ch : qm ] ) , precision electroweak tests ( sec . [ ch : ew ] ) , rare and exotic processes ( sec . [ ch : rare ] ) , charm physics ( sec . [ ch : charm ] ) and neutrino oscillations ( sec . [ ch : osc ] ) . before proceeding , we note that much of the interesting physics involves aspects of cc and nc charm and beauty production ( and hence the motivation for active vertex detectors as targets ) . as well as a ckm physics program that complements those from @xmath83 and @xmath84 factories and from precision @xmath53 boson branching fraction measurements at colliders , @xmath85 and @xmath86 production at a @xmath2mc allows precisions tests of qcd near heavy flavor thresholds , permits sensitive probes for new physics such as flavor - changing neutral currents , and provides a novel , very high statistics sample of charmed hadrons . figures [ cbprod_nu ] and [ cbprod_nubar ] show heavy quark production fractions and indicate that , given the expected multi - billion inclusive event samples , very high statistics can indeed be accumulated for both @xmath86 and @xmath85 final states at sufficiently above the relevant energy thresholds . historically , neutrino experiments have made major contributions to our understanding and verification of the both the qcd theory of strong interactions and the constituent components of protons and neutrons . the extrapolation of present experimental statistics consisting of @xmath87 events to the expected @xmath88 well reconstructed dis events at @xmath2mcs might well provide the best ever experimental laboratory for studying qcd and the structure of the nucleon through a scattering process . both traditional and novel areas for potential study will be discussed in sec . [ sec : qcd ] . they include : ( 1 ) one of the most precise and theoretically sound measurements of the strong coupling constant , @xmath89 ; ( 2 ) stringent consistency checks for the predictions of perturbative qcd ; ( 3 ) detailed flavor and spin dependence nucleon structure functions using both cc and nc probes ; ( 4 ) precise tests of qcd near the @xmath86 and @xmath85 quark heavy flavor transitions ; and ( 5 ) the first systematic studies of qcd in nuclear environments probed by neutrino and antineutrino beams . some of the most important high - rate measurements will involve the cabibbo - kobayashi - maskawa ( ckm ) mixing matrix that characterizes cc weak interactions of quarks . this topic is discussed in detail in sec . [ ch : qm ] . neutrino - nucleon dis offers unique and systematically independent measurements of ckm matrix elements since it uses a high @xmath90 virtual @xmath53 probe coupling directly to quarks rather than relying on the complex interplay of weak and strong interactions that is inherent in hadron decay . for sufficiently high energy and event rates , four of the nine ckm matrix element amplitudes @xmath91 , @xmath92 , @xmath93 and @xmath94 are directly probed through @xmath86 and @xmath85 production . the higher momentum transfers from the external @xmath53 probe allow for a cleaner theoretical interpretation that requires only relatively small corrections from perturbative qcd . as a further theoretical advantage , the measurements are semi - inclusive i.e. summing over all final states with single charm or beauty production and thus do not suffer from uncertainties in hadronic branching ratios . the fractional production rates shown in figures [ cbprod_nu ] and [ cbprod_nubar ] for @xmath3cc @xmath86 and @xmath95 production at high energies are of order @xmath96 and @xmath97 , respectively , where @xmath98 are the @xmath99 cabibbo - kobayashi - maskawa ( ckm ) matrix elements . other significant contributions to heavy flavor production are proportional to @xmath100 and @xmath101 , where @xmath102 are the @xmath103 ckm elements . the relatively clean theoretical interpretations and large samples of flavor - tagged events , particularly for charm production , should allow impressive measurements of the absolute squares for several of the elements in the ckm quark mixing matrix . estimated precisions in determining the ckm matrix elements are summarized in table [ ckm_table ] . perhaps the most interesting potential measurement outlined in sec . [ ch : qm ] is the determination of @xmath104 to better than @xmath105 , perhaps eventually reaching @xmath106 . this is an order of magnitude better than the current uncertainty and might well be better than will be achieved in any other single measurement at , for example , a @xmath83 factory . @xmath83 production should also allow for an extraction of @xmath107 at the few percent level that is systematically different , in both its experimental and theoretical aspects , from studies of decay processes and that is comparably accurate to the anticipated future measurements using decays . the matrix element @xmath108 is already best measured from cc charm production in today s neutrino experiments , based on event samples of several thousands of events . the present accuracy is mainly limited by statistics and uncertainties in charmed hadron production and decay characteristics . it is clear that the accuracy in @xmath91 would be vastly improved from the analysis of hundreds of millions of vertex - tagged charm events in a high performance detector . [ c]\|c\|lll\| & @xmath109 & @xmath110 & @xmath111 + @xmath112 & @xmath113 & @xmath114 & @xmath115 + & @xmath116 & @xmath117 & @xmath118 + & & & + @xmath119 & @xmath120 & @xmath121 & @xmath122 + & @xmath123 & @xmath124 & @xmath125 + & & & + section [ ch : ew ] demonstrates that @xmath2mcs should be able to provide two types of precision measurements of the weak mixing angle @xmath126 : from the ratio of neutral current ( nc ) to charged current ( cc ) dis events and also from neutrino - electron scattering . both types of determinations require a large extrapolation in event statistics and experimental technique from today s best neutrino results . they will allow vigorous consistency checks of the standard model and provide sensitivity to several potential possibilities for new physics . with its huge statistics , the dis measurement of @xmath126 will eventually be systematically limited by theoretical hadronic uncertainties but it should anyway become several times more precise than today s best neutrino measurements , which are already equivalent to about a 100 mev uncertainty on the @xmath53 mass . by contrast , no significant theoretical uncertainties enter into neutrino - electron scattering a simple scattering process between two elementary point particles and so the measurements will be limited only by statistics and experimental ingenuity . one can contemplate neutrino - electron scattering event samples as large as @xmath127 events using a dedicated detector with parameters like those given in table [ tab : events ] . this would correspond to impressive statistical uncertainties in @xmath126 of order @xmath128 and sensitivity to new contact interactions at energy scales up to approximately 25 tev . the biggest experimental challenges may come from normalizing the neutrino beam flux . if the experimental uncertainties could be reduced to the extremely challenging level of the statistical uncertainties then this process holds the potential for measurements of @xmath126 that might potentially be as good as or better than the best current measurements from collider experiments . a @xmath2mc provides a facility for several unique searches for physics beyond the standard model and provides a venue for observing a number of very rare standard model processes , as outlined in sec . [ ch : rare ] . examples of the former include flavor changing @xmath129 , @xmath130 , and @xmath131 neutral currents and isosinglet electron and muon - type neutral heavy leptons . a list of the latter includes @xmath132 annihilation and the scattering of virtual @xmath53 and @xmath54 bosons from quasi - real photons in a nuclear coulomb field ; these serve as weak analogs of @xmath133 and @xmath134 physics . sensitivity to high energy scales through mixing of @xmath53 or @xmath54 with higher mass propagators is limited due to the characteristically weak dependence of the new propagator mass reach as the fourth root of experimental statistics . a @xmath2mc should function as an efficient factory for the study of charm decays , with a clean , well reconstructed sample of several times @xmath127 charmed hadrons produced in @xmath135 neutrino interactions . section [ ch : charm ] points out several interesting physics motivations for charm studies at a @xmath2mc . measurement of charm decay branching ratios and lifetimes are useful both for qcd studies and for the theoretical calibration of the physics analyses on @xmath83 hadrons . charm decays also provide a clean laboratory to search for exotic physics contributions since the standard model predicts tiny branching fractions for rare decays , small @xmath136 asymmetries and slow @xmath137 oscillations . the charge of the final state lepton in cc - induced charm production from neutrinos tags charm quarks vs. antiquarks with high efficiency and purity . this tag is of particular benefit to oscillation and @xmath136 studies , as is the expected precise vertexing reconstruction for the proper lifetime of decays . section [ ch : charm ] shows it to be quite plausible that a @xmath2mc could provide the first observations of both @xmath138 mixing and @xmath136 violation in the charm sector and additionally provide some context for their proper interpretation . the potential for long baseline oscillation experiments is the @xmath2mc topic that is currently of most interest to the high energy physics community . however , both the experimental and theoretical status of neutrino oscillations are in such a state of flux that long - term predictions for @xmath2mcs can be stated only in general terms . long baseline experiments at @xmath2mcs might well provide a definitive follow - up to the recent intriguing evidence for neutrino oscillations . there will presumably already have been some progress in the verification or refutation of today s oscillation signals by the time long baseline @xmath2mcs come on - line . even so , @xmath2mcs will still clearly be important for more probative follow - up studies to characterize the form and phenomenology of any observed oscillation signals . for example , @xmath2mcs might help to determine whether an observed oscillation signal is consistent with mixing between three neutrino families or whether a fourth , sterile neutrino is required , as is discussed in chapter [ ch : osc ] . in the former case , long baseline @xmath2mcs can nail down the values of the mass - squared differences , search for @xmath136 violation and look for matter effects in oscillations in the earth s interior . clearly , the spectrum of possible studies would be richer still in the case of 4-neutrino mixing . starting from slac electron scattering experiments in the late 60 s , proceeding through the cern and fermilab neutrino and muon experiments of the 70 s through 90 s , and continuing with hera experiments still underway , deep inelastic scattering ( dis ) has provided us with an increasingly accurate picture of the partonic structure of the nucleon . moreover , dis has served and still serves as one of the best test - beds for perturbative qcd . a @xmath2mc could take the physics of dis to a new level by : ( 1 ) providing the statistical power to extract all six structure functions for @xmath2 and @xmath139 beams on proton and deuterium targets ; ( 2 ) allowing for low mass , high acceptance spectrometers with vastly improved resolution over present calorimetric detectors ; ( 3 ) creating naturally redundant measurements through simultaneous measurement of electron and muon final scattering final states ; ( 4 ) generating the rate and small beam spot size required for the first polarized neutrino targets ; ( 5 ) permitting use of active vertexing targets for systematic studies of heavy flavor production , and ( 6 ) facilitating the use of a large array of nuclear targets . examples of physics topics that would emerge include : * _ definitive proton parton distribution functions ( pdf ) for _ @xmath140__. _ _ the @xmath140 behavior of parton densities at accessible @xmath2mc @xmath90 controls the cross section behavior of the highest energy scale physics of the tevatron and lhc . understanding subtle deviations caused by new physics requires precise control of pdf systematics . a @xmath2mc will have the statistical power and the systematic redundancy checks to generate a complete pdf set from a single experiment . * _ _ a test of qcd to next - to - next to leading order ( nnlo ) . _ _ few experiments quantitatively test qcd beyond leading order in the coupling @xmath141 . it is frequently the case instead that leading order ( lo ) and next - to - leading order ( nlo ) provide equally good descriptions of the data ; and the nlo calculation is preferred mainly because it reduces theoretical systematic uncertainties in quantities such as @xmath141 . as an inclusive scattering process at space - like momentum transfer , dis is perhaps the phenomenon for which qcd is most rigorously applicable . nlo cross sections are already fully calculated and nnlo computations for several processes exist as well . testing the entire theory at nnlo seems feasible . * _ precise measurements of _ @xmath142 _ _ at moderate _ _ @xmath90 the running of the strong coupling constant is largely determined by excellent measurements of the @xmath143 lepton hadronic decay width at @xmath144 as well as a series of precise measurements at @xmath145 . getting this running right is important ; for example , one of the few experimental pieces of evidence for supersymmetry is its success in getting the strong , weak , and electromagnetic couplings to unify at one scale . given this importance , further precise measurements of @xmath146 at scales between @xmath147 and @xmath148 are valuable . a @xmath2mc provides at least two ways of achieving these : through scaling violations in non - singlet structure functions and through evaluations of the gross - lleweyllyn - smith sum rule . * _ studies of two - scale qcd via _ @xmath2 _ _ and _ _ @xmath149 _ _ heavy flavor production . _ _ a quark @xmath56 is treated as a heavy object in dis if @xmath150 whereas it is instead considered to be a parton when @xmath151 . deep inelastic scattering at a @xmath2mc allows study of the transition of @xmath56 from heavy quark to parton by opening the possibility of measuring quark mass effects at a series of scales in cc and nc scattering . the possibility of using vertexing targets maintains the inclusive nature of measurements by avoiding the need for final state lepton tagging . * _ neutrino spin physics . _ charged lepton scattering experiments from polarized targets show that @xmath57 and @xmath60 type quarks carry very little of the nucleon spin and have hinted at strong polarization effects in gluons and strange quarks . a @xmath2mc creates the first possibility of using polarized targets for neutrinos and brings all of their power for flavor and helicity selection to bear on nucleon spin physics . * _ _ neutrino nuclear physics . _ _ thin nuclear targets at a @xmath2mc can rapidly acquire the statistics to make measurements of the @xmath82 dependence of the @xmath152 structure function for neutrinos that complement those from charged lepton scattering . the first precise measurements of the @xmath82 dependence of @xmath153 will become available as well . invariance principles dictate the general form of @xmath52 ( @xmath44 or @xmath45 ) nucleon scattering . for energies much greater than the final state lepton mass and to leading order in electroweak couplings : @xmath154 , \label{crsec}%\end{aligned}\ ] ] with @xmath155 the fermi coupling constant , @xmath156 the nucleon mass , @xmath157 the neutrino energy , @xmath158 the inelasticity , @xmath159 the bjorken scaling variable , and @xmath90 the negative squared four - momentum transfer to the nucleon target . the plus ( minus ) sign in the final term is conventional for neutrino ( antineutrino ) scattering , and @xmath160 for cc(nc ) scattering . the structure functions @xmath161 , @xmath162 and @xmath163 contain all the information about the internal structure of the target . the cross sections for electron - neutrinos and muon - neutrinos are nearly identical , up to electroweak radiative corrections . the sf depend on the @xmath82 and @xmath54 of the target nucleus , on whether the beam is neutrino or antineutrino , and on whether the scattering is cc or nc . they can be experimentally extracted in principle by measuring the differential cross sections in fixed @xmath159 and @xmath90 bins as a function of @xmath158 and then exploiting the @xmath158 dependences shown in eq . [ crsec ] to fit for @xmath164 , @xmath152 , and @xmath153 . in practice , the reduced @xmath158 coverage created by cuts on final state lepton energies and kinematic constraints limits this procedure , and various model - dependent alternatives have been assumed . for example , the sf @xmath164 has rarely been measured in neutrino scattering ; instead it has been related to @xmath152 through a model for the longitudinal structure function@xmath165 with @xmath166 computed from qcd or taken from charged lepton scattering . charged current interactions , with their observable lepton in the final state , can be much better reconstructed than nc interactions and so we will assume cc sf in the discussion that follows . neutrino - nucleon scattering is the only dis process that can provide measurements of the parity - violating @xmath167 structure functions , apart from the much less precise measurements in a different kinematic regime from hera . the parity - conserving @xmath164 and @xmath152 structure functions for neutrino - nucleon scattering probe different combinations of quarks to the analogous sfs defined for charged lepton dis experiments . a rough summary of the current knowledge of neutrino sf follows . a more complete review may be found in ref . [ csb - rmp - ref ] . 1 . measurements at an accuracy of a few percent exist for @xmath168 , @xmath169 averaged cc sf@xmath170 _ using iron targets _ for @xmath171 @xmath172 and @xmath173 gev@xmath174 . the @xmath159 and @xmath90 ranges are highly correlated by the limited range of beam energies . these measurements assume a model for @xmath166 . uncertainties on @xmath175 are dominated by systematic effects , while @xmath176 errors still contain a significant statistical contribution . measurements at the @xmath177 level exist for : 1 . the @xmath168 , @xmath178 averaged @xmath166 using an iron target ; 2 . the @xmath168 averaged @xmath179 ; 3 . @xmath180 and @xmath181 for @xmath182 and with @xmath183deuterium . no sf - oriented neutrino experiments are currently in operation and no new experiments are planned other than a possible @xmath2mc program . the sf goal for an @xmath2mc is simple : to measure , over as wide a range of @xmath159 and @xmath90 as possible , the six sf of eq . [ crsec ] , particularly for the proton and deuteron but also for other nuclear targets . structure functions provide much of the information used to deduce pdfs . in turn , the pdfs are crucial for all predictions of event rates at the tevatron and lhc . to avoid nuclear complications , @xmath2mc sf should be extracted with proton and deuterium targets . to leading order ( lo ) in qcd , the sf can be expressed in terms of nucleon pdf as , i.e. @xmath184 is @xmath159 times the probability of finding a @xmath57 quark with momentum fraction @xmath159 . this is close to what is actually measured in experiments and is also the form provided in compilations such as pdflib@xcite . ] : @xmath185 , \nonumber\\ f_{2}^{\bar{\nu}n}(x , q^{2 } ) & = 2\left [ u^{n}(x , q^{2})+c^{n}(x , q^{2})+\bar { d}^{n}(x , q^{2})+\bar{s}^{n}(x , q^{2})\right ] , \nonumber\\ xf_{3}^{\nu n}(x , q^{2 } ) & = 2\left [ d^{n}(x , q^{2})+s^{n}(x , q^{2}% ) -\bar{u}^{n}(x , q^{2})-\bar{c}^{n}(x , q^{2})\right ] , \nonumber\\ xf_{3}^{\bar{\nu}n}(x , q^{2 } ) & = 2\left [ u^{n}(x , q^{2})+c^{n}(x , q^{2}% ) -\bar{d}^{n}(x , q^{2})-\bar{s}^{n}(x , q^{2})\right ] ; \label{sfquarks}%\end{aligned}\ ] ] and @xmath159 can be identified as the target s fractional 4-momentum carried by the struck quark . the callan - gross relation holds , @xmath186 implying that @xmath187 and @xmath188 provide redundant parton information . measurements of the eight independent observables of eq . [ sfquarks ] ( two sf for each beam on two targets ) represents more information than is available from charged lepton scattering but is not enough to specify the eighteen independent pdf for each target @xmath189 ( @xmath57 , @xmath60 , @xmath190 , @xmath86 , plus their antiquarks and the gluon for @xmath45 and @xmath191 ) . further constraints emerge from isospin symmetry : @xmath192 the nine extra constraints of eqs . [ isospin ] reduce the number of independent pdf to nine . more reduction occurs if one assumes @xmath193 which is supported by measurements of dimuon production in @xmath194 and @xmath195 scattering@xcite . furthermore , at the @xmath90 accessible to lower energy @xmath2mcs , one can consistently adopt a three - flavor qcd scheme , whence@xmath196 these relations are supported by measurements of nc charm production in @xmath197 and @xmath198 scattering@xcite . this finally leaves six independent pdf : @xmath199 , @xmath200 , @xmath201 , @xmath202 , and the _ valence distributions _ @xmath203 at lo , the four deuterium sf can be combined to yield the total quark and valence quark distributions@xmath204 & = u_{v}(x , q^{2})+d_{v}(x , q^{2}).\end{aligned}\ ] ] the four individual proton sf then allow separation of the @xmath57 and @xmath60 quark contributions . sensitivity to @xmath201 emerges from@xmath205 and , in principle , from comparison to the charged lepton @xmath206,@xmath207 practical implementation of eqs . [ delta xf3 ] and [ 5 - 18 rule ] has been stymied by difficulties in controlling systematic errors , and current measurements of @xmath201 all come from semi - inclusive @xmath208 and @xmath209 charm production . the gluon pdf does not enter directly at lo for cc sf . it affects the qcd evolution of @xmath152 and @xmath164 , appears with next - to - leading order ( nlo ) cross section terms , and enters directly into semi - inclusive double heavy quark production through the nc and cc processes@xmath210 as well as neutrino @xmath211 and @xmath212 production@xcite . at a @xmath2mc , data will be of sufficient precision to probe qcd to nlo , and perhaps to nnlo . in practice , the data will be in the form of differential cross sections in @xmath213 and @xmath158 at different neutrino energies . the general procedure for a qcd analysis of the data consists of : ( 1 ) defining a favorable kinematic region where pqcd is expected to apply to within small corrections ; ( 2 ) choosing a parameterization of the six pdf @xmath214 , where @xmath215 , @xmath216 is a reference @xmath90 and @xmath217 are a set of parameters describing the pdf ; ( 3 ) using pqcd to fit the data over all @xmath90 by varying @xmath218 . the quality of such a fit to the data constitutes an immediate test of qcd . for example , qcd can be verified to nlo if nlo pqcd provides a better fit than lo . while contemporary practice dictates fitting with nlo pqcd , it is worth pointing out that neutrino _ data _ do not convincingly favor the higher order calculation over a lo interpretation . assuming a good pqcd fit , one can then go on to extract the single parameter of qcd itself , which can taken to be the strong coupling constant evaluated at the reference @xmath90 of the experiment @xmath219 . as a technical comment , the fit procedure is admittedly somewhat complex , and the need to treat pdf parameterizations and qcd together arises from kinematic acceptance issues . the @xmath90 value will anyway be limited to @xmath220 and cuts on final state lepton energies may further limit the available @xmath221 phase space , particularly at lower energy @xmath2mcs . more precisely , if @xmath222 and @xmath223 represent the minimum acceptable final state lepton and hadron energies , then for a given @xmath90 , one has @xmath224 , \nonumber\\ e_{had}^{\min}/e & \lesssim y\lesssim1-p_{\ell}^{\min}/e.\end{aligned}\ ] ] this implies that it is impossible to span all of @xmath159 for a fixed @xmath90 in order to extract the pdf of eqs . [ sfquarks ] . an interpolation scheme is needed to connect different regions of @xmath159 and @xmath90 space ; fortunately , pqcd provides just that scheme through the dglap equations@xcite . more important even than the high statistics and potentially improved systematics promised by @xmath2mcs are the richness of evolution tests created by the availability of @xmath225 proton and deuterium sf , even with the limited phase space . examples include : 1 . @xmath226 $ ] : the `` classic '' non - singlet sf s evolution is independent of @xmath227 but suffers uncertainties from charm production . 2 . @xmath228 $ ] : the most precisely measurable sf usefully constrains @xmath227 and cross - checks charged lepton scattering . @xmath229 : a new combination for a @xmath2mc ; this difference is both independent of @xmath202 and charm production . sum rule - tests comprise some of the most accurately calculated observables in qcd . for example , the gross llewellyn smith ( gls ) sum rule @xcite yields : @xmath230 dx\nonumber\\ & = 3\left [ 1-\frac{\alpha_{s}}{\pi}-a(n_{f})\left ( \frac{\alpha_{s}}{\pi } \right ) ^{2}\right . \nonumber\\ & \left . -b(n_{f})\left ( \frac{\alpha_{s}}{\pi}\right ) ^{3}+o\left ( \frac{\alpha_{s}^{4}}{\pi^{4}}\right ) + o^{\prime}\left ( \frac{m^{2}}{q^{2}% } \right ) \right ] , \label{sgls}%\end{aligned}\ ] ] with @xmath231 and @xmath232 are known functions of @xmath90 and the specified number of active flavors @xmath233 used in the pqcd analysis . corrections from order @xmath234 pqcd@xcite and order @xmath235 higher twist effects have been calculated@xcite . a @xmath2mc would further provide the targets and the statistics to evaluate for the first time with a precision sufficient to test qcd through the adler sum@xcite@xmath236 dx,\ ] ] assuming that @xmath237 . structure function analyses in neutrino scattering are a complex business that requires a painstaking attention to systematic error sources . the current best qcd measurements the evolution@xcite of @xmath153 and the gls sum rule@xcite are limited by energy calibration uncertainties in the former and flux - related errors in the latter . calibration uncertainties will be reduced at @xmath2mcs through use of lower mass particle spectrometers that allow better resolution and the possibility of using _ in situ _ sources such as @xmath238 and @xmath211 decay . the ability to simultaneously measure @xmath239 and @xmath240 scattering provides built in cross - checks , and the @xmath239 cc scattered electrons can be measured both magnetically and calorimetrically to cross - calibrate the spectrometer and calorimeter . flux errors will be diminished considerably by the simplicity of the neutrino source compared to @xmath39 decay beams , which permits much more reliable monitoring . high rates will also allow direct flux measurement in the @xmath42 mode through use of the electroweak reactions @xmath241 and @xmath242 . the simple language of the quark parton model must be modified for @xmath52 dis events with a charm or beauty quark in the final state in order to take into account the non - negligible quark mass . this presents both a challenge and an opportunity to test the qcd formalism for making these corrections . perhaps the simplest and most widely used correction scheme is the essentially kinematic `` slow rescaling '' model of georgi and politzer @xcite , which amounts to a redefinition of the scaling variable @xmath159 through @xmath243 where @xmath244 is a heavy quark mass . one of the shortcomings of this prescription is that it fails to address ambiguities in the pdf that arise when one attempts a self - consistent pqcd treatment of heavy quark production . neutral current charm production illustrates this ambiguity . on the one hand , this can be treated as a `` flavor - creation '' process , @xmath245 , which occurs at order @xmath246 in the pqcd expansion . in this case , the appropriate scheme for calculation of qcd radiative corrections involves only three light flavors ( the so - called three - flavor scheme ) , and all of the effects associated with the charm quark are accounted for perturbatively in the hard scattering . on the other hand , the charm quark can be considered to be `` light '' at high energies when @xmath247 so the mass of the charm quark is much smaller than the relevant physical scale . in this case , one can view the charm quark as a parton with a corresponding parton distribution function ( the so - called four - flavor scheme ) . in particular , the lowest order wilson coefficient for charm production in this situation is a `` flavor - excitation '' process , @xmath248 , which is of order @xmath249 . these two viewpoints must be reconciled in a consistent manner in order to avoid double counting . the corresponding calculational problem for pqcd is the presence of a heavy quark mass scale that can be comparable to the @xmath90 of the interaction . in the absence of this complication , the factorization theorem of qcd separates the high scale set by the value of transferred momentum , @xmath250 , from the low scale of hadronic physics , @xmath251 , and operator product expansion techniques can be used to sum the large logarithms of the form @xmath252 that multiply the expansion parameter , @xmath146 . a practical recipe for incorporating heavy quark masses in the pqcd summations is provided by the acot @xcite prescription , which treats the number of active flavors , @xmath233 , as a scale - dependent quantity . in particular , it suggests the use of a three - flavor evolution for @xmath253 and a four - flavor evolution above @xmath254 , with continuity at the break point . in this prescription , the parton distribution functions are labeled by the number of active flavors and the heavy quark parton distribution functions , @xmath255 , vanish for @xmath256 and satisfy the usual @xmath257 qcd evolution equation ( with massless kernel functions ) for @xmath258 . charm and beauty production at @xmath2mcs will include large numbers of events with @xmath90 above , below and around the effective @xmath259 scale and should provide the most stringent tests from any experimental process of the pqcd formalism for heavy quark production . at that point , the experimental precision will likely require a more careful quantum field theoretical treatment of heavy quark masses . available rates at a @xmath2mc ( figs . [ cbprod_nu ] and [ cbprod_nubar ] ) provide access to five different values for @xmath244 : @xmath254 ( from @xmath260 ) , @xmath261 ( from @xmath262 ) , @xmath263 ( from @xmath264 ) , @xmath265 ( from @xmath266 ) , and @xmath267 ( from @xmath268 ) . to the degree a consistent pair of values for @xmath269 and @xmath270 emerges , one would have established theoretical control over these production processes . those mass values could also be compared to what qcd sum rules , heavy quark expansions and lattice qcd yield . they can also be used to calibrate models for heavy quark production that enter into other @xmath2mc analyses , such as the pqcd tests described above and precision @xmath126 measurements in inclusive nc scattering discussed in sec . [ ch : ew ] . other than the ubiquitous high rates , the main experimental asset for a @xmath2mc in these studies will be the opportunity to use vertexing targets . this should allow for lifetime tagging of heavy flavors that minimizes systematic effects due to fragmentation and decay uncertainties . the unique level of quark - by - quark characterization of nucleon structure expected at @xmath2mcs will provide an invaluable reference source for many diverse analyses in collider and fixed target physics including , of course , other precision analyses at @xmath2mcs . precise measurements as @xmath271 are particularly relevant to the modeling of rates for interesting physics processes and backgrounds at hadron colliders because uncertainties at high @xmath159 and at the typical @xmath90 values for @xmath2mcs will evolve to uncertainties at much lower @xmath159 as @xmath90 increases to collider values . uncertainties at high @xmath159 in current nucleon pdf derive from two sources : the ratio @xmath272 as @xmath271 and the role of higher twist corrections . analyses on present leptoproduction data sets that used hydrogen and deuterium targets have been unable to precisely pin down the high-@xmath159 behavior of @xmath273 . for example , qcd fits @xcite to the high-@xmath159 nmc data and the cdf @xmath53-decay asymmetry improve with the inclusion of a simple @xmath159-dependent correction for a @xmath273 ratio that asymptotically approaches @xmath274 rather than zero for @xmath271 . on the other hand , the recent cteq5 global qcd analysis @xcite found that including this correction had little effect on the quality of their fits . besides the statistical and experimental uncertainties in the data sets , a complication with high @xmath159 analyses is the need to model nuclear binding effects in deuterium this issue can be avoided at a @xmath2mc , where a high statistics exposure to a @xmath275 target alone could directly measure the @xmath273 ratio in protons as @xmath271 from the ratio of @xmath276 to @xmath277 cross sections . such a measurement would require only a small correction for the residual sea quark contributions at high @xmath159 . measurement of quark pdfs at high @xmath159 is closely related to the question of the leading power corrections in the qcd perturbative expansion that are known as `` higher twist effects '' . the @xmath278 order higher twist effects are proportional to @xmath279 and reflect the fact that quarks have transverse momentum within the nucleon and that the probe becomes larger as @xmath90 decreases , thus increasing the probability of multi - quark participation in an interaction . as was the case with the @xmath280 ratio , different analyses of higher twist corrections in current data leave unresolved issues that would benefit from new experimental information . an analysis by milsztajn@xcite that combined electroproduction data from slac with bcdms muo - production data found that the relative size of the twist-4 contribution rapidly increased above @xmath281 and was equal in magnitude to the leading @xmath282 term for @xmath283 . the only measurements of this higher - twist term in neutrino experiments have been two low - statistics bubble chamber experiments : in gargamelle @xcite with freon and in bebc with @xmath284 . both bubble chamber analyses are complicated by nuclear corrections at high-@xmath159 . however , both found a twist-4 contribution that is smaller in magnitude and , most significantly , of opposite sign to that of the charged leptoproduction analysis . in contrast , a cteq global qcd analysis that combines neutrino and charged lepton production analyses up to @xmath285 , finds that _ no _ higher - twist term is required for values of @xmath90 down to @xmath286 gev@xmath174 . however , this analysis uses a cut on the invariant mass of the hadronic system , @xmath287 gev , that could exclude the bulk of any twist-4 contribution . most recently , the yang - bodek analysis mentioned above @xcite reanalyzed electroproduction data looking for a higher twist contribution . they find that by incorporating a nlo qcd analysis , as opposed to the lo analysis used by milsztajn , the higher twist contribution becomes much smaller . however , they have also included the @xmath288 model in their extraction of higher twist , which will reduce the size of any extracted higher twist term . from a more theoretical viewpoint , a recent cteq paper by guo and qiu @xcite has predicted the leading @xmath159 dependence to the higher - twist term in @xmath289 in the region of large @xmath159 and found that the higher - twist contribution is _ different _ for @xmath57 and @xmath60 quarks . were this to be true , then the @xmath273 analysis and the leading power correction analysis would be directly intertwined . interestingly enough , this prediction only depends on two non - perturbative parameters and is therefore highly constrained . it could be tested at both cebaf and a @xmath2mc . [ subsec : qcd_pol_intro ] a unique new feature of @xmath2mcs would be the availability of sufficiently intense beams to allow , for the first time , measurements of neutrino scattering off nucleons in polarized targets . this would provide access to the polarized nucleon structure functions for neutrino - nucleon scattering and could answer several currently unresolved questions about the spin structure of the nucleon . polarized structure functions ( psfs ) can be represented approximately by the quark - parton model in terms of the differences between the parton densities of quarks polarized parallel to the nuclear spin and those that are polarized anti - parallel : @xmath290 in the naive ( lo ) quark - parton model , the polarized structure functions @xmath291 and @xmath292 have quark spin content corresponding to the quark content of the parity conserving and parity violating unpolarized structure functions @xmath293 and @xmath167 , respectively : @xmath294 and @xmath295,\\ 2xg_{3}^{\overline{\nu}p}(x , q^{2 } ) & = -[\delta u(x , q^{2})+\delta c(x , q^{2})-\delta\bar{d}(x , q^{2})-\delta\bar{s}(x , q^{2})].\label{pol_g3}%\end{aligned}\ ] ] the other parity - conserving psf , @xmath296 , has no simple interpretation in the lo quark - parton model , as is also the case with the parity - violating @xmath297 . the remaining psf , @xmath298 , is related to @xmath292 via @xmath299 in the naive quark - parton model . the levels of the different polarized quark pdf s can be regarded as quantifying the extent to which a parton of flavor @xmath56 `` remembers '' the polarization of its nucleon parent in interactions at a given @xmath90 . the nucleon spin , ( @xmath300 ) , can be decomposed as : @xmath301 where @xmath302 and @xmath303 are the integrated net quark and gluon helicities along the nucleon spin direction , e.g. , @xmath304 the @xmath305 and @xmath306 represent the relative orbital angular momenta of the quarks and gluons , respectively . to date , the only charged lepton polarized dis data comes from non - collider experiments , running at energies where photon exchange dominates . thus , only the parity conserving polarized structure functions @xmath307 and @xmath308 have been studied so far . the results are surprising@xcite . from the measured first moment of @xmath307 structure function , it can be concluded that only @xmath309 of the nucleon spin is carried by quarks if one assumes @xmath310 , _ i.e. _ , @xmath311 . an additional conclusion in this simple parton model interpretation is that the strange sea is anti - aligned with the nucleon spin , @xmath312 . in the more realistic qcd - enhanced parton model , the qcd evolution of the quark distributions brings in contributions from gluon radiation @xmath313 and pair - production @xmath314 that induce a non - zero gluon spin distribution . this latter conclusion is supported by pqcd analyses of the @xmath90 evolution of the available data on @xmath307 , as is illustrated in fig . [ gluon ] . although the decomposition of fig . [ gluon ] is highly model dependent , the ability of neutrinos to select specific quark flavors should allow a polarized dis experiment at a muon storage ring to greatly constrain such models . an ideal polarized target would have a mass of at least 10 kg , high polarizability , a large fraction @xmath315 of polarizable material , and the capability for polarizing both protons and neutrons . the solid @xmath316 compound material `` ice '' appears to be a promising new technology @xcite . both the @xmath317 and the @xmath318 can be polarized , either in separate experimental runs or together . the expected polarization and dilution are @xmath319 and @xmath320 for hydrogen , and @xmath321=50% and @xmath322 for deuterium . a 7 kg ( density @xmath323=0.11 g/@xmath324 ) polarized target with the qualities mentioned above can be built out of an ice - target that is 20 cm in radius and 50 cm deep , perhaps mounted upstream from a general purpose neutrino detector like that in fig . [ hrdet ] . raw event rates in such a polarized target could be of order @xmath325 events per year , as shown in table [ tab : events ] . a previous study for @xmath2mcs @xcite has shown that a @xmath326 kg target of polarized butanol with @xmath327 polarization could measure the strange sea polarization to about @xmath328 in a one year run in a beam downstream of a 250 gev muon collider with a @xmath329 m straight section . measurement of @xmath291 requires both @xmath2 and @xmath330 beams and follows from the double cross section difference : @xmath331 where @xmath332 denotes , for example , the neutrino scattering off a target @xmath189 with its spin polarized _ parallel _ to the neutrino helicity . this difference should be measured as a function of @xmath159 and @xmath158 to test for effects of unwanted psf such as @xmath296 . for @xmath292 one evaluates the similar quantity @xmath333 cross checks can be made with the target transversely polarized , where the suppression factors for the unwanted psf are different . measured values for @xmath291 from any nucleon target at a @xmath2mc would allow verification of the predicted decomposition : @xmath334 where the factor @xmath335 is model dependent and a common choice is @xmath336 , and one has knowledge of @xmath337 from other data . in contrast to polarized charged lepton scattering , this determination can be done without relying on low energy input from beta decay data augmented by su(3 ) symmetry . for another application , one notes that @xmath292 , @xmath297 and @xmath298 probe only non - singlet combinations of parton densities so that they do not get a contribution from the gluon density . then , for scattering off an isoscalar target : @xmath338 ( this is the polarized target analog of eq . [ delta xf3 ] . ) assuming that @xmath339 , this provides a measure of the level of polarization of the strange quark sea . like @xmath153 , the non - singlet psf , @xmath292 , @xmath297 and @xmath298 , have qcd evolutions containing no contribution from gluons at lowest order . comparison of the non - singlet functions with the singlet sf @xmath291 and @xmath152 should therefore provide an indirect means for measuring the gluon contribution @xmath303 . as was the case for unpolarized targets , final state quark flavor tagging should provide additional semi - inclusive structure functions to augment the information on quark content from inclusive polarized structure functions . probably the most important example will be studies of the strange spin contribution via charm production . figures [ cbprod_nu ] and [ cbprod_nubar ] show that approximately @xmath105 of the events from a 50 gev muon storage ring will have charm in the final state , with @xmath340 of these decaying semi - leptonically . an experiment with 20 million neutrino interactions thus will have @xmath341 semi - leptonic charm events before kinematic cuts , which should be sufficient for a precise measurement of @xmath342 and @xmath343 . nuclear effects in dis have been studied extensively using muon and electron beams but have only been glanced at for neutrinos in low - statistics bubble chamber experiments . neutrino experiments with high statisticss have formerly only been possible using heavy nuclear targets such as iron calorimeters and , for these targets , nuclear effects in @xmath52 interactions have typically been considered as problems to overcome rather than as a source of physics insights . this section reviews the physics of nuclear qcd that is relevant to neutrino interactions and shows that @xmath2mcs could instead provide experimental conditions where a great deal of interesting knowledge could be added to this field by using a variety of heavy nuclear targets as well as @xmath275 and @xmath344 . nuclear studies at a @xmath2mc could use a general purpose detector such as that in fig . [ hrdet ] if it could be designed with interchangeable targets . alternatively , a detector dedicated to nuclear studies could be used , perhaps with a geometry similar to that of the fermilab e-665 tevatron muon experiment@xcite . for example , it could consist of a liquid @xmath275 or @xmath344 target followed by a rotating support of targets with different @xmath82 interspersed with tracking chambers , and then by an appropriate calorimeter / muon spectrometer . a @xmath2mc should be easily capable of supplying the event sample sizes required to examine the predicted nuclear effects . for example , one can consider a one - year exposure of each target to the beam in the 50 gev @xmath2mc scenario of table [ tab : beam_specs ] . then @xmath345 events would be acquired in targets subtending out to the @xmath346 characteristic angular size of the neutrino beam and that had lengths of 18 cm for @xmath344 , 1.4 cm for graphite and only 0.16 cm for tungsten . different types of nuclear effects arise on passing through four distinct regions in bjorken @xmath159 : 1 . `` shadowing '' for @xmath347 , 2 . `` anti - shadowing '' for @xmath348 , 3 . the `` emc effect '' for @xmath349 , and 4 . `` fermi motion '' for @xmath350 . these regions will now be discussed in turn . in the shadowing region , @xmath347 , there are several effects where a neutrino probe could provide different insights to charged lepton probes . considering first the limit as @xmath351 , the vector current is conserved and goes to zero but the axial - vector part of the weak current is only partially conserved ( pcac ) and @xmath289 approaches a non - zero constant value as @xmath351 . according to the adler theorem @xcite , @xmath352 can be related to to @xmath353 at @xmath354 . a @xmath2mc should be able to address the question of what effect a nuclear environment has on the adler theorem . the region of vector meson dominance ( vmd ) is reached in nuclear scattering of charged leptons ( @xmath355 scattering ) in the low-@xmath159 shadowing regime as @xmath90 increases from @xmath356 but remains below of order 10 gev@xmath174 . the physics concept of vmd is the dissociation of a virtual photon into a @xmath357 pair , one of which interacts strongly with the `` surface '' nucleons of the target nucleus . thus the interior nucleons are shadowed by the surface nucleons . neutrino scattering should involve not only a vmd effect ( though now with dissociation of a virtual @xmath53 rather than a photon ) but also additional contributions from axial - vector mesons such as the @xmath358 . further , there should be additional non - perturbative effects that appear as nuclear shadowing ( mainly in large nuclei ) and which involve gluon recombination from nucleons neighboring the struck nucleon that shift the parton distributions towards higher values of @xmath159 . _ et al . _ @xcite predict that the resulting shadowing effects in @xmath359 scattering will be roughly @xmath360 that measured in @xmath355 scattering . a more quantitative analysis by kulagin@xcite uses a non - perturbative parton model to predict that shadowing in @xmath361 scattering to be either equal to or slightly above that in @xmath355 scattering at @xmath362 gev@xmath174 . it also attempts to determine the quark flavor dependence of shadowing effects by separately predicting shadowing for @xmath363 , @xmath363 and @xmath364 . the predictions of kulagin should be testable at a @xmath2mc , as is indicated by fig . [ fig : shadow ] . this shows a simulation for a @xmath2mc data sample of the predicted measured ratios of the values for @xmath365 and @xmath366 on a calcium target divided by those on a deuterium target : @xmath367/f_{2}(x , q^{2})[d_{2}],\\ r_{3a}(x ) & \equiv f_{3}(x , q^{2})[ca]/f_{3}(x , q^{2})[d_{2}],\end{aligned}\ ] ] where the ratios are an event - weighted average over the experimental @xmath90 range . as can be seen , the predicted difference between the shadowing on sea and valence quarks is clearly visible down to @xmath368 . drell - yan experiments have also measured nuclear effects and their results are quite similar to dis experiments in the shadowing region . however , in the anti - shadowing region where @xmath369 makes a brief but statistically significant excursion above @xmath370 in dis , drell - yan experiments see no effect . this could be an indication of difference in nuclear effects between valence and sea quarks . eskola _ et al . _ @xcite have quantified this difference using a model which predicts that the differences between nuclear effects in @xmath364 and @xmath371 should persist through the anti - shadowing region as well . taking the work of kulagin and eskola together implies that nuclear effects in @xmath364 should be quite dramatic , with more shadowing than @xmath371 at lower @xmath159 followed by @xmath372 rising fairly rapidly to yield significant anti - shadowing around @xmath373 . the 50 gev @xmath2mc experiment assumed for figure [ fig : shadow ] should be able to measure anti - shadowing effects and the difference between shadowing effects in @xmath289 and @xmath374 at a statistical level of about 6 standard deviations . analyses from dis experiments of @xmath289 in the `` fermi - motion '' region , @xmath350 , have required few - nucleon correlation models and multi - quark cluster models to fit the data . these models boost the momentum of some of the quarks , producing a high-@xmath159 tail in @xmath289 that is predicted to behave as @xmath375 . however , fits to @xmath376 @xcite and @xmath194 @xcite have obtained two different values for the fitted constant @xmath377 : @xmath378 and @xmath379 , respectively . this was considered surprising because of the expectation that any few - nucleon - correlation or multi - quark effects would have already saturated by carbon . a high statistics data sample from a @xmath2mc , with precise kinematic reconstruction of the bjorken @xmath159 values for each event , could go a long way towards resolving the dependence of the value of @xmath377 on the nucleus and on the lepton probe . the ckm quark mixing matrix comprises a set of fundamental parameters of the standard model that reflect the dynamics shaping the generation of particle masses . furthermore , it provides essential input for predictions on various cp asymmetries in @xmath83 decays which should be sharpened as much as possible in order to increase their sensitivity to the presence of new physics processes . it is likely that @xmath2mcs could make a very substantial contribution to measurements of quark mixing , with systematically unique measurements of four of the nine moduli in the ckm matrix that should be comparable to or better than the best future measurements at other experiments . the relevant moduli are @xmath380 , @xmath381 , @xmath382 , and @xmath383 , to be extracted from charm production off @xmath60 and @xmath190 quarks , and from beauty production off @xmath57 and @xmath86 quarks , respectively . if muon colliders ever reach the 100 tev energy center - of - mass range then their @xmath2mcs could further provide a unique opportunity for precise direct measurements of the ckm elements involving the top quark @xcite . this section is organized as follows . section [ subsec : qm_intro_current ] gives a summary of the current experimental status and future expectations for before the advent of @xmath2mcs . section [ subsec : qm_intro_extraction ] then gives a general overview of the method for extracting ckm information at @xmath384mcs . sections [ sec : qm_tocharm ] and [ sec : qm_tobottom ] then give more specific details on @xmath2mc experimental analyses involving the production of charm and bottom quarks , respectively . a further analysis that may have potential for a measurement of ckm parameters diffractive charmed vector meson production is briefly addressed in subsection [ sec : qm_diffr ] . finally , subsection [ sec : qm_improvements ] summarizes all of these expected improvements from @xmath2mcs to our knowledge of the ckm matrix . table [ ckm_table ] presented the current @xcite experimental values and uncertainties for the absolute squares of the interesting ckm matrix elements for @xmath2mcs : @xmath385 , @xmath386 , @xmath387 and @xmath388 . the first two of these can be more tightly constrained by adding the assumption of three family unitarity , so improved measurements of these elements can be considered as testing these unitarity conditions . the values for @xmath380 and @xmath381 have been extracted from charm production in deep inelastic scattering supplemented by information gathered in semileptonic @xmath318 decays , while @xmath383 and @xmath389 have been obtained from semileptonic @xmath83 decays . in the next few years , _ exclusive _ semileptonic charm decays like @xmath390 @xmath391 @xmath392 @xmath393 will be measured more precisely . however it is not clear at present whether theoretical technologies like qcd simulations on the lattice will improve sufficiently to fully utilize such data for extracting more accurate values of @xmath381 and @xmath380 . at present @xmath383 is measured with about a 5% _ theoretical _ uncertainty and a 5% or better _ experimental _ uncertainty . ( for comparisons with the uncertainties quoted in table [ ckm_table ] recall that the fractional uncertainty in the _ square _ of the modulus is twice that of the modulus itself . ) the two most reliable determinations are from the semileptonic @xmath83 width and the form factor for @xmath394 at zero recoil . one can reasonably expect the theoretical uncertainty in these processes to go down to about @xmath395 @xcite . the situation is considerably less satisfactory for @xmath382 . most of the analyses employed so far contain a large dependence on theoretical models , the accuracy of which is hard to evaluate . one can hope that studies of both exclusive and inclusive semileptonic @xmath83 decays done at the @xmath83 factories will yield values for @xmath382 with a _ theoretical _ accuracy of no more than @xmath327 by 2005 . however , this is not guaranteed . it is highly desirable to improve on this situation by obtaining more precise and reliable values for these ckm parameters . the main reason is that they are theoretically expected to be a consequence of the generation of quark masses which in turn reflects dynamics operating at presumably ultra - high energy scales . it turns out that qualitatively quite distinct `` textures '' for the yukawa couplings assumed to apply at gut scales lead to ckm parameters that , due to differing renormalization effects , are numerically relatively similar at electroweak scales . secondly , various cp asymmetries in @xmath83 decays will be measured presumably to better than 5% over the next ten years ; and their predicted values depend crucially on the size of @xmath387 , @xmath388 ( and @xmath396 ) . to exploit the discovery potential for new physics to the fullest one wants to match up experimental and theoretical accuracy . to be more specific : * one wants at least to confirm the values for @xmath381 and @xmath383 by a _ systematically _ different method . * one would like to determine a precise value for @xmath380 _ without _ imposing three - family - unitarity . * it is an important goal to extract the value of @xmath382 with considerably less than @xmath327 uncertainty . it appears that @xmath2mcs are up to these tasks and possibly more . indeed , if a future @xmath2mc in the 100 tev energy range were ever to be realized , then even top quark production could be studied and this would uniquely allow the extraction @xcite of @xmath396 and @xmath397 in a relatively straightforward way ! the experimental quantities used for all the quark mixing measurements in @xmath2mcs are the production cross sections and kinematic distributions of heavy flavor final states from cc interactions . these are related back to the quark couplings through a parameterization of the initial state quark distributions and a model of the production processes that includes threshold suppressions due to the quark masses . measurements of ckm matrix elements at @xmath2mcs will be analogous to , but vastly superior to , current neutrino measurements of @xmath398 that use dimuon events for final state tagging of charm quarks , which are reviewed elsewhere@xcite . this improvement is due to the overall higher rate of interactions , the ability to tag charm _ inclusively _ by vertexing , and the capability of fully reconstructing the event kinematics for charm decays to hadrons . in principle , the struck quark can be converted into any of the three final state quarks that differ by one unit of charge . in practice , production of the heavy top quark is kinematically forbidden except from neutrinos at energies above about 16 tev , and the production of other quark flavors is influenced by their mass . after correcting for this kinematic suppression , the standard model predicts the probability for the interaction to be proportional to the absolute square of the appropriate element in the ckm matrix . the most relevant theoretical uncertainty lies in the treatment of the production thresholds . measuring these cross sections at different energies and separately for cc and nc reactions should help in understanding the threshold behavior . the extraction of @xmath380 will be the cleanest of the four ckm measurements at @xmath2mcs discussed here and is the one ckm element whose modulus is already best determined in neutrino - nucleon scattering . it is measured from charm production off valence @xmath60 quarks in isoscalar targets , and this valence quark distribution can be accurately determined from nucleon structure function measurements . figures [ cbprod_nu ] and [ cbprod_nubar ] of sec . [ ch : intro ] show that charm production occurs in @xmath399 of the cc interactions from 100 gev neutrinos , and will occur at slightly lower / higher rates for lower / higher energies according to the changing level of mass threshold suppression . this implies that realistic charmed event sample sizes might reach the @xmath127 level . the feature of the @xmath380 analysis that makes it systematically rather clean is that charm production from valence quarks occurs for neutrinos but not for antineutrinos . valence quarks have a harder @xmath159 distribution than sea contributions from both @xmath190 and ( at a lower level ) @xmath60 quarks . it is also helpful that the @xmath190 and @xmath60 seas seen by neutrinos are also closely equal to their antiquark counterparts that are probed by antineutrinos , as would be verified in structure function analyses of the @xmath2mc data set ( sec . [ sec : qcd ] ) . this allows use of the @xmath139 sample as an effective background subtraction for the sea contributions to the charm production by neutrinos . to illustrate , one can measure scattering from an isoscalar target @xmath318 with @xmath178 beams and extract the ratio @xmath400 where @xmath401 represent the charm production cross sections , and @xmath402 denote the total inclusive cc cross sections . to leading order in qcd , this ratio can be computed from@xmath403 where @xmath404 , @xmath405 , and @xmath254 is the charm quark mass . note that @xmath406 depends only on the precisely measured @xmath407 , the well understood valence quark distribution , @xmath408 the charm mass , and @xmath409 . because the valence distribution peaks at relatively high @xmath159 , threshold effects associated with @xmath254 are minimized . next - to - leading - order calculations are more complicated but leave these features intact . as in analyses of today s neutrino experiments , the @xmath60 and @xmath190 contributions to the charm event sample will in practice be separated from one another in a fit involving the measured @xmath159 distributions of the events and with the charm quark mass as a fitted parameter . with such huge statistics , the statistical uncertainty in the measurement will be only at the @xmath128 level so the measurement will be dominated by systematic uncertainties . major contributions to the uncertainty are likely to be : * estimation of the charm tagging efficiency . this error is minimized by the clean , high efficiency charm tagging that is possible with high performance vertexing and a negligible @xmath83 hadron background@xcite . * the charm production model . it is a very attractive feature that the measurement is effectively of the ratio of charm production to the total production in each kinematical bin . therefore , one is sensitive only to the kinematical suppression of charm states and not to the initial parton distribution functions . mass threshold corrections that are applied according to theoretical models can be checked and calibrated from the trends in the data itself over the range of kinematical bins . therefore , this uncertainty should initially decrease with increasing sample size . * estimation of the extent to which the sea is symmetric in quarks versus antiquarks . this uncertainty is lessened because the signal is largely at higher @xmath159 than the background . if helpful , a cut in @xmath159 value could be applied to the fit to further reduce this uncertainty . with experimental handles on the major systematic uncertainties , it may be guessed that the irreducible _ theoretical _ uncertainty on @xmath398 due to violations of quark - hadron duality might be at the percent level . a measurement of @xmath410 presents more challenges because this matrix element always appears in combination with the strange quark pdf @xmath411 , which is difficult , but not impossible , to measure separately . perhaps the best scheme discussed in sec . [ sec : qcd ] assume a known ckm matrix . ] involves a simultaneous analysis of inclusive charged current scattering and semi - inclusive charm production@xcite on deuterium targets . the inclusive cc cross section is of the form@xmath412 where @xmath413 represents the part of the cross section with no charm in the final state ( @xmath85 production is neglected ) . to leading order@xmath414 \left ( 1-\frac{m_{c}^{2}}{2me\xi } \right ) , \label{sigma - nu - charm}\\ \frac{\pi}{2g_{f}^{2}me}\frac{d\sigma_{\mu cx}^{\bar{\nu}d}}{dxdy } & = \left [ \left\| v_{cs}\right\| ^{2}\bar{s}\left ( \xi , q^{2}\right ) + \left\| v_{cd}\right\| ^{2}\bar{q}\left ( \xi , q^{2}\right ) \right ] \left ( 1-\frac{m_{c}^{2}}{2me\xi}\right ) , \label{sigma - bar - charm}\\ \frac{\pi}{2g_{f}^{2}me}\frac{d\sigma_{\mu\not cx}^{\nu d}}{dxdy } & = \left\| v_{ud}\right\| ^{2}q\left ( x , q^{2}\right ) + \left\| v_{us}\right\| ^{2}s\left ( x , q^{2}\right ) \nonumber\\ & + \bar{q}\left ( x , q^{2}\right ) \left ( 1-y\right ) ^{2}% , \label{sigma - nu - light}\\ \frac{\pi}{2g_{f}^{2}me}\frac{d\sigma_{\mu\not cx}^{\bar{\nu}d}}{dxdy } & = \left\| v_{ud}\right\| ^{2}\bar{q}\left ( x , q^{2}\right ) + \left\| v_{us}\right\| ^{2}\bar{s}\left ( x , q^{2}\right ) \nonumber\\ & + q\left ( x , q^{2}\right ) \left ( 1-y\right ) ^{2},\label{sigma - bar - light}%\end{aligned}\ ] ] measuring @xmath415 and @xmath416 gives @xmath417 . the @xmath158 dependencies of the non - charm cross sections allow measurement of @xmath418 and @xmath419 . assuming that @xmath420 and @xmath421 contribute negligible error , and that @xmath422 , one then extracts @xmath201 from the cabibbo suppressed @xmath423 pieces of @xmath424 , and then @xmath100 ( and @xmath96 ) from @xmath416 . this procedure will require very high statistics and extreme care with theoretical and experimental systematic errors as it turns on the @xmath425 term in the cross section , which contributes only @xmath426 of the total cross section . inclusive and charm cross sections must be cross - normalized to high accuracy , and the expressions ( [ sigma - nu - charm]-[sigma - bar - light ] ) must be generalized to include higher order qcd and possible non - perturbative effects . additionally , it may also be possible to measure the total strange quark content of the nucleon in nc `` leading particle tagging '' with a @xmath427 final state @xcite , although no experience with this method yet exists in neutrino physics . in the end , it will likely be difficult to compete with the indirect constraint on @xmath100 from precise measurements of the hadronic width of the @xmath53 , @xmath428 , at lep2 , the tevatron and the lhc . this observable is , to leading order , @xmath429 where @xmath430 is the total hadronic width . because @xmath431 is small and all other terms in the sum can be measured precisely elsewhere , a strong @xmath100 constraint emerges . measurements of @xmath387 and @xmath388 from the two transitions with a @xmath85 quark in the final state @xmath432 and @xmath433 , respectively require analyses that are conceptually similar to those for charm discussed in the preceding subsection . figures [ cbprod_nu ] and [ cbprod_nubar ] show that statistics for cc @xmath85 quark production to be of order @xmath434 events for a total sample of @xmath135 inclusive events , depending on the energy of the muon storage ring and the consequent threshold suppression due to the @xmath85 quark mass . besides the threshold suppression , the main reasons why the @xmath83 production levels are so low for the two processes are that , in the one case , the production from @xmath57 quarks is suppressed by @xmath435 @xmath436 , while production from charm quarks is inhibited by @xmath437 and the extra mass of the @xmath438 final state . given the nearly optimal vertexing geometry possible at a @xmath2mc , separating out much of the fractionally small @xmath85 hadron event sample@xcite from the charm background should be feasible . assuming that the @xmath83 tagging efficiency and purity are well known , then the @xmath85 production analysis should be a relatively straightforward copy of that for charm , involving a simultaneous fit to the @xmath57 and @xmath86 contributions to @xmath83 production from neutrino and antineutrino scattering . in analogy to charm production , the resolving power of the fit arises because antineutrino ( but not neutrino ) interactions give a high @xmath159 contribution from @xmath57 valence quarks , while @xmath433 transitions will be equal for @xmath2 and @xmath139 and typically at lower @xmath159 . it is again easy to see the effect at lo in qcd , as follows . define , in analogy to eq . [ r - charm ] , @xmath439 where now @xmath440 represents the beauty production cross sections . qcd then predicts @xmath441 with @xmath442 and @xmath263 the @xmath85 quark mass . as was the case with charm , @xmath443 depends only on the high @xmath159 valence quarks , which helps reduce the more substantial suppression associated with the higher value of @xmath263 . reasonably accurate estimates of the measurement precisions would require both a more detailed study and a knowledge of the expected event sample size . the precision of the @xmath435 measurement may approach the statistical limit of around @xmath106 for a total neutrino event sample of order @xmath135 events . the @xmath433 transition can ( unlike @xmath444 ) be identified experimentally by the presence of soft @xmath86 quark observed in association with the @xmath95 quark in @xmath2 scattering . at @xmath2mc energies , it is reasonable to assume that @xmath438 production will be dominated by @xmath445gluon fusion , and a good knowledge of the gluon pdf at high @xmath159 will be required . this may well limit the @xmath446 measurement accuracy to the few percent level . it is widely expected that @xmath383 will be known from @xmath83 decays in other experiments to a few percent or better by 2005 @xcite . given this , an alternative and maybe more useful analysis strategy for a @xmath2mc might be to combine the @xmath388 analysis data at the @xmath2mc with the value of @xmath388 as extracted from semileptonic @xmath83 decays in other experiments , with the purpose of analyzing the mass suppression in beauty production close to threshold , or in constraining the gluon pdf . this information could then be used as input to the extraction of @xmath387 as sketched above . it is possible to diffractively produce charmed vector mesons via @xmath445boson pomeron scattering@xcite : @xmath448 the @xmath449 cross section is of order @xmath450 fb / nucleus , while that for @xmath451 should be smaller by a factor of @xmath452 . for @xmath135 cc events , one expects of order @xmath453 @xmath454 events and @xmath455 @xmath456 events . if the @xmath457 and @xmath458 dynamical couplings are identical , the relative rate can measure the ratio of the ckm matrix elements . the experimental signature is fairly unique : a two- or three - prong muon - meson vertex with a @xmath459 or @xmath460 secondary . the cross section will peak at low momentum transfer and there should be no evidence for nuclear breakup . the chief theoretical uncertainty likely has to do with evaluating su(3)-breaking effects in the @xmath445vector meson couplings and this may limit the potential for ckm studies from this process . again , an alternative perspective can be taken : to use the information available on @xmath461 to analyze the @xmath462 pattern of diffractive @xmath463 _ vs. _ @xmath464 production . the particle data group@xcite assigns the following uncertainties to the four ckm parameters that were the main focus of our discussion : latexmath:[\ ] ] a similar equation can be written for direct neutrino - quark interactions . one can use @xmath642 symmetry to relate @xmath2 and @xmath643 couplings : @xmath644 so that @xmath52 interactions can be used to constrain the @xmath640 s in the lagrangian of eq . [ nci ] . a particular example of a high - energy model that leads to a low - energy lagrangian of this type is provided by r - parity - violating ( @xmath645 ) susy , with the lagrangian : @xmath646 at low values of transferred momenta , this lagrangian can be approximated in terms of local four - fermion interactions : @xmath647 assuming that the squarks of first two generations are degenerate and imposing @xmath642 symmetry constraints gives @xmath648 indeed , the best constraint on this coupling , @xmath649 already comes from the analysis of neutrino - nucleon scattering experiments @xcite . data from @xmath2mcs should complement new constraints on the new physics contributions that will become available from new tevatron , lhc and desy analyses . another opportunity for @xmath2mcs to significantly improve on already existing bounds on new physics from neutrino experiments is provided by the ability to search for the existence of neutral heavy leptons @xcite . in several models @xcite , neutral heavy leptons are considered heavy isosinglets that interact and decay by mixing with their lighter neutrino counterparts . the high intensity neutrino beams created by @xmath2mcs provide an ideal setting to search for neutral heavy leptons with a mass below @xmath536 mev . it is postulated that neutral heavy leptons , @xmath650 , could be produced from muon decays when one of the neutrinos from the decay mixes with its heavy , iso - singlet partner . the expression for the number of neutral heavy leptons produced in a muon beam is : @xmath651 here @xmath652 is the number of neutrinos produced from muon decays , @xmath653 is the branching ratio of muons decaying into neutral heavy leptons versus ordinary muon decays , @xmath76 is the distance from the beam - line to the detector , @xmath654 is the length of the detector , @xmath655 is the combined detector efficiency and geometric efficiency , @xmath143 is the @xmath650 lifetime , and @xmath656 is the branching ratio for the neutral heavy lepton decaying via a detectable channel ( mainly , @xmath657 ) . note here that the muon has two possible ways of producing @xmath650 s : @xmath658 the branching ratio for each of these reactions is given by : @xmath659 here @xmath660 is defined as @xmath661 and @xmath662 is defined as the coupling constant between the specific type of neutrino and the neutral heavy lepton . once produced , a neutral heavy lepton of such low mass will either decay via three neutrinos ( @xmath663 ) , or through the channels : @xmath664 the first decay mode involves only the neutral current , whereas the second contains a mixture of both neutral and charged current interactions . the branching ratios for decay processes ( [ decay1 ] ) and ( [ decay2 ] ) have been calculated@xcite . note that the number of @xmath650 s detectable at the @xmath2mc depends roughly on @xmath665 . using the above model with some additional assumptions , one can estimate the number of neutral heavy leptons produced at the @xmath2mc that then decay to two electrons and a neutrino and are detected in an experiment . the plots in fig . [ fig : nhl ] show limits for the coupling constants at a @xmath2mc as a function of the @xmath650 mass and for a number of different muon energies and detector distances . all the plots assume a pure , unpolarized muon beam containing @xmath666 muons / year with straight sections such that 25 percent of the muons will decay to neutrinos pointing towards the detector . the detector parameters are based on the detector for the nutev @xmath650 searches@xcite : 3 meters in diameter and 30 meters in length and with enough resolution to detect the @xmath133 vertex from the @xmath650 decay . finally , it is assumed that backgrounds are negligible . it is seen from fig . [ fig : nhl ] that one achieves the best limits from using relatively low energy muon beams . this is a significant improvement over previous neutral heavy lepton searches , where limits fail to reach below @xmath667 in the low mass region . the @xmath2mc may prove to be an ideal location to continue the search for neutral heavy leptons . the high intensity neutrino beam allows for a neutral heavy lepton search to be sensitive to coupling constants in the low mass region . in addition , such a neutral heavy lepton program could easily interface with an already existing detector utilizing the neutrino beam . it is also clear that a neutral heavy lepton search would receive the most benefit at lower muon energies , and thus would yield best results at lower energy @xmath2mcs . section [ sec : ew_nue ] showed that neutrino - electron elastic and quasi - elastic scattering can be used to extract the electroweak parameter @xmath126 from purely leptonic interactions . this section instead discusses the neutrino - electron annihilation processes of eq . [ reac : nube - e - anhil ] , @xmath668 electron antineutrino - electron annihilation is expected to show considerable complexity once the center of mass energy exceeds the threshold for annihilation into hadronic final states . the process can be compared to @xmath669hadrons but with the @xmath670 and other low - lying vector resonances replaced by the @xmath671 , and @xmath672 . the axial component of the weak current produces coupling to axial vector resonances and , at low energy , to the pion . these latter couplings are absent in electron - positron annihilation and thus the weak annihilation of the electron offers a novel complement to electron - positron physics . direct measurements of @xmath132 annihilation would complement the detailed studies of hadronic tau decays that currently provide some of the most powerful tests of qcd at low momentum transfer . these measurements bridge two asymptotic limits of qcd : the perturbative regime operative at high @xmath90 and the chiral limit @xmath673 that is approached at low @xmath674 , and would provide some important information about the qcd spectral functions used in reducing the theoretical hadronic uncertainties in studies of @xmath136 -violation in kaon decays , particularly in the interpretation of @xmath675 measurements @xcite . unfortunately , since the @xmath676 threshold occurs at @xmath677 tev , there is a long way to go in neutrino energy in order to completely overlap the @xmath674 region probed by tau decays . table [ thresholds ] summarizes neutrino energy thresholds for @xmath190-channel final states . the thresholds for all channels are quite high . nevertheless , one can see that a @xmath678 gev muon storage ring could provide access to the lowest lying final states and explore the interesting region near @xmath679 to provide a clean test of pcac , and a @xmath680 gev muon beam would extend the reach up to the threshold for kaon production . [ c]\|c\|c\|*final state * & @xmath681 * ( gev ) * + @xmath682 & @xmath683 + @xmath684 & @xmath685 + @xmath686 & @xmath687 + @xmath688 & @xmath689 + @xmath690 & @xmath691 + @xmath692 & @xmath693 + @xmath694 & @xmath695 + just as intense neutrino beams open up the possibility of directly exploring weak annihilation of leptons in analogy to @xmath133 annihilation , so too does one acquire access to the analog of two - photon physics : the scattering of virtual @xmath53 or @xmath54 beams from quasi - real photons in the coulomb field of the nucleus . the general reaction is of the form @xmath697 where @xmath698 are fermions , and @xmath699 or @xmath700 . for @xmath701 both cc and nc diagrams contribute . otherwise @xmath702 proceeds through @xmath703 scattering and @xmath704 through @xmath705 scattering . the creation of lepton pairs in the coulomb field of the nucleus is sensitive to @xmath706 interference and provides a direct test of the standard model . previous experimental observations are ambiguous@xcite . the purely coherent part of the cross section for @xmath707 at asymptotically large energies is @xcite @xmath708 @xmath709 is a reaction and nucleus - dependent form factor with @xmath710 @xmath711 gev set by the nuclear size , and @xmath712,@xmath713 are simple functions of either @xmath714 or final state fermion masses . the factor @xmath715 reflects the coherent electromagnetic nature of the process ; and the scale of the cross section is set by @xmath716 . since the effective center - of - mass energy is @xmath717 and @xmath718 , all possible leptonic final states are accessible ( including those with @xmath719 ) to a neutrino beam derived from a 250 gev muon beam . the factor @xmath720 incorporates effects of neutral currents , including interference in the @xmath721 final states@xmath722 a nice electroweak test is to measure @xmath720 through the ratios @xmath723 where @xmath724 will depend only weakly on nuclear form factor and energy . hadronic resonances are also possible . we expect similar states to those produced in @xmath134 collisions , namely @xmath725 , @xmath726 , @xmath727 , @xmath728with @xmath729 and @xmath730 . some cabibbo - allowed examples include @xmath731 unfortunately , some of these hadronic resonances do not provide a unique experimental signature . single pion production , for example , can occur through the diffractive process @xmath732 mediated by a collision of the pion component of the virtual @xmath53 boson with a pomeron from the nucleus or nucleon . experimental backgrounds that have to be controlled for these studies include coherent meson production mediated by @xmath615-pomeron scattering . especially tricky are the diffractive @xmath733 channels where the @xmath734 undergoes two - body @xmath143 decay,@xmath735 also of concern are backgrounds from inclusive charged current charm production , @xmath736 where the charm quark fragments into a charmed hadron which takes nearly all of the event s hadronic energy and then decays semi - leptonically into a final state where leptons carry nearly all the energy . rare processes that could be studied at a @xmath2mc would probe , both directly and indirectly , an energy range from fractions of a gev to above the tev scale . this would provide important information complementary to the existing results in some areas ( e.g. , in low energy qcd studies , fcnc , contact interactions ) and could substantially improve current bounds on the parameters of some new physics models ( e.g. heavy neutral lepton searches ) . a @xmath2mc will constitute a rather impressive charm factory . figures [ cbprod_nu ] and [ cbprod_nubar ] of sec . [ ch : intro ] show that one can expect between @xmath737 and @xmath738 well - reconstructed charm events in a total event sample of @xmath135 events , depending on the @xmath2mc energy . several species of charmed hadrons should be produced , with measured @xcite relative production fractions for the more common charmed hadrons of : @xmath739 also , the @xmath740 and @xmath741 are expected@xcite to have comparable production cross sections to @xmath742 . the charmed - strange baryons @xmath743 and @xmath744 should be produced at levels down by a factor of a few and @xmath745 should be still less common . the ratios of eq . [ cfracs ] are relatively independent of the neutrino energy for energies above 10 gev and are for production from neutrinos ; the corresponding antiparticles containing anti - charm will be produced from antineutrinos in similar ratios , although with differences in the absolute cross sections and kinematic distributions . the large asymmetry between @xmath746 and @xmath81 production is due to the prevalence of @xmath456 production with its preference for decays into @xmath746 . as well as providing good all - around event reconstruction , @xmath2mcs will have two other distinct and important experimental advantages over all other types of charm facilities . firstly , reconstruction of the charm decay vertex should be superior to that at any collider experiment , particularly for the reconstruction of the challenging 1-prong charm decays , as was illustrated by fig . [ vertexing ] . secondly , a uniquely pure and efficient tag of whether the production flavor is charm or anti - charm is provided by the 100% correlated sign of the primary lepton from the interaction : @xmath747 this section discusses several areas for charm decay physics at @xmath2mcs where these experimental capabilities should be important . the theoretical interest of each measurement will be discussed , and brief summaries of the expected experimental techniques and sensitivities at @xmath2mcs will be included . expected relative strengths and weaknesses of @xmath2mcs compared to other future charm facilities will also be touched on . however , detailed numerical predictions for measurement precisions await more extensive feasibility studies than have been performed for this report . there are also possibilities for @xmath83 decay physics using the neutrinos from multi - tev muon colliders@xcite , where @xmath748 production in neutral current interactions should be at the level of @xmath583 of the total cross section . associated production of @xmath749 and @xmath750 can also be studied , however the relevant production cross - sections are suppressed by approximately two orders of magnitude compared to the @xmath748 production cross section . it is clear that , from the point of view of standard model electroweak physics , charm decays represent a decidedly dull affair . first , the relevant ckm parameters are reasonably well known , for the smallness of @xmath383 and @xmath382 constrains @xmath386 and @xmath385 very tightly through three - family unitarity ( see sec . [ ch : qm ] ) . second , @xmath751 oscillations proceed slowly . third , @xmath136 asymmetries are small due to the fact that both decaying and final state particles contain quarks of only the first two generations . finally , rare charm decay rates are tiny and , again , are dominated by long - distance effects . these apparent vices can , however , be turned into virtues . since the weak dynamics apparently hold no secrets , one can employ charm decays as a laboratory to study qcd in the interface of perturbative and non - perturbative dynamics . also , precisely because the standard model promises us no drama in charm decays , one can conduct searches for @xmath751 oscillations , @xmath136 violation and rare charm decays as probes for new physics with almost no background from the standard model . improved measurements of charm decays are needed for phenomenological and theoretical reasons even in the _ absence _ of new physics , for the following reasons : * to improve the data base needed for analyzing @xmath83 decays one needs more precise measurements of the _ absolute _ branching ratios of charm hadrons ; * measurements of the leptonic decay rates @xmath752 are required for determining the meson decay constants ; these decay constants give us quantitative insight into the dynamics of heavy - light bound state systems and can be used for tuning the lattice qcd methods and a more reliable evaluation of @xmath753 oscillations ; * more precise studies of _ inclusive _ semileptonic @xmath318 , @xmath734 , @xmath754 , etc . decays would provide us with valuable novel insights into the inner workings of qcd and at the same time sharpen our tools for a quantitative treatment of @xmath83 decays . as the discussion about the charm content in the final state of @xmath83 decays illustrates , a significant bottleneck in the detailed analysis of beauty decays of the @xmath755 type is currently caused by the uncertainties in the absolute branching ratios of charm hadron decays to specified final states , in particular of @xmath734 , @xmath754 and @xmath756 . a @xmath2mc should be well suited to obtaining these branching ratios , as we now discuss . this information will be useful even if it is obtained only after the next generation of @xmath83 experiments have accumulated their samples . as what is typically the less difficult part of the measurements , the expected excellent particle identification and event reconstruction at @xmath2mcs should give good capabilities for determining relative branching ratios for each hadron . the more difficult task of obtaining the production normalization factors to convert these to absolute branching ratios should then be achieved by fitting the experimental decay length distributions in a procedure that was studied for the cosmos ( e803 ) neutrino experiment at fermilab . the cosmos technique@xcite envisions fitting normalization factors to the several known decay exponentials one for each charmed hadron species in the observed neutral and charged distributions for the variable @xmath757 , with @xmath60 the charmed hadron distance to the decay vertex and @xmath191 its reconstructed momentum . it is helpful that the exponential decay constants in this variable are well separated for both the charged and neutral hadron distributions : @xmath758 and @xmath759 auxiliary information for the fit will be available from particle identification in the detector . in particular , the presence of a proton in the final state will reliably indicate the decay of a baryon rather than a meson . to test the method for the cosmos environment , exponential fits were performed @xcite for simulated decay length distributions from approximately 14 000 reconstructed @xmath81 , @xmath460 and @xmath742 charm decays . the fitted statistical uncertainties for the three species were 3.4% , 12% and 5.4% , respectively . these simulations show that statistical uncertainties would be negligible for such a fit at a @xmath2mc , which would have several orders of magnitude more events . the uncertainties in the charm production rates would instead be dominated by uncertainties in modeling the level of vertexing inefficiencies . hopefully , these uncertainties could also be made small due to the favorable vertexing geometry shown in fig . [ vertexing ] and to the considerable potential for using the data itself to estimate the inefficiencies . another area where @xmath2mcs can be expected to make significant or even unique contributions is in the analysis of final states that contain _ more than one _ neutral hadron , e.g. , @xmath760 even all neutral final states like @xmath761 might become observable . such neutral - rich channels are rather elusive for the usual @xmath133 annihilation and photoproduction experiments . a @xmath2mc could access these modes through the expected sample of @xmath87 nc - produced @xmath762 events . vertex tagging one of the charmed hadrons would allow a search for such decay modes in the other . filling in these ` white spots ' in the map of charm decays would close or at least narrow the gap between exclusive and inclusive decays and thus can provide us with important lessons on how quark - hadron duality is realized in subclasses of total decays . for example : a quark based description leads to the prediction that the ( cabibbo suppressed ) inclusive rates driven by @xmath763 and @xmath764 should practically coincide since @xmath765 , @xmath592 @xmath766 @xmath254 . yet exclusive channels like @xmath767 and @xmath768 do not at all follow this expectation ! duality suggests that a ( near ) equality will emerge for @xmath769 @xmath770 @xmath771 . testing this expectation requires the measurement of final states with neutrals . experimental studies of multi - body decays with more than one neutral meson in the final state ( in particular , dalitz plot analyses ) also allow us to have different handles on the studies of direct @xmath136-violation in d - decays in and beyond the standard model @xcite as well as on the dynamics of hadronic resonances governing these transitions ( see , e.g. the e791 analysis @xcite ) . the primary goal behind measuring leptonic decays , @xmath773 , or the cabibbo suppressed versions , @xmath774 , with @xmath775 , is the desire to extract the decay constants @xmath776 and @xmath777 . these quantities are important probes of heavy meson wave functions . in addition , these decay constants have been extracted from monte carlo simulations of qcd on the lattice with estimated uncertainties of about 20 percent on their absolute values and about 10 percent on their ratio . improvements are expected for future lattice calculations . for the proper evaluation of these calculations , one wants to calibrate them against experimental results of similar accuracy . currently , the branching ratios for @xmath778 transitions have been measured by the cleo collaboration with large uncertainties , @xmath779 and @xmath780 . no measurement is currently available for other @xmath318 mesons although there is an upper bound : @xmath781 . this can be explained by the @xmath782 ckm suppression factor for the @xmath81 leptonic decays relative to those of the @xmath734 . once the absolute values of @xmath777 or @xmath776 are known experimentally with about @xmath783 accuracy or better then one will be able to feel more confident about extrapolating to the decay constants in the @xmath83 system , @xmath784 and @xmath785 , which are crucial quantities for a quantitative understanding of @xmath753 oscillations and the extraction of @xmath396 from them . observing and measuring these transitions has always represented a highly nontrivial experimental challenge ( and much more so for @xmath786 ) , so the potentially exceptional performance for observing 1-prong d decays at @xmath2mc s could allow them to make a significant contribution here even down the line . as a secondary goal one might even perform a detailed comparison of the rates for @xmath787 and @xmath788 as a probe for new physics in the form of a non - minimal higgs sector , for charged higgs exchanges would affect the latter much more than the former . heavy quark expansions ( hqe ) allow the treatment of inclusive heavy flavor decays , including their non - perturbative aspects @xcite . in addition to total decay widths , other central quantities are inclusive semileptonic branching ratios and decay spectra for the different meson and baryon species . these techniques provide the basis for some of the most reliable methods for extracting @xmath383 and @xmath382 in @xmath83 decays . obviously one wants to cross check these methods in a system where the ckm parameters are known , namely the charm system , by testing how precisely @xmath381 and @xmath380 can be extracted from semileptonic charm decays . in addition one can extract the size of the matrix elements of four - fermion operators that are of direct relevance in _ beauty _ decays and at the same time provide important calibration points for lattice simulations of qcd . no data of sufficient detail are available . the @xmath83 factories ( cleo , babar and belle ) will significantly improve the situation , but might not achieve the desired experimental accuracy . furthermore , it turns out that comparing neutrino with charged lepton spectra in semileptonic decays provides us with particularly probing insights . one has to keep the following in mind . since the expansion parameter is @xmath789 with @xmath790 gev , one has to allow for uncalculated higher order contributions to modify the results significantly in charm decays . to have a handle on this complication , one needs to be able to perform detailed comparisons of the lepton spectra separately in @xmath746 , @xmath81 , @xmath734 and @xmath754 decays , which should be possible at @xmath2mc s . focus and selex data will presumably yield precise lifetimes for @xmath791 baryons , but quite possibly not for the @xmath745 . the latter is presumably the shortest lived hadron in the single charm sector , with @xmath792 sec ; due to its different spin structure its lifetime is affected by different matrix elements than for the other baryons . it is also quite unclear whether next generation experiments like lhc - b and btev can measure such a short lifetime with good accuracy . a @xmath2mc thus could make a relevant measurement that would serve as an a posteriori calibration of some theoretical tools . furthermore , a whole new spectroscopy could be entered into , namely that of baryons carrying two units of charm : @xmath793 $ ] . in principle , radiative inclusive ( and exclusive ) decays can also be studied . the predicted branching ratio for the short - distance contribution is tiny , @xmath794 @xcite , although two - loop qcd corrections could bring it up to @xmath795 @xcite . this could have made it a sensitive probe of new physics as these processes occur in the standard model only at one loop . unfortunately , the problem is that the long - distance effects can actually completely dominate this decay , enhancing it up to @xmath796 , and these enhancements can not be estimated model - independently . the phenomenon of meson - antimeson mixing has been studied both experimentally and theoretically for a long time as it provides an extremely sensitive test of the standard model as well as its various possible extensions . this is especially true for @xmath798 mixing , as was already indicated . to study such oscillations one must tag separately the flavor of the produced meson and of the decaying meson . the charge of the primary lepton from a cc interaction uniquely tags the production sign of the charm quark method at an @xmath2mc . this should easily be the cleanest and most efficient tag . it can be checked by a more conventional alternative method involving production of the charged @xmath456 mesons and studies of the decay chain @xmath799 @xcite , where anti - correlation studies of the charge of @xmath800 and decay products of @xmath318 would reveal whether mixing took place . for charmed mesons with tagged production flavor , @xmath751 oscillations are most cleanly probed through ` wrong - sign ' semileptonic decays with the branching ratios : @xmath801 for @xmath802 , @xmath803 the @xmath318 width and @xmath804 the difference in the @xmath746 and @xmath805 mass widths . in principle , one can determine the flavor of the final state through charged kaons ; mis - tags that happen due to doubly cabibbo suppressed decays can be eliminated using a time - dependent analysis , as discussed below . the most recent experimental limits , which are from fixed target experiments at fnal and from cleo at cesr and combine tagging through ` wrong ' sign leptons and kaons , read : @xmath806 where @xmath807 with @xmath808 denoting the strong phase shift between @xmath809 and @xmath810 ( see @xcite for the recent analysis ) , and @xmath811 where @xmath812 in the standard model . since possible new physics effects or hadronic uncertainties will affect these experiments differently , a careful analysis to extract the true values of @xmath813 and @xmath804 from the data should be performed @xcite . the @xmath83 factories at cornell , slac and kek will refine the search for @xmath751 oscillations to an expected sensitivity of @xmath814 @xmath558 few@xmath815 @xcite . while the standard model undoubtedly predicts slow @xmath751 oscillations @xmath816 , @xmath817 @xmath818 there is considerable uncertainty in the numerical predictions . a _ conservative _ standard model bound is given by @xcite @xmath819 bolder predictions have been made that @xmath816 and @xmath817 can not exceed @xmath583 @xcite and therefore @xmath820 within the standard model . on the other hand , new physics could enhance @xmath816 up to , and actually even above , the present bound , @xmath821 without violating any other limit and while leaving @xmath817 unaffected . examples of such new physics processes include various supersymmetric models @xcite ( including susy models with quark - squark alignment that actually require @xmath813 close to the current experimental bound ) @xcite , models with singlet up quarks @xcite , various leptoquark models @xcite , and multiscalar models with @xcite , and without @xcite , natural flavor conservation . any experimental effort to lower the current limit on @xmath813 is essential in determining the available parameter space for many possible extensions of the standard model ! the cleanest way to probe for @xmath751 oscillations is to analyze the time evolution of transitions into ` wrong - sign ' leptons : @xmath822 here we have invoked the @xmath823 rule of the standard model which makes oscillations the only source for wrong - sign leptons . since one is embarking on a search for new physics , one should generalize equation [ doscsm ] to allow for a violation of the @xmath823 rule , giving : @xmath824@xmath825 , \label{d0loscnonsm}\ ] ] where @xmath826 denotes the ratio of @xmath827 to @xmath828 amplitudes , @xmath829 relates mass and flavor eigenstates , and the oscillating functions multiplying the usual @xmath830 term have been expanded in powers of the proper time @xmath482 since @xmath816 , @xmath817 @xmath818 . the @xmath827 term has no @xmath482 dependence beyond that of @xmath830 , the pure oscillation term has a @xmath831 dependence , while the interference between the two generates a term linear in @xmath482 . the violation of the @xmath823 rule arises even within the standard model for the decays @xmath832 due to doubly cabibbo suppressed transitions ( dcst ) producing the direct decay @xmath832 , with a branching ratio @xmath833 @xcite , and thus mimicking the signal for @xmath834 mixing . the equation corresponding to eq . [ d0loscnonsm ] is : @xmath835@xmath836@xmath837 , \label{dkpi}%\ ] ] where @xmath838 is the fraction of wrong - sign decays . one can also search for lifetime differences in certain well - chosen @xmath839 decay channels in order to probe the contributions to oscillations from the @xmath840 term of eq . [ xdyd ] . with @xmath136 invariance holding ( at least ) to good approximation , @xmath136 eigenstates can be treated as mass eigenstates . while @xmath841 , @xmath842 will then exhibit @xmath843 , @xmath844 , @xmath845 , @xmath846 , @xmath847 etc . will be controlled by @xmath848 , where @xmath843 [ @xmath848 ] denotes the width for the @xmath136 even [ odd ] state and @xmath850 furthermore the width for @xmath851 is approximately given by @xmath852 @xcite . there is a wide field of potential @xmath136 violation in @xmath318 decays that can be discussed in close qualitative analogy to @xmath83 decays . @xmath136 asymmetries that necessarily involve @xmath751 oscillations can arise in final states that are @xmath136 eigenstates , like @xmath853 or @xmath854 : @xmath855@xmath856 with @xmath857 and @xmath858 , one arrives at an asymmetry of only around @xmath859 , which would likely be too small to measure even at a @xmath2mc . yet with new physics one conceivably has @xmath860 and @xmath861 , leading to an asymmetry that could be as large as of order 1% . likewise , one can search for cp violation by comparing the proper time distribution of eq . [ dkpi ] for the doubly cabibbo suppressed transitions @xmath832 with that for @xmath805 decays : @xmath862@xmath863@xmath864 , \ ] ] where @xmath865 in such new physics scenarios one would expect a considerably enhanced asymmetry perhaps as large as @xmath866 but at the cost of smaller statistics . hoping for an asymmetry of several percent is more realistic , though . effects of that size would unequivocally signal the intervention of new physics ! one should note that these rough estimates are based on @xmath867 which would correspond to @xmath868 . this implies that , even if oscillations have not been found on the @xmath869 level in semileptonic @xmath746 decays , a @xmath136 asymmetry of several percent ( or conceivably ten percent ) could still be encountered in @xmath832 ! _ direct _ cp violation can occur as well . there are actually two types of effects : differences between partial rates for @xmath136 conjugate transitions @xmath870 and asymmetries in final state distributions such as , e.g. , dalitz plot populations . strong final state interactions play an important part in both cases : in the former they must induce the phase shifts that are essential to make a difference observable ; in the latter they can very significantly affect the observable asymmetry . the existence of resonances in the neighborhood of the charmed meson mass is proof that hadron dynamics is active in this energy region and will affect the weak decays of charmed particles . the good news is that whenever there are @xmath136 violating weak phases one can count on final state interactions to make them observable . the bad news is that interpreting a signal as evidence for new physics will pose a highly nontrivial theoretical challenge . the standard model with the ckm @xmath871 can induce direct @xmath136 asymmetries only in cabibbo suppressed channels . model - dependent estimates usually predict direct @xmath136 asymmetries to be of the order of @xmath583 but , exceptionally , they could reach the @xmath872 level @xcite . a measurement of the branching ratios for all related channels in particular also those with neutral hadrons in the final state , as sketched above would enable us to constrain the strong phase shifts quite significantly . a @xmath2mc will have a significant advantage in this respect ! one special feature of @xmath2mcs is represented by the production of @xmath754 in cc and nc reactions : @xmath874 this allows novel studies of various @xmath754 decay form factors with a @xmath90 range extending to well above @xmath875 . yet even more intriguing and promising would be a detailed analysis of the final state in its semileptonic decays : @xmath876 with the parent @xmath86 quark being left - handed one expects the @xmath754 to emerge in a highly polarized state . the usual valence quark description actually suggests that the @xmath754 polarization is completely carried by its @xmath86 quark ; i.e. , a left - handed @xmath86 quark fragments into a left - handed @xmath754 . yet even with unpolarized @xmath754 one can form an experimentally observable @xmath873 odd correlation @xmath877 connecting spin and momentum of the daughter hyperon with the lepton momentum . in a general experimental process , observing a non - vanishing value for a @xmath873 odd correlation does not automatically establish that @xmath873 ( and @xmath136 ) invariance is violated since in general final state interactions could fake such an effect . however , this problem does not occur for eq . [ lcsldec ] since it can not be affected by either strong or electromagnetic final state interactions ! this is analogous to the well known situation in @xmath878 ( @xmath770 @xmath879 ) . like there , the ckm @xmath871 can not generate an observable effect here , yet certain new physics scenarios can . an effect of order @xmath872 is not inconceivable , particularly if the channel @xmath880 could be studied @xcite . the research program at a @xmath2mc is likely to improve our knowledge and understanding of charm decays quite significantly even ten years from now : * it would fill out many white spots on our map of @xmath318 , @xmath754 , @xmath756 and @xmath745 decays by measuring many new relative and absolute branching ratios , including for final states with more than one neutral particle . * it would allow the measurement of @xmath881 , @xmath882 in a very clean environment . * it could probe for @xmath751 oscillations and for @xmath136 asymmetries involving them with superbly clean systematics . it would significantly improve on the sensitivity that can be obtained at @xmath83 factories for such phenomena . * it would enable us to search for direct @xmath136 asymmetries in many different channels and at the same time provide us with information that could help us in properly interpreting a signal . in a modern theoretical context , one generally expects nonzero neutrino masses and associated lepton mixing @xcite . experimentally , there has been accumulating evidence for such masses and mixing . all solar neutrino experiments ( homestake , kamiokande , superkamiokande , sage , and gallex ) show a significant deficit in the neutrino fluxes coming from the sun @xcite . this deficit can be explained by oscillations of the @xmath531 s into one or more other weak eigenstates , with @xmath883 of the order @xmath859 ev@xmath884 for solutions involving the mikheev - smirnov - wolfenstein ( msw ) resonant matter oscillations @xcite or of the order of @xmath885 ev@xmath884 for vacuum oscillations . accounting for the data with vacuum oscillations ( vo ) requires almost maximal mixing . the msw solutions include one for small mixing angles ( sma ) and one with essentially maximal mixing ( lma ) . another piece of evidence for neutrino oscillations is the atmospheric neutrino anomaly , observed by kamiokande @xcite , imb @xcite , superkamiokande @xcite with the highest statistics , and by soudan @xcite and macro @xcite . this data can be fit by the inference of @xmath886 oscillations with @xmath887 ev@xmath884 @xcite and maximal mixing @xmath888 . the identification @xmath889 is preferred over @xmath890 at about the @xmath891 level @xcite , and the identification @xmath892 is excluded by both the superkamiokande data and the chooz experiment @xcite . in addition , the lsnd experiment @xcite has reported observing @xmath893 and @xmath894 oscillations with @xmath895 ev@xmath884 and a range of possible mixing angles , depending on @xmath896 . this result is not confirmed , but also not completely ruled out , by a similar experiment , karmen @xcite . inclusion of the signal reported by lsnd with the other two pieces of evidence would imply three distinct mass differences and hence four neutrinos . some proposals for the form of the mixing matrix invoke only 3 generations of neutrinos to account for all signatures , while others invoke a fourth sterile neutrino . a number of fits have been made to the existing neutrino data . the fit by the superkamiokande collaboration to its data yields a minimum in the @xmath897 at @xmath898 , with an allowed region of @xmath899 . in terms of the basic angles in the lepton mixing matrix , this implies that @xmath900 is close to @xmath901 and allows a small , nonzero @xmath902 , consistent with the bound from chooz . as will be discussed below , a major physics capability of the muon storage ring / neutrino factory is the ability to measure @xmath902 . there are currently intense efforts to confirm and extend the evidence for neutrino oscillations in all of the various sectors solar , atmospheric and accelerator . some of these experiments are running ; in addition to superkamiokande and soudan-2 , these include the sudbury neutrino observatory , sno , and the k2k long baseline experiment between kek and kamioka . others are in development and testing phases , such as boone , minos , the cern - gran sasso program , kamland , and borexino @xcite . among the long baseline neutrino oscillation experiments , the approximate distances are @xmath903 km for k2k , @xmath904 km for both minos , from fermilab to soudan and the proposed cern - gran sasso experiments . the sensitivity of these experiments is projected to reach down roughly to the level @xmath905ev@xmath884 . experiments that are planned as part of this program include @xcite icanoe and opera @xcite . although they are expected to begin operation after minos , they will involve somewhat different detector designs and plan to focus on establishing @xmath143 appearance . this , then , is the program of research for the next several years . although a neutrino factory based on a muon storage ring will turn on several years after this near - term period in which k2k , minos , and the cern - gran sasso experiments will run , we believe that it has a valuable role to play , given the very high - intensity neutrino beams of fixed flavor - pure content , including , in particular , @xmath531 and @xmath239 beams as well as the conventional @xmath240 and @xmath508 beams . the potential of the neutrino beams from a muon storage ring is that , in contrast to a conventional neutrino beam @xmath906 decay , is primarily @xmath240 with some admixture of @xmath531 s and other flavors from @xmath84 decays , the neutrino beams from the muon storage ring would high extremely high purity : @xmath521 beams would yield 50 % @xmath240 and 50 % @xmath239 , and so forth for the charge conjugate case of @xmath540 beams . furthermore , these could be produced with extremely high intensities , of order @xmath666 to @xmath38 neutrinos per year . given the form of the oscillation probabilities , a neutrino beam for an oscillation experiment would optimally be made from the lower end of the spectrum of energies considered in this report . because of the lower requirements on beam focusing and acceleration , making a muon storage ring for the purposes of an oscillation experiment has been proposed as a first step towards developing a muon collider . energies being considered range from 20 gev to 50 gev , and the geometry of the final muon ring is very different from a traditional collider ring , in that the straight section that points to a neutrino experiment comprises between 25 and 40% of the `` circumference '' of the ring . the types of neutrino oscillations that can be searched for with the neutrino factory based on a muon storage ring , along with the final state charged lepton species for neutrino - nucleon dis that tags the interacting neutrino flavor , are listed below for the case of the @xmath42 beam from @xmath521 , decaying as @xmath907 : 1 . @xmath908 , @xmath909 ( survival ) ; 2 . @xmath894 , @xmath910 ( appearance ) ; 3 . @xmath911 , @xmath912 , @xmath913 ( appearance@xmath914 ) ; 4 . @xmath915 , @xmath916 ( survival ) ; 5 . @xmath917 , @xmath918 ( appearance ) ; 6 . @xmath919 , @xmath920 ; @xmath921 ( appearance@xmath914 ) ; where the asterisks denote that the tau appearance signatures may be somewhat indirect in involving detection of the leptonic daughters of tau decays rather than the decay vertices of the taus themselves . it is clear from the list of processes above that , since the beam contains both neutrinos and antineutrinos , the only way to determine what the parent neutrino was is to measure the charge of the final state lepton . the @xmath922 oscillation will produce a wrong - sign @xmath923 as will the @xmath911 oscillation followed by @xmath143 decay to @xmath923 . the easiest wrong - sign lepton signatures to detect arise from the oscillations @xmath917 , giving a @xmath540 , and from @xmath919 , giving a @xmath924 which will decay part of the time to @xmath540 . if one is searching for @xmath143 final states , muon storage ring energies above @xmath589 gev should be used to minimize the threshold kinematic suppression . to get a rough idea of how the sensitivity of an oscillation experiment would scale with energy and baseline length , recall that the event rate in the absence of oscillations is simply the neutrino flux times the cross section . first of all , neutrino cross sections in the region above about 10 gev ( and slightly higher for @xmath143 production ) grow linearly with the neutrino energy . secondly , the beam divergence is a function of the initial muon storage ring energy ; this divergence yields a flux , as a function of @xmath925 , the angle of deviation from the forward direction , that goes like @xmath926 . combining this with the linear @xmath927 dependence of the neutrino cross section and the overall @xmath928 dependence of the flux far from the production region , one finds that the event rate goes like @xmath929 to set the scale , consider an experiment that sees @xmath930 30 gev muon decays in a straight section pointed at a detector 2800 km away . in the absence of oscillations , the @xmath931 charged current rate would be 52,500 ( 22,600 ) events per 10 kton @xcite . figure [ fig : manyflux ] shows the relative @xmath531 and @xmath932 statistics for two configurations : one is a 20 gev @xmath540 ring with a detector at 2800 km , the other is a 50 gev @xmath540 ring with a detector at 9100 km , which is close to the distance from either fermilab or cern to kamiokande . = now recall the general formula for the probability of a two - species neutrino oscillation in vacuum , say of @xmath933 : @xmath934 where the @xmath935 term is small , it can be expanded to give a factor of @xmath936 that cancels the @xmath76 dependence in equation [ eventrate ] and reduces the @xmath927 dependence to a linear factor . underlying these considerations of optimal energy is the fact that , even if one designs an experiment for @xmath143 appearance , the overall event rate of detected @xmath143 s may be rather small . it is quite likely that , by the time a neutrino factory turns on , @xmath937 and @xmath938 will be known at the 10 - 30% level . although a neutrino factory could undoubtedly improve the precision on those two parameters , the novel physics that can be addressed is a determination of @xmath902 , and the sign of @xmath939 . by using matter effects , and a comparison of @xmath940 versus @xmath941 oscillations ( by switching the muon storage ring from @xmath521 to @xmath540 ) , the sign of @xmath937 can be determined . with the advent of the muon storage ring , the distances one can place detectors will become large enough that , for the first time , matter effects can be exploited in accelerator - based oscillation experiments . simply put , matter effects are the matter - induced oscillations which neutrinos undergo along their flight path through the earth from the source to the detector . given the typical density of the earth , matter effects are important for the neutrino energy range @xmath942 gev and @xmath943 ev@xmath884 values that are relevant for @xmath2mc long baseline experiments . follow - up studies to initial discussion @xcite of matter - induced resonant neutrino oscillations include an early study of these effects that assumed three neutrino generations @xcite and a discussion of the sensitivity of an atmospheric neutrino experiment to small @xmath944 due to the long baselines and the necessity of taking into account matter effects was discussed@xcite . many analyses were performed in the 1980 s of the effects of resonant neutrino oscillations@xcite on the solar neutrino flux , and matter effects in the earth @xcite , and of matter effect on atmospheric neutrinos@xcite . work continues@xcite on matter effects relevant to atmospheric neutrinos . early studies of matter effects on long baseline neutrino oscillation experiments @xcite have been extended to cover neutrino factories@xcite , @xcite in recent papers by one of the authors ( rs ) and i. mocioiu , calculations were presented of the matter effect for parameters relevant to the possible long baseline neutrino experiments envisioned for the muon storage ring / neutrino factory @xcite . in particular , these authors compared the results obtained with constant matter density along the neutrino path versus results obtained by incorporating the actual density profiles . they studied the dependence of the oscillation signal on both @xmath945 and on the angles in the leptonic mixing matrix , and commented on the influence of @xmath946 and cp violation on the oscillations . additional recent studies are listed in the references @xcite . in order to measure oscillation parameters that describe the transitions above , one would ideally want a detector that could identify the existence and flavor of any outgoing lepton from the neutrino interaction , as well as the hadronic and leptonic energy in the event . when measuring very small oscillation probabilities , however , backgrounds must be taken into account . naively one would think that simply detecting a muon of opposite charge to that in the storage ring is a signal for the electron neutrino oscillating . however , pions and kaons are produced copiously in neutral and charged current neutrino interactions , and if one decays to a muon before it interacts in the detector this can constitute a significant background . at higher energies charmed mesons are also produced which decay immediately to muons 10% of the time . ultimately , detectors will need to be designed that have sufficient resolution on both the energies and the angles of the final state lepton and hadronic shower to be able to remove these backgrounds . although detectors exist that could identify all of the final state leptons and their charges , the challenge is to make them on the several kiloton scale . if the largest @xmath944 is in the lsnd region , there will undoubtedly be more work done to optimize relatively low mass detectors that emphasize tau appearance . however , the only detectors which have thus far been proposed on the 10 - 40 kiloton scale are for detecting wrong sign muon events . the two detector technologies that have been considered in detail for oscillation experiments for a muon storage ring will now be discussed in turn : one is a magnetized sampling calorimeter such as the one used by minos @xcite and the other is a liquid argon time projection chamber ( tpc ) combined with a muon spectrometer such as the one proposed by the icanoe @xcite collaboration for the cern to gran sasso neutrino beam . magnetized sampling calorimeters consist simply of alternating layers of magnetized steel and readout , where the readout traditionally consists of scintillator and/or drift chambers . the charged particle efficiency of the readout planes can be close to 100% so the performance of the sampling calorimeter depends primarily on the sampling frequency of the detector . a steel / scintillator sandwich with sampling every 5 cm of steel would have a fractional hadron energy @xmath947 resolution of approximately @xcite @xmath948}}.\ ] ] with fine enough transverse segmentation , the hadron angular resolution is dominated by the hadron energy resolution . the muon energy and angular resolution are expected to be much better than for the hadronic shower . although separating @xmath531 charged current events from neutral current events is difficult and determining the charge of the outgoing electron impossible in this detector , a muon in the final state can be easily and efficiently detected , and its charge , momentum , and initial outgoing angle can be determined once the muon traverses enough steel to be spatially separated from the hadronic shower . kinematic cuts can be made on the muon momentum and its component transverse to the hadronic shower to reduce the background from charm production . with signal efficiencies from 25 to 30% , the backgrounds can be reduced to a level of @xmath859 to @xmath949 , depending on the neutrino energy . at higher energies the backgrounds are larger but the faster improvement in the background rejection actually causes a reduction in the background contributions to the analysis @xcite . the icanoe - type detector would consist of a large volume of liquid argon instrumented with time projection chambers ( tpc s ) , followed by a much thinner volume of magnetized steel where a muon s charge and momentum can be determined . the tpc would have very small wire spacing ( 3 mm ) and would act much like an electronic bubble chamber . electron neutrino charged current interactions could be distinguished from neutral current interactions , although the electron charge could not be measured . by breaking up the event samples into four distinct classes right sign muons , wrong sign muons , electron - like events and neutral current events one could fit all four distributions simultaneously to determine oscillation parameters . @xmath950 s might also be identified on a statistical basis by looking at the acoplanarity distribution in the event sample . the energy and angular resolutions of all the final state particles would be extremely good , e.g. , @xmath951}}%\ ] ] for the hadronic energy and 150 mrad for hadron shower angles . however , the ability of this detector to see wrong - sign muons would depend primarily on the segmentation between the liquid argon and the magnetized spectrometer and on the thickness of the spectrometer itself . the thinner one makes the spectrometer , the more likely one is to have backgrounds from charge misidentification . the thicker the spectrometer , the less room there is for the liquid argon in a given volume . the thinner the liquid argon , the higher the acceptance for low energy muons ( since muons lose approximately 210 mev / m in liquid argon @xcite ) , but the less target volume one has overall . clearly optimization of this geometry is needed , and will depend somewhat on the energy of the muon storage ring . although the two types of detectors have different strengths , detailed studies @xcite have shown that both either would be adequate to make precise measurements of @xmath952 , @xmath953 , and @xmath954 ; and to extend the sensitivity of @xmath955 by 1 - 2 orders of magnitude in the scenario where the largest @xmath944 is described by the atmospheric neutrino anomaly . alternate technologies must be employed to achieve electron or tau identification event - by - event , or electron or tau charge measurements . if lsnd is confirmed and the largest @xmath944 would suggest baselines on the order of tens of km , then a much higher premium will be placed on designing detectors that can do tau and electron charge determination , and they will not have to be as massive . at these short baselines , detectors on the 1 kiloton scale could be quite adequate to make precision measurements on @xmath956 and @xmath911 . even if lsnd is not confirmed , efforts to make massive tau and electron charge identification detectors should not be abandoned since these two channels still comprise a large part of the mixing matrix and should be researched to confirm our understanding of neutrino mixing . one category of new detectors uses thin ( @xmath957 @xmath48 m ) sheets of emulsion combined with low - density ( @xmath958 @xmath48 m ) spacers , and thin sheets of metal to give the detector mass . with emulsion one can measure the kink that occurs when a tau decays by comparing the slope of a track before and after the spacer . such a geometry , with lead as the mass , is described in reference @xcite . this would be very useful for identifying taus and electrons . however , for charge identification one needs to introduce a magnetic field . this could be done using an extremely large external magnet , such as the one used in atlas , and thin steel plates , or by using a coil and magnetized steel to make the mass @xcite . since the overhead for analysis of each event is high in this sort of detector , one would place it in a region where the tau appearance probability is maximized . in conclusion , neutrino masses and mixing are generic theoretical expectations . the seesaw mechanism naturally yields light neutrinos , although its detailed predictions are model - dependent and may require a lower mass scale than the gut mass scale . current atmospheric neutrino data is consistent with maximal mixing in the relevant channel , which at present is favored to be @xmath911 . even after the near - term program of experiments by k2k , minos , the cern - gran sasso experiments , and mini - boone , a high - intensity neutrino factory generating @xmath959 neutrinos per year will add greatly to our knowledge of the neutrino masses and mixing matrix . ideally , the muon storage ring should be coupled with two long - baseline neutrino oscillation experiments , located at different baselines , that can take advantage of matter effects to amplify certain transitions and with a massive detector that will identify @xmath48 s and @xmath960 s with charge discrimination . in particular , it should be able to measure @xmath961 and @xmath938 to the level of several per cent and also give important information about the sign of @xmath962 and about @xmath963 . beams from @xmath2mcs have the potential to provide vast improvements over today s conventional neutrino beams from @xmath39 decays . they are much more intense , have a much smaller transverse extent and produce precisely predictable beam spectra . oscillation experiments may extend even to intercontinental baselines , while unprecedented event statistics approaching of order @xmath580 to @xmath135 precisely constructed dis events will open new regimes of neutrino interaction physics . the extraordinary rates will enable use of active vertexing targets surrounding by a high resolution spectrometer and calorimeter . the first polarized targets for neutrino scattering can be substituted for special studies . highlights of @xmath2mc physics program include : * substantially extending the reach of accelerator - based experiments to study neutrino oscillations ; * measurements of the ckm quark mixing matrix elements @xmath964 , @xmath104 , @xmath965 , and @xmath107 in inclusive high @xmath90 scattering , with few percent accuracies achievable for the first two ; * a realistic opportunity to determine the detailed quark - by - quark structure of the nucleon ; * mapping out the quark - by - quark spin structure with polarized targets , and , perhaps , determining the gluon contribution to the nucleon s spin ; * some of the most precise measurements and tests of perturbative qcd ; * tests of the electroweak theory through measurements of @xmath576 with fractional uncertainties approaching @xmath128 ; * a new realm to search for exotic physics processes ; * a charm factory with unique and novel capabilities ; * a new laboratory for the study of nuclear physics with neutrino beams . the potential experimental capabilities of @xmath2mcs reach so far beyond present neutrino programs that physics surprises not touched upon in this report can be expected . to see , we must build the machines . we thank janet conrad , keith ellis , michelangelo mangano , michael shaevitz and don summers for helpful information and comments . this work was performed under us department of energy contract numbers de - ac02 - 98ch10886 , de - ac02 - 76ch03000 , de - fg02 - 91er40684 , de - fg02 - 91er40685 and de - fg03 - 99er41093 , under us national science foundation contract numbers phy 96 - 05080 , nsf 97 - 22101 , phy 98 - 13383 and phy 00 - 87419 and under the auspices of the illinois board of higher education . b.j . king , _ assessment of the prospects for muon colliders _ , paper submitted in partial fulfillment of requirements for ph.d . , columbia university , new york ( 1994 ) , available from http://xxx.lanl.gov/ as physics/9907027 . _ a feasibility study of a neutrino source based on a muon storage ring _ norbert holtkamp and david finley , march , 2000 . available at`http://www.fnal.gov / projects / muon_collider / nu / study / report / machine_report/ ` . king , _ mighty murines : neutrino physics at very high energy muon colliders _ , proc . hemc99 workshop studies on colliders and collider physics at the highest energies : muon colliders at 10 tev to 100 tev ; montauk , ny , september 27-october 1 , 1999 , hep - ex/0005007 . m. goncharov , _ et al . _ ( nutev collaboration ) , `` precise measurement of dimuon production cross - sections in muon neutrino fe and muon antineutrino fe deep inelastic scattering at the tevatron '' , e - print number hep - ex/0102049 , _ submitted to phys d. , _ february , 2001 . s. a. larin and j. a. vermaseren , phys . * b259 * , 345 ( 1991 ) . v. m. braun and a. v. kolesnichenko , nucl . * b283 * , 723 ( 1987 ) . g. g. ross and r. g. roberts , phys . * b322 * , 425 ( 1994 ) [ hep - ph/9312237 ] . s. l. adler , phys . rev . 143 , 1144 ( 1966 ) . h. georgi and h. d. politzer , phys . rev . d * 14 * , 1829 ( 1976 ) . m. a. aivazis , j. c. collins , f. i. olness and w. tung , energies , phys . rev . d * 50 * , 3102 ( 1994 ) [ hep - ph/9312319 ] . u. k. yang and a. bodek , phys . lett . * 82 * , 2467 ( 1999 ) [ hep - ph/9809480 ] . h. l. lai _ et al . _ [ cteq collaboration ] , eur . . j. * c12 * , 375 ( 2000 ) [ hep - ph/9903282 ] . m. virchaux , and a. milsztajn , phys . lett . * b 274 * , 221 ( 1992 ) .. h. deden and et al . [ gargamelle neutrino collaboration ] , nucl . * b85 * , 269 ( 1975 ) . x. guo and j. qiu , hep - ph/9810548 . an ice target will be used in the laser electron gamma - ray source ( legs ) experiment at brookhaven national laboratory , ` http://www.legs.bnl.gov ` . deborah a. harris and kevin s. mcfarland , _ a small target neutrino deep - inelastic scattering experiment at the first muon collider _ , proc . workshop on physics at the first muon collider and front end of a muon collider , fermilab , november 6 - 9 , 1997 , hep - ex/9804010 . c. boros , j. t. londergan and a. w. thomas , phys . d * 59 * , 074021 ( 1999 ) [ hep - ph/9810220 ] . s. a. kulagin , hep - ph/9812532 . k. j. eskola , v. j. kolhinen , p. v. ruuskanen and c. a. salgado , nucl . . * a661 * , 645 ( 1999 ) [ hep - ph/9906484 ] . a. benvenuti _ et al . _ , c63 ( 1994 ) , 29 . a concise review with a critical overview of the literature is given in : i. bigi , m. shifman and n. uraltsev , ann . nucl . part . sci . * 47 * , 591 ( 1997 ) [ hep - ph/9703290 ] , with references to earlier work j. goldman , _ a next - to - leading - order qcd analysis of charged current event rates in @xmath52 deep inelastic scattering at the tevatron _ , kansas state university ph.d . thesis , unpublished ( 2000 ) . k. abe _ et al . _ [ sld collaboration ] , phys . rev . * 78 * , 2075 ( 1997 ) . k. abe _ et al . _ [ sld collaboration ] , `` a high - precision measurement of the left - right z boson cross - section asymmetry , '' [ hep - ex/0004026 ] . a. czarnecki and w. j. marciano , int . j. mod . phys . * a15 * , 2365 ( 2000 ) [ hep - ph/0003049 ] , phys . d * 53 * , 1066 ( 1996 ) [ hep - ph/9507420 ] . s. sarantakos , a. sirlin and w. j. marciano , nucl . * b217 * , 84 ( 1983 ) . c. h. llewellyn smith , nucl . * b228 * , 205 ( 1983 ) . e. a. paschos and l. wolfenstein , phys . d * 7 * , 91 ( 1973 ) . , nutev collaboration , proceedings of the xxxiiird rencontres de moriond ( 1998 ) , ` hep - ex/9806013`;j . nutev collaboration , proceedings of the dis98 , brussels , belgium , eds . g.h.coremans and r. roosen , world scientific , 588 ( 1998 ) . j. f. donoghue and e. golowich , phys . d49 * , 1513 ( 1994 ) [ hep - ph/9307262 ] . j. f. donoghue and e. golowich , phys . * b478 * , 172 ( 2000 ) [ hep - ph/9911309 ] . d. geiregat _ et al . _ [ charm - ii collaboration ] , phys . b * 245 * , 271 ( 1990 ) . r. belusevic and j. smith , phys . rev . * d37 * , 2419 ( 1988 ) . y. grossman , z. ligeti and e. nardi , nucl . * b465 * , 369 ( 1996 ) [ hep - ph/9510378 ] . a. j. buras and r. fleischer , hep - ph/9704376 . d. zeppenfeld and k. cheung , hep - ph/9810277 . m. gronau , c.n . leung and j.l . rosner , phys . * d29 * , 2539 ( 1984 ) . r. n. mohapatra and g. senjanovic , phys . * d23 * , 165 ( 1981 ) . d. wyler and l. wolfenstein , nucl . * b218 * , 205 ( 1983 ) . l. m. johnson , d. w. mckay and t. bolton , phys . rev . * d56 * , 2970 ( 1997 ) [ hep - ph/9703333 ] . a. vaitaitis _ et al . _ [ nutev collaboration ] , phys . rev . * 83 * , 4943 ( 1999 ) . x. h. guo and a. w. thomas , phys . d * 61 * , 116009 ( 2000 ) [ hep - ph/9907370 ] . e. m. aitala _ et al . _ [ e791 collaboration ] , charmed mesons , phys . * b462 * , 401 ( 1999 ) [ hep - ex/9906045 ] . a second e791 rare charm decay paper has recently been accepted by prl and is available at los alamos as hep - ex 0011077 . * e. m. aitala _ et al . _ [ e791 collaboration ] , d0 > v l+ l- and h h l l , hep - ex/0011077 . m. a. shifman and m. b. voloshin , sov . j. nucl . * 41 * , 120 ( 1985 ) . j. chay , h. georgi and b. grinstein , phys . * b247 * , 399 ( 1990 ) . i. i. bigi , n. g. uraltsev and a. i. vainshtein , phys . * b293 * , 430 ( 1992 ) [ hep - ph/9207214 ] . b. blok , l. koyrakh , m. shifman and a. i. vainshtein , phys . rev . * d49 * , 3356 ( 1994 ) [ hep - ph/9307247 ] . a. f. falk , m. luke and m. j. savage , phys . rev . * d53 * , 6316 ( 1996 ) [ hep - ph/9511454 ] . g. burdman , e. golowich , j. l. hewett and s. pakvasa , phys . rev . * d52 * , 6383 ( 1995 ) [ hep - ph/9502329 ] . c. greub , t. hurth , m. misiak and d. wyler , phys . * b382 * , 415 ( 1996 ) [ hep - ph/9603417 ] . j. anjos , _ et al . * 60 * , ( 1988 ) 1239 . r. godang _ et al . _ [ cleo collaboration ] , phys . lett . * 84 * , 5038 ( 2000 ) [ hep - ex/0001060 ] . h. park , hep - ex/0005044 . a. f. falk , y. nir and a. a. petrov , `` strong phases and d0 anti - d0 mixing parameters , '' jhep * 9912 * , 019 ( 1999 ) [ hep - ph/9911369 ] . j. m. link _ et al . _ [ focus collaboration ] , phys . b485 * , 62 ( 2000 ) [ hep - ex/0004034 ] . s. bergmann , y. grossman , z. ligeti , y. nir and a. a. petrov , phys . * b486 * , 418 ( 2000 ) [ hep - ph/0005181 ] . for a comprehensive review of @xmath83 physics at the @xmath967 , see p. f. harrison and h. r. quinn [ babar collaboration ] , _ the babar physics book : physics at an asymmetric b factory , _ slac - r-0504 , 1998 . . i. bigi and n. g. uraltsev , nucl . * b592 * , 92 ( 2000 ) [ hep - ph/0005089 ] . e. golowich and a. a. petrov , phys . * b427 * , 172 ( 1998 ) [ hep - ph/9802291 ] . for a review see : g. burdman , in : `` workshop on the tau / charm factory '' , argonne national lab , 1995 , aip conference proceedings no . 349 , p.409 , or a. a. petrov , hep - ph/0009160 . j. ellis and d. v. nanopoulos , phys . * b110 * , 44 ( 1982 ) . f. gabbiani , e. gabrielli , a. masiero and l. silvestrini , nucl . * b477 * , 321 ( 1996 ) [ hep - ph/9604387 ] . y. nir and n. seiberg , phys . lett . * b309 * ( 1993 ) 337 ; m. leurer , y. nir and n. seiberg , nucl . phys . * b420 * ( 1994 ) 468 . s. pakvasa and h. sugawara , phys . lett . * b73 * ( 1978 ) 61 ; t.p . chen and m. sher , phys . rev . * d35 * ( 1987 ) 3484 ; l. hall and s. weinberg , phys . rev . * d48 * ( 1993 ) r979 ; d. atwood , _ et al . _ phys . rev . * d55 * ( 1997 ) 3156 . fits and references to the homestake , kamiokande , gallex , sage , and super kamiokande data include n. hata and p. langacker , phys . rev . * d56 * 6107 ( 1997 ) ; j. bahcall , p. krastev , and a. smirnov , phys . rev . * d58 * , 096016 ( 1998 ) ; j. bahcall and p. krastev , phys . b436 * , 243 ( 1998 ) ; j. bahcall , p. krastev , and a. smirnov , phys . * d60 * , 093001 ( 1999 ) , http://www.sns.ias.edu/ jnb/ , and m. gonzalez - garcia , p. de holanda , c. pena - garay , and j. w. f. valle , nucl . b , in press ( hep - ph/9906469 ) . recent super kamiokande data is reported in super kamiokande collab . , y.fukuda et al . , phys . lett . * 82 * , 1810 , 243 ( 1999 ) . for recent experimental reviews , see , e.g. , l. dilella , hep - ex/9912010 ; h. robertson , hep - ex/0001034 , and talks at the workshop on the next generation nucleon decay and neutrino detector nnn99 , stony brook ( sept . 1999),http://superk.physics.sunysb.edu / nnn99/scientific_program/. d. ayres , t. gaisser , a. k. mann , and r. shrock , in _ proceedings of the 1982 dpf summer study on elementary particles and future facilities _ , snowmass , p. 590 ; d. ayres , b. cortez , t. gaisser , a. k. mann , r. shrock , and l. sulak , phys . rev . * d29 * , 902 ( 1984 ) . m. campanelli , a. bueno and a. rubbia , hep - ph/9905240 . d. dooling , c. giunti , k. kang and c. w. kim , phys . d * 61 * , 073011 ( 2000 ) [ hep - ph/9908513 ] . i. mocioiu , r. shrock , in the proceedings of the workshop on the next generation nucleon decay and neutrino detector nnn99 , stony brook ( sept . 1999 ) , hep - ph/9910554 . i. mocioiu and r. shrock , phys . d * 62 * , 053017 ( 2000 ) [ hep - ph/0002149 ] . bilenky , c. giunti , w.grimus , phys.rev.d58 , 033001 ( 1998 ) ; k. dick , m. freund , m. lindner , a. romanino , nucl . phys . * b562 * , 29 ( 1999 ) ; m. tanimoto , phys . lett . * b462 * , 115 ( 1999 ) ; a. donini , m. b. gavela , p. hernandez and s. rigolin , nucl . phys . * b574 * , 23 ( 2000 ) [ hep - ph/9909254 ] ; m. koike and j. sato , phys . d * 61 * , 073012 ( 2000 ) [ hep - ph/9909469 ] ; p. f. harrison and w. g. scott , phys . lett . * b476 * , 349 ( 2000 ) [ hep - ph/9912435 ] ; m. freund , m. lindner , s. t. petcov and a. romanino , nucl . * b578 * , 27 ( 2000 ) [ hep - ph/9912457 ] . the minos experiment , ` http://www-numi.fnal.gov:8875 ` .	conceptual design studies are underway for muon colliders and other high - current muon storage rings that have the potential to become the first true `` neutrino factories '' . muon decays in long straight sections of the storage rings would produce precisely characterized beams of electron and muon type neutrinos of unprecedented intensity . this article reviews prospects the for these facilities to greatly extend our capabilities for neutrino experiments , largely emphasizing the physics of neutrino interactions . muon colliders , muon storage rings , neutrino factories ; * pacs13.15.+g *
this work was supported by the national basic research program of china under grants nos . 2012cb821305 , 2010cb923200 and 2013cb922403 , the national natural science foundation of china under grants nos . 11374375 , 11204043 , 11274399 and 61078027 , and the ph.d . programs foundation of ministry of education of china under grant nos . b.a.m . appreciates hospitality of the sun yat - sen university ( guangzhou , china ) .	we demonstrate that in - bulk vortex localized modes , and their surface half - vortex ( horseshoe " ) counterparts self - trap in two - dimensional ( 2d ) nonlinear optical systems with @xmath0-symmetric photonic lattices ( pls ) . the respective stability regions are identified in the underlying parameter space . the in - bulk states are related to truncated nonlinear bloch waves in gaps of the pl - induced spectrum . the basic vortex and horseshoe modes are built , severally , of four and three beams with appropriate phase shifts between them . their stable complex counterparts , built of up to 12 beams , are reported too . nonlinear spatially periodic systems support diverse types of self - trapped in - gap states . in particular , spatial gap solitons @xcite originate from the interplay between the periodicity and nonlinearity . further , surface gap solitons @xcite appear at the interface between a uniform medium and a photonic lattice ( pl ) built into a nonlinear material . extended self - trapped waves with steep edges also exist in these settings , being related to truncated nonlinear bloch waves @xcite . modes of the latter type provide a link between extended nonlinear bloch waves @xcite and tightly localized gap solitons @xcite . recently , a great deal of interest has been drawn to the realization of parity - time ( @xmath0 ) symmetry in optics . originally , this concept was developed in quantum mechanics , where it was demonstrated that , beyond the conventional hermitian hamiltonians , their @xmath0-symmetric non - hermitian counterparts may also give rise to purely real ( hence physically relevant ) spectra @xcite . following the similarity between quantum mechanics and paraxial optics @xcite , @xmath0-symmetric optical systems with complex refractive indices @xcite have been extensively studied theoretically @xcite and experimentally @xcite . in this context , @xmath0-symmetric pls play an important role . taking into account the non - orthogonality of the respective eigenmodes , their coupled - mode description had to be reformulated via the variational principle @xcite . the light propagation in @xmath0-symmetric pls embedded into linear media were analyzed preliminarily @xcite . further , it has been found that 1d and 2d spatial gap solitons exist in pls built into a nonlinear material @xcite . although bloch waves @xcite and gap solitons @xcite were studied before in the context of some @xmath0-symmetric pls , the comprehensive study of self - trapped states in 2d @xmath0-symmetric systems combining lattices and nonlinearity was not reported yet . in particular , such self - trapped nonlinear states may serve as a necessary link between spatially localized gap solitons and extended nonlinear bloch waves under the @xmath0 symmetry . self - trapped vortices and surface modes are of great interest in the context of the @xmath0-symmetric settings . indeed , the study of nonlinear surface modes pinned on the interface of a @xmath0-symmetric system opens a way to explore the interplay between surface effects , the nonlinearity , and the @xmath0-symmetry . on the other hand , the analysis of localized vortices supported by @xmath0-symmetric pls should shed light on the cooperation and competition of the @xmath0-symmetry with the azimuthal instability and spatial periodicity . in this work , we show the existence of in - bulk and surface self - trapped states in 2d nonlinear systems with @xmath0-symmetric pls . in particular , we report in - bulk solitary vortices and novel half - vortex surface modes . stable half - vortex surface modes appear as horseshoes `` pinned on the interface between a uniform linear medium and a nonlinear medium with built - in @xmath0-symmetric pl . the in - bulk vortices and surface horseshoes '' have a common linear stability region at intermediate values of propagation constants . we consider the light propagations in two nonlinear systems with @xmath0-symmetric pls : a uniform setting of the nonlinear @xmath0-symmetric pl , and a composite setting of the nonlinear @xmath0-symmetric pl at the left side ( @xmath1 ) and a uniform linear medium at the right side ( @xmath2 ) . assuming that the light propagates along the @xmath3-axis , the amplitude of the electromagnetic field is written as @xmath4 , with carrier wavenumber @xmath5 and frequency @xmath6 . with the effective refractive index including contributions from the complex pl and the kerr effect , @xmath7 , the amplitude obeys the nonlinear schrdinger equation with the complex potential , @xmath8 e \notag \\ & & + 2k_{0}^{2}\left [ n_{0}^{\mathrm{pl}}\left ( n^{\mathrm{r}}+in^{\mathrm{i}}\right ) + n_{0}^{\mathrm{pl}}n^{\mathrm{nl}}\|e\|^{2}\right ] e=0 . \label{eq : one}\end{aligned}\]]here , @xmath9 is a constant , @xmath10 is the background refractive index , @xmath11 and @xmath12 are real and imaginary ( gain / loss ) parts of the spatial modulation of the local index , and @xmath13 is the kerr coefficient . similarly , the light propagation in the linear uniform medium obeys the paraxial equation @xmath14 e=0 , \label{eq : two}\]]with the respective real refractive index , @xmath15 . we normalize the equations by defining @xmath16 , @xmath17 , @xmath18 , and @xmath19 z}$ ] with an arbitrary scaling factor @xmath20 . the accordingly rescaled form of eqs . ( [ eq : one ] ) and ( [ eq : two ] ) is@xmath21with @xmath22 , and @xmath23 $ ] . the complex @xmath0-symmetric potential , @xmath24 , is chosen as@xmath25 , \\ w(\xi , \eta ) & = & \theta \{\sin [ \sqrt{2}(\eta -\xi ) ] + \sin [ \sqrt{2}(\eta+\xi ) ] \},\end{aligned}\]]with amplitudes @xmath26 and @xmath27 of the modulation of the real and imaginary parts of the refractive index . this pl is a @xmath28 counterclockwise rotation of the one considered in refs . @xcite . the configurations of pls are shown by the white - blue circles in the @xmath29-plane . its band - gap structure can be derived by using the plane - wave expansion method based on the floquet - bloch theorem , see fig . [ fig : one ] ( a ) . -symmetric photonic lattice : @xmath30 is the propagation constant , and @xmath31 , @xmath32 are bloch wavenumbers in @xmath33 and @xmath34 directions . ( b , c ) intensity profiles of self - trapped modes for propagation constant @xmath35 in the uniform and truncated systems , respectively . the white dot - dash line depicts the interface in the truncated system . ( d ) power @xmath36 and ( e ) the real part of the instability growth rate , @xmath37 , of the self - trapped modes versus the propagation constant , @xmath38 . green triangles and red circles in ( d ) correspond to the modes shown in ( b ) and ( c ) , respectively . the green dashed and red solid lines in ( e ) represent , severally , in - bulk and surface self - trapped modes . ( f ) the density profile at @xmath39 , evolved from the initial self - trapped mode ( c ) with @xmath40 noise . parameters are @xmath41 , @xmath42 and @xmath43 . ] to combine eqs . ( [ eq : seven ] ) and ( [ eq : eight ] ) into a single equation , we define a step function , @xmath44 at @xmath45 and @xmath46 at @xmath47 : @xmath48 gq+u(\xi ) \|q\|^{2}q=0 . \label{eq : jia_1}\end{aligned}\]]the stationary solution with real propagation constant @xmath38 is looked for as @xmath49 , where complex function @xmath50 obeys equation @xmath51 gu \nonumber\\ & + & u(\xi)\|u\|^{2}u - bu=0.\label{eq : five}\end{aligned}\ ] ] to find the stationary self - trapping solutions , we used numerical simulations with the modified squared - operator method @xcite . while the existence and stability of the simplest single - beam solitons in the present setting is quite evident , as the first step of the analysis we produce double - beam self - trapped states . for @xmath35 and @xmath43 , the in - bulk and surface double modes are displayed in fig . [ fig : one](b , c ) . due to the presence of the interface , the intensity of the surface self - trapped states is larger than the in - bulk ones , at the same propagation constant . the dependence of the total power , @xmath52 , on the propagation constant @xmath38 demonstrates that the power of the surface modes is also larger than that of the in - bulk ones , see fig . [ fig : one](d ) . different from the single - beam solitons @xcite , both surface and in - bulk self - trapped states in the semi - infinite gap do not exist near the first bloch band in fig . [ fig : one](d ) . stability of the self - trapped modes was investigated by means of the linearization for small perturbations . to a given stationary state , @xmath53 , the perturbation is added as @xmath54e^{ib\zeta } $ ] with infinitesimal @xmath55 @xcite , where @xmath56 and @xmath57 are two perturbation eigenfunctions , @xmath58 is the corresponding growth rate , and the star ( @xmath59 ) stands for the complex conjugate . from eq . ( [ eq : jia_1 ] ) , the following linearized equations are derived : @xmath60f \nonumber\\ & & + u(\xi ) u^{2}g , \\ + i\delta g&=&\left [ \nabla _ { \perp } ^{2}+u(\xi ) r^{\ast } ( \xi,\eta)u+\psi+2u(\xi ) \|u\|^{2}\right]g \nonumber\\ & & + u(\xi)\left(u^{\ast}\right)^{2}f,\end{aligned}\ ] ] with @xmath61g - b$ ] . as usual , the self - trapped mode is linearly unstable if there is an eigenvalue with @xmath62 . as seen in fig . [ fig : one](e ) , the two - beam in - bulk and surface self - trapped modes have a common stable region , with @xmath63 , at intermediate values of propagation constants @xmath38 . the predicted stability of the modes has been verified in direct simulations of eq . ( [ eq : jia_1 ] ) with @xmath40 random noise added as an initial perturbation , see an example for @xmath35 in fig . [ fig : one](f ) . . ( c , d ) the intensity profile and phase distribution for the three - beam surface self - trapped state at @xmath64 . ( e ) power @xmath36 of the self - trapped states versus @xmath38 . ( f ) the real part of the instability growth rate , @xmath65 , versus @xmath38 . green triangles and the red circles correspond to the modes shown in ( a ) and ( c ) , respectively . parameters are @xmath41 , @xmath42 , and @xmath43 . ] adding more beams with phase shifts between them , one can construct composite vortices . for an example , a composite vortex with the total phase circulation of @xmath66 may appear as a four - beam complex with the off - site vortex core in the center and the phase shifts @xmath67 between adjacent beams @xcite . we have found that the composite vortices can exist in the system of the uniform setting . the intensity profile and phase distribution of a typical stable four - beam vortex in the uniform setting system are shown in fig . [ fig : two](a ) and ( b ) for @xmath64 . near the interface in of the composite setting system , there are no complete vortex modes , while there appear essentially new surface modes , in the form of _ half - vortices _ ( horseshoes `` ) , built of three beams , see figs . [ fig : two](c ) and ( d ) . the dependence of the power @xmath36 on the propagation constant @xmath38 shows that , although the half - vortex mode ( c ) is built of three beams , its power @xmath36 is larger ( near the first bloch band ) than that of the in - bulk vortex mode ( a ) , which is composed of four beams , see fig . [ fig : two](e ) . the linear stability analysis shows that there exists a common stability region at intermediate values of propagation constants @xmath38 for the in - bulk vortices and surface horseshoes '' , see fig . [ fig : two](f ) . , @xmath68 ( b ) . profiles of the 7-beam surface states : ( c ) at @xmath69 , ( d ) at @xmath68 . ( e ) power @xmath36 of the states versus @xmath38 . ( f ) the real part of the instability growth rate , @xmath37 , versus @xmath38 . green triangles correspond to ( a ) and ( b ) , and red circles to ( c ) and ( d ) . parameters are @xmath41 , @xmath42 and @xmath43 . ] on top of the simple few - beam self - trapped states , like the conservative 2d nonlinear systems @xcite , our @xmath0-symmetric systems can also support complex multi - beam ones built of up to 12 beams , see fig . [ fig : three ] . due to the interaction between individual beams , their intensity is larger at the center of the structure , the intensity difference gradually vanishing with the increase of propagation constant @xmath38 . near the interface in the truncated system , there are no beams located in the linear medium , while the near - interface beams become stronger , see fig . [ fig : three](c , d ) . power @xmath36 increases with propagation constant @xmath38 for both the in - bulk and surface self - trapped states , see fig . [ fig : three](e ) . results of the linear - stability analysis for these states , displayed in fig . [ fig : three](f ) , reveal a common stability region for the in - bulk and surface modes , at intermediate values of propagation constants @xmath38 . noise . ( a ) the density profile at @xmath70 evolves from the the in - bulk vortex self - trapped nonlinear waves in fig . [ fig : two ] ( a ) . ( b ) the corresponding phase distribution for ( a ) . ( c ) the density profile at @xmath70 evolves from the in - bulk multi - beam self - trapped mode in fig . [ fig : three ] ( b ) . ( d ) the density profile at @xmath70 evolves from the surface multi - beam self - trapped mode in fig . [ fig : three ] ( d ) . ] by simulating the beam propagation with @xmath40 random noise , we have verified the stability of the vortex modes , as shown in fig . [ fig : two ] , and of multi - beam ones , see fig . [ fig : three ] . in particular , fig . [ fig : four](b ) demonstrates that the phase distribution of the input vortex mode keeps the phase - winding structure . thus , the direct simulations corroborate the predictions of the linear - stability analysis . in conclusion , we have found several novel species of in - bulk and surface self - trapped states in 2d kerr - nonlinear optical systems with @xmath0-symmetric pls ( photonic lattices ) . these include stable in - bulk localized vortices and surface half - vortices ( horseshoes " ) . the self - trapped modes are related to truncated nonlinear bloch waves , the surface modes being linked with the truncated in - bulk ones . along with the basic vortex and half - vortex states , which are built , respectively , of four and three constituent beams . the stable multi - beam self - trapped states , composed of up to 12 constituents , have been found too . the formation of these surface modes results from the interplay of the surface effects , nonlinearity , and the @xmath0-symmetry . due to the surface - enhanced reflection , the discrete diffraction is stronger in the direction perpendicular to the interface than in the direction parallel to it @xcite , therefore the surface modes feature stronger nonlinearity , which is necessary to balance the diffraction .
globular clusters are important tracers of the chemical and dynamical evolution of the milky way . since the orbits of globular clusters may have kept their characteristics from the early times of the formation of the milky way , we may extract from their kinematics some clues for the understanding of the origin of the milky way . however so far for only some 40 globular clusters absolute proper motions exist ( dinescu et al . 1999 ) , which partly are of low quality ( e.g. dauphole et al . 1996 ) . nevertheless on the basis of a sample of 26 globular clusters dauphole et al . ( 1996 ) found an indication of a metallicity gradient among the halo clusters and a mean rotation of about + 40 km s@xmath1 for the complete globular cluster sample . additional data especial for southern clusters were determined by dinescu et al . ( 1999 and references herein ) . the results of the hipparcos mission offer new possibilities for the determination of absolute proper motions of globular clusters and other objects of interest for galactic kinematics . since previous studies have used mainly extragalactic objects as proper motion zero points , the use of hipparcos reference stars offers a unique possibility for the determination of absolute proper motions at low galactic latitutes without suitable extragalactic background objects . geffert et al . ( 1997 ) have determined absolute proper motions of 10 globular clusters with respect to hipparcos . these were combined with data of 5 other clusters and the kinematics of this sample was studied by brosche et al . ( 1997 ) and odenkirchen et al . a complete discussion of these data together with additional proper motions was given recently in dinescu et al . ( 1999 ) . m10 is a globular cluster located very near to the direction to the galactic center ( @xmath2 , @xmath3 ) , but at a distance of only 4.3 kpc from the sun . due to its metallicity of @xmath4 = -1.52\;$]dex ( harris 1996 ) it belongs to the halo group of clusters according to zinn s ( 1985 ) classification . the preliminary kinematical data of m10 ( geffert et al . 1997 ) indicate a disk like motion , the most disk like motion found in the group of globular clusters , whose proper motions were determined using hipparcos ( odenkirchen et al . 1997 ) . however , this preliminary proper motion has the problem that only three to four hipparcos stars could be used for the reduction of the plates . this made the previous solution very uncertain . here we have included new photographic material ( from shanghai ) and recent ccd observations to get a new determination of the absolute proper motion of m10 . moreover , we use in addition to the hipparcos catalogue the act catalogue ( urban et al . the act catalogue , due to its denser coverage of the sky , allows the use of more reference stars for the determination of the absolute proper motions . lrrr plate no . & epoch & emulsion & telescope + 149 & 1902.6 & & dr hl + 406 & 1905.6 & & r sh + 262 & 1916.5 & & dr hl + 309 & 1917.5 & & dr hl + 312 & 1917.5 & & dr hl + 314 & 1917.5 & & dr hl + cl57016 & 1957.5 & 103a - o & r sh + cl57031 & 1957.6 & 103a - o & r sh + cl57033 & 1957.6 & 103a - o & r sh + 1232 & 1976.6 & iia - o / bg 25 & dr hl + 1874 & 1994.5 & iia - o / bg 25 & dr hl + 1875 & 1994.5 & iia - o / bg 25 & dr hl + + [ t_photmat ] table [ t_photmat ] lists the photographic plates used in this work , while table [ t_ccdobs ] indicates the observational data of the ccd frames . the plates were taken either with the ( @xmath5 m , @xmath6 m ) double refractor of bonn , now located at hoher list observatory , or with the ( @xmath7 m , @xmath8 m ) refractor of shanghai . the ccd frames were taken with the wwfpp camera ( reif et al . 1994 ) at the 1.23 m telescope at calar alto and with a similar camera ( holicam ) at hoher list observatory . the use of the complete material allows a nearly uniform coverage of the epoch difference of 92 years . the limiting magnitude of the plates is of the order of @xmath9mag , while the corresponding one of the ccd frames ranges from @xmath10 to 19 mag . the refractor plates of bonn were mainly scanned at the pds 2020gm of university of mnster . the plates r1874 and r1875 with lower limiting magnitude have been measured at the ascorecord of hoher list observatory . on these plates only 80 stars for each plate were measured . stars were extracted and rectangular coordinates @xmath11 and @xmath12 were determined from the pds measurements using standard procedures ( e.g. tucholke 1994 ) . the first epoch plates of the bonn refractor contained scratches and reseau lines , which led to problems for a significant number of the stars . therefore for some of these stars no rectangular coordinates could be obtained . the plates from shanghai were scanned at the pds 1010ms of the purple mountain observatory , chinese academy of sciences ( see also wang et al . 1999 ) . the determination of the rectangular coordinates @xmath11 , @xmath12 of the stars on the ccd frames was performed for the observations from hoher list by standard ccd reduction techniques ( daophot , iraf ) routines . magnitudes , @xmath11 and @xmath12 positions were determined via psf fit . the observations from calar alto were reduced earlier ( geffert et al . 1994 ) by the imex routine of the iraf program package . the astrometric reduction was performed by a central overlap algorithm . due to the small field of the ccd frames we had to use only the plates in the first step of the reduction . a catalogue of positions and proper motions of 450 stars was established in this first step . in the following steps of the reduction the ccd frames were included . while for the plates only quadratic polynomials of the rectangular coordinates had to be taken into account , third order polynomials were necessary for the reduction of the ccd frames . the third order polynomials for the reduction of the ccd frames had to be used due to the distortion of the optics of the focal reducer of the wwfpp camera ( geffert et al . from the different position and time pairs we determined for each star for a certain epoch the mean position and the proper motion using least squares technique . all stars with proper motion errors larger than 4 mas / yr were omitted . the final catalogue contains 532 positions and proper motions of stars in the region of m10 . the median of the internal errors was about @xmath13mas / yr . from a plot of the proper motions versus magnitude no magnitude equation was found in our data . lclr telescope & epoch & filter & no . of frames + 1 m hoher list & 1996 & @xmath14 , @xmath15 & 10 + 1.23 m calar alto & 1994 & @xmath16 & 5 + [ t_ccdobs ] lr@ r@ r@lr@ r@ r@lrrrrrrr & & & & & & & & + & [ @xmath17 & @xmath18 & @xmath19 & ] & [ @xmath20 & @xmath21 & @xmath22 & ] & & & & & & + 1 & 16 & 54 & 34&.488 & @xmath23 & 45 & 20&.57 & 0.004 & 0.04 & @xmath24 & @xmath25 & 1.7 & 1.2 + 2 & 16 & 54 & 46&.910 & @xmath26 & 52 & 53&.73 & 0.006 & 0.02 & 3.5 & 5.8 & 3.3 & 0.6 + 3 & 16 & 54 & 47&.085 & @xmath23 & 23 & 06&.77 & 0.002 & 0.01 & @xmath27 & 4.4 & 1.1 & 0.3 + 4 & 16 & 54 & 48&.326 & @xmath23 & 13 & 34&.70 & 0.002 & 0.03 & @xmath28 & @xmath29 & 0.7 & 0.7 + @xmath30 & & & & & @xmath31 & @xmath31 & @xmath31 & @xmath31 + we have performed two independent reductions of the m10 data with reference stars from hipparcos ( esa 1997 ) and the act catalogue ( urban et al . although the hipparcos stars proper motions are more accurate , they seemed to be of only limited usefulness for our work , since the majority of the plates contained only four hipparcos stars , while about 17 stars could be used from the act catalogue . therefore we consider both solutions as equivalent . table [ t_cdsdat ] ( the complete table is available in electronic form ) gives the catalogue of our positions and proper motions for the complete field with respect to the act catalogue . the limiting magnitude of this catalogue is about @xmath9mag , which corresponds to the limiting magnitude of the first epoch plates . the size of the field is approximately @xmath32 centered on m10 . we chose the act solution for the catalogue in table [ t_cdsdat ] . the proper motions may be transfered to the hipparcos system by adding @xmath33mas / yr and @xmath34mas / yr to the proper motions from table [ t_cdsdat ] . for the determination of the membership we will also use the solution based on the act catalogue , while for the determination of the absolute proper motion of m10 we will take the mean of both solutions . ccds have been used in astrometry since several years , e.g. for the determination of parallaxes , double stars and for meridian circle observations ( see references in geffert 1998 ) . however most of these observations are based on ccd observations with fields of @xmath35 and smaller . since our study uses ccd observations of fields with a size at least 20@xmath3620 it seems necessary to evaluate the accuracy , which may be obtained with such telescope detector combinations . while the ccd frames of calar alto were already tested in an earlier study ( geffert et al . 1994 ) , we will concentrate here on the hoher list observations . in a first step we have compared positions of stars from pairs of ccd frames . the positions of one frame were transformed by an affine transformation to a second frame and the rms of the differences were calculated for each coordinate . under the assumption that both frames contribute with equal weight to the differences , we have calculated from the rms the mean uncertainty of one position on one frame . these are given in table [ t_ccdacc ] for several ccd pairs with nearly identical limiting magnitudes . in this comparison we included all stars which were detected on the ccd frames . lrrrr frames & & & lim . mag . & of stars + & [ mas ] & [ mas ] & & + 4236/4237 & 62 & 64 & 16.5 & 1373 + 4238/4239 & 120 & 120 & 18.5 & 3378 + 4240/4241 & 63 & 67 & 16.2 & 1081 + 4242/4243 & 100 & 80 & 17.0 & 1980 + [ t_ccdacc ] in a second test we consider only the stars which contribute to the final catalogue of our investigation . these are stars at the brighter end of the magnitude distribution . reducing the plates with one catalogue ( act or hipparcos ) leads to positions of the stars of each plate / ccd frame in a common system . the position and proper motion of each star in our final solution described in sect . 2 are determined by a fit to the positions and epochs of the different plates / ccd frames for each star . the mean position and proper motion of each star is then used to update the positions of each star for the epoch of the individual plates / ccd frames . for each plate the mean and rms of the positional differences from the initial positions are determined . the rms will give an indication of the accuracy of each individual plate / ccd frame . table [ t_posdiff ] summarizes for the hoher list frames the mean and standard deviations for each frame . lrrrrr ccd frame & & & & & colour + & [ mas ] & [ mas ] & [ mas ] & [ mas ] & + 4236 & @xmath37 & 70 & @xmath38 & 90 & @xmath15 + 4237 & @xmath39 & 70 & @xmath40 & 70 & @xmath15 + 4238 & @xmath41 & 80 & @xmath42 & 90 & @xmath15 + 4239 & @xmath39 & 100 & @xmath43 & 50 & @xmath15 + 4240 & @xmath41 & 80 & @xmath42 & 50 & @xmath14 + 4241 & @xmath44 & 40 & @xmath45 & 50 & @xmath14 + 4242 & @xmath46 & 90 & @xmath47 & 50 & @xmath14 + 4243 & @xmath48 & 40 & @xmath39 & 90 & @xmath14 + [ t_posdiff ] table [ t_posdiff ] shows a slight difference in the @xmath15 and @xmath14 frames of the order of a few mas . nevertheless the systematic deviations between the @xmath15 and @xmath14 frames are small . the positional accuracy of each star is of the order of one tenth of a pixel . this value is a little bit larger with respect to other studies ( e.g. geffert 1998 ) . the reason may be the crowding in the region of m10 . in general , the accuracy of one single frame is of the order of the accuracy of one refractor plate , which justfies the use of ccd frames for the second epoch observation . figure [ f_vppdia ] gives the vector - point - plot diagram of all our proper motions indicating the concentration of the cluster stars . for the separation of cluster and field stars we have concentrated our investigation on stars up to the limiting radius of 14@xmath49 ( webbink 1988 ) around the cluster centre . 284 stars remained in the data sample . for these stars a bivariate gaussian fitting to the proper motions was performed using the method of sanders ( 1971 ) . table [ t_fitpar ] gives the result of the analysis . the cluster standard deviations are larger than the internal errors of the proper motion . this difference is clearly not an indication of a possible detection of internal motions of stars in the cluster . more likely these differences are caused by crowding effects and undetected systematic errors of the different telescopes used in this investigation . m10 is a globular cluster with three slow variable stars ( clement et al . 1985 ) and several uv bright stars ( zinn et al . 1972 ; harris et al . our sample contains only the variable stars v2 and v3 ( see references in clement et al . the other special stars are located in crowded regions , below the limiting magnitude , or exhibit too large proper motion errors . both variable stars v2 , v3 have according to our investigation a membership probability of 97% . this confirms the earlier suggestion ( clement et al . 1985 ) that these stars are of w - virginis type , although v3 has a very unusual period of 7.8 days . lrlr parameter & value & parameter & value + @xmath50 & 193 & @xmath51 & 88 + @xmath52 & @xmath53 & @xmath54 & @xmath55 + @xmath56 & @xmath57 & @xmath58 & @xmath59 + @xmath60 & 1.8 & @xmath61 & 7.0 + @xmath62 & 2.0 & @xmath63 & 7.1 + [ t_fitpar ] the first absolute proper motion of m10 based on the hipparcos system was given in geffert et al . ( 1997 ) , which was later used in odenkirchen et al . ( 1997 ) and dinescu et al . . however , as mentioned in odenkirchen et al . ( 1997 ) , the result of m10 was only preliminary due to the small number of hipparcos stars used for the reduction of the plates . the use of additional plates from shanghai has led to a significant lower proper motion in declination . however , even for this sample of plates the solution is only moderately stable : omitting one reference star leads to an absolute proper motion of m10 , which differs from the original solution by about more than the mean uncertainties of the absolute proper motions . fortunately , this is not the case for the denser act catalogue . in summary , the use of the act catalogue resulted in a more stable solution , while the use of hipparcos stars provided the more direct link to an extragalactic reference frame . the catalogues based on the act catalogue ( urban et al . 1998 ) and on the hipparcos catalogue ( esa 1997 ) contributed therefore with equal weight to the final absolute proper motion of the globular cluster m10 . we obtained a mean absolute proper motion of @xmath64mas / yr and @xmath65mas / yr . together with the distance from the sun of 4.3 kpc and a radial velocity of @xmath66 km s@xmath1 ( harris 1996 ) we have calculated the velocity components in a system of galactic standard at rest ( @xmath67,@xmath14,@xmath68 ) , peri and apogalactic distances @xmath69 , @xmath70 , eccentricity @xmath71 and the @xmath0-component of the angular momentum of the orbit using a simple logarithmic galactic mass model ( see e.g. dauphole et al . these data are shown in table [ t_m10dat ] . figure [ f_m10orb ] gives the orbit of m10 integrated over 10 gyr backwards according to the method of allen & santilln ( 1991 ) using the programme of odenkirchen et al . ( 1997 ) . lrrrrrrr object & & & & & & & + & & & & & & & + m10 & @xmath72 & @xmath73 & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 + [ t_m10dat ] = 6.0 cm our results demonstrate the importance of the new astrometric catalogues ( hipparcos , act ) for the determination of absolute proper motions of objects located in the galactic plane . for these objects no extragalactic link of the proper motion system is available due to the absense of suitable objects . we show that the differences between the solutions based on the act catalogue and the corresponding one based on the hipparcos catalogue are small . although this result has to be checked using data of other fields , it would offer a range of applications for the act catalogue because of its larger number of stars . moreover , even reaching only moderate accuracy on the wide field ccd frames with respect to small field ccd observations , our results indicate that these observations may supersede the photographic plates in near future . the new data of m10 show an eccentricity of the orbit of 0.41 , which would place the cluster now in the range of eccentricities from 0.4 to 0.8 found for the majority of the globular clusters ( odenkirchen et al . the rotational velocity changed to @xmath79 km s@xmath1 with respect to the @xmath80 km s@xmath1 found in odenkirchen et al . according to its moderate eccentricity and rotational velocity , m10 belongs rather to the halo class of objects . this is in line with its metallicity . however , the @xmath0-distance from the galactic plane does not exceed 3 kpc , which would be more characteristic for a thick disk object . as noted already in dinescu et al . ( 1999 ) recent age determinations ( hurley et al . 1989 ; richer et al . 1996 ; chaboyer et al . 1996 ; buonano et al . 1998 ) agree that m10 is an old halo cluster . together with ngc 6626 , ngc 6752 ( dinescu et al . 1999 ) and m71 , which has according to geffert & maintz ( 2000 ) a much higher age , m10 establishes a group of high age globular clusters with orbits of a disk or thick disk character . this result would imply that at the time when the oldest globular clusters were born , the milky way had already a gas distribution with a disk component , where the gas enabled the formation of clusters . on the other hand the recent age determination by rosenberg et al . ( 1999 ) places m10 in the middle of the age distribution of globular clusters . piotto & zoccali ( 1999 ) have found an unusual steep luminosity function of m10 in comparison with other clusters . our data indicate that this luminosity function was generated more likely by internal dynamics of the cluster than by the motion in the galaxy . if the luminosity function would be affected by the crossing of the cluster through the galactic plane , one would rather expect a more flat luminosity function for m10 due to the evaporation of faint stars as it was found recently for the globular cluster ngc 6712 ( de marchi et al . 1999 ) . we are indebted to h .- j . tucholke ( bonn ) for measuring some of the refractor plates . it is a pleasure to thank k.s . de boer ( bonn ) for helpful discussions . this research has made use of the simbad database ( see wenger et al . 2000 ) , operated at cds , strasbourg , france . geffert m. , 1998 , ccd astrometry . in : p. brosche et al . ( eds ) ` the message of the angles - astrometry from 1798 to 1998 ' , proc . of the spring meeting of the astronomische gesellschaft , verlag harri deutsch , thun und frankfurt am main , p. 181 , chen l. , wu z.y . , gupta a.c . , geffert m. , 1999 , a&as 142 , 373 webbink r.f . , 1988 , kinematics of the galactic globular cluster system . in : grindlay j.e . , davis philip a.g . proc iau symp . 126 , the harlow - shapley symposium on globular cluster systems in galaxies , iau symp 126 , reidel , dordrecht , p. 49	we present the first proper motion study of m10 ( ngc 6254 ) . absolute proper motions of about 532 stars in the field of the globular cluster m10 were determined with respect to hipparcos and act reference stars . in addition to photographic plates of bonn and shanghai also wide field ccd observations as second epoch plates were used . the wide field ccd observations show an accuracy comparable to that of the photographic plates . a good coincidence of the solutions based on reference stars from hipparcos and from act was found . our final proper motions allow a sufficient separation of cluster and field stars . two population ii cepheids were confirmed to be members of m10 . the absolute proper motion of m10 was determined and combined with its distance from the sun and its radial velocity . the space motion and metallicity of m10 indicates the characteristics of a halo object with an orbit reaching to a maximal @xmath0-distance of less than 3 kpc .
pks 1830@xmath0211 @xcite consists of two @xmath3 @xcite blazar images separated by 1 and lensed by a @xmath4 @xcite spiral galaxy @xcite . pks 1830@xmath0211 was observed in the radio ( e.g. , * ? ? ? * ; * ? ? ? * ) , infrared ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , x - rays @xcite , and gamma - rays @xcite , and the spectral energy distribution of pks 1830@xmath0211 @xcite shows two emission bumps with one in the infrared and the other between the hard x - ray and gamma - ray bands . the gravitational lens pks 1830@xmath0211 is complicated in many aspects , and one of them is the x - ray absorption . the x - ray absorption was first detected by rosat @xcite . later asca observations @xcite show that the x - ray absorption is variable ; these authors favored a model in which microlensing was the cause of variability . recently , pks 1830@xmath0211 was observed by _ chandra _ and _ xmm - newton _ for five epochs , which enabled us to study the nature of the x - ray absorption in detail combining the angular resolution of _ chandra _ and the large collecting area of _ xmm - newton_. since the two images are resolved by _ chandra _ , we can test the microlensing model predictions and study the x - ray absorption separately for the two lensed images . @xcite found that the differential absorption at the lens galaxy between the two images is @xmath5 . in this paper , we study the time evolution of the absorption in the system from the five _ chandra_and _ xmm - newton _ observations and confirm that the x - ray absorption toward pks 1830@xmath0211 is variable . moreover , we rule out the microlensing interpretation of the observed variability . we observed pks 1830@xmath0211 twice with _ chandra _ and three times with _ xmm - newton_from 2000 to 2004 . the details of the observations are listed in table [ tab : obslog ] . the _ chandra _ data were reduced with the ` ciao3.2 ` software tools provided by the _ chandra _ x - ray center ( cxc ) following the standard threads on the cxc website . only events with standard _ asca _ grades of 0 , 2 , 3 , 4 , and 6 were used in the analysis . we improved the image quality of the data by removing the pixel randomization applied to the event positions by the standard pipeline . in addition , we applied a sub - pixel resolution technique @xcite to the events on the s3 chip of acis where the quasar images are located . this allowed us to resolve the two lensed images of the blazar . the _ xmm - newton _ data were reduced using the standard analysis software ` sas6.0 ` . we used the tasks ` epchain ` and ` emchain ` from ` sas ` to reduce the pn and mos data , and filtered the events with patterns @xmath6 and @xmath7 for the pn and mos chips , respectively . we fitted the spectra of pks 1830@xmath0211 using ` xspec v11.3.1 ` @xcite in the 0.358 kev observed energy range for _ chandra_spectra and in the 0.3510 kev range for _ xmm - newton _ spectra . in all of our models , we fixed the galactic absorption at @xmath8 @xcite . we analyzed the _ chandra _ spectra of individual images a and b. we modeled the spectra with a power law modified by the galactic absorption and the absorption at the lens redshift ; the fitting results are listed in table [ tab : indi ] . we experimented with two models where the power law photon indices for the two images were allowed to be different ( model 1 ) and where they were constrained to be the same ( model 2 ) . we assumed that the excess absorption ( above galactic ) arises in the lens galaxy at z=0.886 ( we test this assumption later ) . in both the models , the column densities at the two _ chandra _ epochs for image b are consistently larger than those of image a. in addition , the absorption in the first epoch is consistently larger than that in the second epoch for both images , except for image b in model 1 where it is similar within errors . the photon indices obtained from model 1 fits are similar for the two images and in both epochs , partially due to the large error bars on the parameter caused by the low signal - to - noise ratio of the individual spectra . the photon indices obtained from model 2 show @xmath9 difference from epoch to epoch , which again is not significant . the difference between the absorption in the two images occurs simply because the lines of sight of the two images intersect different parts of the lens galaxy @xcite . the difference between the absorption at the two epochs indicates that the x - ray absorption is variable . the origin of the absorption variability could be either at the lens redshift or at the source redshift . we note that the differential absorption between images b and a is similar in the two epochs , especially in model 2 . this is suggestive of absorption variability occurring at the source redshift . we then analyzed the spectra of the combined images ab for the _ chandra_and _ xmm - newton _ observations . we first fitted the spectra of pks 1830@xmath0211 of the five epochs with a power - law model modified by neutral absorption from the milky way and the lens galaxy . the fitting results are listed in table [ tab : spec ] ( model 3 ) . we note that the absorption at the lens redshift for this model should be treated as an averaged absorption for the two lines of sight . we also fitted the co - added three _ xmm - newton _ spectra to obtain a higher signal - to - noise ratio spectrum , and the results are also listed in table [ tab : spec ] . the higher s / n of _ xmm - newton_spectra allows better constraints on the power - law spectral index . we find that the spectral index varies between the _ chandra _ and the _ xmm - newton_observations , though variations within the _ chandra _ epochs and _ xmm - newton _ epochs are smaller . comparing the fitting results of model 3 and model 2 , the spectral index obtained by fitting the combined images ab for the _ chandra _ observations are consistent with results from individual spectral fits in [ sec : indi ] . the absorption at the lens is also variable , especially when comparing the second _ chandra _ observation with other epochs . we further test the variability of the spectra by comparing a model with no spectral variability ( with the exception of normalization ) and another model with variable absorption and spectral index . we found that the model with the spectral variability produced a better fit with a @xmath10 improvement of 56.1 ( a null probability of @xmath11 by the f - test ) when jointly fitting the five x - ray observations . when we only consider the three _ xmm - newton _ observations , the corresponding improvement is 14.5 with a null probability of 0.0097 given by the f - test . the largest absorption variation is @xmath12 between the second _ chandra _ epoch and the first _ xmm - newton _ epoch . although the variable spectral index in blazars is common ( e.g. , * ? ? ? * ) , the absorption at the lens galaxy is unlikely to vary on the time scales of years . as discussed in [ sec : indi ] , it is more likely that the variable absorption component is at the source redshift close to the agn , where a short time - scale variability is possible and may be expected . the next model we tried is a power - law modified by three absorption components , the galactic absorption and the absorptions at the lens and the source . we fixed the absorption at the lens as @xmath13 , the smallest absorption column density detected in the five epochs from the previous model . this is the largest absorption column density at lens we can set to ensure no absorption variability at the lens galaxy . the fitting results are listed in table [ tab : spec ] ( model 4 ) . again we detected variability of both the photon index and the absorption at the source redshift . using the same test that we described in the previous model , we found that the model allowing spectral variability improved the @xmath10 by 59.8 with a null probability of @xmath14 for jointly fitting the five x - ray observations , and an improvement of 14.7 with a null probability of 0.01 for jointly fitting the three _ xmm - newton _ observations . the largest absorption variation for this model is @xmath15 between the second _ chandra _ epoch and the first _ xmm - newton _ epoch . the variable absorption in this model can be naturally associated with outflows from the central engine . although , this model produces comparable fits for the two _ chandra _ spectra compared with model 3 , where there is no absorption at the source redshift , the fits for the _ xmm - newton _ spectra with a higher s / n are worse than model 3 with @xmath10 increases of 5.8 , 37.2 , 13.0 for the three _ xmm - newton _ spectra . we also tested whether fitting the absorption component at the source with a warm absorber or a partially covering absorber would improve the fit . while the warm absorber model produces a worse fit , the partial covering model ( table [ tab : spec ] , model 5 ) produces comparable fits to model 3 . for the partial covering model , the model allowing spectral variability produces improvement of the @xmath10 by 79.7 with a null probability of @xmath16 for jointly fitting the five x - ray observations , and an improvement of 14.1 with a null probability of 0.01 for jointly fitting the three _ xmm - newton _ observations . the largest absorption variation for this model is @xmath17 , again , between the second _ chandra _ epoch and the first _ xmm - newton _ epoch . as a final improvement to our model , we assume that the lines of sight of the two lensed images pass through different locations in the lens galaxy and thus have different absorption column densities . this differential absorption is measured as @xmath5 @xcite . therefore , we have a model composed of two power - law components and each of them is modified by three absorptions from galactic , lens , and source . the two power - law components representing the two lensed images have the same photon index but different normalizations . the normalization ratio between images a and b is constrained as @xmath18 by taking the hard x - ray flux ratios ( 38 kev ) from the _ chandra _ observation . the galactic absorption and the absorption at the source are the same for the two power - law components , and the absorbers at the source are assumed to be partially covering the continuum . the absorber column densities at the lens are different for the two power - law components by the amount given by @xcite . we fit the two _ chandra _ spectra and the co - added _ xmm - newton _ spectrum simultaneously with this model . we constrained that the absorption at lens did not vary . the fitting results are listed in table [ tab : spec ] ( model 6 ) , and the spectra are shown in figure [ fig : spec ] . the fitting results indicate that the absorption at the source has varied by roughly @xmath19 . we detected the variable absorption , variable power - law photon index , and variable flux for the blazar pks 1830@xmath0211 . the observed variability between _ chandra _ and _ xmm - newton _ spectra could be , in part , due to imperfect cross - calibration between the two . however for high s / n spectra , the cross - calibration between _ chandra _ and _ xmm - newton _ can yield a spectral difference of @xmath20 0.03 , whereas our measured differences are between @xmath20 0.150.3 . in addition , the higher s / n _ xmm - newton _ spectra show a statistically significant variability within the three _ xmm - newton _ observations ( @xmath10 improvement of 14.5 , 14.7 , and 14.1 for models 3 , 4 , and 5 with spectral variability ) . we also found the spectral variability between the two _ chandra _ observations , where the second _ chandra _ epoch has consistently lower absorption column densities than the first observation . this is better demonstrated in 3.1 , where we analyzed the individual spectra of each image . we plot the variations of absorption in figure [ fig : nh ] , photon index variations in figure [ fig : gamma ] , and the photon index against unabsorbed flux in figure [ fig : gammaf ] . it is well known that blazars are variable , so the observed spectral variability is not surprising in itself . we also found that the photon index and flux are correlated ( figure [ fig : gammaf ] ) . this relation has been observed in several gamma - ray loud agns ( foschini et al . pks 1830@xmath0211 provides an additional example of a source following this correlation . the variable absorption is more intriguing . the variable absorption toward pks 1830@xmath0211 was detected previously with _ observations @xcite , although interpreted differently . @xcite obtained a variation amplitude of @xmath21 when modeling the absorption at the lens , which is consistent with our measurement of @xmath22 ( model 3 ) . we note that the variation detected by @xcite is on the timescale of 10 days while that found by us is on the timescale of years . it is possible that the variable absorption is at the source redshift . depending on the nature of the absorber , we obtained variations of @xmath23 and @xmath24 for a fully covering absorber ( model 4 ) and a partially covering absorber ( model 5 , 6 ) , respectively . the other possibilities are microlensing models . @xcite proposed a microlensing interpretation to explain the apparent variable absorption . the basic idea is that the absorption does not vary but is different for the two images , and the microlensing produces flux changes between the two images which appears as variable absorption . as we discussed in the spectral fits to individual _ chandra _ images ( [ sec : indi ] ) , the absorption varies in individual images between the two epochs , which can not be attributed to this simple microlensing model . for this reason , we rule out the simple microlensing model as the cause of variable absorption in pks 1830@xmath0211 . more complicated microlensing models are required which must involve significant emission size differences between the soft and hard x - ray regions to interpret the data . however , even these models will have difficulty explaining the similar differential absorptions between images a and b for the two _ chandra _ epochs . microlensing will only produce uncorrelated changes between the images . variable absorption thus appears to be a robust result , independent of model . the variability can either come from the lens or from the source redshift . if the variable absorption is at the lens , a variability amplitude of @xmath22 can sufficiently parameterize the _ chandra _ and _ xmm - newton _ spectra without invoking any absorption at the source . recently , variable mg ii absorption has been detected in an intervening absorber in front of a gamma - ray burst @xcite . the observed mg ii column density changes as the burst expands geometrically @xcite . similarly , if the size of the blazar changes with luminosity , it may produce variable absorption at the lens . another mechanism that can produce variable absorption at the lens galaxy is proposed by @xcite , where microlensed images , slightly shifted from the original image position , intersect slightly different positions at the lens , producing variable absorption . however , in pks 1830@xmath0211 , the differential absorption of the two images in the two _ chandra _ epochs is similar , arguing against the variability at the lens . it is far more likely that the variable absorption is intrinsic to the blazar , simply because there are other agns with variable intrinsic absorption , including blazars ( see below ) . while the absorption toward the gravitational lens pks 1830@xmath0211 is very complicated , with the sequence of _ chandra _ and _ xmm - newton _ observations , a converging picture is emerging . besides the galactic absorption of @xmath25 @xcite , there is also an intrinsic absorption and absorption from the lens galaxy . the absorption column densities in the lens galaxy are different for the two lensed images by @xmath26 @xcite , and this is consistent with the differential extinction measurement between the two images and a galactic dust - to - gas ratio . the amplitude of the variability of the intrinsic absorption is model dependent with amplitudes of @xmath1@xmath2 . recently , the intrinsic absorption for blazars has been reported in several cases , such as gb b1428 + 4217 @xcite , pmn j0525@xmath03343 @xcite , and rbs 315 ( piconcelli & guainazzi 2005 ; see also tavecchio et al . 2007 ) . in this paper , we present variable intrinsic absorption toward gravitationally - lensed blazar pks 1830@xmath0211 . the large variation in the absorption column density can be interpreted as outflows similar to those detected in bal qsos . however , as pks 1830@xmath0211 is a blazar , this outflow must be in the polar direction . recent numerical simulations of accretion flows in agns @xcite discussed the cases of polar outflows ; our observations of pks 1830@xmath0211 might be providing an example in support of such models . in addition , it is also possible that the intrinsic absorption variation is caused by the interaction between the blazar jet and its environment , or the variation from the geometric configuration of the jet . more intense monitoring of this system or further variability studies using a large sample of blazars are needed to better constrain the nature of their x - ray spectral variation . hao , h. , stanek , k. z. , dobrzycki , a. , matheson , t. , bentz , m. c. , kuraszkiewicz , j. , garnavich , p. m. , howk , j. c. , calkins , m. l. , worthey , g. , modjaz , m. , & serven j. 2006 , apj letter submitted , astro - ph/0612409 mori , k. , tsunemi , h. , miyata , e. , baluta , c. , burrows , d. n. , garmire , g. p. , & chartas , g. 2001 , in asp conf . ser . 251 , new century of x - ray astronomy , ed . h. inoue & h. kunieda ( san francisco : asp ) , 576 cccc 2000 - 06 - 26 & _ chandra _ & hetgs & 47471 + 2001 - 06 - 25 & _ chandra _ & hetgs & 51219 + 2004 - 03 - 10 & _ xmm - newton _ & none & 8328 + 2004 - 03 - 24 & _ xmm - newton _ & none & 27004 + 2004 - 05 - 05 & _ xmm - newton _ & none & 21356 + 2000 - 06 - 26 & a & @xmath27 & @xmath28 & 100.5(108 ) + 2000 - 06 - 26 & b & @xmath29 & @xmath30 & 129.6(106 ) + 2001 - 06 - 25 & a & @xmath31 & @xmath32 & 140.7(120 ) + 2001 - 06 - 25 & b & @xmath33 & @xmath34 & 113.8(116 ) + 2000 - 06 - 26 & a & @xmath27 & @xmath35 & 229.6(215 ) + 2000 - 06 - 26 & b & @xmath36 & & + 2001 - 06 - 25 & a & @xmath37 & @xmath38 & 255.2(237 ) + 2001 - 06 - 25 & b & @xmath39 & & + date & 2000 - 06 - 26 & 2001 - 06 - 25 & 2004 - 03 - 10 & 2004 - 03 - 24 & 2004 - 05 - 05 & + ( @xmath40 ) & @xmath41 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & @xmath46 + @xmath47 & @xmath48 & @xmath49 & @xmath50 & @xmath51 & @xmath52 & @xmath53 + unabsorbed flux & @xmath54 & @xmath55 & @xmath56 & @xmath57 & @xmath58 & @xmath59 + @xmath60 & 365.6(305 ) & 365.6(337 ) & 151.8(152 ) & 447.3(396 ) & 314.3(301 ) & 843.7(743 ) + ( @xmath61 ) & @xmath62 & @xmath63 & @xmath64 & @xmath65 & @xmath66 & @xmath67 + @xmath47 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath53 & @xmath51 + unabsorbed flux & @xmath72 & @xmath55 & @xmath56 & @xmath73 & @xmath58 & @xmath59 + @xmath60 & 369.9(305 ) & 365.6(337 ) & 157.6(152 ) & 484.5(396 ) & 327.3(301 ) & 899.3(743 ) + ( @xmath61 ) & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 + covering factor & 0.44 ( fixed ) & 0.44 ( fixed ) & @xmath80 & @xmath81 & @xmath82 & @xmath83 + @xmath47 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath87 & @xmath88 + unabsorbed flux & @xmath72 & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 + @xmath60 & 370.6(305 ) & 365.9(337 ) & 137.4(151 ) & 447.4(395 ) & 315.6(300 ) & 827.0(742 ) + @xmath94 ( @xmath40 ) & + ( @xmath61 ) & @xmath95 & @xmath75 & & & & @xmath96 + covering factor & + @xmath47 & @xmath97 & @xmath98 & & & & @xmath99 + @xmath60 & +	we present x - ray spectral analysis of five _ chandra _ and _ xmm - newton_observations of the gravitationally - lensed blazar pks 1830@xmath0211 from 2000 to 2004 . we show that the x - ray absorption toward pks 1830@xmath0211 is variable , and the variable absorption is most likely to be intrinsic with amplitudes of @xmath1@xmath2 depending on whether or not the absorber is partially covering the x - ray source . our results confirm the variable absorption observed previously , although interpreted differently , in a sequence of _ asca _ observations . this large variation in the absorption column density can be interpreted as outflows from the central engine in the polar direction , consistent with recent numerical models of inflow / outflows in agns . in addition , it could possibly be caused by the interaction between the blazar jet and its environment , or the variation from the geometric configuration of the jet . while the spectra can also be fitted with a variable absorption at the lens redshift , we show that this model is unlikely . we also rule out the simple microlensing interpretation of variability which was previously suggested .
magnetohydrostatic equilibrium describes the balance of the magnetic and thermal pressures across individual sunspots @xcite . along the direction parallel to the solar surface , the local environment of a sunspot markedly shifts from the cooler , convection inhibited , umbral region , hosting multi - kilogauss magnetic fields , to the convectively turbulent , hotter , external plasma of the quiet sun with considerably lower magnetic flux density . the mechanism responsible for this inverse relationship between magnetic field strength and temperature remains unclear @xcite since umbrae host a wide range of thermal - magnetic field distributions @xcite . interesting , however , is the empirical relationship between peak magnetic field strength and minimum intensity for different umbrae @xcite . support a single power law to express this relationship . as such , the total magnetic pressure of a sunspot always _ more than _ compensates for the corresponding reduction in thermal pressure , leading to a predictably larger spatial extent for umbrae with greater field strengths @xcite and lower minimum intensities @xcite . however , to date , no study of the temporal stability of the relationship claimed by norton and gilman has been completed for a statistically significant number of umbrae . we are interested in whether this equilibrium structure is influenced by solar cycle progression . the recent study of 3931 umbrae by revealed a solar - cycle oscillation in average umbral intensity between 1992 and 2003 . this agreed well with the mean intensity variation observed by as well as both the mean intensity and magnetic field changes reported by ; though , both of those reports neglected the necessary size parameter , as addressed by . importantly , the intensity oscillation reported by @xcite occurs without any evident corresponding umbral size variation . this thus agrees with the constant umbral size spectrum shape observed over multiple solar cycles by . since the observed relationship between intensity and size did not change , the reported mean intensity oscillation insinuates that the magnetic field - size relationship changes over the solar cycle if the relationship claimed by @xcite is indeed constant . a cycle variation in the umbral magnetic twist might influence this relationship , but has not yet been observed @xcite . as this is the only significant evidence for the local magnetic properties of sunspot umbrae being influenced by the evolution of the global solar field , further exploration is imperative as it may have implications to both the solar dynamo @xcite and umbral magnetic field orientation @xcite . it is worth noting , though , that reported a significantly lower number of sunspot groups with strong fields in odd numbered solar cycles compared to each cycle s preceding even numbered cycle pointing to a possible solar cycle influence on sunspot group generation . we address this topic more completely than previous authors by revisiting the data from the kitt - peak spectromagnetograms ( section [ s : data ] ) used by to exploit its measurements of magnetic field for the first time in tandem with the intensity measurements . automating our sunspot detection ( section [ s : selection ] ) yields the greatest number of umbrae so far measured for this type of study . we explore the temporal variability of the relationships between peak magnetic field strength , minimum intensity , and umbral size ( section [ s : results ] ) to discuss the alleged dependence of sunspot structure on solar cycle progression . we analyze 2771 full - disk daily maps of line - of - sight ( los ) magnetic flux density and continuum intensity produced by the nasa / nso spectromagnetograph ( spm ) @xcite between 1993 and 2004 at the kitt peak vacuum telescope ( kpvt ) . each map results from the on - line analysis of scanned long - slit spectra in both states of circular polarization centered on the 8688 fe i absorption line ( land g @xmath0 1.66 ) as described in . values for los magnetic flux and continuum intensity are therefore spatially and temporally synchronous . the full - disk maps have a field of view of @xmath1 and a pixel size of @xmath2 . los magnetic flux measurements from spm are utilized despite problems alluded to by . new observations from the spectropolarimeter onboard hinode @xcite as well as more precise infrared measurements @xcite have shown that many synoptic magnetographs have underestimated magnetic flux , especially within active regions . mdi underestimates los field values in the penumbra by almost a factor of two @xcite . values from spm compare well with gong+ , but both are 2040 percent lower than mdi @xcite . in accordance with , we multiply the spm magnetic flux values by 1.41 to calibrate their values with mdi . we proceed with our use of the spm magnetic flux values since they are crucial to the study at hand , but we are careful to compare their results with other measurements . as always , stray light is a considerable problem when discussing umbral intensity and it also can affect the determination of the magnetic field . we fully acknowledge that for a more precise description of the intensity of a sunspot , a stray light correction is necessary . daily stray - light corrections are unfortunately impossible with the spm for it is a scanning slit device in which the limb is observed at different times than the center of the disk . , though , studied the long term variation of stray - light within the instrument and found a very weak temporal trend ( about 0.5% ) . furthermore , as the spm measures the continuum at infrared wavelengths , intensity measurements are expected to be less affected by stray - light than instruments operating at visible wavelengths . thus , while stray light does affect our complete determination of the sunspot intensity , its effect is constant enough to allow us to comment on its solar cycle variability . we identify umbrae on the full - disk spm continuum maps using a thresholding method @xcite with a single definition for the umbra - penumbra boundary as in and . the contrast of features on each map is the ratio of the measured intensity to the local intensity . we define the local intensity using the legendre polynomial fits of observed continuum intensity with respect to the cosine of the heliocentric angle ( @xmath3 ) which are recorded in each data file header . this method reduces inconsistencies in determining the local photospheric intensity due to large multi - umbrae active regions as well as local brightenings due to diverted energy flux and plage regions around the sunspot @xcite . our umbra - penumbra boundary definition is set at a contrast value of 0.62 for the spm maps of continuum intensity near 8688 , which is slightly lower than that used by , 0.655 , and , 0.68 . this differs also from the variable umbra - penumbra boundary defined by , which will be discussed in section [ s : conclusions ] . our method for including complex spots is consistent with the approach used by other authors @xcite . multiple umbrae that share the same penumbra within single complex sunspots are treated as individual umbrae . we require these umbrae to be separated by at least one @xmath4 arcsec pixel with a penumbral intensity value , and each umbra must contain at least two observable pixels with umbral intensity values . our conservative intensity threshold value counts connected umbrae as one . furthermore , we find only 4.5% of the umbrae selected to be closely spaced , where we consider closely spaced to mean the smaller umbrae of a two - umbrae pair is less than two times its radius away from the nearest pixel of the larger umbra . we , in general , do not attempt to locate umbrae that are close to the limb ( @xmath5 ) since magnetic field values are less reliable , projection effects are difficult to remove consistently , and the wilson depression conceals a portion of the umbrae . we also verify the selection of each umbra using the synchronous maps of magnetic field . we require that each umbra have a peak magnetic flux greater than 50 gauss , which is more than three sigma deviations from the background noise of the spm magnetograms . in total , we locate 12967 umbrae between 1993 and 2003 . for each umbra we glean values for the umbral area and radius , as well as the minimum continuum intensity and its corresponding magnetic flux density . the umbral area simply corresponds to its spatial coverage by counting the number of pixels observed and correcting for angular foreshortening . we calculate an effective umbral radius ( later referred to as simply the radius ) by assuming that each umbra is circular . using the solar ephemeris for each individual spm map , angular units are converted into physical units ( 10@xmath6 km ) . the use of microhemispheres in section [ s : size_dist ] is also corrected for the annual variation in the apparent solar semidiameter . all values of los magnetic flux and continuum intensity presented here reflect the value measured at the darkest point in each sunspot umbra , which nearly always corresponds to the peak umbral magnetic field strength . this dark umbral core evolves the least during its formation and evolution ( socas - navarro 2003 ) and is least affected by umbral dots , which affect the intensity distribution of the umbra but only minimally affect the determined umbral radius . furthermore , in the dark umbral core the magnetic field is known to be the most vertically directed which allows us to adopt the vertical field approximation , @xmath7 where @xmath8 . the heliocentric angle , @xmath9 , is the angular distance between the observer s line of sight and the local radial vector . in choosing to measure minimum intensity ( * i * ) , corresponding magnetic flux density ( * b * ) , and umbral size ( * r * ) , we parameterize a much more complicated system into merely three interdependent variables . we are interested not only in whether the interdependencies of these variables change over the solar cycle , but also whether the average umbral intensity and magnetic field vary as suggested by . for this , we begin with a size spectrum analysis of the observed umbrae to address whether the solar cycle influences the distribution of umbrae generated at a particular phase . a regression analysis is then performed to test the temporal stability of the b r , and i r relationships . we finish by showing the temporal variation of the mean intensity and magnetic field as in @xcite and carefully use the relationships derived to show that they are consistent with a temporal variation in size . as a much greater number of smaller umbrae appear on the solar disk than larger umbrae , our regression analysis requires binning to ensure that the smaller spots do not dominate the information available in larger spots . thus , prior to performing any regression analysis , the data is binned using the horizontal ( independent ) variable into bins of equal width . using the levenberg - marquardt least - squares procedure , the regression is derived between the mean values of the independent and dependent variables . the standard error of each mean is also carefully propagated into the analysis and is reflected in the regression error . an analysis of the spectrum of umbral sizes has been carried out in the manner described by . it is important to address the spectrum of umbral sizes over the solar cycle rather than merely the average size . deriving the size spectrum of umbrae has traditionally been used to comment on the generating mechanism for sunspot umbrae , whether that be flux - tube fragmentation or coalescence . in a similar fashion , we utilize the size spectrum as a means to explore the influence of the solar cycle on the generation of umbrae . as @xcite remark , however , the spectrum derived is not identical to the distribution of umbral areas at a particular time . rather , it is the statistical probability of locating a spot of a particular size on the solar disk at any given time . thus , it can not account for daily , monthly , or even annual variations of the mean umbral size , especially during solar minimum when solar activity is reduced . furthermore , unlike mere averaging of umbral sizes , the spectrum is influenced less by the ability to consistently select small spots ( comparable in size to the atmospheric seeing limit ) . the report of gave convincing evidence for a stable size spectrum , independent of cycle phase , between 1917 - 1982 . though , since we are studying a solar cycle not included in that study , it is important to address the spectrum observed for we can not assume one cycle is identical to another . for our selected umbrae , we follow nearly the same procedure as @xcite to determine the log - normal distribution defined by @xmath10 where a is the umbral area ( in microhemispheres ) , @xmath11 is the spectrum mean , @xmath12 is the geometric standard deviation , and dn / da is the size distribution function . the distribution function and its error , which is conservatively taken to be the square root of the number of umbrae in each bin , is divided by the number of maps measured as in the @xcite study . we employ the levenberg - marquardt least - squares fit procedure to determine the fit parameters for our observed size distribution over different portions of the solar cycle . the error in the observed distribution is propagated into the fit . .size spectrum lognormal fit parameters [ cols="<,^,>,>,>,>,>",options="header " , ] the minimum umbral intensity versus umbral radius is presented in figure [ fig : size_int ] . the data are binned by radius into bins of equal width ( 250 km ) before any regression analysis is attempted . thus , each data point represents the mean value of the umbral intensities in a given bin . the small hatted error bars give the standard error of each mean while the larger unhatted error bars show the observed spread of the data by illustrating the 10- to 90- percentile range of the data . the standard error of each mean is typically small and is propagated into the regression analysis . we find that a 2nd order polynomial , as opposed to a power law relation , lowers the reduced chi - square of the fit . similar to the previous sections , this analysis is performed for various portions of the solar cycle as well as the whole data set . no significant variation of the relationship is found over the solar cycle , which agrees with the analysis of and . for a quadratic fit of the form @xmath13 where r is the umbral radius in seconds of arc and i@xmath14 is normalized to the surrounding photosphere , we find these best fit parameters for r @xmath15 13000 km : @xmath16 , j@xmath17 , and k@xmath18 . the numeric values given in parentheses reflect the one sigma deviation of the fit parameters . the relationship is also plotted in figure [ fig : size_int ] . additionally , the functions determined by and are provided in the figure . as each of these studies observes the continuum intensity at different wavelengths , the functions given by the authors have been altered using the planck function to reflect an lte approximation for the expected intensity of each previous study at the kpvt / spm wavelength ( see section [ s : mag_int ] ) . we find some consistency between the kpvt / spm data and the stray - light corrected mdi data analyzed by . for larger umbrae , the kpvt / spm values are larger than the mdi values suggesting a greater influence of stray light on these umbrae . for smaller umbrae , kpvt / spm gives a lower mean minimum intensity than the mdi data . the differences may be due to stray light effects , or different umbral radius threshold intensities , which change with wavelength as well as different author s determinations . we do not find a meaningful comparison between the presented data and the function provided by . finally , we address the previous reports of a temporal trend in the mean minimum intensity and mean magnetic field strength by examining these parameters as a function of time in the current data set . the filled diamonds in fig . [ fig : cycle_evol ] display the mean value of magnetic flux ( top ) and minimum intensity ( bottom ) for annual bins between 1993 and 2004 . the error bars give the standard error of each mean . the oscillation in mean minimum umbral intensity as reported by remains evident in our study though the oscillatory amplitude is reduced . a similar oscillation is notable in the mean magnetic field strength in the top plot . although @xcite found no significant temporal variation in the mean umbral size of their selection , our larger sample size reveals that the selection does exhibit a mean umbral size variation , as shown in figure [ fig : cycle_evol ] . while the variation is smaller than the error in the size spectrum analysis given in section [ s : size_dist ] , different bins show significant deviation from others . putting this to task , we evaluate the expected variation in the umbral minimum intensity and corresponding maximum magnetic flux with time based only on the umbral size and the previously determined time - independent relationships ( equations [ eq : size_mag ] and [ eq : size_int ] ) . the correlation coefficient between the measured values of intensity and the calculated ones is 0.91 . between the observed magnetic flux values and the calculated ones ( excluding error ) , the correlation coefficient is 0.65 . we plot these values in figure [ fig : cycle_evol ] as filled squares . the error bars associated with these calculated values reflect both the standard error of the annual bin as well as the one sigma error in the determined relationship between umbral radius and the two parameters shown . the one sigma error for the real values and the calculated values are comparable in the figure . for the mean minimum umbral intensity oscillation , we find that changes in umbral radius can account for the perceived oscillation . and for the magnetic field , all the values calculated from the umbral radii ( except 2003 ) are within the one sigma error of the measured values . the unchanging relationships of b r and i r fully account for the variations of mean umbral intensity and mean magnetic flux over the solar cycle due to a varying mean umbral radius . thus , the mean maximum magnetic field strength for the observed umbrae as well as the mean minimum umbral intensity do not give evidence for a solar cycle change in the equilibrium balance between sunspot umbrae and the quiet sun . our analysis of more than 10000 umbrae observed between 1993 and 2003 gives insight into the temporal stability of bulk umbral structure over the solar cycle , which has been debated since @xcite s initial report in 1978 of a cycle dependent change in sunspot intensity . we find the following : 1 . the relationships of b r , and i r exhibit power law , linear , and quadratic relationships , respectively , and display no significant modulation during the solar cycle . 2 . minimum intensity measurements support umbral limb - darkening observed in the same spectral region by and can be corrected . the umbral size spectrum of the selected umbrae agrees with the conclusion of . the umbra size spectrum shape remains constant in time . the periodic variation of the mean umbral intensity and magnetic flux density over the solar cycle can be explained by a corresponding , yet nearly insignificant , change in mean umbral radius . our last result directly conflicts with who reported a solar cycle oscillation in mean umbral intensity without a corresponding mean umbral radius oscillation . that claim relied solely upon no variation observed within annually binned data whose standard errors were large primarily due to a lower number of selected umbrae . we expand the selection and address the dependencies on umbral radius directly , unlike all reports claiming a solar cycle variation in umbral intensity and/or magnetic field @xcite . we also test whether our umbral limb - darkening correction and the change in the umbra - penumbra boundary convention affect our result , and no change is evident . we find the unvarying relationship claimed by to be a fairly consistent representation of the bulk magnetohydrostatic equilibrium for sunspots over the solar cycle ; however , we also find that considerable scatter plagues the robust use of this relationship even for infrared measurements . along with the lack of a significant change in the size distribution of umbrae generated by the sun , we must conclude that no compelling evidence has yet been found to suggest that the equilibrium balance and/or mean umbral strength is influenced by the progression of the solar cycle . though , we add that significant annual variations in the mean umbral radius , as testified by the previously reported oscillations in minimum umbral intensity and maximum magnetic field , do exist . this claim is further supported by who studied the ratio of the total daily sunspot area to the number of sunspots between 1874 to 2004 . this ratio exhibited a prominent periodicity consistent and in phase with the solar cycle . both of these claims may point to an amplification phase of the toroidal field strength followed by a decay phase . still , however , no change in the generated size distribution , as described by , has been found and the oscillations seen in this paper are not large enough to enact a significant change in the observed size spectrum . we express thanks to bill livingston for providing his infrared measurements for this work . thanks are also extended to harry jones with whom we discussed the operation and measurements of the kpvt spectromagnetograph and jack harvey for his careful reading of the manuscript . we also thank the anonymous referee for helpful suggestions . the nso / kitt peak data used here were produced cooperatively by nsf / noao , nasa / gsfc and noaa / sec . this work was supported by the national solar observatory . alfvn , h. : 1943 , _ ark . mat . fys . _ * 29 * , 1 . albregtsen , f. , maltby , p. : 1978 , _ nature _ * 274 * , 41 . albregtsen , f. , maltby , p. : 1981 , _ solar phys . _ * 71 * , 269 . albregtsen , f. , joras , p.b . , maltby , p. : 1984 , _ solar phys . _ * 90 * , 17 . avrett , e.h . : 1981 , in : cram , l.e . , thomas , j.h . _ the physics of sunspots _ , aura , sacramento peak observatory , sunspot , 235 . biermann , l. 1941 , _ vierteljahrsschr . ges . _ * 76 * , 194 . bogdan , t.j . , gilman , p.a . , lerche , i. , howard , r. : 1988 , _ astrophys . j. _ * 327 * , 451 . brandt , p.n . , schmidt , w. , steinegger , m. : 1990 , _ solar phys . _ * 129 * , 191 . brants , j.j . , zwaan , c. : 1982 , _ solar phys . _ * 80 * , 251 . chou , d .- y . : 1987 , _ astrophys . j. _ * 312 * , 955 . deinzer , w. : 1965 , _ astrophys . j. _ * 141 * , 548 . giovanelli , r.g . , slaughter , c. : 1978 , _ solar phys . _ * 57 * , 255 . jones , h.p . , duvall , t.l . , jr . , harvey , j.w . , mahaffey , c.t . , schwitters , j.d . , simmons , j.e . : 1992 , _ solar phys . _ * 139 * , 211 . jones , h.p . : 1996 , spie * 2804 * , 110 . jones , h.p . , branston , d.d . , jones , p.b . , wills - davey , m.j . : 2000 , _ astrophys . j. _ * 529 * , 1070 . jones , h.p . , ceja , j.a . : 2001 , in sigwarth , m. ( ed . ) , _ advanced solar polarimetry theory , observation , and instrumentation _ , cs-236 , astron . , san francisco , 87 . kopp , g. , rabin , d. : 1992 , _ solar phys . _ * 141 * , 253 . leonard , t. , choudhary , d.p . : 2008 , _ solar phys . _ * 252 * , 33 . li , k.j . , qiu , j. , su , t.w . , gao , p.x . : 2005 , _ astrophys . j. _ * 621 * , l81 . livingston , w. : 2002 , _ solar phys . _ * 207 * , 41 . livingston , w. , harvey , j.w . , malanushenko , o.v . , webster , l. : 2006 , _ solar phys . _ * 239 * , 41 . livingston , w. : 2009 , private communication . maltby , p. , avrett , e.h . , carlsson , m. , kjeldseth - moe , o. , kurucz , r.l . , loeser , r. : 1986 , _ astrophys . j. _ * 306 * , 284 . martnez pillet , v. , vazquez , m. : 1993 , _ astron . astroph . _ * 270 * , 494 . mathew , s.k . , martnez pillet , v. , solanki , s.k . , krivova , n.a . : 2007 , _ astron . astroph . _ * 465 * , 291 . moon , y .- j . , kim , y .- h . , park , y .- d . , ichimoto , k. , sakurai , t. , chae , j. , cho , k.s . , bong , s. , suematsu , y. , tsuneta , s. , katsukawa , y. , shimojo , m. , shimizu , t. , shine , r.a . , tarbell , t.d . , title , a.m. , lites , b. , kubo , m. , nagata , s. , yokoyama , t. : 2007 , _ publ . . japan _ * 59 * , 625 . norton , a.a . , gilman , p.a . : 2004 , _ astrophys . j. _ * 603 * , 348 . penn , m.j . , livingston , w. : 2006 , _ astrophys . j. _ * 649 * , l45 . penn , m.j . , macdonald , r.k.d . : 2007 , _ astrophys . j. _ * 662 * , l123 . pevtsov , a.a . , canfield , r.c . , latushko , s.m . : 2001 , _ astrophys . j. _ * 549 * , l261 . rast , m.p . , fox , p.a . , lin , h. , lites , b.w . , meisner , r.w . , white , o.r . : 1999 , _ nature _ * 401 * , 678 . redi , i. , solanki , s.k . , livingston , w. : 1995 , _ astron . astroph . _ * 302 * , 543 . schlter , a. , temesvry , s. : 1958 , _ electromagnetic phenomena in cosmical physics _ * 6 * , 263 . schuessler , m. : 1980 , _ nature _ * 288 * , 150 . thornton , c.e . , jones , h.p . : 2002 , _ bull . am . * 288 * , 1243 . wesolowski , m.j . , walton , s.r . , chapman , g.a . : 2008 , _ solar phys . _ * 248 * , 141 .	measurements of maximum magnetic flux , minimum intensity , and size are presented for 12967 sunspot umbrae detected on the nasa / nso spectromagnetograms between 1993 and 2004 to study umbral structure and strength during the solar cycle . the umbrae are selected using an automated thresholding technique . measured umbral intensities are first corrected for a confirming observation of umbral limb - darkening . log normal fits to the observed size distribution confirm that the size spectrum shape does not vary with time . the intensity magnetic flux relationship is found to be steady over the solar cycle . the dependence of umbral size on the magnetic flux and minimum intensity are also independent of cycle phase and give linear and quadratic relations , respectively . while the large sample size does show a low amplitude oscillation in the mean minimum intensity and maximum magnetic flux correlated with the solar cycle , this can be explained in terms of variations in the mean umbral size . these size variations , however , are small and do not substantiate a meaningful change in the size spectrum of the umbrae generated by the sun . thus , in contrast to previous reports , the observations suggest the equilibrium structure , as testified by the invariant size - magnetic field relationship , as well as the mean size ( _ i.e. _ strength ) of sunspot umbrae do not significantly depend on solar cycle phase .
quantum teleportation is an important and vital quantum information processing task where an arbitrary unknown quantum state can be replicated at a distant location using previously shared entanglement and classical communication between the sender and the receiver . a remarkable application of entangled states having many ramifications in information technology , quantum teleportation can also be combined with other operations to construct advanced quantum circuits useful for information processing@xcite . the original teleportation protocol of bennett et al.@xcite for an unknown qubit using an epr pair has been generalized to the case of non - maximally entangled or a noisy channel between the sender and the receiver@xcite . the loss of fidelity for teleportation using non - maximally entangled channels could be compensated by schemes for probabilistic teleportation@xcite . the first experimental demonstration of quantum teleportation was reported by bouwmeester et al.@xcite . quantum teleportation is also possible for systems corresponding to infinite dimensional hilbert spaces@xcite . the teleportation process for continuous variables was originally formulated in terms of wigner functions@xcite and has also been extended in terms of characteristic functions@xcite of the quantum systems involved . schemes for obtaining optimal fidelity of teleportation using gaussian@xcite as well as non - gaussian@xcite resource states have been devised . the first experiment of continuous variable teleportation was performed by furusawa et al.@xcite . since then there have been further improvements in the fidelity of teleportation obtained in experiments@xcite . recently , an experimental characterization of continuous variable quantum communication channels has been established by shared entanglement together with local operations and classical communications@xcite . since quantum entanglement is fragile and is easily destroyed in distribution , establishing entanglement between quantum systems at distant locations , and transporting entanglement from one location to another are rather challenging tasks . various ingenious methods have been proposed to accomplish these , such as by using entanglement swapping protocols@xcite , quantum repeaters by combining operations of swapping with entanglement purification@xcite , and by continuous measurements@xcite . for continuous variable systems , some protocols for entanglement swapping@xcite , establishing entanglement between distant stations through teleportation@xcite , testing the efficiency of teleportation with the aid of a third party@xcite , and combining teleportation with cloning@xcite have been proposed . but no protocol for the explict teleportation of an entangled continuous variable state exists in the literature , akin to a similar scheme for discrete variables for teleporting a two - qubit entangled state , that has been presented recently@xcite . the aim of this work is to propose an explicit scheme for the teleportation of an unknown two - mode entangled state of continuous variables from one party ( alice ) to the other distant party ( bob ) . for this purpose we first show how an entangled state of four modes can be generated and shared by alice and bob with the help of linear amplifiers and beam - spitters . our protocol for teleportation can then proceed in the usual way with alice making measurements on her side and communicating their results classically to bob who in turns makes a local operation to obtain the teleported entangled two - mode state . the communication of four bits of information from alice to bob is required , similar to the case of the protocol for teleporting entangled states of two qubits@xcite . we compute the entanglement of the teleported state with bob , and also the fidelity of teleportation as functions of the squeezing parameters of the states generated by the source and the teleportation amplifiers . a trade - off between the entanglement of the teleported state and the fidelity of teleportation is observed with respect to the squeezing parameters . ) to be teleported . bob has two amplifiers ta1 and ta2 and two beam - splitters bs1 and bs2 using which he generates a four - mode state ( @xmath0 ) . he keeps two of these modes @xmath1 and @xmath2 with himself , and sends the remaining two modes @xmath3 and @xmath4 to alice . using the beam - splitters bs3 and bs4 alice combines her modes @xmath5 and @xmath6 with those sent by bob , and performs four measurements on the output modes @xmath7 . she then communicates her results to bob who uses these to apply a unitary transformation to displace the modes @xmath1 and @xmath2 . the final teleported state is found in the modes @xmath8 and @xmath9.,width=302 ] [ f1 ] our protocol is as follows . let alice hold the source parametric amplifier sa3 whose two output entangled modes are to be teleported . bob possesses two teleportation amplifiers ta1 and ta2 which are required as compulsory accessories for our protocol of teleportation of two - mode entangled states . alice s task is to teleport the entangled state of the modes @xmath5 and @xmath6 as signal and idler originating from her sa3 to bob . but prior to this one requires to set up a four - mode entangled state to be shared by alice and bob . for this purpose , consider the modes @xmath10 and @xmath11 coming out of the amplifier ta1 , and the modes @xmath12 and @xmath13 coming out of ta2 . the covariance matrix of the four modes @xmath14 is given by @xmath15 , with @xmath16 , and we assume that the two amplifiers are similar , i.e. , @xmath17 with @xmath18 where @xmath19 with @xmath20 being the squeezing parameter and @xmath21 the amplifier phase . bob uses two beam - splitters ( bs1 and bs2 ) represented by the matrix @xmath22 as @xmath23 where , @xmath24 is a @xmath25 identity matrix , to mix the modes ( @xmath26 ) and ( @xmath27 ) , respectively . the cm corresponding to the four output modes ( @xmath0 ) of the two beam - splitters are given by @xmath28 bob then supplies the modes @xmath4 and @xmath3 to alice and keeps the remaining modes @xmath1 and @xmath2 with himself . hence , alice and bob share a four - mode entangled state to be used for the teleportation protocol . alice has the two entangled modes @xmath5 and @xmath6 originating from her source amplifier sa3 to teleport , in addition to the two modes @xmath4 and @xmath3 which bob has sent her . the combined six - mode state ( four modes with alice and two with bob ) are represented by the cm @xmath29 where @xmath30 , @xmath31 , @xmath32 , @xmath33 , with @xmath34 being the squeezing parameter of the two - mode state ( @xmath5 and @xmath6 ) with alice , and @xmath35 being the phase of amplifier sa3 . to proceed further with the teleportation protocol , alice uses two beam - spiltters bs3 and bs4 represented by @xmath36 to combine the modes ( @xmath4 and @xmath5 ) and ( @xmath3 and @xmath6 ) , respectively in order to obtain the four output modes ( @xmath7 ) . alice then makes measurements on these four modes . without loss of generality , let us assume that her choice of measurements leads to the results ( @xmath37 ) , respectively , which she communicates to bob . bob then uses these measurement results to displace the state of the modes @xmath1 and @xmath2 with him , by applying the unitary transformation @xmath38 to obtain finally the modes @xmath8 and @xmath9 . the gains @xmath39 on the classical measurements have been chosen such that the resultant matrix @xmath40 ( with @xmath41 defined below ) has all the elements either @xmath42 , @xmath43 or @xmath44 . the combined physical processes ( the beam splitters used by alice , the measurements performed by alice , and the unitary transformation performed by bob ) takes the cm @xmath45 to @xmath46 where @xmath47 incorporates the measurements made by alice . note that the measurements on the four modes @xmath48 with the choice of outputs @xmath49 that we have made , corresponds to integrating out these variables , and hence in the cm ( [ measure ] ) the corresponding rows are deleted . thus , the teleported signal and idler can be found as the modes @xmath8 and @xmath9 with bob , represented by the cm @xmath50 where @xmath51 the cm @xmath52 can also be expressed as @xmath53 where @xmath54 and @xmath55 is the @xmath56 identity matrix . for @xmath57 , and in the limit @xmath58 , one obtains @xmath59 . it can be shown that the gaussian state with cm @xmath60 and the gaussian state with cm @xmath61 are equivalent under local linear unitary bogoliubov operations ( llubos)@xcite . thus , in the limit of ideal input squeezing , the two - mode eintangled state is teleported perfectly . further , if the output states from the amplifiers ta1 and ta2 are coherent states , i.e. , @xmath62 , then the variances of the modes @xmath5 and @xmath6 are increased by twice the level of the vacuum noise . thus our teleportation protocol in this special case reproduces the results obtained by tan@xcite for a scheme of teleporting a single - mode state . for the final teleported state is plotted versus the squeezing @xmath34 of the two - mode state generated from the source amplifier sa3 ( @xmath63-axis ) , and the squeezing @xmath20 of the two two - mode states from the teleportation amplifiers ta1 and ta2 , respectively ( @xmath64-axis).,width=340 ] [ f2 ] in order to check that the teleported modes @xmath8 and @xmath9 indeed represent an entangled pair , we compute the symplectic eigenvalues of the partial transpose of the cm @xmath52 given by eq.([covmat5 ] ) . for the modes @xmath8 and @xmath9 to be entangled , the smallest symplectic eigenvalue of @xmath65 has to be less than one@xcite , i.e. , @xmath66 . for simplicity , we assume that the phase of the amplifier sa3 is chosen such that @xmath67 and @xmath68 , and similarly , the phases of ta1 and ta2 are such that @xmath69 . then @xmath70 is given by @xmath71 the magnitude of entanglement in the teleported state is given by the logarithmic negativity defined as @xmath72 \label{entang}\ ] ] @xmath73 is plotted in fig.2 versus the squeezing parameters @xmath34 corresponding to the two - mode state @xmath74 to be teleported , and @xmath20 corresponding to the states @xmath75 and @xmath76 originating from the amplifiers ta1 and ta2 respectively . one sees that the magnitude of entanglement of the teleported output two - mode state obtained by bob goes up with increased squeezing of the states coming from both the source and the teleportation amplifiers . is plotted versus the squeezing @xmath34 of the state to be teleported ( @xmath63-axis ) , and the squeezing @xmath20 of the output modes of the amplifiers ta1 and ta2 ( @xmath64-axis).,width=340 ] [ f3 ] we finally compute the fidelity of the teleported entangled state . the fidelity can be obtained from the expression@xcite given by @xmath77+\delta}-\sqrt{\delta } } \label{fidel}\ ] ] where @xmath78 - \frac{1}{4})(\mathrm{det}[\sigma_{out } ] - \frac{1}{4 } ) \label{defs3}\ ] ] with @xmath79 and @xmath80 in the present case given by eqs.([covmat3 ] ) and ( [ covmat5 ] ) respectively . we plot the fidelity of teleportation versus the squeezing parameters in fig.3 . we notice that a maximum fidelity of @xmath81 is possible in this scheme for an initial two - mode coherent state generated by the source amplifier . the fidelity stays nearly constant with variation of the squeezing of the states from the teleportation amplifiers . thus coherent states generated by the teleportation amplifiers can also be used to implement this protocol . however , increased squeezing of the initial state from the source amplifier leads to the loss of fidelity . the latter feature is in sharp contrast to the magnitude of entanglement of the final two - mode state with bob , which increases with the increase of squeezing of alice s two - mode state . this result seems to support the contention of johnson et al.@xcite that the average fidelity may not be a good indicator of the quality of quantum teleportation . though their results were obtained in the context of teleportation of single modes together with the physical transport of modes between three parties@xcite , it might be more appropriate to use the measure of ` entanglement fidelity'@xcite to quantify , where possible , the ability of a process to preserve entanglement . to summarize , we have presented the first explicit protocol for teleportation of two - mode entangled squeezed states . our scheme is accomplished by the creation of a four - mode state shared initially by two distant parties through beams generated by two teleportation amplifiers and combined by two beam splitters . teleportation takes place with the usual local measurements , unitary operations and the classical communication of four bits parallel to the case involving entangled states of discrete variables@xcite . if coherent states from teleportation amplifiers are used to create the four - mode state shared by the two parties for enabling teleportation , the variance of the entangled two - mode teleported state increases by twice the level of the vacuum noise . in general , the entanglement obtained for the teleported state increases with the squeezing of the initial two - mode state which however leads to loss of fidelity of teleportation . it might be worthwhile to explore other schemes of teleportation to check whether fidelity of teleported entangled states could be substantially incresed . m. a. nielsen and i. l. chuang , _ quantum computation and qauntum information _ , ( cambridge university press , cambridge , 2000 ) ; s. l. braunstein and a. k. pati , _ quantum information with continuous variables _ , ( kluwer academic publishers , dordrecht , 2003 ) .	we consider two - mode squeezed states which are parametrized by the squeezing parameter and the phase . we present a scheme for teleporting such entangled states of continuous variables from alice to bob . our protocol is operationalized through the creation of a four - mode entangled state shared by alice and bob using linear amplifiers and beam splitters . teleportation of the entangled state proceeds with local operations and the classical communication of four bits . we compute the fidelity of teleportation and find that it exhibits a trade - off with the magnitude of entanglement of the resultant teleported state . 0.5 cm
we consider the semilinear singularly perturbed problem @xmath1 , \label{uvod1 } \\ y(0)&=0,\ y(1)=0 , \label{uvod2}\end{aligned}\ ] ] where @xmath2 we assume that the nonlinear function @xmath3 is continuously differentiable , i.e. that @xmath4\times \mathbb{r}\right)$ ] , for @xmath5 and that @xmath3 has a strictly positive derivative with respect to @xmath6 @xmath7\times \mathbb{r}\:\ : \quad ( m = const ) . \label{uvod3}\ ] ] the solution @xmath6 of the problem exhibits sharp boundary layers at the endpoints of @xmath8 $ ] of @xmath9 @xmath10 width . it is well known that the standard discretization methods for solving are unstable and do not give accurate results when the perturbation parameter @xmath0 is smaller than some critical value . with this in mind , we therefore need to develop a method which produces a numerical solution for the starting problem with a satisfactory value of the error . moreover , we additionally require that the error does not depend on @xmath0 ; in this case we say that the method is uniformly convergent with respect to @xmath0 or @xmath0-uniformly convergent . numerical solutions @xmath11 of given continuous problems obtained using a @xmath0-uniformly convergent method satisfy the condition @xmath12 where @xmath6 is the exact solution of the original continuous problem , @xmath13 is the discrete maximum norm , @xmath14 is the number of mesh points that is independent of @xmath0 and @xmath15 is a constant which does not depend on @xmath14 or @xmath0 . we therefore demand that the numerical solution @xmath11 converges to @xmath6 for every value of the perturbation parameter in the domain @xmath16 with respect to the discrete maximum norm @xmath17 the problem has been researched by many authors with various assumptions on @xmath18 . various different difference schemes have been constructed which are uniformly convergent on equidistant meshes as well as schemes on specially constructed , mostly shishkin and bakvhvalov - type meshes , where @xmath0-uniform convergence of second order has been demonstrated , see e.g. @xcite , as well as schemes with @xmath0-uniform convergence of order greater than two , see e.g. @xcite . these difference schemes were usually constructed using the finite difference method and its modifications or collocation methods with polynomial splines . a large number of difference schemes also belongs to the group of exponentially fitted schemes or their uniformly convergent versions . such schemes were mostly used in numerical solving of corresponding linear singularly perturbed boundary value problems on equidistant meshes , see e.g. @xcite . less frequently were used for numerical solving of nonlinear singularly perturbed boundary value problems , see e.g. @xcite . our present work represents a synthesis of these two approaches , i.e. we want to construct a difference scheme which belongs to the group of exponentially fitted schemes and apply this scheme to a corresponding nonequidistant layer - adapted mesh . the main motivation for constructing such a scheme is obtaining an @xmath0-uniform convergent method , which will be guaranteed by the layer - adapted mesh , and then further improving the numerical results by using an exponentially fitted scheme . we therefore aim to construct an @xmath0-uniformly convergent difference scheme on a modified shishkin mesh , using the results on solving linear boundary value problems obtained by roos @xcite , oriordan and stynes @xcite and green s function for a suitable operator . this paper has the following structure . section [ sec1 ] . provides background information and introduces the main concepts used throughout . in section [ sec2 ] . we construct our difference scheme based on which we generate the system of equations whose solving gives us the numerical solution values at the mesh points . we also prove the existence and uniqueness theorem for the numerical solution . in section [ sec3 ] . we construct the mesh , where we use a modified shiskin mesh with a smooth enough generating function in order to discretize the initial problem . in section [ sec4 ] . we show @xmath0-uniform convergence and its rate . in section [ sec5 ] . we provide some numerical experiments and discuss our results and possible future research . * notation . * throughout this paper we denote by @xmath19 ( sometimes subscripted ) a generic positive constant that may take different values in different formulae , always independent of @xmath14 and @xmath0 . we also ( realistically ) assume that @xmath20 . throughout the paper , we denote by @xmath21 the usual discrete maximum norm @xmath22 as well as the corresponding matrix norm . consider the differential equation ( [ uvod1 ] ) in an equivalent form @xmath23 , \label{konst1}\ ] ] where @xmath24 and @xmath25 is a chosen constant . in order to obtain a difference scheme needed to calculate the numerical solution of the boundary value problem , using an arbitrary mesh @xmath26 we construct a solution of the following boundary value problem @xmath27 for @xmath28 it is clear that @xmath29,\ i=0,1,\ldots , n-1.$ ] the solutions of corresponding homogenous boundary value problems @xmath30 for @xmath31 , are known , see @xcite , i.e. @xmath32 for @xmath31 , where @xmath33,\:\beta=\frac{\sqrt{\gamma}}{\epsilon},\:h_i = x_{i+1}-x_i.$ ] the solution of is given by @xmath34,\ ] ] where @xmath35 is the green s function associated with the operator @xmath36 on the interval @xmath37 $ ] . the function @xmath35 in this case has the following form @xmath38 where @xmath39 clearly @xmath40 $ ] . it follows from the boundary conditions ( [ konst32 ] ) that @xmath41 @xmath42 @xmath43 hence , the solution @xmath44 of on @xmath45 $ ] has the following form @xmath46 the boundary value problem @xmath47 has a unique continuously differentiable solution @xmath48 . since @xmath49 on @xmath50 $ ] , @xmath31 , we have that @xmath51 , for @xmath52 using this in differentiating ( [ konst7 ] ) , we get that @xmath53+y_{i+1}\left[-\left ( u_i^{ii}\right)'(x_i ) \right ] & \\ & = \dfrac{\partial}{\partial x}\left[\int_{x_i}^{x_{i+1}}{g_i(x , s)\psi(s , y(s))ds}-\int_{x_{i-1}}^{x_{i}}{g_{i-1}(x , s)\psi(s , y(s))ds } \right]_{x = x_i}. & \label{konst10}\end{aligned}\ ] ] since we have that @xmath54 equation becomes @xmath55,\end{aligned}\ ] ] for @xmath56 and @xmath57 . we can not in general explicitly compute the integrals on the rhs of ( [ konst13 ] ) . in order to get a simple enough difference scheme , we approximate the function @xmath58 on @xmath59\cup \left[x_i , x_{i+1 } \right]$ ] using @xmath60 where @xmath61 are approximate values of the solution @xmath6 of the problem ( [ uvod1])([uvod2 ] ) at points @xmath62 we get that & _ i-1 - ( + ) _ i + _ i+1 & + & = , + & = , for @xmath56 and @xmath63 . using equation , we get that @xmath64 for @xmath56 and @xmath63 , where @xmath65 using the scheme ( [ konst17 ] ) we form a corresponding discrete analogue of f_0&:=_0=0 , & [ konst18a ] + f_i&:= = 0 , & [ konst18b ] + f_n&:=_n=0 , [ konst18c ] where @xmath56 . the solution @xmath66 of the problem , i.e. @xmath67 where @xmath68 is an approximate solution of the problem . [ teo21 ] the discrete problem ( [ konst18a])([konst18c ] ) has a unique solution @xmath11 for @xmath69 . also , for every @xmath70 we have the following stabilizing inequality @xmath71 we use a technique from @xcite and @xcite , while the proof of existence of the solution of @xmath72 is based on the proof of the relation : @xmath73 where @xmath74 is the frchet derivative of @xmath75 . the frchet derivative @xmath76 is a tridiagonal matrix . let @xmath77.$ ] the non - zero elements of this tridiagonal matrix are h_0,0=&h_n , n=1 , + h_i , i=&0 , + h_i , i+1=&>0 , where @xmath78 hence @xmath79 is an @xmath80 matrix . moreover , @xmath79 is an @xmath81matrix since & \|h_i , i\|-\|h_i , i-1\|-\|h_i-1,i\| = & + & 4 m . & consequently @xmath82 using hadamard s theorem ( see e.g. theorem 5.3.10 from @xcite ) , we get that @xmath75 is an homeomorphism . since clearly @xmath83 is non - empty and @xmath84 is the only image of the mapping @xmath75 , we have that has a unique solution . + the proof of second part of the theorem [ teo21 ] is based on a part of the proof of theorem 3 from @xcite . we have that @xmath85 for some @xmath86 . therefore @xmath87 and finally due to inequality we have that @xmath88 since the solution of the problem changes rapidly near @xmath89 and @xmath90 , the mesh has to be refined there . various meshes have been proposed by various authors . the most frequently analyzed are the exponentially graded meshes of bakhvalov , see @xcite , and piecewise uniform meshes of shishkin , see @xcite . here we use the smoothed shishkin mesh from @xcite and we construct it as follows . let @xmath91 be the number of mesh points and @xmath92 and @xmath93 are mesh parameters . define the shishkin mesh transition point by @xmath94 let us chose @xmath95 for simplicity in representation , we assume that @xmath96 , as otherwise the problem can be analyzed in the classical way . we shall also assume that @xmath97 is an integer . this is easily achieved by choosing @xmath98 and @xmath14 divisible by 4 for example . the mesh @xmath99 is generated by @xmath100 with the mesh generating function @xmath101,\\ p(t - q)^3 + \frac{\lambda}{q}t & t\in[q,1/2],\\ 1-\varphi(1-t ) & t\in[1/2,1 ] , \end{array } \right . \label{mreza2}\ ] ] where @xmath102 chosen such that @xmath103 i.e. @xmath104 + note that @xmath105 $ ] with @xmath106 therefore we have that the mesh sizes @xmath107 satisfy @xmath108 in this section we prove the theorem on @xmath0-uniform convergence of the discrete problem . the proof uses the decomposition of the solution @xmath6 to the problem ( [ uvod1])([uvod2 ] ) to the layer @xmath109 and a regular component @xmath110 given by @xcite the solution @xmath6 to problem ( [ uvod1])([uvod2 ] ) can be represented as @xmath111 where for @xmath112 and @xmath113 $ ] we have that @xmath114 [ teorema2 ] [ remark2 ] note that @xmath115 for @xmath116 $ ] and @xmath117 for @xmath118.$ ] these inequalities and the estimate imply that the analysis of the error value can be done on the part of the mesh which corresponds to @xmath116 $ ] omitting the function @xmath119 , keeping in mind that on this part of the mesh we have that @xmath120 an analogous analysis would hold for the part of the mesh which corresponds to @xmath118 $ ] but with the omision of the function @xmath121 and using the inequality @xmath122 from here on in we use @xmath123 and @xmath124 where @xmath125 we begin with a lemma that will be used further on in the proof on the uniform convergence . on the part of the modified shishkin mesh where @xmath126 $ ] @xmath127 $ ] , assuming that @xmath128 , for @xmath129 we have the following estimate @xmath130 [ lema1 ] we are using the decomposition from theorem [ teorema2 ] and expansions , . for the regular component @xmath110 we have that & ( + ) ^-1 \|-\| + & \|r_i \| + & + \| \| + \|\| . [ 15 ] & first we want to estimate the expressions containing only the first derivatives in the rhs of inequality ( [ 15 ] ) . from the identity @xmath131 and the inequalities @xmath132 , @xmath133 , we get the inequality @xmath134 , which yields that @xmath135 , @xmath136 using inequality ( [ 17 ] ) together with ( [ regularna ] ) , we get that @xmath137 now we want to estimate the terms containing the second derivatives from the rhs of ( [ 15 ] ) . using inequality ( [ regularna ] ) , after some simplification , we get that @xmath138 @xmath139 for the layer component @xmath109 , first we have that & ( + ) ^-1 \|-\| & & + & \| \| & & + & + \| \| . & & [ 22 ] the first term of the rhs of ( [ 22 ] ) can be bounded by & \|\| & + & ^2\| s(^-_i)+s(^+_i)\| . [ 23 ] for the second term of the rhs of ( [ 22 ] ) we get that & \| \| + & \|s_i-1-s_i\| + \|s_i - s_i+1\| , [ 24 ] @xmath140 in the first expression of the rhs of ( [ 24 ] ) we have the term @xmath141 although this ratio is bounded by @xmath142 , this quotient is not bounded for @xmath143 when @xmath144 this is why we are going to estimate this expression separately on the transition part and on the nonequidistant part of the mesh . in the case @xmath145 , using the fact that @xmath146 and @xmath147 and the fact that the function @xmath148 takes values from the interval @xmath149 when @xmath150 , we have that @xmath151 when @xmath152 , we can use @xmath153 and + @xmath154 for @xmath155 therefore @xmath156 using equations ( [ mreza31 ] ) , ( [ mreza32 ] ) and ( [ 18])([27 ] ) , we complete the proof of the lemma . now we state the main theorem on @xmath157uniform convergence of our difference scheme and specially chosen layer - adapted mesh . the discrete problem on the mesh from section [ sec2 ] . is uniformly convergent with respect to @xmath0 and @xmath158 where @xmath6 is the solution of the problem ( [ uvod1 ] ) , @xmath11 is the corresponding numerical solution of ( [ konst18a]) and @xmath15 is a constant independent of @xmath14 and @xmath0 . we shall use the technique from @xcite , i.e. since we have stability from theorem [ teo21 ] , we have that @xmath159 and since implies that @xmath72 , it only remains to estimate @xmath160 . let @xmath161 . the discrete problem ( [ konst18a]) can be written down on this part of the mesh in the following form f_0y= & 0 , & & + f_iy = & ( 3a_i+d_i+d_i+1 ) ( y_i-1-y_i ) -(y_i - y_i+1 ) + & . - ( d_i+d_i+1 ) ] + = & + = & ( 2 + 2(h_i))(y_i-1-y_i-(y_i - y_i+1 ) ) + & . -2((h_i)-1 ) ] , for @xmath162 . using the expansions and , we get that f_iy = & + & = & , for @xmath163 and hence @xmath164 for @xmath161 . + now let @xmath165 we rewrite equations ( [ konst18a]) as f_iy= & ( d_i+d_i+1)(y_i-1-y_i-(y_i - y_i+1 ) ) & + & + 4 ( a_i(y_i-1-y_i)-a_i+1(y_i - y_i+1 ) ) + & - .(d_i+d_i+1 ) ] . we estimate the linear and the nonlinear term separately . for the nonlinear term we get & \| ( d_i+d_i+1)\| & & + & = \|f(x_i-1,y_i-1)+2f(x_i , y_i)+f(x_i+1,y_i+1)\|=^2\|y_i-1 + 2y_i+y_i+1\| . for the linear term we get & \| ( d_i+d_i+1)+4\| + & \| y_i-1-y_i-(y_i - y_i+1)\|+ \|a_i(y_i-1-y_i)-a_i+1(y_i - y_i+1 ) \| . [ teo8 ] for the first term in the rhs of ( [ teo8 ] ) we get @xmath166 while for the second term in the rhs of ( [ teo8 ] ) , using and , we get that @xmath167 hence , we get that @xmath168 for @xmath129 . + the proof for @xmath169 is analogous to the case @xmath161 and the proof for @xmath170 is analogous to the case @xmath171 in view of remark [ remark2 ] and lemma [ lema1 ] . finally , the case @xmath172 is simply shown since @xmath173 , @xmath174 and @xmath175 for @xmath176 $ ] . in this section we present numerical results to confirm the uniform accuracy of the discrete problem ( [ konst18a]). to demonstrate the efficiency of the method , we present two examples having boundary layers . the problems from our examples have known exact solutions , so we calculate @xmath177 as @xmath178 where @xmath179 is the value of the numerical solutions at the mesh point @xmath180 , where the mesh has @xmath14 subintervals , and @xmath181 is the value of the exact solution at @xmath180 . the rate of convergence ord is calculated using @xmath182 where @xmath183 tables 1 and 2 give the numerical results for our two examples and we can see that the theoretical and experimental results match . consider the following problem , see @xcite @xmath184 the exact solution of this problem is given by @xmath185 [ primjer1 ] the nonlinear system was solved using the initial condition @xmath186 and the value of the constant @xmath187 . in the analysis of examples 1 and 2 from section [ sec5 ] and the corresponding result tables , we can observe the robustness of the constructed difference scheme , even for small values of the perturbation parameter @xmath0 . note that the results presented in tables 1 and 2 already suggest @xmath0-uniform convergence of second order . the presented method can be used in order to construct schemes of convergence order greater than two . in constructing such schemes , the corresponding analysis should not be more difficult that the analysis for our constructed difference scheme . in the case of constructing schemes for solving a two - dimensional singularly perturbed boundary value problem , if one does not take care that functions of two variables do not appear during the scheme construction , the analysis should not be substantially more difficult then for our constructed scheme . in such a case it would be enough to separate the expressions with the same variables and the analysis is reduced to the previously done one - dimensional analysis . the authors are grateful to nermin okii ' c and elvis barakovi ' c for helpful advice . helena zarin is supported by the ministry of education and science of the republic of serbia under grant no . 174030 . 28 zh.vychisl . mat i mat . * 9 * , 841859 ( in russian ) ( 1969 ) proceedings of the 7th international conference on operational research , croatian or society , 197208 ( 1999 ) . numer.math . * 56 * , 675693 ( 1990 ) novi sad j.math . * 33 * , 2 , 173180 ( 2003 ) j.math . novi sad * 33 * 1 , 145162 ( 2003 ) novi sad j. math . , * 28 * 3 , 4149 ( 1998 ) linss , t. : layer - adapted meshes for reaction - convection - diffusion problems , springer - verlag , berlin , heidelberg ( 2010 ) numer . math . * 43 * , 175198 ( 1984 ) mathematics of computation , * 47 * 176 , 555570 ( 1986 ) . siam , philadelphia , usa ( 2000 ) journal of computational and applied mathematics * 29 * , 69 - 77 ( 1990 ) sov.j.numer.anal.math.modelling * 3 * 393407 ( 1988 ) numer . math . * 50 * , 519531 ( 1987 ) mathematics of computation , * 65 * 215 , 10851109 ( 1996 ) indian j.pure appl . math . , * 27 * 10 , 10051016 ( 1996 ) review of research faculty of science - university of novi sad , * 13 * ( 1983 ) review of research faculty of science - university of novi sad , * 23 * 2 , 363379 ( 1993 ) cmam * 4 * , 368383 ( 2004 ) novi sad journal of mathematics , * 33 * 1 , 133143 ( 2003 )	in this paper we are considering a semilinear singular perturbation reaction diffusion boundary value problem , which contains a small perturbation parameter that acts on the highest order derivative . we construct a difference scheme on an arbitrary nonequidistant mesh using a collocation method and green s function . we show that the constructed difference scheme has a unique solution and that the scheme is stable . the central result of the paper is @xmath0-uniform convergence of almost second order for the discrete approximate solution on a modified shishkin mesh . we finally provide two numerical examples which illustrate the theoretical results on the uniform accuracy of the discrete problem , as well as the robustness of the method . example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore
second order leapfrog or splitting methods are a class of widely used time - symmetric , explicit and symplectic integration algorithms for hamiltonian systems . these characteristics make them a standard tool for very long integrations , as they preserve the phase space structure and first integrals of the system . being time - symmetric , second order leapfrog algorithms have an error expansion that only contains even powers of the timestep . this fact makes them convenient for use within extrapolation schemes , such as the gragg bulirsch stoer ( gbs ) method @xcite , which are often used when very high accuracy is required @xcite . the main problem with leapfrog methods is the fact that they can only be constructed for systems where the hamiltonian separates into two or more parts , where the flow of each part can be separately integrated @xcite . a solution to this problem for partitioned systems of the type @xmath0 was presented in @xcite . by means of auxiliary velocity coordinates , the equations of motion were transformed into a separable form and thus amenable for integration with a leapfrog method . the method , called auxiliary velocity algorithm ( ava ) , can also be used for nonconservative systems as well . in this paper we propose an improved extension of the ava method , applicable for hamiltonian and non - hamiltonian cases where all equations of motion depend on both coordinates and momenta in general . we first briefly introduce leapfrog integration methods , and outline their properties . next , we demonstrate how the phase space of general hamiltonian systems can be extended and a new hamiltonian constructed so that the equations of motion are brought into a separated form . we then construct symmetric leapfrog integrators for the equations . these include maps that mix the extended phase space , which we find to be a requirement for good long term behaviour . finally , we investigate how the extended phase space can be projected back to the original number of dimensions so that extra accuracy can be gained in the process . we then show how the same principle can be applied to nonconservative systems as well . we apply the obtained leapfrog methods to illustrative example cases : hamiltonian geodesic flow , and a forced van der pol oscillator . in many applications to classical physics , such as gravitational interaction of point masses , the hamiltonian function @xmath1 of the system can be separated into two parts @xmath2 where @xmath3 is the kinetic energy , and @xmath4 is the potential energy . in these cases , the hamiltonian equations of motion read @xmath5 where @xmath6 . the equations for coordinates can then be directly integrated , if the momenta are kept constant , and vice versa . the solutions can be combined in a time - symmetric manner to obtain the two forms of the archetypal second order leapfrog , also known as the strmer verlet method , or strang splitting @xcite : @xmath7 and @xmath8 where @xmath9 , @xmath10 , @xmath11 and @xmath12 is the timestep . equations and can also be written as @xmath13 and @xmath14 where @xmath15 , @xmath16 and @xmath17 are the hamiltonian vector fields of @xmath18 and @xmath19 , @xmath20 is the phase space flow along the vector field @xmath21 , @xmath22 is the symplectic form given in local coordinates , @xmath23 is the @xmath24 identity matrix and @xmath25 is the exponential mapping from a lie algebra @xmath26 to the corresponding lie group . here the lie algebra is the algebra of smooth , real - valued functions on the phase space , with the lie product given by the poisson brackets @xmath27 . the group action ( written multiplicatively in equations and ) on the phase space manifold of the corresponding lie group is the phase space flow of the associated vector field . now , a reverse application of the baker campbell hausdorff ( bch ) formula on equation yields @xmath28 \\ & \quad\quad= \exp(h{\widehat{h } } ) , \end{split}\ ] ] where then @xmath29 and similarly for with @xmath30 . equations - are of interest for a number of reasons . first , the flows of hamiltonian vector fields @xmath31 are symplectic transformations for smooth functions @xmath32 @xcite , and thus preserve the geometric structure of the phase space , and all first integrals . since @xmath18 and @xmath19 are smooth , the leapfrog method has these properties as well . equation shows on the other hand that leapfrog integrators exactly solve a hamiltonian problem that is asymptotically related to the original one , with a perturbed hamiltonian @xmath33 . another desirable property of the second order leapfrogs is ( relatively ) easy composition of the basic second order method to yield methods of higher order @xcite . if @xmath34 is the numerical flow of a time - symmetric second order leapfrog , then @xmath35 can be shown to be a method of higher order for certain choices of @xmath36 and @xmath37 , with @xmath38 for time - symmetric methods such as the second order leapfrog @xcite . one particular example is the sixth order composition @xmath39 from @xcite ( composition ` s9odr6a ` in the paper ) , which we will use in section [ sc:5.2 ] . the second order leapfrog is also useful when used within an extrapolation scheme , such as the gbs scheme @xcite . using an extrapolation scheme does in principle destroy the desirable properties of the leapfrog , since the substeps require a change in timestep which destroys symplecticity , and the final linear combination of the symplectic maps is also not symplectic in general . in practice , the increase in accuracy per computational work spent often offsets this . for a comprehensive review of splitting methods in contexts not limited to hamiltonian ordinary differential equations ( odes ) , see @xcite , for geometric integration methods in general , see @xcite , and for extrapolation and other conventional methods for general odes , see @xcite . in the general case , the hamiltonian @xmath40 does not separate into additive parts , and the approach of the previous section can not be used . we can partially circumvent this problem in the following manner . we first extend the phase space @xmath41 ( essentially @xmath42 , when always operating in local coordinates ) by joining it with another identical copy , giving an extended phase space @xmath43 . in local coordinates @xmath44 , the symplectic form @xmath45 on this phase space is @xmath46 where @xmath47 is the symplectic form on @xmath41 and @xmath48 is the kronecker product . we then introduce a new hamiltonian @xmath49 on this extended phase space with @xmath50 where now @xmath51 , equal to the hamiltonian function of the original system . hamilton s equations for the two parts of this split hamiltonian are then [ eq : h1 ] @xmath52 and [ eq : h2 ] @xmath53 we see that the derivatives of @xmath54 and @xmath55 depend only on @xmath56 and @xmath57 and vice versa , and the equations can be integrated to yield the actions of @xmath58 and @xmath59 . we can now apply the results from the previous section to find @xmath60 where @xmath12 , which gives the leapfrog algorithms [ eq : lf212 ] @xmath61 and [ eq : lf121 ] @xmath62 over one timestep @xmath63 . the leapfrog methods and then exactly solve the hamiltonian flows of the related hamiltonians @xmath64 if we now consider a hamiltonian initial value problem , with the initial values @xmath65 and set @xmath66 , @xmath67 , we see that the equations and give identical derivatives and identical evolution for both pairs @xmath68 and @xmath69 , equal to the flow of the original hamiltonian system @xmath70 . the numerical leapfrog solutions and then solve closely related hamiltonian problems given by and . we can write the problem in the form @xmath71 where the component vector fields @xmath72 are hamiltonian . we also define operators @xmath73 , @xmath74 , and similarly for @xmath75 and @xmath76 , where @xmath77 is a translation along the vector field @xmath32 . from this we see that we can also construct split solutions like @xmath78 where the last equality defines a shorthand notation based on the symmetry of the operator . pairs @xmath79 and @xmath80 commute , so e.g. @xmath81 original and auxiliary variables are also interchangeable , since they initially have the same values . as such , the only unique symmetric second order leapfrog compositions are ( with the above shorthand ) [ eq : all_lfs ] @xmath82 up to a reordering of commuting operators , and switching @xmath83 . from the leapfrogs , the only one symplectic in the extended phase space is , as well as its conjugate method @xmath84 . it is _ not _ , however symplectic as a mapping @xmath85 within the original phase space if the final @xmath86 are obtained from @xmath87 by any pairwise choice . however , operating in the extended phase space and projecting only to obtain outputs , without altering the state in the extended phase space , leads to an algorithm that preserves the original hamiltonian with no secular growth in the error . this is not entirely surprising , since the algorithms have similarities with partitioned multistep methods , which can exhibit good long term behaviour despite a formal lack of symplecticity @xcite . by virtue of being leapfrogs , the algorithms are also explicit , time - symmetric , have error expansions that contain only even powers of the timestep and can be sequentially applied to form higher order methods . the idea behind equation can be generalised further . introducing yet another copy of the phase space with coordinates @xmath88 leads to a hamiltonian of the form @xmath89.\ ] ] the hamiltonian gives three sets of separable equations of motion that can be integrated with a leapfrog method . as with the leapfrogs , only the leapfrogs obtained from sequential applications of the flows of the different hamiltonians in equation give a method symplectic in the extended phase space . again , no simple pairwise choice of the extended phase space variables leads to a method that is symplectic in the original phase space . in general , using @xmath90 sets of variables in total , with a phase space @xmath91 , one can use a hamiltonian of the form @xmath92 where @xmath93 where @xmath94 is a cyclic permutation of the indexes of the momenta . the equations of motion of any @xmath95 can be integrated with a leapfrog , as is the case for their symmetric combination @xmath96 . however , we will not investigate this general case in the paper . while equations and can be integrated with a leapfrog , the fact that they are coupled only through the derivatives is a problem . the hamiltonian vector field of one solution at a point depends on the other and vice versa , but not on the solution itself . since the two numerical flows will not in general agree with each other or the exact flow , the derivative function for one numerical flow at one point will end up depending on a different point of the other solution , and both solutions may diverge with time . this problem is quickly confirmed by numerical experiments . an seemingly straightforward solution for this problem would be to introduce feedback between the two solutions , in the form of _ mixing maps _ @xmath97 , @xmath98 . we now amend the leapfrogs to obtain , e.g. @xmath99 so that the resulting algorithm is still symmetric , since at the last step , @xmath100 can be subsumed into the projection map described below . if @xmath101 are symplectic , then the leapfrogs that are symplectic on the extended phase space , i.e. , retain this character . there is no need to restrict @xmath101 to strictly symplectic maps , however , since the extended phase space leapfrogs are not symplectic when restricted to the original phase space in any case . without this restriction , potentially attractive candidates might stem from e.g. symmetries of the extended hamiltonian and its exact solution . for example , for , permutations of coordinates , @xmath102 , or momenta , @xmath103 , do not change the exact solution for given ( equal ) initial conditions . permuting both switches the component hamiltonians , but since they are equal , this has no effect for the exact solution . we will find that for the numerical method , the permutations can be beneficial . a related problem is how to project a vector in extended phase space back to the dimension of the original . this should be done in a manner that minimizes the error in the obtained coordinates , momenta and original hamiltonian . in addition , the final algorithm should be symplectic , or as close to symplectic as possible . to this end , we introduce a _ projection map _ @xmath104 . in principle , the projection map could be used in two different ways . the first is to obtain formal outputs at desired intervals , while the algorithm always runs in the extended phase space . the second is to use the projection map after each step , and then copy the projected values to create the extended set of variables for the next step . the final algorithm @xmath105 over @xmath106 steps would then be either of @xmath107 where @xmath108 is function composition , and @xmath109 is the cloning map @xmath110 . it should be emphasized that in equation , the projection map is included only to obtain the formal output after @xmath106 steps , while the current state @xmath111 is preserved , and used to continue the integration . in contrast , the algorithm can evidently be considered as a mapping @xmath112 in the original phase space , and as such represents a conventional method such as a partitioned runge kutta method , as seen in section [ sc:3.2 ] . unfortunately , for hamiltonian problems , it leads to secular increase of the error in the original hamiltonian . in the hamiltonian case , it seems that operating in the extended phase space is necessary , and as such , is to be used . to take a first stab at determining a suitable choice for @xmath101 and @xmath113 , we turn to analyzing the error in the hamiltonian function and deviations from symplecticity . we start the search of suitable candidates for @xmath101 and @xmath113 from symmetric linear maps of the form @xmath114 and @xmath115 where @xmath98 , and all matrix elements are real . we then look for the coefficients that give the best results according to conservation of hamiltonian function or symplecticity . we can thus use , say , @xmath116 take two steps to make @xmath100 and @xmath113 independent , and expand @xmath117 in terms of @xmath63 , where @xmath118 , and look at the coefficients . in this example case , the zeroth order coefficient of reads @xmath119 with @xmath120 to make the cofficient identically zero , we need to have @xmath121 for all the maps @xmath101 and @xmath113 , which makes them linear interpolations between the original and auxiliary coordinates and momenta . with the substitutions , also the first order coefficient of becomes identically zero the same substitution zeroes the coefficients of up to and including the second order . the third order coefficient of becomes independent of the map matrix elements , and as such we will focus on the expansion of . the second order coefficient is @xmath122,\ ] ] where @xmath123 and the derivatives @xmath124 are @xmath125-linear operators @xmath126 where @xmath127 , @xmath128 , and juxtaposition in implies contraction . from here , there are several choices available . we will consider some interesting combinations . choosing @xmath129 makes the second order term identically zero . taking @xmath130 and @xmath131 then makes also the third order term be identically zero , and as such apparently gives an additional order of accuracy compared to the standard leaprog . however , despite this , these choices lead to poor long term behaviour , which numerically is quickly apparent . if @xmath132 and @xmath133 , so that the momenta are permuted , then choosing @xmath134 will also identically zero the second order coefficient . in this case , the third order coefficient ca nt be brought to zero , but the numerical long term behaviour found in the problem of section [ sc:5.1 ] is good , if @xmath135 is chosen so that there is a permutation symmetry between coordinates and momenta . numerically , the best results were obtained with a choice @xmath136 , @xmath137 and @xmath138 , @xmath139 . this necessitates @xmath140 to zero the second order coefficient . we conclude that the long term behaviour of the method is not evident from considering the conservation of the hamiltonian alone . in addition to the conservation of the hamiltonian , we are interested in the conservation of the symplectic form , or the symplecticity of the method . in local coordinates , the condition for symplecticity of an integration method @xmath141 over one step @xmath63 is @xmath142 where @xmath143 is the jacobian of the map @xmath34 @xcite . we consider first symplecticity in the extended phase space @xmath144 . it is clear that if @xmath145 and @xmath100 are identity maps , then the method is symplectic . however , we know that this does not lead to good numerical results in the long term . to investigate other possibilities , we again apply the method for two steps and expand the left side of equation in terms of @xmath63 . the first order term gives two independent conditions @xmath146 \\ & \quad\quad\times\left[- ( { { \beta } _ { { m}_1}}(1 - 2 { { \beta } _ { { m}_2}})^2 ) + { { \beta } _ { { m}_1}}^2(1 - 2 { { \beta } _ { { m}_2}})^2 + ( -1 + { { \beta } _ { { m}_2 } } ) { { \beta } _ { { m}_2}}\right ] \\ & \quad + \left[1 - 2 { { \alpha } _ { { m}_1}}(1 - 2 { { \alpha } _ { { m}_2}})^2 + 2 { { \alpha } _ { { m}_1}}^2(1 - 2 { { \alpha } _ { { m}_2}})^2 - 2 { { \alpha } _ { { m}_2 } } + 2 { { \alpha } _ { { m}_2}}^2\right ] \\ & \quad\quad\times\left[1 - 2 { { \beta } _ { { m}_1}}(1 - 2 { { \beta } _ { { m}_2}})^2 + 2 { { \beta } _ { { m}_1}}^2(1 - 2 { { \beta } _ { { m}_2}})^2 - 2 { { \beta } _ { { m}_2 } } + 2 { { \beta } _ { { m}_2}}^2\right ] = 1 \end{split } \label{eq : scond01}\\ \begin{split } & 2\left[1 - 2 { { \alpha } _ { { m}_1}}(1 - 2 { { \alpha } _ { { m}_2}})^2 + 2 { { \alpha } _ { { m}_1}}^2(1 - 2 { { \alpha } _ { { m}_2}})^2 - 2 { { \alpha } _ { { m}_2 } } + 2 { { \alpha } _ { { m}_2}}^2\right ] \\ & \quad\quad\times\left[- ( { { \beta } _ { { m}_1}}(1 - 2 { { \beta } _ { { m}_2}})^2 ) + { { \beta } _ { { m}_1}}^2(1 - 2 { { \beta } _ { { m}_2}})^2 + ( -1 + { { \beta } _ { { m}_2 } } ) { { \beta } _ { { m}_2}}\right ] \\ & \quad + 2\left[- ( { { \alpha } _ { { m}_1}}(1 - 2 { { \alpha } _ { { m}_2}})^2 ) + { { \alpha } _ { { m}_1}}^2(1 - 2 { { \alpha } _ { { m}_2}})^2 + ( -1 + { { \alpha } _ { { m}_2 } } ) { { \alpha } _ { { m}_2}}\right ] \\ & \quad\quad\times\left[1 - 2 { { \beta } _ { { m}_1}}(1 - 2 { { \beta } _ { { m}_2}})^2 + 2 { { \beta } _ { { m}_1}}^2(1 - 2 { { \beta } _ { { m}_2}})^2 - 2 { { \beta } _ { { m}_2 } } + 2 { { \beta } _ { { m}_2}}^2\right ] = 0 . \end{split}\label{eq : scond02}\end{aligned}\ ] ] solving any coefficient from these requires that none of the others is @xmath147 . as such , simply averaging the extended variable pairs leads to destruction of symplecticity in the extended phase space already in the zeroth order . on the other hand , any combination of @xmath148 makes equations and identically true . of these , combinations with @xmath149 are exactly symplectic , since the corresponding maps @xmath101 are . other combinations give a non - zero cofficient at second order . to investigate symplecticity in the original phase space @xmath41 , we append the projection map @xmath113 . in this case , the zeroth and first order terms are zero independely of the coefficients of the maps @xmath101 and @xmath113 . the second order term is a cumbersomely lengthy function of the map components and derivatives of the hamiltonian . however , it is reduced to zero by those substitutions from @xmath148 that have @xmath150 or @xmath151 . in this case , if we also put @xmath152 , the first non - zero term is of the fifth order . in the case that @xmath153 or @xmath154 , setting @xmath152 does zero out the second order term , as well as the third order term , but not the fourth . the combination of identity maps @xmath101 with @xmath130 and @xmath131 , shown above to conserve hamiltonian to an extra order of accuracy , leads to a non - zero error already in the third order . the method , which gives the best results for the application in section [ sc:5.1 ] in both accuracy and long term behaviour , is the one with @xmath155 , @xmath137 and @xmath156 . interestingly , these choices give a non - zero contribution already at second order , and this is not affected by subsituting in the specific choice of hamiltonian for the problem . many common integration methods can be written in some general formulation as well , which can give additional insight to the numerical behaviour of the algorithm . here , we will consider partitioned runge - kutta ( prk ) methods . prk methods form a very general class of algorithms for solving a partitioned system of differential equations @xmath157 where @xmath158 , @xmath159 , @xmath32 and @xmath160 may be vector valued . a partitioned runge kutta algorithm for system can be written as @xmath161 where @xmath162 , @xmath163 and @xmath164 , @xmath165 are the coefficients of two ( possibly different ) runge kutta methods , respectively . the second order leapfrog can be written as a prk algorithm using the coefficients in table [ tb : lfcoeff ] . if @xmath166 and @xmath167 are functions of only @xmath159 and @xmath158 , respectively , the resulting algorithm is explicit . in the extended phase space the equations of motion and can be written as @xmath168 with @xmath169 , @xmath170 and @xmath171 . if the maps @xmath101 are identity maps , the leapfrogs and can be written as prk algorithms with coefficients from table [ tb : lfcoeff ] as well . if this is not the case , the final result for @xmath172 will involve both @xmath173 and @xmath174 , and similarly for @xmath175 , but in general during the integration , @xmath176 at the beginning of any given step . the resulting @xmath172 and @xmath175 will both also involve a mix of @xmath177 and @xmath178 , as well . this can not be obtained with a prk scheme . if , however , @xmath179 , we can write a prk scheme for the _ first _ step , where @xmath180 , and also @xmath181 . the resulting coefficients are listed in table [ tb : ephcoeff ] . .the butcher tableaus for the second order leapfrog as a partitioned runge kutta system . [ cols="^,^,^ " , ] to assess the long term behaviour of the methods , we did another integration for 3000 orbital periods , and investigated the error in the conservation of the hamiltonian . figure [ fig : geo_hami ] shows the absolute error for the first 10 orbits and the maximum error up to a given time during the integration for the whole run . while the lsode method is very accurate , we see that it eventually displays a secular power law increase in the maximum error with time , typical for nonsymplectic algorithms . the symplectic implicit midpoint method shows no secular growth in the error , and neither does the method @xmath182 with either projection @xmath183 . this result is not completely unexpected , since symmetric non - symplectic methods can also display behaviour similar to symplectic ones , particularly for quadratic hamiltonians , such as in this case @xcite . for the @xmath182 method with @xmath184 ( solid line ) and @xmath185 ( dotted line ) , implicit midpoint method ( dashed line ) , and the lsode method ( dash - dotted line ) . _ left : _ absolute error , for @xmath186 first orbits . _ right : _ maximum absolute error up to given time during integration . note the different @xmath158-axis scaling . , title="fig:",scaledwidth=50.0% ] for the @xmath182 method with @xmath184 ( solid line ) and @xmath185 ( dotted line ) , implicit midpoint method ( dashed line ) , and the lsode method ( dash - dotted line ) . _ left : _ absolute error , for @xmath186 first orbits . _ right : _ maximum absolute error up to given time during integration . note the different @xmath158-axis scaling . , title="fig:",scaledwidth=50.0% ] for this particular problem , the method @xmath182 with @xmath184 produces excellent results . the resulting orbit is much more closely aligned with the lsode solution than the orbit given by a basic symplectic integrator , the implicit midpoint method , and there is no secular growth in the error of the hamiltonian . it is notable also that since the method is explicit , the number of vector field evaluations required and thereby the computing time used is much less than for the implicit midpoint method , or the lsode method . as such , the @xmath182 could be used with a smaller timestep to obtain even better results relative to the other methods . we test the extended phase space method also on a non - conservative system , the forced van der pol oscillator @xcite @xmath187 which can be written in the equivalent form @xmath188 where @xmath189 parametrizes the non - linearity of the system , and @xmath190 are the amplitude and period of the sinusoidal forcing . the van der pol oscillator is essentially a damped non - linear oscillator that exhibits a limit cycle . for our test , we set @xmath191 , @xmath192 and @xmath193 ; a choice for which the oscillator is known to exhibit chaotic behaviour @xcite . as initial conditions , we take @xmath194 . to integrate the system , we use split systems of types @xmath195 ( method 1 in the following ) @xmath196 ( method 2 ) , in the same symmetric shorthand as in equations . method 1 is of type while method 2 is of type . for both methods , we employ the 6th order composition coefficients from to yield a 6th order method . in this case , we use the method , and set @xmath197 so that after one composited step , the original and auxiliary variables are averaged . for the mixing maps , we take @xmath198 and @xmath199 . as in the previous section , we compare these methods to the implicit midpoint method , iterated to @xmath200 precision in relative error , and the lsode solver with a relative accuracy parameter of @xmath201 and absolute accuracy parameter of @xmath200 . we propagate the system until @xmath202 using a timestep @xmath203 . figure [ fig : vdp_orbits ] shows the numerical orbits of the four methods in the phase space @xmath204 . the behaviour of the system is characterized by slow dwell near points @xmath205 , and quick progression along the curved paths when not . in figure [ fig : vdp_errors ] we have plotted the maximum absolute errors in @xmath158 and @xmath159 up to given time with respect to the lsode method . we find that all the methods display a secular growth in the coordinate errors , with method 1 performing best , followed by method 2 and the implicit midpoint method . methods 1 and 2 show similar qualitative behaviour as the lsode solution , while the midpoint method shows a clear divergence . these results needs to be contrasted with the amounts of vector field evaluations , which are @xmath206 ( method 1 ) , @xmath207 ( method 2 ) , @xmath207 ( implicit midpoint ) and @xmath208 ( lsode ) . the number of evaluations is roughly similar for each method , and as such method 1 is rather clearly the best of the constant timestep methods , with lsode likely the best overall . , @xmath192 and @xmath193 , integrated until @xmath202 , with method 1 ( solid line , top left ) , method 2 ( dotted line , top right ) , the lsode method ( dash - dotted line , bottom left ) , and the implicit midpoint method ( dashed line , bottom right ) . , scaledwidth=100.0% ] perhaps the most obvious benefit of splitting methods is that they re explicit , which eliminates all problems inherent in having to find iterative solutions . specifically , if evaluating the vector field , or parts of it , is very computationally intensive , then explicit methods are to be preferred , as they only require a single evaluation of the vector field for each step of the algorithm . leapfrog methods can also be composited with large degrees of freedom , making it possible to optimize the method used for the specific task at hand . more subtly , when the problem separates to asymmetric kinetic and potential parts , different algorithmic regularization schemes can be used to yield very powerful ( even exact ) , yet simple integrators @xcite . however , for the case when the hamiltonian remains unspecified , algorithmic regularization seems to yield little benefit , since both parts of the new hamiltonian are essentially identical . this is problematic , since the leapfrog integration of an inseparable hamiltonian typically leads to wrong results only when the system is in `` difficult '' regions of the phase space , such as near the schwarzschild radius for the case in section [ sc:5.1 ] , where the derivatives have very large numerical values and the expansions - may not converge for the chosen value of timestep . this is exactly the problem that algorithmic regularization solves , and it would be greatly beneficial if such a scheme could be employed even for the artificial splitting of the hamiltonian in . despite the lack of algorithmic regularization , the extended phase space methods seem promising . the results in section [ sc:5.1 ] demonstrate that the extended phase space methods can give results comparable to an established differential equation solver , lsode , but with less computational work . more importantly , the results are superior to a known symplectic method , the implicit midpoint method . the results in section [ sc:5.2 ] are less conclusive with respect to the lsode method , but clear superiority versus the implicit midpoint method is still evident . we find this encouraging , and believe that the extended phase space methods should be investigated further . obvious candidate for further research is the best possible form and use of the mixing and projection maps . the optimal result is likely problem dependent . another issue that would benefit from investigation is how to find algorithmic regularization schemes for the split , preferably with as loose constraints on the form of the original hamiltonian as possible . finally , whether useful integrators can be obtained from the splits of types - should be investigated . we have presented a way to construct splitting method integrators for hamiltonian problems where the hamiltonian is inseparable , by introducing a copy of the original phase space , and a new hamiltonian which leads to equations of motion that can be directly integrated . we have also shown how the phase space extension can be used to construct similar leapfrogs for general problems that can be reduced to a system of first order differential equations . we have then implemented various examples of the new leapfrogs , including a higher order composition . these methods have then been applied to the problem of geodesics in a curved space and a non - linear , non - conservative forced oscillator . with these examples , we have demonstrated that utilizing both the auxiliary and original variables in deriving the final result , via the mixing and projection maps , instead of discarding one pair as is done in @xcite and @xcite , can yield better results than established methods , such as the implicit midpoint method . the new methods share some of the benefits of the standard leapfrog methods in that they re explicit , time - symmetric and only depend on the state of the system during the previous step . for a hamiltonian problem of the type in section [ sc:5.1 ] they also have no secular growth in the error in the hamiltonian . however , the extended phase space methods leave large degrees of freedom in how to mix the variables in the extended phase space , and how to project them back to the original dimension . as such , there is likely room for further improvement in this direction , as well as in the possibility of deriving a working algorithmic regularization scheme for these methods . in conclusion , we find the extended phase space methods to be an interesting class of numerical integrators , especially for hamiltonian problems . hairer , e. , lubich , c. , wanner , g. : geometric numerical integration : structure - preserving algorithms for ordinary differential equations , springer series in computational mathematics , vol 31 . springer , berlin ( 2006 ) radhakrishnan , k. , hindmarsh , a. : description and use of lsode , the livermore solver for ordinary differential equations . nasa , office of management , scientific and technical information program ( 1993 )	we present a method for explicit leapfrog integration of inseparable hamiltonian systems by means of an extended phase space . a suitably defined new hamiltonian on the extended phase space leads to equations of motion that can be numerically integrated by standard symplectic leapfrog ( splitting ) methods . when the leapfrog is combined with coordinate mixing transformations , the resulting algorithm shows good long term stability and error behaviour . we extend the method to non - hamiltonian problems as well , and investigate optimal methods of projecting the extended phase space back to original dimension . finally , we apply the methods to a hamiltonian problem of geodesics in a curved space , and a non - hamiltonian problem of a forced non - linear oscillator . we compare the performance of the methods to a general purpose differential equation solver lsode , and the implicit midpoint method , a symplectic one - step method . we find the extended phase space methods to compare favorably to both for the hamiltonian problem , and to the implicit midpoint method in the case of the non - linear oscillator .
in the seminal discovery of the quantum hall effect ( qhe ) in graphene@xcite , a most striking point is that the graphene qhe is regarded as a realization of the _ anomaly_@xcite in the massless dirac particle , where each dirac cone carries 1/2 of the usual qhe topological number@xcite . namely , in place of the usual hall conductivity , @xmath1 ( in units of @xmath2 , with the spin degrees of freedom dropped here ) , we have @xmath3 . the standard understanding is that the honeycomb lattice has a couple of dirac cones at k and k points in the brillouin zone , so that we just divide the above formula by 2 to have @xmath4 for the contribution from each valley . the half integers are thus hidden in the total hall conductivity . since the nielsen - ninomiya theorem@xcite dictates that dirac cones should always appear in pairs as far as the chiral symmetry is present , we can pose an important question : is it really impossible to resolve the half - integer components ? in terms of field theory ( as opposed to lattice models ) , the situation is simple : the hall conductivity for a massive ( @xmath5 ) dirac particle , in zero magnetic field , is given by @xmath6 when the fermi energy @xmath7 is in the mass gap@xcite . this can be readily shown by calculating berry s connection . if we consider a field theoretical model concerning a _ dirac fermion at @xmath8 with a gap @xmath9 , the hamiltonian is given by @xmath10 , where @xmath11 , @xmath12@xcite . if we denote the eigenstate having an energy @xmath13 as @xmath14 , and its berry connection as @xmath15 , the total berry curvature over the @xmath16 plane ( represented as @xmath17 ) is @xmath18 , where @xmath19 , which is the flux from a magnetic monopole at the origin with @xmath20 and @xmath21 . we can then see that we have 1/2 since just the half the total flux of the magnetic monopole passes through the plane @xmath22 . this implies that the topological change in the quantum ground state ( filled dirac sea ) is characterized by @xmath23@xcite . generically speaking , a topological quantum phase transition is naturally accompanied by a sign change in the mass in the effective , low - energy dirac fermions . quantum hall plateau transition is a typical example , and the topological insulators such as the quantum spin hall system are also described along this line , where the spin - orbit interaction induces the sign change in the mass@xcite . while the quantization into the half odd integers is conceptually interesting , if we go over to lattice models , however , we have a periodicity in the brillouin zone , which implies that the topological numbers should always be integers as dictated by the celebrated but inescapable thouless - kohmoto - nightingale - den nijs ( tknn ) formula@xcite . the integer total hall conductivity for graphene with a pair of dirac cones is an example of this . if we turn to a wider class of lattice models , we can go around the nielsen - ninomiya theorem . for instance , we can have a lattice model that has an odd number of massless dirac cones , but even in that case we still end up with integer hall conductivities , which may be regarded as due to hidden massive dirac fermions required to exist for the topological consistency in the lattice system to be guaranteed . the massive dirac fermions ( sometimes regarded as spectators"@xcite ) are hidden in the high - energy region , which do not appear in the low - energy physics except that they make sure that the the hall conductivity are topologically protected to be integers . in another manipulation of dirac cones@xcite , a quantum hall effect in zero total magnetic field has been considered , where the hall conductivity is shown to take the value of @xmath24 even in zero field in a model containing complex hoppings in the situation where the dirac cones are made massive . it is thus rather difficult to confirm half - odd - integers in the hall conductivity for a single dirac cone . in this rapid communication , we want to shed light to this problem by posing the following question : if we can manipulate the energies of the multiple dirac points , can the half integer series confirmed through a systematic behavior of the total hall conductivity ? namely , we shall construct a lattice model , where the two dirac points are shifted with the massless cones preserved for both of the dirac points . we shall identify , by directly computing the topological chern number for systematically varied relative position of the dirac cones , that each dirac cone indeed has a half - odd - integer series ( @xmath0 ) when @xmath7 traverses ( now shifted sets of ) landau levels belonging to the two dirac cones . the picture is further endorsed , via the bulk - edge correspondence , from the behavior of the edge modes for the shifted dirac cones for finite systems . a strategy to conceive a model in which two dirac cones are preserved but mutually shifted in energy is the following . a simplest solution is to add a term that is proportional to @xmath25 ( unit matrix ) in the space of pauli matrices with a @xmath26-dependent coefficient . thus we can introduce a hamiltonian , @xmath27_{\alpha,\beta}\hat{c}_{\bm{k}\beta},\\ \notag h^{\mathrm{gr}}_{\bm{k}}=t_0\big[(1+\cos{k_1}+\cos{k_2})\sigma_1+(\sin{k_1}+\sin{k_2})\sigma_2\big],\end{gathered}\ ] ] where @xmath28 creates an electron in @xmath26- space , and @xmath29 s are pauli matrices with @xmath30 denoting their components . in this hamiltonian , we have added , on top of the nearest - neighbor hopping @xmath31 , an extra @xmath25 term with hopping @xmath32 as a coefficient . this only affects the band dispersion as an additive , @xmath26-dependent term , which implies that we lift the degeneracy between k and k points if the @xmath26-dependent term has different values at k and k. a simplest choice is @xmath33 . if we go back to the real space , the tight - binding model is as depicted in fig.[fig : model ] , which has extra second - neighbor hoppings . the added hopping has to be only between @xmath34-@xmath34 and @xmath35-@xmath35 for the dirac cone to be preserved , and they have to be complex for the degeneracy between k and k ( mutually time - reversal pairs ) to be lifted . so the model with a complex hopping is rather fictitious , but we do accomplish shifted dirac cones as depicted in fig.[fig : dirac ] . in this model the chiral symmetry [ i.e. , the presence of an operator ( @xmath36 in the case of graphene ) that anticommutes with the hamiltonian ] is broken , since appearance of @xmath25 in @xmath37 invalidates @xmath38 . nevertheless , the addition of @xmath25 preserves the shape of dirac cones , along with the species doubling . if we expand hamiltonian ( [ eq : hamiltonian ] ) around @xmath8 ( k or k points ) , we have @xmath39,\ ] ] where @xmath40 correspond to k ( k ) , @xmath41 is ( half ) the shift , @xmath42 the fermi velocity , and @xmath43 . so the effective theory around these points are a pair of dirac equations at k and k with a shift @xmath44 in energy as desired . the landau levels in a magnetic field @xmath45 expected from the effective dirac eq . ( [ eq : df ] ) are @xmath46 [ @xmath47 , @xmath48 . however , we should of course go back to the original lattice model for obtaining the landau levels and , especially , the hall conductivity . the calculation of the hall conductivity , which is a topological chern number , requires a bit of care , since we have to question the behavior of the topological number around @xmath49 which is just where the hole and electron branches meet in the dirac cone dispersions . this implies we have to sum over all the contributions from the dirac sea " to get the total hall conductivity . this ordinarily poses a numerically difficult problem , but we can overcome the problem with a method that employs a non - commutative berry s connection@xcite and its integration ( chern number ) over the brillouin zone with a technique developed in the lattice gauge theory@xcite . ) and ( b ) for the present model with a shifted dirac cones ( @xmath50 ) we display the numerically calculated chern numbers against @xmath7 ( upper panels ) , along with the landau levels for each of the two dirac cones , where the chern numbers are indicated for @xmath7 in each of the gaps ( lower panels ) . the result is for the flux @xmath51 . [ fig : llchern],title="fig:"][fig : llchern0 ] ) and ( b ) for the present model with a shifted dirac cones ( @xmath50 ) we display the numerically calculated chern numbers against @xmath7 ( upper panels ) , along with the landau levels for each of the two dirac cones , where the chern numbers are indicated for @xmath7 in each of the gaps ( lower panels ) . the result is for the flux @xmath51 . [ fig : llchern],title="fig:"][fig : llchern3 ] the result for the chern number in the present model obtained with the above method is displayed in fig.[fig : llchern ] , which is the key result in the present work , and reveals an intriguing feature . for comparison , let us first look at fig.[fig : llchern ] ( a ) , which is just the result for the ordinary graphene with the usual graphene landau levels at @xmath52 . the numerical result for the chern number is simply the graphene qhe , where the well - known chern number sequence of @xmath53 for the honeycomb lattice coincides with a sum over the two , half - integer contributions from the degenerate pair of dirac fields . ; the horizontal axis ) is varied , with the chern number indicated for each of the landau gap . the shading represents the chern numbers expected from the effective theory ( see text ) , where the boundaries between different shades ( chern numbers ) almost exactly coincide with the numerical result . accumulation of lines around @xmath54 corresponds to the van hove singularity , which shifts quadratically with @xmath32 . the result is for a magnetic field of @xmath51 , and two vertical lines indicate , respectively , the situation for figs.[fig : llchern](a ) and [ fig : llchern](b ) . [ fig : landaulevel ] ] if we now turn to the result for the present lattice model in fig.[fig : llchern ] ( b ) , the crucial question is : does the result conform to the expectation from the effective dirac model where the sum of two half - odd - integer series ( @xmath0 ) with a shift in energy between them should give the total hall conductivity as @xmath7 traverses shifted sets of landau levels ( as displayed in the lower panel in the figure ) . the numerical result for the chern number in the lattice model is seen to agree with this . namely , in a striking contrast to the ordinary graphene where each qhe step has a jump of 2 in the chern number , the present model exhibits a jump of 1 at each step . this , along with the positions of these jumps , perfectly fit with the positions and the associated chern numbers of the shifted landau levels as the figure indicates . the agreement is rather surprising , since there is no obvious reason why the superposition of effective field theory for the vicinities of k and k and the lattice model should have the same chern numbers . thus , although we have still no half integers for the total hall conductivity ( since a sum of two half - odd integers is an integer ) , we have indirectly confirmed the half integers . since no half integers can possibly appear according to tknn , this is indeed _ as best as _ we can confirm the half - integer property of each dirac cone . we expect such a decomposition of the hall conductivity into the contributions from each dirac cone to systematically occur when we vary the energy separation ( @xmath55 ) between the two dirac points . fig.[fig : landaulevel ] displays the loci of landau levels ( solid lines ) and the chern number for each gap between the landau levels when @xmath56 is varied . the result is just what is expected from the above decomposition , as indicated by the result for the two dirac fields ( shaded regions ) , which almost exactly coincides with the result for the lattice model . the qhe state is a topological state , which immediately implies that edge states should appear at the boundaries of a sample , where an intimate relation exists between the edge topological number ( i.e. , the edge hall conductivity ) and the bulk topological number ( the bulk hall conductivity)@xcite . so it is an intriguing as well as necessary test to look into how this property appears in the edge - state in the present model . fig.[fig : edge ] displays a numerical result for the energy spectrum against @xmath57 for a finite sample with zig - zag edges along @xmath58 . the spectrum comprises the bulk landau levels ( horizontal lines ) and a series of edge modes ( identified as curves that traverse between landau levels ) . according to the bulk - edge correspondence in a topological argument @xcite , the bulk chern number should coincide with the number of pairs of ( right and left ) edge modes in the landau gap considered . in fig.[fig : edge ] we can clearly recognize that the bulk dirac landau levels ( @xmath59 with the shift ) ( horizontal lines ) are accompanied by right and left edge modes ( upturning for the electron branch , down - turning for the hole branch ) . this occurs around each of k and k. if we count the number of pairs of edge modes , we have , for each of the series , the beautiful sequence @xmath60 ( as opposed to the sequence @xmath61 for graphene , see fig.10 of @xcite ) both in the electron or hole branches . the two series of pairs of edge modes arise with the energy shift between k and k as they should . we can thus reinforce our argument , via the bulk - edge correspondence , that we can decompose the contribution from each of the dirac cone that carries the bulk chern number of 1/2 ( fig.[fig : llchern ] ) . in the present model for a finite sample with zigzag - bearded edges @xcite . horizontal solid ( dashed ) lines correspond to bulk landau levels corresponding to the shifted dirac fermions by @xmath62 @xmath63 . curves are the edge modes whose amplitudes are confirmed to be localized along the zigzag right ( red ) or bearded left ( blue ) edge . the result is for a magnetic field @xmath64 with @xmath65 . ] we have demonstrated that each dirac fermion contributes the half odd integer series ( @xmath0 ) to the hall conductivity when the fermi energy traverses the landau levels in a lattice model in which the energy between the two dirac points is systematically shifted . the picture is also confirmed , via the bulk - edge correspondence , in the edge mode of the dirac fermions . recently it is recognized that massless dirac fermions can be realized as surface states of the three - dimensional topological ( quantum spin hall ) systems@xcite . there , the two - dimensional surface of the gapped three - dimensional topological insulator accommodates gapless ( massless dirac ) fermions , which is a manifestation of the bulk - edge correspondence@xcite . the massless dirac fermions are again doubled for the topological consistency , but the doubling partner exists at the other side of the system in this case . thus , whether we can decompose the topological numbers into contributions from each dirac cone is becoming a realistic question . it is an interesting future problem to consider how the present toy model can be realized in actual materials , where the surfaces of 3d topological insulators may be one possible avenue . another possible realization of our model is a cleverly arranged cold atoms for an optical lattice where non - abelian gauge structure can arise from the intra - atomic degree of freedom with a fine - tuning of desired parameters@xcite . 99 k. s. novoselov _ et al _ , nature ( london ) * 438 * , 197 ( 2005 ) ; a. k. geim and k. s. novoselov , nature mater . * 6 * , 183 ( 2007 ) . y. zhang , y. w. tan , h. l. stormer , and p. kim , nature ( london ) * 438 * , 201 ( 2005 ) . a. h. castro neto _ et al _ , rev . phys . * 81 * , 109 ( 2009 ) . a. j. niemi , g. w. semenoff , and y .- s . wu , nucl . b * 276 * , 173 ( 1986 ) ; a. j. niemi and g. w. semenoff , phys . * 51 * , 2077 ( 1983 ) ; a. n. redlich , phys . lett . * 52 * , 18 ( 1984 ) . d. j. thouless , m. kohmoto , m. p. nightingale , and m. den nijs , phys . rev . lett . * 49 * , 405 ( 1982 ) . y. hatsugai , phys . lett . * 71 * , 3697 ( 1993 ) . h. b. nielsen and m. ninomiya , nucl . b * 185 * , 20 ( 1981 ) ; * 193 * , 173 ( 1981 ) . m. v. berry , proc . london , ser . a * 392 * , 45 ( 1984 ) . f. d. m. haldane , phys . lett . * 61 * , 2015 ( 1988 ) . m. oshikawa , phys . b * 50 * , 17357 ( 1994 ) . y. hatsugai , m. kohmoto , and y .- s . wu , phys . b * 54 * , 4898 ( 1996 ) . et al _ , science * 318 * , 766 ( 2007 ) . c. l. kane and e. j. mele , phys . * 95 * , 146802 ( 2005 ) . m. z. hasan and c. l. kane , rev . phys . * 82 * , 3045 ( 2010 ) . y. hatsugai , j. phys . . jpn . * 73 * , 2604 ( 2004 ) . t. fukui , y. hatsugai , and h. suzuki , j. phys . * 74 * , 1674 ( 2005 ) . y. hatsugai , phys . b * 48 * , 11851 ( 1993 ) . y. hatsugai , t. fukui , and h. aoki , phys . b * 74 * , 205414 ( 2006 ) ; eur . phys . 148 * , 133 ( 2007 ) . n. goldman , a. kubasiak , a. bermudez , p. gaspard , m. lewenstein , and m. a. martin - delgado , phys . * 103 * , 035301 ( 2009 ) .	while the quantum hall effect in graphene has been regarded as a realization of the _ anomaly _ associated with the massless dirac particle carrying half the usual topological integer , this is hidden due to the doubling of the dirac cones . in order to confirm the half - integer contribution from each dirac cone , here we theoretically consider a lattice model in which the relative energy between the two dirac points is systematically shifted . with an explicit calculation of the topological ( chern ) number , we have demonstrated that each dirac cone does indeed contribute to the hall conductivity as the half odd integer series ( @xmath0 ) when the fermi energy traverses the ( shifted sets of ) landau levels . the picture is also endorsed , via the bulk - edge correspondence , from the edge mode spectrum for the present model .
a binary matrix @xmath0 satisfies the consecutive - ones property ( c1p ) if its columns can be ordered in such a way that , in each row , all 1 entries appear consecutively . the c1p has been studied in relation to a wide range of problems , from theoretical computer science @xcite to genome mapping ( see @xcite and references there ) . the c1p can be naturally described in terms of covering hypergraph edges by walks . assume a binary matrix @xmath0 is the incidence matrix of a hypergraph @xmath1 , where columns represent vertices and rows encode edges ; then @xmath0 is c1p if and only if @xmath1 can be covered by a path that contains all vertices and where every edge appears as a contiguous subpath . deciding if a binary matrix is c1p can be done in linear time and space ( see @xcite and references there ) . if a matrix is not c1p , a natural approach is to remove the smallest number of rows from this matrix in such a way that the resulting matrix is c1p . this problem , equivalent to an edge - deletion problem on hypergraphs that solves the hamiltonian path problem , is np - complete , although fixed - parameter tractability ( fpt ) results have recently been published . at a high level of abstraction , genome assembly problems can be seen as graph or hypergraph covering problems : vertices represent small genomic sequences , edges encode co - localisation information , and one wishes to cover the hypergraph with a set of linear walks ( or circular walks for genomes with circular chromosomes ) that respect co - localisation information . such walks encode the order of elements along chromosomal segments of the assembled genome . one of the major issues in genome assembly problems concerns _ repeats_- genomic elements that appear , up to limited changes , in several locations in the genome being assembled . such repeats are known to confuse assembly algorithms and to introduce ambiguity in assemblies @xcite . modeling repeats in graph theoretical models of genome assembly can be done by associating to each vertex a _ multiplicity _ : the multiplicity of a vertex is an upper bound on the number of occurrences of this vertex in linear / circular walks that cover the hypergraph , and thus a vertex with a multiplicity greater than @xmath2 can traversed several times in these walks ( _ i.e. _ , encodes a repeat as defined above ) . this hypergraph covering problem naturally translates into a variant of the c1p , called the c1p with multiplicity ( mc1p ) that received little attention until recently , when it was investigated in several recent papers in relation to assembling ancestral genomes that describee both hardness and tractability results for decision and edge - deletion problems @xcite . in the present paper , we formalize the previously studied c1p and mc1p notions in terms of _ covering of assembly hypergraphs _ by linear and circular walks and edge - deletion problems ( section [ sec : preliminaries ] ) . next , we describe new tractability results for decision and edge - deletion problems ( section [ sec : results ] ) : we show that deciding if a given assembly hypergraph admits a covering by linear and circular walks that respects the multiplicity of all vertices is fpt and we describe polynomial time algorithms for decision and edge - deletion problems for families of assembly hypergraphs which encode information allowing us to clear ambiguities due to repeats . we conclude with several open questions ( section [ sec : conclusion ] ) . [ def : hypergraph ] an _ assembly hypergraph _ is a quadruple @xmath3 where @xmath4 is a hypergraph and @xmath5 are three mappings such that @xmath6 , @xmath7 , @xmath8 where @xmath9 is either a sequence on the alphabet @xmath10 where each element appears at least once , or @xmath11 ( the empty sequence ) . from now , we consider that @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 . a vertex @xmath18 such that @xmath19 is called a _ repeat _ ; @xmath20 is the set of repeats and @xmath21 . edges s.t . @xmath22 are called _ adjacencies _ ; from now , without loss of generality , we assume that @xmath23 if @xmath24 is an adjacency . edges s.t . @xmath25 ( resp . @xmath26 ) are called _ intervals _ ( resp . _ triples _ ) . we denote the set of adjacencies ( resp . weights of adjacencies ) by @xmath27 ( resp . @xmath28 ) and the set of intervals ( resp . weights of intervals ) by @xmath29 ( resp . @xmath30 ) . an interval is _ ordered _ if @xmath31 ; an assembly graph with no ordered interval is _ unordered_. from now , unless explicitly specified , our assembly hypergraphs will be unordered and unweighted . we call @xmath32 the _ multiplicity _ of @xmath18 . [ def : adjacencygraph ] an assembly hypergraph with no interval is an _ adjacency graph_. given an assembly hypergraph @xmath33 , we denote its _ induced adjacency graph _ by @xmath34 for every @xmath35 , as adjacencies are unordered . ] . [ def : compatibility ] let @xmath36 be an assembly hypergraph and @xmath37 ( resp . @xmath38 ) a linear ( resp . circular ) sequence on the alphabet @xmath39 . an unordered interval @xmath24 is _ compatible _ with @xmath37 ( resp . @xmath38 ) if there is a contiguous subsequence of @xmath37 ( resp . @xmath38 ) whose content is equal to @xmath24 . an ordered interval @xmath24 is compatible with @xmath37 ( resp . @xmath38 ) if there exists a contiguous subsequence of @xmath37 ( resp . @xmath38 ) equal to @xmath40 or its mirror . [ def : assembly ] an assembly hypergraph @xmath3 admits a _ linear assembly _ ( resp . _ mixed assembly _ ) if there exists a set @xmath41 of linear sequences ( resp . linear and/or circular sequences ) on @xmath39 such that every edge @xmath42 is compatible with at least one sequence of @xmath41 , and every vertex @xmath18 appears at most @xmath32 times in @xmath41 . the weight of an assembly is @xmath43 . an assembly as defined above can naturally be seen as a set of walks ( some possibly closed in mixed assemblies ) on @xmath1 such that every edge of @xmath44 is traversed by a contiguous subwalk . in the following , we consider two kinds of algorithmic problems that we investigate for different families of assembly hypergraphs and genome models , a decision problem and an edge - deletion problem . * the _ assembly decision problem _ : given an assembly hypergraph @xmath45 @xmath46 and a genome model ( linear or mixed ) , does there exist an assembly of @xmath47 in this model ? * the _ assembly maximum edge compatibility problem _ : given an assembly hypergraph @xmath33 and a genome model , compute a maximum weight subset @xmath48 of @xmath44 such that the assembly hypergraph @xmath49 admits an assembly in this model . [ def : repeatcluster ] let @xmath36 be an assembly hypergraph maximal repeat cluster _ is a connected component of the hypergraph whose vertex set is @xmath20 and edge set is @xmath50 . as outlined in the introduction , vertices in an assembly hypergraph represent genomic elements , each with an associated copy number @xmath32 , while edges and their order ( for intervals ) encode hypothetical co - localisation information , each with an associated weight . linear and/or circular sequences of vertices defining an assembly represent the order of these genomic elements along chromosomal segments , the circular ones representing circular chromosomes . a maximal repeat cluster encodes a group of elements that are believed to appear in several locations of the genome to assemble , although different occurrences might differ in terms of content and/or order ( see @xcite for example ) . such repeated structures cause ambiguity in genome assemblies based solely on adjacencies ; for example , if @xmath51 , with @xmath52 and @xmath53 , and @xmath54 , then there are essentially three possible linear assemblies ( @xmath55 ) , while adding the ordered interval @xmath56 leads to a single possible assembly . when no repeats are allowed ( @xmath57 ) , the assembly decision problem in the linear genome model is equivalent to asking if a binary matrix has the c1p , which can be solved in @xmath58 time and space . the set of all linear assemblies can be encoded into a compact data structure , the _ pq - tree_. in the mixed genome model , the problem can also be solved in linear time , as it reduces to testing the circular c1p for every connected component of the overlap graph of the matrix . the _ pc - tree _ , a slightly modified pq - tree , can be used to encode all mixed genome assemblies . we summarize some of these results in the following theorem and refer to @xcite for a survey on these questions . [ thm : cis1 ] the assembly decision problem can be solved in @xmath58 time and space when @xmath59 , in the linear and mixed genome models . in the linear genome model , the assembly maximum edge compatibility problem is hard for adjacency graphs it solves the problem of computing a set of paths that cover a maximum number of edges of the graph but fpt results have recently appeared @xcite . tractability results are less general when repeats are allowed , as shown below . [ thm : adadjacencies1]_@xcite _ ( 1 ) the assembly decision problem can be solved in time and space @xmath58 for adjacency graphs ( @xmath60 ) in the linear and mixed genome models . ( 2 ) in both genome models , the assembly decision problem is np - hard if @xmath61 and @xmath62 . the principle of the proof for ( 1 ) is that an adjacency graph admits a valid assembly if and only if every vertex has at most @xmath63 neighbours and , in the linear model , if every connected component @xmath38 satisfies @xmath64 . this result , combined with the use of pq - trees on the assembly hypergraph without its repeats , can be extended slightly in the linear genome model . [ thm : adadjacencies2]_@xcite _ the assembly decision problem can be solved in polynomial time and space in the linear genome model for unordered assembly hypergraphs where , for every edge @xmath24 containing a repeat , either @xmath24 is an adjacency or @xmath24 is an interval that contains a single repeat @xmath65 and there exists an edge @xmath66 . finally , to the best of our knowledge , the following is the only tractability result for edge - deletion problems when repeats are allowed , limited to adjacency graphs and the mixed genome model . [ thm : amedadjacencies1]_@xcite _ ( 1 ) the assembly maximum edge compatibility problem can be solved in polynomial time and space in the mixed genome model for adjacency graphs ( @xmath60 ) . ( 2 ) the assembly maximum edge compatibility problem is np - hard in the mixed genome model if @xmath61 , even if @xmath57 . we first show that the assembly decision problem is fpt with respect to parameters @xmath67 and @xmath68 . then we describe positive results for the case where the induced adjacency graph @xmath69 is assumed to admit an assembly and specific families of intervals are added to clear ambiguities caused by repeats . we discuss the practical implications of our positive results at the end of the section . the assembly decision problem can be solved in space @xmath70 and time @xmath71 in the linear and mixed genome models . the principle of the proof is , for the given assembly hypergraph @xmath72 and @xmath73 here as the weight does not impact decision problems and we deal with unordered hypergraphs . so , we eliminate both mappings from our notation . ] , to build another assembly hypergraph @xmath74 such that @xmath75 for all @xmath76 , by making @xmath77 copies of each @xmath78 and considering each possible set @xmath79 of choices of 2 neighbors for each of these copies . @xmath80 can then be checked for the existence of an assembly with theorem [ thm : cis1 ] . the sets @xmath79 of choices are made in such a way that @xmath81 has an assembly if and only if , for at least one of these sets @xmath79 of choices , @xmath82 has an assembly . finally , if @xmath83 and @xmath68 are fixed , we prove that there is a fixed number of such sets @xmath79 . let @xmath84 be the set of copies we shall introduce for each @xmath78 ( and @xmath85 ) , @xmath86 be the _ neighborhood _ of @xmath18 in @xmath1 , that is the set of vertices belonging to edges containing @xmath18 , and @xmath87 be the `` new neighborhood '' from which we choose neighbors for vertices in @xmath88 . we represent each set of possible choices of 2 neighbors of each @xmath89 with a mapping @xmath90 , where @xmath91 . let @xmath92 be the collection of these mappings ( itself a mapping @xmath93 where @xmath94 ) . we can now state the full algorithm as follows . 1 . for each @xmath95 , make @xmath77 copies of @xmath65 , which defines the set @xmath88 . let @xmath96 . 2 . for each @xmath97 , choose @xmath98 neighbours from @xmath99 , thus defining @xmath100 for every @xmath95 . this also defines @xmath79 as the collection of mappings @xmath100 over all @xmath95 . 3 . construct a new assembly hypergraph @xmath101 with @xmath102 , @xmath75 for all @xmath103 , and @xmath104 defined as follows : ( 1 ) for each @xmath89 , @xmath105 , @xmath106 for some @xmath107 , add @xmath108 and @xmath109 to @xmath110 ( @xmath79-edges ) and ( 2 ) for each @xmath111 , add an edge @xmath112 containing @xmath113 . 4 . for each @xmath114 adjacent to a vertex of @xmath115 , let @xmath116 be the unique path in @xmath117 s.t . @xmath118 and @xmath119 add all of @xmath120 to @xmath121 for each @xmath112 such that @xmath122 . 5 . use theorem [ thm : cis1 ] on @xmath123 . output yes and exit if @xmath80 admits an assembly in the chosen genome model iterate over all possible sets of neighbour choices @xmath79 in step 2 . output no if no @xmath123 admits an assembly in the chosen genome model . [ [ algorithm - correctness . ] ] algorithm correctness . + + + + + + + + + + + + + + + + + + + + + + the premise for the algorithm is the following claim , which we state and prove below . @xmath81 has an assembly if and only if , for some @xmath79 , @xmath80 has an assembly . [ lem : mapping ] first , if @xmath81 has the assembly @xmath124 , in @xmath124 , we replace each occurrence of a vertex @xmath78 by copies @xmath89 where @xmath125 . let this new assembly be called @xmath126 . each such @xmath127 is adjacent to at most 2 other distinct vertices . we consider the mapping @xmath79 which maps each such @xmath127 to its two neighbours in this assembly @xmath126 . if we can establish that the hypergraph obtained from this mapping and the new edges we introduce admits @xmath126 as an assembly , we are done . to decide if @xmath80 has an assembly , we first note that any set of covering walks on @xmath80 is a set of paths ( we can not visit the same vertex twice because @xmath75 for all @xmath128 ) . since @xmath124 is a covering walk of @xmath81 , by splitting the vertices of @xmath20 into distinct copies , we ensure that no vertex of @xmath80 is visited twice by @xmath126 . now , let us look at the set of edges @xmath110 . if all of them are covered as contiguous subsequences in @xmath126 , we are done . we show this by the following observations . 1 . in @xmath124 , every edge @xmath24 occurs as a contiguous subsequence . let @xmath121 be the edge in @xmath80 corresponding to @xmath24 . then , by definition of @xmath126 , @xmath121 must occur in it as a contiguous subsequence . 2 . for each @xmath129 for some @xmath78 , we defined @xmath106 using the assembly @xmath124 . so , we definitely get both adjacencies @xmath130 in @xmath126 . so , @xmath126 must be an assembly for @xmath80 , which implies that @xmath80 has an assembly . conversely , if the graph @xmath80 has an assembly , it contains all vertices @xmath131 , and occurrences of each @xmath132 for all repeat vertices @xmath78 . if we remove the subscripts , _ i.e. _ , @xmath127 becomes @xmath65 for all @xmath133 , we get an assembly @xmath124 , which we claim is an assembly for @xmath81 , as @xmath124 will have the following properties . 1 . every vertex @xmath134 appears at least once , and at most @xmath32 times . 2 . for every edge @xmath135 consisting only of vertices in @xmath131 , we get a contiguous occurrence of @xmath111 , which is the corresponding edge in @xmath81 . 3 . for every edge @xmath111 , such that @xmath136 for some @xmath137 , there is an edge @xmath112 such that @xmath89 has two neighbours and @xmath138 . in this case , we get a contiguous occurrence of @xmath121 including @xmath127 . removing the subscripts gives us a contiguous occurrence of @xmath24 in the new assembly @xmath124 . so , @xmath124 contains occurrences of every edge @xmath111 in @xmath81 as contiguous subsequences , which proves that @xmath124 is an assembly for @xmath81 . this proves the claim . this proof holds for both genome models as theorem [ thm : cis1 ] considers them both . [ [ algorithm - complexity . ] ] algorithm complexity . + + + + + + + + + + + + + + + + + + + + + the space complexity follows obviously from the construction of @xmath80 . the choice of neighbours can be made in at most @xmath139 ways for each new vertex . so , in total , we get at most @xmath140 possible mappings @xmath93 . the procedure on each @xmath141 can be done in time @xmath142 , since we just need to check its neighbours , which are at most @xmath98 . doing so for all vertices in @xmath143 takes time at most @xmath144 . the final step , checking for the existence of an assembly for a given @xmath80 , can be done in @xmath145 time , since we add at most @xmath146 new edges , and @xmath147 new vertices . now , we assume we are given an assembly hypergraph @xmath148 whose induced adjacency graph @xmath69 is known to have a mixed assembly . to state our result , we extend slightly the notion of compatibility : an unordered interval @xmath24 is said to be _ compatible _ with @xmath69 if there exists a walk in @xmath149 whose vertex set is exactly @xmath24 . we consider the interval compatibility problem defined below . the _ assembly maximum interval compatibility problem _ : given an assembly hypergraph @xmath33 such that @xmath69 admits a mixed assembly , compute a maximum weight subset of @xmath29 , @xmath150 , such that @xmath151 admits a mixed assembly . [ thm : maxcompatibility1 ] let @xmath152 be a weighted assembly hypergraph such that @xmath69 admits a mixed genome assembly , and each interval is a triple containing at most one repeat and compatible with @xmath69 . the assembly maximum interval compatibility problem in the mixed genome model can be solved for @xmath47 in linear space and @xmath153 time . the proof proceeds in two stages : we first show that repeat - free triples , as well as triples whose non - repeat vertices form an adjacency , must always be included in a maximum weight compatible set of triples . then , we present an algorithm which uses the adjacency compatibility algorithm of mauch et al . @xcite to decide which of the remaining triples to include . from now , we denote by @xmath154 a maximum weight subset of @xmath29 such that @xmath155 admits a mixed assembly . if a triple @xmath156 satisfies @xmath157 , with @xmath158 , then @xmath159 . as @xmath24 is assumed to be compatible with @xmath69 by hypothesis , there is a walk on these three vertices in @xmath160 . as a walk on three non - repeat vertices is a path , w.l.o.g we assume that the adjacencies in the path are @xmath161 and @xmath162 ( the argument holds by symmetry for the other cases ) . then , in any mixed assembly of @xmath69 , in order to contain both adjacencies , and to make sure that @xmath163 appears exactly once in the assembly , the assembly must contain @xmath24 , in the order @xmath164 . so , it must be included in @xmath154 , as @xmath154 is a maximum weight subset of @xmath165 . if a triple @xmath156 satisfies @xmath166 , with @xmath167 , @xmath168 and @xmath169 , then @xmath159 . for the triple @xmath24 to be compatible with @xmath69 , @xmath65 needs to be adjacent to at least one of @xmath170 and @xmath163 . assume , w.l.o.g , that @xmath171 . if @xmath69 admits a mixed assembly , both @xmath161 and @xmath172 must occur in a path or a cycle . furthermore , since @xmath173 , these two adjacencies must occur in the same path or cycle , in the order @xmath174 . this is an occurrence of @xmath24 as a contiguous sequence , which implies that such a triple must occur in every assembly of @xmath47 , and must be included in @xmath154 . we are now left with the set @xmath175 of triples @xmath166 such that @xmath65 is a repeat and @xmath176 , which means that @xmath65 is adjacent to both @xmath170 and @xmath163 , and we need to find a maximum weight subset of triples of this form . to do this , we rely on the optimal edge - deletion algorithm designed by mauch et al . @xcite for adjacency graphs as shown below . 1 . initialize an empty set @xmath177 and @xmath178 . 2 . for every @xmath179 : a. add an adjacency @xmath180 to @xmath177 , label @xmath181 with the triple @xmath24 , and set @xmath182 . b. remove @xmath183 and @xmath184 from @xmath48 , if present . 3 . for every remaining adjacency @xmath185 , set @xmath186 . 4 . apply the linearization algorithm ( theorem [ thm : amedadjacencies1 ] ) @xcite on @xmath187 . add the triples corresponding to the labels of the adjacencies from @xmath177 retained by the linearization algorithm to @xmath154 . _ algorithm correctness . _ given a triple @xmath166 with a repeat vertex @xmath65 and no adjacency @xmath188 , we consider a candidate mixed assembly of @xmath47 containing the elements of @xmath24 contiguously . in such an assembly , we would encounter the consecutive substring @xmath189 . we can contract this substring and label the newly formed adjacency @xmath161 , signifying that there is a path of length @xmath98 between @xmath170 and @xmath163 which passes through @xmath65 and contains no other vertices , _ i.e. _ , it encodes the triple @xmath24 . so , we construct the new assembly hypergraph ( an adjacency graph ) by deleting the adjacencies @xmath190 and @xmath172 and encoding the path containing @xmath24 into the adjacency @xmath161 added to @xmath177 . the optimal edge - deletion algorithm from @xcite computes a maximum weight set of adjacencies @xmath191 such that the assembly graph @xmath192 has a mixed assembly , where @xmath193 and @xmath194 are the restrictions of @xmath195 and @xmath196 to @xmath197 . in this assembly , we can replace every @xmath198 by the corresponding triple @xmath24 and the two corresponding adjacencies from @xmath27 . note that none of the adjacencies from @xmath27 are discarded during linearization since they are weighted so that discarding any one would be suboptimal when compared to discarding the entire set of adjacencies from @xmath177 . so the assembly obtained by this process will contain all the edges from @xmath27 , as well as a maximum weight set @xmath199 such that every @xmath200 is present . this implies that we computed a maximum weight compatible set of triples from @xmath175 . _ algorithm complexity . _ checking the compatibility of a triple @xmath24 with @xmath69 can be done in constant time , since we just need a @xmath201-step graph search from any vertex @xmath202 , and proceed until we find a path connecting all @xmath201 vertices in @xmath24 . we can also check the number of repeats in @xmath24 in constant time . to deal with triples from the set @xmath175 , the new assembly hypergraph can obviously be constructed in @xmath203 time and space , and contains @xmath204 vertices and @xmath205 edges . so the optimal edge - deletion algorithm is the main component of the process , and is based on a maximum weight matching algorithm of time complexity @xmath153 @xcite . related to this theorem , we have the following corollary . [ thm : maxcompatibility2 ] let @xmath152 be an assembly hypergraph such that @xmath69 admits a mixed genome assembly , maximal repeat clusters are all of size @xmath2 , and each interval is an unordered compatible triple . the assembly maximum interval compatibility problem in the mixed genome model can be solved for @xmath47 in linear space and @xmath153 time . we already know that we can find a maximal weight compatible subset @xmath199 if there is no @xmath156 containing more than @xmath2 repeat . we now show that for the current problem , a triple @xmath206 , where @xmath207 and @xmath208 are repeats , and @xmath209 , can also be included in the set @xmath154 if it is compatible with @xmath69 . note that @xmath207 and @xmath208 can not have an adjacency between them , since the size of a maximal cluster can not exceed @xmath2 . so , for @xmath24 to be compatible , the corresponding adjacencies will be @xmath210 and @xmath211 . for @xmath69 to have a mixed assembly which contains both adjacencies , the assembly must contain @xmath24 in the order @xmath212 . this is a contiguous appearance of the elements of @xmath24 , and it must occur in every mixed assembly . it can thus be included in @xmath154 . theorem [ thm : maxcompatibility1 ] concludes the proof . [ def : repeatinterval ] let @xmath36 be an assembly hypergraph . an interval @xmath213 is an _ ordered repeat spanning interval _ for a maximal repeat cluster @xmath214 if @xmath215 with @xmath216 , @xmath217 and @xmath218 , where @xmath219 is a sequence on the set @xmath220 , containing every element at least once . the subset of ordered repeat spanning intervals in @xmath165 is denoted by @xmath221 [ thm : repeat_spanning_compatibility ] let @xmath222 be an assembly hypergraph such that every repeat @xmath95 is either contained in an adjacency , or it is contained in an interval @xmath223 of one of the following forms . 1 . @xmath24 is an ordered repeat spanning interval . @xmath65 is the only repeat in @xmath24 , @xmath224 , and @xmath225 . the assembly decision problem in the linear genome model can be solved for @xmath47 in polynomial time and space . the basic idea of the proof is to realize the sequence @xmath40 for every repeat spanning interval @xmath226 by creating unique copies of the repeats in @xmath24 and decreasing the multiplicity accordingly . this leads to an assembly graph that can then be checked using theorem [ thm : adadjacencies2 ] . formally we define an extended assembly hypergraph , @xmath227 , as follows ( we omit @xmath228 from the notation , since we are addressing a decision problem ) . 1 . @xmath229 , @xmath230 , @xmath231 , @xmath232 , @xmath233 . 2 . for every repeat spanning interval @xmath234 . 1 . let @xmath235 , possibly @xmath236 for @xmath237 ( the @xmath127 are repeats ) . 2 . for @xmath133 from @xmath2 to @xmath238 1 . add a unique vertex @xmath239 to @xmath240 , with multiplicity @xmath241 , 2 . add an adjacency @xmath242 to @xmath48 for @xmath243 , 3 . decrease @xmath244 by 1 . 3 . add edges @xmath245 and @xmath246 to @xmath48 . 4 . if the adjacencies @xmath247 and @xmath248 are present , add them to @xmath177 . 3 . check if the assembly hypergraph , @xmath249 admits a linear genome assembly using theorem [ thm : adadjacencies2 ] . @xmath47 admits a valid genome assembly in the linear genome model if and only if @xmath250 for every repeat @xmath251 and @xmath252 admits one . assume @xmath252 admits an assembly @xmath253 . by construction , every repeat @xmath65 of @xmath20 maps to a subset of @xmath240 composed of @xmath65 and the vertices added when reading occurrences of @xmath65 in the ordered repeat spanning intervals of @xmath29 . for a repeat @xmath254 , let @xmath255 be this subset of @xmath240 and @xmath256 the inverse map . by construction , the adjacencies added to @xmath48 when reading the order @xmath40 of an interval @xmath24 , when the inverse map is applied @xmath256 to their vertices , define a walk in @xmath47 corresponding exactly to @xmath40 , which allows us to unambiguously translate the set of linear walks on @xmath257 defining @xmath253 into a set of linear walks @xmath41 on @xmath1 . this implies that every edge of @xmath44 is compatible with @xmath41 ( as defined in def . [ def : compatibility ] ) , and we only need to consider potential problems caused by multiplicities . assume that for every repeat @xmath251 one has @xmath250 and that for every @xmath258 , @xmath259 appears at most @xmath260 times in an assembly of @xmath252 , _ i.e. _ , exactly @xmath2 time , since @xmath261 for all @xmath258 . for a vertex @xmath134 such that @xmath262 , by construction @xmath263 , so an assembly of @xmath252 also satisfies the constraints of an assembly of @xmath47 for @xmath18 . for a repeat @xmath251 , the number of occurrences of elements of @xmath264 in @xmath253 is at most @xmath265 . by construction , @xmath266 , so assuming that @xmath250 implies that the constraint on @xmath77 is satisfied in the linear walks on @xmath41 . now , consider @xmath47 admits an assembly @xmath41 in the linear genome model . by definition , for every repeat spanning interval @xmath24 , @xmath40 appears as a walk in @xmath41 . by replacing the repeats in such a walk by new vertices with multiplicity @xmath2 as done in step 2.b of the algorithm above , one clearly obtains an assembly @xmath267 for @xmath252 , and the identity @xmath266 ensures that @xmath250 . _ _ the polynomial time and space complexity follows from theorem [ thm : adadjacencies2 ] , since the the construction of @xmath252 results in an assembly hypergraph with the structure in which no two repeats are contained in an interval ( the repeat spanning intervals being resolved ) , and if an interval @xmath185 contains a repeat @xmath65 , there exists an edge @xmath268 in @xmath48 , since we added them directly from @xmath47 . the following corollary follows easily from the previous theorem . [ cor : repeatspanning ] let @xmath152 be an assembly hypergraph such that each interval is an ordered repeat spanning interval . the assembly decision problem in the mixed and linear genome models can be solved for @xmath47 in @xmath269 time and space . we make the same construction as in theorem [ thm : repeat_spanning_compatibility ] . the extended assembly graph @xmath270 we create now is composed entirely of adjacencies , since @xmath271 . an application of theorem [ thm : adadjacencies1 ] completes the proof . the time and space complexities follow immediately from the linear time and space complexities stated in theorem [ thm : adadjacencies1 ] and from the size of @xmath270 . the results above have interesting practical implications that we outline now . first , corollary [ cor : repeatspanning ] shows that , if provided with ordered repeat spanning intervals , one can check for the existence of an assembly in both genome models . ordered repeat spanning intervals can be obtained in practice in several ways , such as mapping the elements of @xmath39 onto related genomes @xcite or long reads ( see appendix for more details ) . the tractability of the assembly decision problem , with linear time and space complexities , makes it possible to combine it with the tractability result of theorem [ thm : amedadjacencies1 ] to select a subset of adjacencies , followed by a greedy heuristic for the assembly maximum interval compatibility problem . note also that the condition on the unordered intervals in the statement of theorem [ thm : repeat_spanning_compatibility ] allows one to account for the important notion of _ telomeres _ @xcite . regarding theorem [ thm : maxcompatibility1 ] , it can be used to partially clear the ambiguities caused by repeats in assembly hypergraphs where triples are obtained from mate - pairs of reads from sequencing libraries defined with inserts of length greater than the length of repeats@xcite . if all maximal repeat clusters are `` collapsed '' into a single vertex ( with the maximum multiplicity among all initial repeats of the cluster ) , such mate - pairs spanning repeat clusters define the triples . solving the assembly maximum interval compatibility problem allows us to specify the locations of the different occurrences of the spanned repeat clusters in the assembled genome , thus resolving part of the ambiguity due to repeats and leaving only the internal structure of each repeat cluster ( content and order ) unresolved . in the present work , we presented a set of positive results on some hypergraph covering problems motivated by genome assembly questions . to the best of our knowledge , these are the first such results for handling repeats in assembly problems in an edge - deletion approach , as previous results focused on superstring approaches @xcite , and these new methods have been applied on real data @xcite . moreover , the initial results we presented suggest several open problems . first , our results about triples assume that they are compatible with @xmath69 ( _ i.e. _ , appear as walks in @xmath160 ) ; we conjecture that similar positive results can be obtained when relaxing this condition ( in particular when triple elements might not appear in the same connected component ) . next , our edge - deletion positive results assume that @xmath69 admits a genome assembly , and only intervals are considered for being deleted . this leads to a two - stage assembly process where adjacencies are deleted first , followed by intervals . it remains open to see if both adjacencies and limited families of intervals can be considered jointly . also of interest would be to see if the size of maximal repeat clusters or of intervals can be used as parameters for fpt results . regarding repeat - spanning intervals , it can be asked if one can relax the total order structure @xmath73 to account for uncertainty ; for example , if they are defined from the comparison of pairs of related genomes , it might happen that specific rearrangements lead to conserved genome segments that can be described by partial orders @xcite , which opens the question of solving the assembly decision problem with partial orders to describe repeat - spanning intervals . along the same line , it might happen that intervals spanning only prefixes or suffixes of repeat occurrences ( called _ repeat - overlapping intervals _ ) can be detected , and the tractability of the assembly decision problem with such intervals is open ; we conjecture it is fpt in the number of such intervals . finally , _ gaps _ , that can be described in terms of binary matrices , as entries @xmath272 appearing between entries @xmath2 , appears naturally in genome scaffolding problems @xcite ; the notion of gaps can naturally be described , for graphs , in terms of _ bandwidth _ and has been extended to binary matrices / hypergraphs in @xcite . very limited tractability result exist when gaps are allowed , whether it is for graphs @xcite or hypergraphs @xcite , none considering repeats , which opens a wide range of questions of practical importance . in this appendix , we describe how the assembly hypergraph relates to practical genome assembly problems . our initial motivation for investigating the algorithmic problems described in this paper follows from earlier computational paleogenomics methods developed to compute genome maps and scaffolds for ancestral genomes @xcite . in this problem , the vertex set @xmath39 represents a set of @xmath204 ancestral genomic markers , obtained either through whole genome alignment @xcite , the analysis of gene families @xcite , or the sequencing of an ancient genome @xcite . the function @xmath273 encodes the multiplicity , that is an upper bound on the allowed number of copies of each marker in potential assemblies . for ancestral genomes , it can be obtained from traditional parsimony methods @xcite . an edge @xmath274 encodes the hypothesis that @xmath275 appear _ contiguously _ in an assembly of the elements of @xmath39 . for ordered intervals , that are edges @xmath24 , such that @xmath25 and @xmath276 , @xmath40 encodes a total ordering information about the genomic elements they contain . in computational paleogenomics , edges and intervals ( including order ) can be obtained from the comparison of pairs of genomes related to the ancient genome that is being assembled . the function @xmath228 is a weight that can be seen as a confidence measure on every edge ( the higher , the better ) , that can be based on phylogenetic conservation . more generally , the assembly hypergraph is a natural model for genome mapping problems @xcite . however , the assembly hypergraph also allows us to formalize other assembly problems . for example , in the _ scaffolding _ problem @xcite , @xmath39 would represent _ contigs _ and @xmath273 can be obtained by methods based on the reads depth of coverage @xcite . co - localization information can be obtained from mate - pairs libraries with an insert that is short with respect to the minimum contig length , thus describing adjacencies , while ordered intervals can be obtained from mapping contigs onto long reads @xcite or related genome sequences @xcite . the assembly hypergraph can also be used to model the problem of assembling short reads into contigs , although contig assembly is generally based on eulerian superstring approaches @xcite instead of edge deletions approaches . in this problem , the vertices @xmath39 represent short sequence elements , such as reads in the overlap graph approach @xcite or @xmath238-mers ( substrings of length @xmath238 ) in the widely used de bruijn graph approach @xcite . the function @xmath273 can here again be obtained from the reads depth of coverage . adjacencies follow from overlaps between elements of @xmath39 , whose statistical significance , combined with the read quality for example , can be used to define @xmath228 . intervals can here again be obtained from mapping short reads on long reads . finally , it is important to remember that genomic segments are _ oriented _ along a chromosome , due to the double stranded nature of most genomes . the algorithms we described in the present paper can handle this problem in a very easy way . each genomic element is represented by two vertices , one for each extremity , with an adjacency linking them ( called a _ required _ adjacency , while adjacencies between extremities of different elements are called _ inferred _ adjacencies ) . a compatible assembly then needs to be composed of linear or circular walks where required adjacencies alternate with inferred adjacencies . this property can be handled naturally by the decision algorithms ( see @xcite ) , and also by the optimization algorithms by weighting each required adjacency by a weight greater than the cumulative weight of all inferred adjacencies . also , triples that overlap repeats need to be replaced by quadruples containing both extremities of a same initial genomic element , which can be handled by our algorithms ( full details will be given in the complete version of our work ) .	the consecutive - ones property ( c1p ) is a classical concept in discrete mathematics that has been used in several genomics applications , from physical mapping of contemporary genomes to the assembly of ancient genomes . a common issue in genome assembly concerns repeats , genomic sequences that appear in several locations of a genome . handling repeats leads to a variant of the c1p , the c1p with multiplicity ( mc1p ) , that can also be seen as the problem of covering edges of hypergraphs by linear and circular walks . in the present work , we describe variants of the mc1p that address specific issues of genome assembly , and polynomial time or fixed - parameter algorithms to solve them .
out - of - equilibrium dynamics following a quantum quench is a topic of intense study at present . the theme is pursued primarily along two lines . the first one is about the equilibration and thermalization mechanism of a quantum system @xcite , a fundamental yet still open issue in statistical physics . the second one is about the the real - time dynamical behavior of a many - body system @xcite , which is highly non - trivial in the regime where the quasi - particle picture breaks down . among all the models investigated so far , the bose - hubbard model takes a special position . as a paradigmatic strongly - correlated model , it can be realized accurately with cold atoms in optical lattices , and especially , the parameters can be controlled ( e.g. changed suddenly ) to a high degree @xcite . this nice property makes it an ideal candidate to investigate quantum quench dynamics both theoretically and experimentally . up to now , in the few theoretical works on the quench dynamics of the bose - hubbard model @xcite , the state of the system before the quench is always assumed to be the ground state of the initial hamiltonian . that is , the system is assumed to be at zero temperature initially . however , in this paper we shall start from a thermal equilibrium state . one should note that this scenario is actually more experimentally relevant . because in current experiments , one generally gets not a single tube of cold atoms , but instead a two - dimensional array of one - dimensional lattices for the cold atoms @xcite . in other words , an _ ensemble _ of one - dimensional bose - hubbard models is obtained in one shot . moreover , in view of the fact that the cold atoms are at finite temperatures necessarily @xcite , it is reasonable to start from a thermal state described by a canonical ensemble density matrix [ see eq . ( [ rhoi ] ) below ] . as emphasized by linden _ et al_. @xcite , in the pursuit of thermalization , it is important to distinguish the two closely related but inequivalent concepts of equilibration and thermalization . the latter is much stronger and has the trademark feature of the boltzmann distribution , whereas the former refers only to the stationary property of the density matrix of a ( sub)system or some physical observables . it is highly possible that a system equilibrates but without thermalization . this is actually the case for the bose - hubbard model . as revealed both in previous works ( zero temperature case ) @xcite and in the present paper ( finite temperature case ) , the bose - hubbard model thermalizes only if the quench amplitude is not so large , at least at the finite sizes currently accessible . however , it will be shown below that in a much wider range of parameters , some generic physical observables equilibrate very well . among them are the populations on the bloch states , which are ready to measure by the typical time - of - flight experiment @xcite . remarkably , this is actually guaranteed for these quantities in the thermodynamic limit , i.e. , when the size of the system gets large enough . the equilibration behavior of the physical observables imposes a question . it is ready to recognize that the equilibration of the physical observables is largely an effect of interference cancelation . it never means that the density matrix has suffered any dephasing or decoherence . actually , the density matrix evolves unitarily and in the diagonal representation of the hamiltonian , its elements simply rotate at constant angular velocities . a natural question is then , does the time - dependence of the density matrix has any chance to exhibit it , given that it is almost absent in the average values of the physical observables ? this leads us to consider giving the system a second quench . the concern is , would the system yield different long - time behaviors if the second quench comes at different times ? it turns out that the answer depends on whether the intermediate hamiltonian is integrable or non - integrable . in the former case , the density matrix shows repeated appreciable recurrences and thus the dependence on the second quench time is apparent . in the latter case , on the contrary , the density matrix shows no sign of recurrence and quantitatively similar long - time dynamics is observed for quenches at different times . this paper is organized as follows . in sec . [ sec2 ] , the setting of the problem and the basic approaches are given . in sec . [ sec3 ] , the dynamics after a single quench is studied . the time - averaged density matrix and the real - time evolution of some physical observables are investigated in detail . based on the observation in this section , we proceed to study the scenario of a second quench in sec . [ sec4 ] . finally , we summarize the results in sec . the time - dependent hamiltonian of the bose - hubbard model is ( @xmath1 throughout this paper ) @xmath2 here @xmath3 is the number of sites ( the total atom number will be denoted as @xmath4 ) and @xmath5 ( @xmath6 ) is the creation ( annihilation ) operator for an atom at site @xmath7 . note that here periodic boundary condition is assumed . the parameters @xmath8 and @xmath0 are the nearest - neighbor hopping strength and the on - site atom - atom interaction strength , respectively . note that the dynamics of the system depends only on the ratio @xmath9 , thus we will set @xmath10 throughout . we say the system is quenched if @xmath0 is changed suddenly at some time from one value to another value . experimentally , for cold atoms in an optical lattice , this can be realized by using the feshbach resonance . assume that initially the parameter @xmath0 is of value @xmath11 ( the corresponding hamiltonian is denoted as @xmath12 ) , and the system is at thermal equilibrium and of inverse temperature @xmath13 . denote the @xmath14-th eigenvalue and eigenstate of @xmath12 as @xmath15 and @xmath16 , respectively . the initial density matrix of the system is then @xmath17 is the partition function and @xmath18 is the probability of occupying the eigenstate @xmath19 . note that @xmath20 is the dimension of the hilbert space @xmath21 . the density matrix at time @xmath22 is given formally as @xmath23 , with @xmath24 $ ] . here @xmath25 means time ordering . the hamiltonian @xmath26 is invariant under the translation @xmath27 . this indicates that the total quasi - momentum of the system @xmath28 , where @xmath29 is the creation operator for an atom in the @xmath30-th bloch state , is conserved . this property implies that if the full hilbert space is decomposed into @xmath3 subspaces according to the values of @xmath31 , i.e. , @xmath32 , the hamiltonian and the density matrix are always block - diagonal with respect to the @xmath31-subspaces , i.e.,@xmath33 and @xmath34 @xcite . it is then possible to study the dynamics in each subspace individually ( which saves a lot of computational resource ) and then gather the information together ( note that for the expectation values of quantities like @xmath35 , there are contributions from each subspace ) . here it is necessary to mention that though we should have done the gathering or averaging process for many quantities studied below , we would rather not do so , because it is observed that the system behaves quantitatively similar in all the @xmath31-subspaces @xcite . a single @xmath31-subspace captures the overall behavior very well . therefore , our strategy is to focus on some specific @xmath31-subspace ( @xmath36 actually ) and take the normalization @xmath37 . _ it is understood that in the following all hamiltonians , density matrices , eigenvalues , and eigenstates refer to those belonging to this specific @xmath31-subspace . we will drop the superscript @xmath31 for notational simplicity . _ suppose at time @xmath38 the system is quenched by changing the value of @xmath0 from @xmath11 to @xmath39 , which is then held on forever . the hamiltonian later will be denoted as @xmath40 , and the eigenvalues and eigenstates associated will be denoted as @xmath41 and @xmath42 , respectivley . in the representation of @xmath43 , the density matrix at time @xmath22 is then simply ( in this paper @xmath44 means quantum state averaging while @xmath45 means time averaging ) @xmath46 where @xmath47 is the dimension of the specific @xmath31-subspace . it will prove useful to define the time - averaged density matrix @xmath48 the time - averaged density matrix is of great relevance for our purposes . first , it is both time - independent and variable - independent . second , the time - averaged value of an arbitrary operator @xmath49 is given simply by @xmath50 . that is , the time - averaged density matrix contains the overall information of the dynamics of the system . actually , as we will see later , for some quantities which fluctuate little in time , the time - averaged density matrix tells almost a complete story . third , the process of averaging over time is a process of relaxation in the sense that the entropy associated with @xmath51 is definitely no less than that with the density matrix at an arbitrary time , i.e. , @xmath52 . this is a corollary of the klein inequality @xcite and is reasonable since @xmath53 contains all the information of @xmath51 while the inverse is invalid . the equality also means that @xmath54 will never be damped , and time - averaging is essential . note that when @xmath55 , generally there is no degeneracy between the eigenvalues of @xmath40 . therefore the time - averaged density matrix is simply diagonal in the basis of @xmath56 , i.e. , @xmath57 with @xmath58 being the population on the eigenstate @xmath59 . in the special case of @xmath60 , the hamiltonian reduces to @xmath61 , with @xmath62 . in this case , each eigenvalue is of the form @xmath63 , under the constraints @xmath64 and @xmath65 , and there can be level degeneracy . however , we can always make some unitary transforms in each degenerate subspace to make sure that @xmath51 is in the form of ( [ diag ] ) . since the time - averaged density matrix provides an overall information of the dynamics of the system , we look into it first . in fig . [ fig1 ] , we consider the scenario of starting from the same initial condition ( @xmath66 , @xmath67 ) but quenching to six different values of @xmath39 @xcite . in each panel , the logarithms of @xmath68 are plotted against the eigenvalues @xmath69 ( red dots ) . we have compared @xmath51 with a canonical ensemble density matrix @xmath70 , which is defined as @xmath71 under the condition @xmath72 . here @xmath73 , the final inverse temperature , is the only fitting parameter . in fig . [ fig1 ] , the green dots which form a straight line correspond to @xmath70 . we see that @xmath51 exhibits many interesting features . in the case of @xmath60 , @xmath51 agrees well with @xmath70 throughout the spectrum . in the case of @xmath74 , @xmath51 agrees well with @xmath70 in the lower part of the spectrum , while deviates from it significantly in the higher part of the spectrum . but overall the two are in good agreement since the weight of the higher part is small . the case of @xmath75 is somewhat the reverse of the @xmath74 case . it is in the lower part of the spectrum that @xmath76 fluctuates wildly . in the higher part @xmath76 goes almost linearly . since the weight is dominated by the lower part , @xmath70 is not a good approximation of @xmath51 . in the strong interaction limits of @xmath77 , another feature takes the place . as a whole the red dots do not fall close to a single straight line , but they do form some stripes , and the stripes are almost parallel with a common slope close to @xmath78 . it is easy to recognize that each stripe corresponds to a bump in the density of states of @xmath40 . in order to understand the various features in fig . [ fig1 ] , we rewrite @xmath68 as @xmath79 where @xmath80 is a probability distribution @xcite associated with @xmath81 . note that @xmath82 is an intrinsic property of @xmath83 independent of @xmath78 . we have tried to characterize the distribution @xmath82 by its mean @xmath84 , its second central moment @xmath85 , and its third central moment @xmath86 , which are defined as follows , [ mean2 ] @xmath87 these quantities are presented in fig . these data enable us to understand fig . suppose for a distribution @xmath82 with @xmath88 , we define a gaussian distribution @xmath89 which shares the same mean and variance with @xmath90 but has vanishing third central moment . replacing @xmath90 in eq . ( [ pm2 ] ) by @xmath91 , we get an approximation of @xmath68 , @xmath92 in fig . [ fig1 ] , @xmath93 are represented by the blue dots . we see that as a whole @xmath93 is a good approximation of @xmath68 , except at the lower part of the spectrum in fig . the reason is clear the @xmath94 s there are the largest throughout all the figures , which indicates that the corresponding distributions @xmath90 are wide and asymmetric and thus can not be well approximated with a gaussian distribution . now we can understand the good fittings in figs . [ fig1]c and [ fig1]d . in these two cases , @xmath84 is almost a linear function of @xmath69 , and @xmath85 does not vary so much , therefore the exponent in eq . ( [ pmprime ] ) goes almost linearly with @xmath69 . the situation is similar in the higher part of the spectrum in fig . [ fig2]b , and therefore we have a good linear fitting for the higher spectrum part in fig . [ fig1]b . in contrast , in fig . [ fig2]e , @xmath84 varies wildly for adjacent @xmath69 , therefore we see in fig . [ fig1]e large fluctuations about the straight line . as for the parallel stripes in figs . [ fig1]a and [ fig1]f , they are also understandable in terms of figs . [ fig2]a and [ fig2]f , where @xmath84 form parallel stripes . it is numerically checked and can be argued that the slopes of the stripes are almost unity . actually we have @xmath95 where @xmath96 . note that in the limit of large @xmath97 , the kinetic term in the hamiltonian ( [ h ] ) can be viewed as a perturbation to the second interaction term . the spectrum of the latter is highly degenerate and consists of integral multipliers of @xmath39 . the effect of the perturbation is to mix up the eigenstates of the interaction hamiltonian with different eigenvalues and smooth the spectrum . that is why there are bumps in the density of states in figs . [ fig1]a and [ fig1]f and two adjacent bumps are placed roughly @xmath39 apart . by perturbation theory , it is easy to show that the second term in eq . ( [ slope ] ) varies on the order of @xmath98 among eigenstates belonging to the same bump . therefore , approximately we have @xmath99 for each bump and this explains why the stripes in figs . [ fig2]a and [ fig2]f are of slope unity . in turn it explains [ with the help of eq . ( [ pmprime ] ) ] why we have the parallel stripes in figs . [ fig1]a and [ fig1]f , and especially the slopes are approximately @xmath78 . it seems in fig . [ fig1 ] that @xmath70 is a good approximation of @xmath51 only when @xmath100 is small . in fig . [ fig3 ] , we employ the tools of distance @xmath101 , fidelity @xmath102 , and relative entropy @xmath103 ( for the definitions see @xcite ) between two density matrices to quantify the difference or resemblance between @xmath70 and @xmath51 . there it is clear that only in the range of @xmath104 , we have @xmath105 , which means @xmath51 is close to @xmath70 . in the subsequent subsection we will see that only in this range the expectation values of some generic physical observables according to @xmath51 and @xmath70 agree well . the distance @xmath101 and fidelity @xmath102 between @xmath70 and @xmath51 , and the relative entropy of @xmath70 with respect to @xmath51 , as functions of @xmath39 . the initial state is the same as in fig . [ fig1].,scaledwidth=35.0% ] we now proceed to study the time evolution of the system after the quench . in fig . [ fig2 ] , we show the time evolution of the populations on the bloch states @xmath106 . the six sub - figures correspond to those in fig . [ fig1 ] respectively . for all the @xmath39 s and all the @xmath30 s , @xmath107 equilibrate to their average values after a transient time , which is relatively longer in the cases of @xmath75 and @xmath108 . in the special case of @xmath60 , there is no fluctuation at all . the reason is simply that in this case , @xmath109 are conserved . we see that the time - averaged values of @xmath110 predicted by @xmath51 ( @xmath111 ) and @xmath70 ( @xmath112 ) agree relatively well in the cases of @xmath60 and @xmath113 . this is consistent with the closeness between @xmath51 and @xmath70 for these two values of @xmath39 , as revealed in fig . [ fig1 ] and fig . [ fig3 ] . here we would say the system thermalizes well in the @xmath74 case , however , we would refrain making the same statement for the @xmath60 case . the reason will be clear in the next section . figure [ fig4 ] is about a finite - sized system with some specific initial condition . however , here we have some general statements . we argue that in the thermodynamic limit ( @xmath114 with @xmath115 fixed ) , as long as initially the system is at finite - temperature thermal equilibrium and described by a canonical ensemble density matrix as ( [ rhoi ] ) , we should see steady behaviors of the physical variables like @xmath109 . let @xmath116 and let @xmath117 . the ensemble - averaged value of @xmath118 at time @xmath22 is @xmath119,\ ] ] where @xmath120 . its time - averaged value is @xmath121 here note that for a generic hamiltonian @xmath40 , there is no level degeneracy . the time - averaged value of @xmath122 is @xcite @xmath123 note that here it is assumed that there is no degeneracy of energy gaps . thus we have for the variance of @xmath124 in time , @xmath125 , @xmath126 since @xmath118 is semi - positive definite and bounded , we have @xmath127 . thus we have @xmath128 here we note that the summation is the square of the frobenius norm of @xmath53 in the representation of @xmath129 , which is invariant in all representations and is preserved by an arbitrary unitary evolution @xcite . explicitly , we have @xmath130 we argue that this quantity , which depends only on the initial state , decays exponentially with the size @xmath3 . let @xmath131 increase with @xmath14 . we have @xmath132 as @xmath133 . here in the @xmath134 relation we used the fact the ground state energy @xmath135 of @xmath12 scales linearly with @xmath3 and so does the free energy @xmath136 of the initial state @xcite . the coefficients @xmath137 and @xmath138 are independent of @xmath3 . moreover , it is easy to see that @xmath139 for any @xmath78 , with the equality taken only in the limit of @xmath140 or @xmath141 , and @xmath142 increases monotonically with @xmath143 . this makes sure that @xmath144 would not grow exponentially with @xmath3 and transcend unity . with ( [ bound1 ] ) and ( [ bound2 ] ) , we get an upper bound for @xmath145 , @xmath146 where @xmath147 is some constant . the upper bound of @xmath145 helps us determine an upper bound for the probability of finding @xmath124 deviating away from the mean @xmath148 by a distance larger than @xmath149 . actually , following reimann @xcite , using the chebyshev inequality @xcite , we have @xmath150 for a fixed value of @xmath149 , the upper bound decreases exponentially with the size of the system according to ( [ bound3 ] ) . it then follows the statement above . here some comments are worthy . though in the derivation above we have in mind a sudden quench , it is easy to see that the conclusion actually applies to any type of quench ( e.g. , the hamiltonian can be changed continuously over some period , as in @xcite , or quenched multiple times as in sec . [ sec4 ] below ) , as long as after some point the hamiltonian is never changed again . the reason lies in that the frobenius norm of the density matrix @xmath54 is conserved under unitary evolutions , and thus is independent of the historical or the final values of @xmath26 , but is determined entirely by the initial state . as for the operator @xmath118 , only the properties of semi - positive - definiteness and boundedness are used . thus similar conclusions can apply to other operators such as @xmath151 and @xmath152 , or operators in other models . finally , it should be mentioned that the conclusion relies on the fact that the quantity in eq . ( [ fnorm ] ) is bounded by some exponentially decreasing function , which is the case only at finite temperatures ( @xmath153 ) . at zero temperature , the quantity in eq . ( [ fnorm ] ) is always equal to unity and thus the problem is still open . it is shown in fig . [ fig4 ] that after a finite transient time , the physical variables equilibrate to their average values exhibiting minimal fluctuations . moreover , it has been proven that the amplitudes of the fluctuations will decrease exponentially with the size of the system . therefore , the observation is that the system , described by the density matrix @xmath54 , is almost indistinguishable from a system described by the time - averaged density matrix @xmath51 , as far as the simple realistic physical variables are concerned . this is remarkable . because though @xmath54 evolves unitarily and suffers no loss of information of @xmath53 , it behaves as if it were fully decoherenced . the question is then , to what extent can we hold onto this proposition ? is it possible to distinguish @xmath54 and @xmath51 , or @xmath154 and @xmath155 ( @xmath156 ) , by some means ? motivated by this problem , we have considered the scenario of giving the quenched system a second quench . that is , after the first quench at @xmath38 which changes @xmath0 from @xmath11 to @xmath39 , at time @xmath157 , the system is quenched again by changing the value of @xmath0 from @xmath39 to @xmath158 , which is then held on forever . the concern is , would the long - time dynamics of the system depends on the specific time @xmath159 ? denote the hamiltonian associated with @xmath158 as @xmath160 . the density matrix of the system later is given by @xmath161 ( for @xmath162 ) . as before , we are interested in the long - time averaged value of @xmath54 , @xmath163 since it has been shown and proven above that the dynamics of the system is to a large extent captured by the time - averaged density matrix . here the subscript indicates the dependence on the time @xmath159 . it is also useful to define the average of @xmath164 with respect to @xmath159 , @xmath165 the second equality means that @xmath166 is actually the time - averaged density matrix associated with an initial state @xmath51 [ see eqs . ( [ def_bar_rho ] ) and ( [ diag ] ) ] and a hamiltonian @xmath160 . one purpose of defining @xmath166 is to set a reference state independent of @xmath159 . to gain an overall idea of the dependence on @xmath159 of the long - time dynamics , we have studied the distance between @xmath164 and @xmath166 @xcite , and the time - averaged value of @xmath167 , @xmath168 as functions of @xmath159 . note that the average value of @xmath169 with respect to @xmath159 is given by @xmath166 , @xmath170 this is another reason for defining @xmath166 . the quantities @xmath171 and @xmath172 are shown in fig . eight pairs of @xmath173 are examined with the same initial condition as in fig . we see that for all cases with @xmath55 , both @xmath101 and @xmath174 set down to their average values quickly . however , for the special case of @xmath60 , both @xmath101 and @xmath175 display repeated recurrences , without any sign of equilibration . the situation is the reverse of that in fig . [ fig4 ] , where @xmath176 does not show any fluctuations in the case of @xmath60 . this phenomenon is due to the recurrence of the density matrix @xmath54 to @xmath53 @xcite . from eq . ( [ rhot ] ) , we see that in the representation of @xmath129 , the @xmath177-th off - diagonal element of @xmath54 rotates at an angular frequency of @xmath178 . in the generic case of @xmath179 , the energy gaps @xmath178 are quite random and incommensurate , and thus recurrence of the density matrix is rare . more precisely , the span between two times when all the matrix elements of @xmath54 get ( nearly ) in phase again is extraordinarily large . on the contrary , in the special case of @xmath60 , all eigenvalues and hence all the energy gaps @xmath178 are integral combinations of the few basic frequencies @xmath180 , and thus the probability of recurrence is much higher to demonstrate that the sharp peaks in figs . [ fig5]c and [ fig5]d are due to recurrences of the density matrix @xmath54 to @xmath53 , we define the figure - of - merit of recurrence , @xmath181 where the prime means the summation is over @xmath182 such that @xmath183 . it is clear that @xmath184 and @xmath185 when and only when all the off - diagonal elements get in phase . in figs . [ fig6]a and [ fig6]b , which share the same parameters as figs . [ fig5]c and [ fig5]e respectively , we have shown @xmath186 together with @xmath171 and @xmath174 . in fig . [ fig6]a , we see that every time @xmath171 and @xmath172 get close to their values at @xmath187 , @xmath186 shows a peak . in other words , there is a strong positive correlation between @xmath186 and @xmath171 and @xmath172 . in comparison , in fig . [ fig6]b , @xmath186 drops quickly from unity to less than @xmath188 and remains low all the time , and in turn @xmath171 and @xmath189 do not show any recurrence . to further consolidate the connection between the recurrence of @xmath54 and that of @xmath171 and @xmath174 , we have considered the case of @xmath190 . in this case , if @xmath60 , all the basic frequencies @xmath180 are commensurate , and thus there exist perfect recurrences , as shown in fig . there we see clearly that @xmath171 and @xmath172 return to their original values at @xmath187 periodically , and this happens when and only when @xmath191 returns to unity . however , once @xmath192 is set nonzero ( see fig . [ fig6]d ) and thus the commensurability of the energy gaps is destroyed , the situation returns to that in fig . [ fig6]b . the fact revealed in fig . [ fig5 ] and fig . [ fig6 ] is quite interesting . the long - time dynamics of the system is sensitive or insensitive to the exact time when the second quench is applied , depending on whether the intermediate hamiltonian @xmath40 is integrable ( @xmath60 ) or non - integrable ( @xmath55 ) . in the integrable case , @xmath172 exhibits large fluctuations and repeated recurrences . the system retains the memory of the initial state under the control of the hamiltonian @xmath40 . by contrast , in the non - integrable case , @xmath193 go over to their average values ( predicted by @xmath166 ) after a transitory period , showing little dependence on @xmath159 afterwards . combined with fig . [ fig4 ] , the picture is that @xmath54 evolving under the control of a non - integrable hamiltonian , not only yields the expectation values of @xmath194 as if it were @xmath51 , but even responds to the second quench as if it were @xmath51 . in fig . [ fig7 ] , we have checked this picture by studying the real time evolution of @xmath195 with @xmath196 under the double - quench scenario . the eight figures shown correspond to those in fig . [ fig5 ] respectively . for each pair of @xmath173 , we have studied the evolution of @xmath197 for several different values of @xmath159 . we see that in all the cases with @xmath55 , as long as @xmath159 is larger than the transient time , which can be roughly read from fig . [ fig5 ] , the later evolution of @xmath198 is quantitatively independent of @xmath159 . on the contrary , in the case with @xmath60 , the later values of @xmath199 vary wildly for different values of @xmath159 . here it is instructive to combine fig . [ fig4 ] and fig . [ fig7 ] and compare . in the @xmath55 cases , there is a sense of typicality @xcite . the density matrix @xmath54 governed by @xmath40 is surely non - stationary . however , for @xmath54 at different times , they yield almost the same expectation values for the observables , and moreover , they share almost the same response to the same quench . in the case of @xmath60 , what fig . [ fig7 ] reveals is a good complement to that in fig . it demonstrates that it is inappropriate to say that the system thermalizes in this case , even though the density matrices and expectation values of the observables agree since according to one s everyday experience , a system in thermal equilibrium should not show any time dependence . we have studied the quench dynamics of the bose - hubbard model both analytically and numerically . the issues of thermalization and equilibration are investigated thoroughly . on the thermalization side , which concerns whether the quenched system behaves like a canonical ensemble , it is found that this is the case only for small - amplitude quenches ( at least for the finite - sized system investigated ) . however , the time - averaged density matrix does manifest many interesting features in different regimes . these features are self - consistently understood after a study of the overlaps between the eigenstates of @xmath12 and @xmath40 . here we would like to say that it is urgent and would be very helpful to develop some analytical tools so that some general relations between the eigen - systems of @xmath12 and @xmath40 can be established . these tools and relations would also be useful to determine whether the non - thermalizationphenomenon observed is just a finite - size effect . on the equilibration side , which is about whether physical observables relax to stationary values without appreciable fluctuations , the result is that this is indeed the case for quantities as @xmath200 which are of most interest . moreover , it is proven analytically that for these quantities the fluctuations in time will decay exponentially with the size of the system . therefore , the overall picture is that generally the system equilibrates but without thermalization . the second quench reveals something more subtle . first , the subsequent dynamics depends or not on the second quench time @xmath159 according to @xmath60 or not . the underline reason is the recurrence or not of the initial density matrix , which in turn has its root in the eigenvalue statistics of the hamiltonian @xmath40 . this effect leaves us the impression that a non - integrable hamiltonian has more `` dephasing power '' than an integrable one . possibly it can be a tool to check the integrability of a hamiltonian . second , in the case of @xmath55 , it is found that the system described by @xmath154 responds to the second quench as if it were @xmath51 for @xmath159 larger than the transient time . this means that we can take the equilibration more serious@xmath154 and @xmath51 not only yield almost the same expectation values for the generic physical variables but also yield almost the same dynamics after a quench . moreover , the fact that the transient time is short indicates that the intermediate hamiltonian @xmath40 , which is non - integrable , is effective in `` dephasing '' the initial density matrix . in another perspective , the dynamics of the system is sensitive to the fluctuations of @xmath0 . this has the implication that in future experiments , accurate control of @xmath0 would be a necessity to interpret the results correctly . we are grateful to h. t. yang , z. x. gong , y .- h . chan , l. m. duan , and d. l. zhou for stimulating discussions and valuable suggestions . j. m. z. is supported by nsf of china under grant no .	we study quench dynamics of the bose - hubbard model by exact diagonalization . initially the system is at thermal equilibrium and of a finite temperature . the system is then quenched by changing the on - site interaction strength @xmath0 suddenly . both the single - quench and double - quench scenarios are considered . in the former case , the time - averaged density matrix and the real - time evolution are investigated . it is found that though the system thermalizes only in a very narrow range of the quenched value of @xmath0 , it does equilibrate or relax well in a much larger range . most importantly , it is proven that this is guaranteed for some typical observables in the thermodynamic limit . in order to test whether it is possible to distinguish the unitarily evolving density matrix from the time - averaged ( thus time - independent ) , fully decoherenced density matrix , a second quench is considered . it turns out that the answer is affirmative or negative according to the intermediate value of @xmath0 is zero or not .
in 1984 , an unconditionally secure key distribution protocol using quantum resources was proposed by bennett and brassard @xcite . the scheme , which is now known as bb84 protocol drew considerable attention of the cryptography community by its own merit as it offered unconditional security , which was unachievable by any classical protocol of key distribution . however , the relevance of bb84 quantum key distribution ( qkd ) protocol and a set of other schemes of qkd were actually established very strongly in 1994 , when the seminal work of shor @xcite established that rsa @xcite and a few other schemes of classical cryptography @xcite would not remain secure if a scalable quantum computer is built . the bb84 protocol , not only established the possibility of obtaining unconditional security , but also manifested enormous power of quantum resources that had been maneuvered since then . specifically , this attempt at the unconditional security of qkd was followed by a set of protocols for the same task @xcite . interestingly , the beautiful applications of quantum mechanics in secure communication did not remain restricted to key distribution . in fact , it was realized soon that the messages can be sent in a secure manner without preparing a prior key @xcite . exploiting this idea various such schemes were proposed which fall under the category of secure direct quantum communication ( @xcite and references therein ) . the schemes for secure direct quantum communication can be categorized into two classes on the basis of additional classical communication required by the receiver ( bob ) to decode each bit of the transmitted message- ( i ) quantum secure direct communication ( qsdc ) @xcite and ( ii ) deterministic secure quantum communication ( dsqc ) @xcite . in the former , bob does not require an additional classical communication to decode the message , while such a classical communication is involved in the latter ( see @xcite for review ) . it is worth noting that in a scheme of qsdc / dsqc meaningful information flows in one direction as it only allows alice to send a message to bob in an unconditionally secure manner using quantum resources and without generation of a key . however , in our daily life , we often require two way communication ( say , when we speak on a telephone ) . interestingly , a modification of one of the first few qsdc schemes ( i.e. , ping - pong scheme @xcite ) led to a new type of protocol that allows both alice and bob to communicate simultaneously using the same quantum channel . this scheme for simultaneous two way communication was first proposed by ba an @xcite and is known as quantum dialogue ( qd ) . due to its similarity with the task performed by telephones , a scheme for qd are also referred as quantum telephone @xcite or quantum conversation @xcite scheme , but in what follows , we will refer to them as qd . due to its practical relevance , schemes of qd received much attention and several new schemes of qd have been proposed in the last decade @xcite . however , all these schemes of qd , and also the schemes of qsdc and dsqc , mentioned here are restricted to the two - party scenario . this observation led to two simple questions- ( i ) do we need a multiparty qd for any practical purpose ? and ( ii ) if answer of the previous question is yes , can we construct such a scheme ? it is easy for us ( specially for the readers of this paper and the authors of the similar papers who often participate in conferences and meet as members of various committees ) to recognize that conferences and meetings provide examples of situation where multiparty dialogue happens . specifically , in a conference a large number of participants can exchange their thoughts ( inputs , which may be viewed as classical information ) . although , usually participants of the conference / meeting are located in one place , but with the advent of new technologies , tele - conferences , webinar , and similar ideas that allow remotely located users to get involved in multiparty dialogue , are becoming extremely popular . for the participants of such a conference or meeting that allows users to be located at different places , desirable characteristics of the scheme for the conference should be as follows- ( a ) a participant must be able to communicate directly with all other participants , or in other words , every participant must be able to listen the talk / opinion delivered by every speaker as it happens in a real conference . ( b ) a participant should not be able to communicate different opinion / message to different users or user groups . ( c ) illegitimate users or unauthorized parties ( say those who have not paid conference registration fees ) will not be able to follow the proceedings of the conference . it is obvious that criterion ( c ) requires security and a secure scheme for multiparty quantum dialogue satisfying ( a)-(c ) is essential for today s society . we refer to such a scheme for multiparty secure communication that satisfies ( a)-(c ) as ascheme for quantum conference ( qc ) because of its analogy with the traditional conferences ( specially with the tele - conferences ) . the analogy between the communication task performed here and the traditional conference can be made clearer by noting that wikipedia defines conference as `` a conference is a meeting of people who confer about a topic '' @xcite . similarly , oxford dictionary describes a conference as `` a linking of several telephones or computers , so that each user may communicate with the others simultaneously '' @xcite . this is exactly the task that the proposed protocol for qc is aimed to perform using quantum resources and in a secure manner . thus , qc is simply a conference , which is an @xmath0-party communication , where each participant can communicate his / her inputs ( classical information ) using quantum resources to remaining @xmath1 participants . however , it should be made clear that it is neither a multi - channel qsdc nor a multi - channel qd scheme . to be precise , one may assume that each participant maintains private quantum channels with all other participants and uses those to communicate his / her input to others via qsdc or qd . this is against the idea of a conference , as in this arrangement , a participant may send different information / opinion to different participants , in violation of criterion ( b ) listed above . the fact that to the best of our knowledge , no such scheme for multiparty secure quantum communication exists has motivated us to introduce the notion of qc and to aim to design a scheme for the same . here it would be apt to note that although no scheme for qc is yet proposed , various schemes for other multiparty quantum communication tasks have already been proposed . for example , quantum schemes for voting @xcite , auction @xcite , and e - commerce @xcite are necessarily expected to be multiparty quantum communication schemes . interestingly , there are a few schemes for all these tasks proposed in the past ( @xcite and references therein ) . another recently discussed multiparty task isquantum key agreement ( qka ) ( @xcite and references therein ) , where the final key is generated by the contribution of all the parties involved , and a single or a few parties can not decide the final key . for instance , a multiparty qka scheme @xcite was proposed in the recent past , in which encoded qubits travel in a circular manner among all the parties . in fact , most of these multiparty quantum communication schemes , except qka , can be intrinsically viewed as a ( many ) sender(s ) sending some useful information in a secure manner to a ( many ) receiver(s ) under the control of a third party . further , all these schemes can be broadly categorized as secure multiparty quantum communication and secure multiparty quantum computation . though the line between the two is very faint to distinguish and categorize a scheme among one of them , qka and e - commerce may be considered in the former , while voting and auction fall under the latter . some efforts have also been made to introduce a notion of qc as a multiparty quantum communication task . however , earlier ideas of qc can be viewed as special cases of the notion of qc presented here and they are not sufficient to perform a conference as defined above in analogy with the definition provided in oxford dictionary and other sources . bose , vederal and knight @xcite proposed a generalized entanglement - swapping - based scheme for multiparty quantum communication that led to a set of quantum communication schemes related to qc , viz . , cryptographic conference @xcite , conference key agreement and conference call @xcite , and a scheme where many senders send their messages to single receiver via generalized superdense coding @xcite . in cryptographic conference , all parties share a multipartite entangled state . they perform measurement in the computational or diagonal basis , and the results of those measurements in which the bases chosen by all the users coincide are used to establish the secret key which will be known to all the users within the group . a similar notion of conference key agreement was used in @xcite , where a generalized notion of dense coding was used . clearly the notion of conference is weaker here , and in our version of conference such keys can be distributed easily if all the users communicate random bits instead of meaningful messages . recent success of designing the above mentioned schemes for multiparty quantum communication further motivated us to look for a scheme for qc . a two party analogue of qc can be considered as qd , where both parties can communicate simultaneously . the group theoretic structure of ba - an - type qd schemes has been discussed in ref . the group theoretic structure discussed in @xcite will be exploited here to introduce the concept of qc . further , an asymmetric counterpart of the ba - an - type qd scheme is proposed in the recent past @xcite . following which we will also introduce and briefly discuss an asymmetric qc ( aqc ) , where all the parties involved need not to send an equal amount of information . with the recent interest of quantum communication community on quantum internet @xcite and experimental realization of multiparty quantum communication schemes @xcite , the motivation for introducing a qc or aqc scheme can be established . remaining part of the paper is organized as follows . [ sec : ba - an - protocol ] is dedicated to a brief review of qd and the group theoretic approach of qd for the sake of completeness of the paper , which has been used in the forthcoming sections to develop the idea of qc . two general schemes for the task of qc have been introduced in sec . [ sec : quantum - conferencepro ] . in the next section , we have considered a few specific examples of both these schemes . the feasibility of an aqc scheme has also been discussed in sec . [ sec : examples - and - possible ] . finally , the security and efficiency of the proposed schemes have been discussed in sec . [ sec : security - analysis ] before concluding the paper in sec . [ sec : conclusion ] . [ [ section ] ] it would be relevant to mention that some of the present authors had presented the general structure of qd protocols in @xcite and established that the set of unitary operators used by alice and bob must form a group under multiplication . the group structure has also been found to be suitable for the asymmetric qd schemes @xcite , where alice and bob use encoding operations from different subgroups of a modified pauli group , like @xmath2 . this particular abelian group ( @xmath3 ) is of order 4 under multiplication and is called a modified pauli group as we neglect the global phase in the product of any two elements of this group , which is consistent with the quantum mechanics ( for detail see @xcite ) . the generalized group @xmath4 can be formed by @xmath0-fold tensor products of @xmath3 , i.e. , @xmath5 . in the original qd protocol @xcite , the encoding is done by alice and bob , respectively , using the same set of operations @xmath6 from the modified pauli group @xmath3 . the entire scheme of ba an @xcite can be summed up in the formula @xmath7 @xmath8 , where @xmath9 are the bell states . it is required that all the possible final states obtained after alice s and bob s encoding operations should remain orthonormal to each other and also with the initial state . once the initial and final states are known to both the legitimate users , they can exploit knowledge of their own encoding operation to extract each other s message . interestingly , alice and bob encode information with the same operators , say , @xmath10 for 00 , @xmath11 for 01 , @xmath12 for 10 , and @xmath13 for 11 . in this scenario , alice obtains a unique bijective mapping from the composite encoding of alice and bob ( @xmath14 ) to bob s operation ( @xmath15 ) using her unitary operation ( @xmath16 ) . this is obvious where there are only 2 parties , we may ask , is it possible to extend this scheme for qd to design a scheme for multiparty conference ? let us examine two cases with 3 parties : in case 1 : when all the parties encode the same bits say , 00 i.e. , they apply @xmath17 and @xmath18 ; and in case 2 : when one of them encodes the same bits used in case 1 , i.e. , 00 and other two will encode the similar bits but other than 00 , say 01 @xmath19 , i.e. , they apply @xmath20 and @xmath18 , respectively . in these two cases , the resultant state is always the same as what was prepared initially , and none of the parties can deterministically conclude each others encoding . in fact , there will be many such cases , hence , ba an s original protocol for qd can not be generalized directly to design a scheme for multiparty conference . to design a scheme for qc , we will use the idea of disjoint subgroups introduced by some of the present authors in the recent past @xcite . disjoint subgroups refer to subgroups , say @xmath21 and @xmath22 , of a group @xmath4 such that they satisfy @xmath23 . thus , except identity @xmath21 and @xmath22 do not contain any common element . the modified pauli group @xmath3 has 3 mutually disjoint subgroups : @xmath24 @xmath25 and @xmath26 . whenever there are more than two parties , we can encode using disjoint subgroups of operators , i.e. , each party may be allowed to encode with a unique disjoint subgroup . for example , if alice , bob and charlie want to set up a qc among them , then alice can encode using @xmath27 , bob can encode using @xmath28 and charlie can encode using @xmath29 the use of disjoint subgroups circumvents the limitations of the original two - party qd scheme and provides a unique mapping required for multiparty conversation . in what follows , we have proposed two protocols to accomplish the task of a qc scheme . [ [ section-1 ] ] here , we have designed two multiparty quantum communication schemes where prior generation of key is not required . these schemes may be used for qc , i.e. , for multiparty communication of meaningful information among the users . additionally , it is easy to observe that these schemes naturally reduce to the schemes for multiparty key distribution if the parties send random bits instead of meaningful messages . let us start with the simplest case , where @xmath30 parties send their message to @xmath31th party . this can be thought of as a multiparty qsdc . suppose all the parties decide to encode or communicate @xmath32-bit classical messages . in this case , each user would require a subgroup of operators with at least @xmath33 operators . in other words , each party would need at least a subgroup @xmath21 of order @xmath33 of a group @xmath34 . here , we would like to propose one such multiparty qsdc scheme . step 1.1 : : first party alice be given one subgroup @xmath35 to encode her @xmath32-bit information . similarly , other parties ( say bob and charlie ) can encode using subgroups @xmath36 , and @xmath37 , and so on for @xmath30th party diana , whose encoding operations are @xmath38 . + all these subgroups are pairwise disjoint subgroups , i.e. , they are chosen in such a way that @xmath39 . as the requirement for encoding operations to be from disjoint subgroups has been already established beforehand . + additionally , here we assume that all the parties do nothing ( equivalent to operator identity ) on their qubits for encoding a string of @xmath32 zeros . as identity is the common element in the set of encoding operations to be used by each party it will be convenient to consider this as a convention in the rest of the paper . step 1.2 : : nathan ( the @xmath31th party ) prepares an @xmath0-qubit entangled state @xmath40 ( with @xmath41 ) . + it is noteworthy that maximum information that can be encoded on the @xmath42-qubit quantum channel is @xmath42 bits and here @xmath30 parties are sending @xmath32 bits each . in other words , after encoding operation of all the @xmath30 parties the quantum states should be one of the @xmath43 possible orthogonal states . step 1.3 : : nathan sends @xmath44 qubits ( @xmath45 ) of the entangled state @xmath40 to alice in a secure manner , who applies one of the operations @xmath46 ( which is an element of the subgroup of operators available with her ) on the travel qubits to encode her message . this will transform the initial state to @xmath47 . subsequently , alice sends all these encoded qubits to the next user bob . step 1.4 : : bob encodes his message which will transform the quantum state to @xmath48 . finally , he also sends the encoded qubits to charlie in a secure manner . step 1.5 : : charlie would follow the same strategy as followed by alice and bob . in the end , diana receives all the encoded travel qubits and she also performs theoperation corresponding to her message to transform the state into @xmath49 . she returns all the travel qubits to nathan . step 1.6 : : nathan can extract the information sent by all @xmath30 parties by measuring the final state using an appropriate basis set . + it may be noted that nathan can decode messages sent by all @xmath30 parties , if and only if the set of all the encoding operations gives orthogonal states after their application on the quantum state , i.e. , @xmath50 are orthogonal for all @xmath51 . in other words , after the encoding operation of all the @xmath30 parties the quantum states should be a part of a basis set with @xmath43 orthogonal states for unique decoding of all possible encoding operations . this scheme can be viewed as the generalization of ping - pong protocol @xcite to a multiparty scenario , where multiple sender s can simultaneously send their information to a receiver . in a similar way , if all the senders wish to send and receive the same amount of information , then all of them can also choose to prepare their initial state @xmath40 independently and send it to all other parties in a sequential manner . subsequently , all of them may follow the above protocol faithfully to perform @xmath31 simultaneous multiparty qsdc protocols . in fact , @xmath31 simultaneous multiparty qsdc schemes of the above form will perform the task required in an ideal qc scheme . however , as each sender has to encode his secret multiple times ( @xmath52 times ) , it would allow him to encode different information in each round . though it may be advantageous in some communication schemes , where a sender is allowed to send different bit values to different receivers , but is undesirable in a scheme for qc . specifically , to stress on the relevance of a scheme that allows each sender to encode different bits to all the receivers , we may consider a situation where each party ( or a few of them ) publicly asks a question , and the receivers answer the question independently ( for an analogy think of a panel discussion in television ) . in this case , all the receivers may have different opinions ( say one may agree with some of them and may not with the remaining ) about various questions being asked . as far as a scheme for qc is concerned , protocol 1 described here would work under the assumption of semi - honesty . specifically , a semi - honest party may try to cheat , but he / she would follow the protocol faithfully . this assumption would enable us to consider that each party is encoding the same information every time . in what follows . we will establish that such an assumption is not required . specifically , in protocol 2 , we aim to design a genuine qc scheme , which does not require the semi - honesty assumption to restrict a user from sending different information to different receivers . here , we will attempt to design an efficient qc scheme , which can be thought of as a generalized qd scheme . in analogy of the original ba - an - type qd scheme , we will need the set of encoding operations for the @xmath31th party ( nathan ) . here , firstly we propose the protocol which is followed by a prescription to obtain the set of operations for @xmath31th party , assuming a working scheme designed for the protocol 1 . step 2.1 : : same as that of step 1.1 of protocol 1 with a simple modification that also provide nathan a subgroup @xmath53which enables him to encode a @xmath32-bit message at a later stage . + the mathematical structure of this subgroup will be discussed after the protocol . step 2.2 : : same as step 1.2 of protocol 1 . step 2.3 : : same as step 1.3 of protocol 1 . step 2.4 : : same as step 1.4 of protocol 1 . step 2.5 : : same as step 1.5 of protocol 1 . step 2.6 : : nathan applies unitary operation @xmath54 to encode his secret and the resulting state would be @xmath55 @xmath56 . step 2.7 : : nathan measures @xmath57 using the appropriate basis as was done in step 1.6 of protocol 1 and announces the measurement outcome . now , with the information of the initial state , final state and one s own encoding all parties can extract the information of all other parties . + it is to be noted that the information can be extracted only if the set of all the encoding operations gives orthogonal states after their application on the quantum state , i.e. , all the elements of @xmath58 are required to be mutually orthogonal for @xmath59 . in other words , after the encoding operation of all the @xmath31 parties the set of all possible quantum states should form a @xmath43 dimensional basis set . nathan s unitary operation can be obtained using the fact that the remaining @xmath30 parties have already utilized the channel capacity . hence , his encoding should be in such a way that after his encoding operation @xmath54 , the final quantum state should remain an element of the basis set in which the initial state was prepared . however , the bijective mapping between the initial and final states present in protocol 1 would disappear here . this is not a limitation . it is actually a requirement . this is so because , in contrast to protocol 1 where the initial and final states are secret , in protocol 2 , the choice of the initial state and the final state are publicly broadcasted . existence of a bijective mapping would have revealed all the secrets to eve . this condition provides us a mathematical advantage . specifically , it allows us to construct the set of unitary operations that nathan can apply . to do so we need to use the information about the disjoint subgroups of operators that are used by other parties . the procedure for construction of nathan s set of operations is described below . for simplicity , let us write the encoding operations of all the parties as follows : @xmath60 here , @xmath61 corresponds to the binary value of the decimal number @xmath62 , and it represents the classical information to be encoded by user x * * * ( listed in column 1 ) using the the operator @xmath63 ( listed in @xmath64th column in the row corresponding to the user x * * * ) . for example , to encode @xmath65 alice would use the operator @xmath66 , whereas for the same encoding bob and charlie would use @xmath67 and @xmath68 , respectively . further , we would like to note that by construction operators @xmath69 as @xmath70 is an element of the modified pauli group , and it is assumed that the encoding operations of the different users are chosen from the disjoint subgroups of the modified pauli groups in such a way that the product of operations listed in any column is identity , i.e. , @xmath71 this implies that if all the parties encode the same secret then the final state and the initial state would be the same . to illustrate this we may consider following example @xmath72 from eqs . ( [ eq : condition])-([eq : example ] ) , it is clear that the choice of encoding operations of the other users ( i.e. , @xmath73 would uniquely determine @xmath74 . further , it is assumed that the encoding operations used by different users to encode @xmath61 are selected in a particular order that ensures @xmath75 and particular choice of @xmath76 for example , this condition implies that if alice s operators satisfy @xmath77 then bob and charlie would be given the encoding operators in an order that satisfy @xmath78 and @xmath79 respectively , and the same ordering of operators will be applicable to all other users . now , using the above mentioned facts and convention , we need to establish that @xmath80 forms a group under multiplication . ( [ eq : condition])and the self reversibility of the elements @xmath70 lead to following identity- @xmath81 this may be used to establish the closure property of the group @xmath80 as @xmath82 . this is so because the pauli operators commute with each other under the operational definition of multiplication used in defining the modified pauli group . all the remaining properties of the group follows directly from the nature of pauli operators used to design @xmath83 . thus , it is established that the generalized multiparty qsdc scheme can be modified to a generalized qd scheme . it will be interesting to obtain the original ba an s qd scheme as a limiting case as follows . @xmath84 this particular case and all the discussions leave us with @xmath85 which is identical with alice s operations . in table [ tab : conference ] , we have provided a list comprising of the number of participants in the qc and the number of cbits they want to encode . the table explicitly mentions different multipartite states or quantum channels that can be utilized for the same . .[tab : conference ] various possibilities of qc scheme with a maximum number of @xmath31 parties each encoding @xmath32 bits using a group of unitary operators with at least @xmath43 elements . the quantum states suitable in each case and corresponding number of travel qubits are also mentioned . [ cols="^,^,^,^,^ " , ] the proposed qc scheme may also be extended to an asymmetric counterpart of the qc scheme , where each party may not be encoding the same amount of information . one such easiest example is a lecture , where the orator speaks most of the time while the remaining users barely speak . in such cases , the parties sending redundant bits to accommodate the qc scheme may choose an aqc scheme . to exploit the maximum benefit of such schemes a party encoding more information than others ( say alice ) should prepare ( and also measure ) the quantum state ( in other words , start the qc scheme ) . in this case , the choice of unitary operations by each party would also become relevant and alice should use a subgroup of higher order than the remaining users . for instance , in a 3-party scenario , alice may use a @xmath86 from row 2 of table [ tab : multiparty - quantum - conference . ] to encode 2 bits message , while the remaining three users may choose @xmath87 and @xmath88 , respectively . it is worth noting here that the security of the qc scheme discussed in the following section ensures the security of the aqc scheme designed here as well . further , the proposed schemes can also be easily modified to obtain corresponding schemes for controlled qc , where an additional party ( who is referred to as the controller ) would prepare the quantum channel in such a way that the qc task can only be accomplished after the controller allows the other users to do so @xcite . controlled qc can be achieved in various ways . for example , the controller may prepare the initial state and keep some of the qubits with himself , and in absence of the measurement outcome of the corresponding qubits the other legitimate parties would fail to accomplish the task @xcite . the same feat can also be achieved by the controller without keeping a single qubit with himself by using permutation of particles @xcite . thus , it is easy to generalize the proposed schemes for qc to yield schemes for controlled qc . such a scheme for controlled qc would have many applications . for example , a direct application to that scheme would be quantum telephone where the controller can be a telephone company @xcite that provides the channel to the respective users after authentication . thus , the present scheme can be used to generalize the scheme proposed in @xcite and thus to obtain a scheme for multiparty quantum telephone or quantum teleconference . additionally , the multiparty communication schemes proposed here can be reduced to schemes for secure multiparty quantum computation . interestingly , a recently proposed secure multiparty computation scheme designed for quantum sealed - bid auction task @xcite can be viewed as a reduction of the protocol 1 proposed here . therefore , we hope that the proposed schemes may also be modified to obtain solutions of various other real life problems . a qc protocol is expected to confront the disturbance attack ( or denial of service attack ) , the intercept - and - resend attack , the entangle - and - measure attack , man - in - the - middle attack and trojan - horse * * attack by implementing the bb84 subroutine strategy ( for detail see @xcite ) , which allows senders to insert decoy qubits prepared randomly in @xmath11-basis or @xmath13-basis in analogy with bb84 protocol and to reveal the traces of eavesdropping by comparing the initial states of the decoy qubits with the states of the same qubits after measured by the receivers randomly using @xmath11-basis or @xmath13-basis . in fact , quantum communication of all the qubits from one party to other , as mentioned in both the protocols ( for example , in step 1.3 ) , is performed in a secure manner . to accomplish the secure communication of message qubits using bb84 subroutine , an equal number of decoy qubits ( the number of decoy qubits are required to be equal to the number of message qubits traveling through the channel ) are inserted randomly in the string of travel qubits . on the authenticated receipt of this enlarged sequence of travel qubits , the sender discloses the positions of the decoy qubits and those qubits are then measured by the receiver randomly in @xmath11-basis or @xmath13-basis . subsequent comparison of the initial states and the measurement outcomes reveals the error rate . if the computed error rate is obtained below a tolerable limit , then the quantum communication of message qubits is considered to be accomplished in a secure manner @xcite , and the steps thereafter are followed . therefore , the above mentioned attacks on the proposed schemes can be defeated simply by adding decoy qubits and following bb84 subroutine . further , bob s intimation by alice that she has sent her qubits and bob s acknowledgment of the receipt of qubits , via an authenticated classical channel , is necessary to avoid the unwanted circumstances under which eve pretends as the desired party . there also exist some technical procedures to circumvent the trojan - horse attack ( @xcite and references therein ) . as a scheme of qc incorporates multiusers we have discussed below the security in two scenarios where ( 1 ) an outsider ( eve ) attacks the protocol , or ( 2 ) an insider ( one or some of the legitimate users ) attacks the protocol . further , all the attacks and counter measures mentioned in this section are applicable on both the schemes , unless specified . in the * entangle - and - measure attack * , eve entangles her qubit @xmath89 with the travel qubit in the channel . eve can extract the information by performing the @xmath13-basis measurement on her ancillae . to counter this attack , the decoy states @xmath90 , @xmath91 , @xmath92 and @xmath93 are randomly inserted and when they are examined for security , then eve is detected with probability @xmath94 when she attacks @xmath90 and @xmath91 states , otherwise the states remain separable for @xmath92 and @xmath93 . consequently , the total detection probability of eve is @xmath95 taking into account that the probability of generation of each decoy qubit state is @xmath96 . in the * intercept - and - resend attack * , eve prepares some fresh qubits and swaps one of her qubits with the accessible qubit in the channel when @xmath97th user sends it to @xmath98th user . thereafter , eve retrieves her qubit during their communication from @xmath98th user to @xmath99th user and obtains the encoding of @xmath98th user by performing a measurement on her qubits . this attack will also be defended by incorporating decoy qubits . however , eve may modify her strategy to measure the intercepted qubits randomly in either the computational or diagonal basis before sending the freshly prepared qubits corresponding to the measurement outcomes . it is evident that eve s measurement of the decoy qubits will produce disturbance if she measures in the wrong basis . let @xmath0 be the total number of travel qubits such that @xmath100 are decoy and message qubits each . eve intercepts @xmath44 qubits which will entail both decoy and message . without a loss of generality , we assume that half of the @xmath44 qubits are decoy and the other half are message qubits . since the security check is performed on the decoy qubits alone , we are interested in the @xmath101 decoy qubits which eve measures in her lab out of the @xmath100 decoy qubits in the channel . the fraction of qubits measured by eve out of the total decoy qubits is given by @xmath102 . from which the information gained by eve is @xmath103 this implies that @xmath104 times the correct basis will be chosen by eve . the error induced by eve is observed by alice and bob only when bob measures in the same basis as of alice and is @xmath105 . the amount of information bob receives is given by @xmath106)$ ] , where @xmath107 $ ] is the shannon binary entropy . the security is ensured until @xmath108 . one can calculate the fraction @xmath109 for secure communication with the tolerable error rate @xmath110 ( @xcite and references therein ) . eve s success probability is @xmath111 and it would decrease with the increasing value of @xmath44 as @xmath112 . * information leakage attack * is inherent in the qd schemes , and consequently , is applicable to protocol 2 proposed here as well . it refers to the information gained by eve about the encoding of the legitimate parties by analyzing the classical channel only . in brief , the leakage can be thought of as the difference between the total information sent by both the legitimate users and the minimum information required by eve to extract that information ( i.e. , eve s ignorance ) . the mathematical prescription for an average gain of eve s information is @xmath113 where @xmath114 is the total classical information all the legitimate parties have encoded ; and @xmath115 is eve s ignorance after the announcement of the measurement outcome and is averaged over all the possible measurement outcomes as @xmath116 , with the conditional entropy @xmath117 . if the party authorized to prepare and measure the quantum state selects the initial state randomly and sends it to all the remaining users by using a standard unconditionally secure protocol for qsdc or dsqc then the leakage can be avoided as it increases the @xmath115 , and thus decreases @xmath118 to zero corresponding to no leakage @xcite . * participant attack * is possible in both the schemes proposed here . * * in the first scheme , a participant can send different cbits to different members unless we assume semi - honest parties . although this scheme is advantageous in certain applications , like sealed bid auction ( where this attack is detected in post - confirmation steps ) @xcite or where each participant wants to encode different values to respective participant , but in the conference scenario where it is required that each participant encodes the same message to all other participants then this attack is prominent , and it is wise to follow the second scheme , which is free from the assumption of semi - honest parties . in the second scheme , the authorized party ( authorized to prepare and measure the quantum state ) encodes his information at the end just before performing the joint measurement and announcing the outcome . if he wants to cheat he can disclose an incorrect measurement outcome corresponding to his modified encoding once he comes to know others encoding . this action can be circumvented , and we can implement this protocol either with a trusted party or we can randomly select any two participants and run the scheme twice considering that respective party encodes same information . another solution would be that the initiator sends the hash value of his message at the beginning to all the remaining users , and if the hash value of his encoding revealed at the last do not match with that of the initially sent hash value , then he had cheated and will be certainly identified . * collusion attack * is a kind of illegal collaboration of more than one party who are not adjacent to each other , to cheat other members of a group to learn their encoding ( precisely of those who are in between them ) . the proposed schemes are circular in nature . in this type of an attack , the attackers generate an entangled state and circulate the same number of fake qubits as that of the travel qubits . the attackers at the end already possess the home photons of the fake qubits circulated by the first attacker and performs a joint measurement to learn the encoding of the participants in between them . it will be more effective if @xmath98th and @xmath119th participants collude . this is so , as both of them get the access of the travel particles at least once after knowing the secret of all the remaining parties . this attack can be averted by breaking the larger circle into @xmath120 sub - circles such that if less than @xmath120 attackers collude , they will not be able to cheat ( see @xcite for details ) . this attack and the solution are applicable in both the proposed schemes . the qubit efficiency of a quantum communication scheme is calculated as @xmath121 where @xmath122 bits of classical information is transmitted using @xmath123 number of qubits , and an additional classical communication of @xmath124 bits @xcite . in the first qc scheme , @xmath125 @xmath126 , and @xmath127 as each party sends @xmath32 bits and prepares @xmath0-qubit entangled state and @xmath44 decoy qubits in each round of quantum communication . therefore , the efficiency is calculated to be @xmath128 . similarly , the qubit efficiency of the second qc scheme among @xmath31 parties such that each party encodes @xmath32 bits can be computed by noting that in this case @xmath125 @xmath129 , and @xmath130 . here , @xmath131 as the classical communication of @xmath0 cbits is associated with the broadcast of the measurement outcome by the authorized party . thus , the qubit efficiency is obtained as @xmath132 . from the @xmath133 one can easily calculate the qubit efficiency of various possible qc schemes detailed in table [ tab : conference ] . for example , one can check that the qubit efficiency of a two party qc with each party encoding 2 bits ( which is ba an s qd protocol ) using bell state as quantum channel is 67% . similarly , the qubit efficiency for a qc scheme involving three parties sending 1 bit each with bell state as the quantum channel can be obtained as 43% . hence , we find that for the same initial state as quantum channel the efficiency decreases as the number of parties increases and / orthe number of encoded bits decreases . in summary , the notion of qc is introduced as a multiparty secure quantum communication task which is analogous with the notion of classical conference , and two protocols for secure qc are designed . the proposed protocols are novel in the sense that they are the first set of protocols for qc , as the term qc used earlier were connected to communication tasks that were not analogous to classical conference . further , it is shown that protocols proposed here can be reduced to protocols for qc proposed earlier considering much weaker notion of conference . one of the proposed protocols can be viewed as a generalization of the ping - pong protocol for qsdc , whereas the other one can be viewed as a generalization of the schemes for qd . it is noted that protocol 1 composes number of rounds of multiple - sender to single receiver secure direct communication , which accomplishes the task of qc under the assumption of semi - honesty of the users . however , this semi - honesty assumption is not required for protocol 2 , which is proposed here as multiple - sender to multiple - receiver scheme , where the task is performed in a single round . subsequently , both the proposed schemes are elaborated with the help of an explicit example . we have discussed the utility and applications of these protocols in different scenarios . specifically , the proposed schemes may be reduced to a set of multi - party qkd and qka schemes , if the parties involved in qc send random bits instead of meaningful messages . further , feasibility and significance of the controlled and asymmetric counterparts of the proposed qc schemes have also been established . the modified versions of the proposed schemes may also be found useful in accomplishing some real - life problems , whose primitive is secure multiparty computation . for example , one can employ the proposed schemes for voting among the five countries having power of veto in united nations , where it is desired that the choice of a voter is not influenced by the choice of the others . the proposed scheme can also be extended to obtain a dynamic version of qc , where a participant can join the conference once it has started and leave it before its termination . such a generalization is possible using the method introduced by some of the present authors in ref . @xcite . further , the effect of various types of markovian and non - markovian noise on the schemes proposed here can be investigated easily using the approach adopted in @xcite . security of the proposed schemes has been established against various types of insider and outsider attacks . further , the qubit efficiency analysis established that protocol 2 is more efficient than protocol 1 . further , one can easily observe that the proposed schemes are much more efficient compared to a simple minded scheme that performs the same task by using multiple two - party direct communication schemes , which will again work only under the assumption of semi - honest users . finally , we have also presented a set of encoding operations suitable with a host of quantum channels for performing the qc schemes for number of parties . this provides experimentalists a freedom to choose the encoding operations and the quantum state to be used as quantum channel as per convenience . further , experimental realization of quantum secure direct communication scheme , which can demonstrate protocols , like quantum dialogue , quantum authentication , has been successfully performed in @xcite , and it paves way for experimental realization of qc . keeping these facts in mind , we conclude this paper with a hope that the schemes proposed here and/or their variants will be realized in the near future . * acknowledgment : * ab acknowledges support from the council of scientific and industrial research , government of india ( scientists pool scheme ) . cs thanks japan society for the promotion of science ( jsps ) , grant - in - aid for jsps fellows kt and ap thank defense research & development organization ( drdo ) , india for the support provided through the project number erip / er/1403163/m/01/1603 . 10 bennett , c. h. , brassard , g. : quantum cryptography : public key distribution and coin tossing . in proceedings of the ieee international conference on computers , systems , and signal processing , bangalore , india , pp . 175 - 179 ( 1984 ) shor , p. w. polynomial - time algorithms for prime factorization and discrete logarithms on a quantum computer , in proceedings of 35th annual symp . on foundations of computer science , santa fe , ieee computer society press . ( 1994 ) thapliyal , k. , pathak , a. : applications of quantum cryptographic switch : various tasks related to controlled quantum communication can be performed using bell states and permutation of particles . quantum inf . process . * 14 * , 2599 - 2616 ( 2015 ) sharma , v. , thapliyal , k. , pathak , a. , banerjee , s. : a comparative study of protocols for secure quantum communication under noisy environment : single - qubit - based protocols versus entangled - state - based protocols . inf . process . * 15 * , 4681 ( 2016 )	a notion of quantum conference is introduced in analogy with the usual notion of a conference that happens frequently in today s world . quantum conference is defined as a multiparty secure communication task that allows each party to communicate their messages simultaneously to all other parties in a secure manner using quantum resources . two efficient and secure protocols for quantum conference have been proposed . the security and efficiency of the proposed protocols have been analyzed critically . it is shown that the proposed protocols can be realized using a large number of entangled states and group of operators . further , it is shown that the proposed schemes can be easily reduced to protocol for multiparty quantum key distribution and some earlier proposed schemes of quantum conference , where the notion of quantum conference was different . keywords : quantum conference , quantum cryptography , secure quantum communication , multiparty quantum communication .
since various systems may be governed by the same fundamental law , universality exists in nature . the early indications of universality in quantum mechanic three - body systems was discovered before 1970.@xcite the first strong evidence was given by efimov in 1970 on three weakly bound identical bosons.@xcite he found that , when the scattering length is sufficiently long , a sequence of loosely bound states , the `` efimov states '' , will appear . the properties of these states are governed by a universal law , not depend on the dynamic details of the 3-boson systems . in this paper , another kind of universality that exists in tightly bound quantum mechanic few - body systems is revealed and confirmed numerically . the hamiltonian of identical particles is invariant under the operations of a set of symmetry groups @xmath0 , @xmath1 , @xmath2 ( including the permutation group).@xcite consequently , the eigenstates @xmath3 are classified according to the representations of these groups , where @xmath4 denotes a set of coordinates and @xmath5 is a serial number . let @xmath6 be an element of @xmath0 , @xmath7 be that of @xmath8 , and @xmath9 denotes a special zone in the high - dimensional coordinate space . when @xmath10 , the effects of @xmath11 and @xmath7 might be equivalent so that @xmath12 . for an example , when @xmath9 is the zone of the squares ( sq ) , @xmath6 is a rotation about the normal of the sq by @xmath13 , and @xmath7 is a cyclic permutation of particles , then @xmath6 and @xmath7 are equivalent in @xmath9 . making use of the representations of groups , the equivalence leads to a set of homogeneous linear equations @xmath14 \psi _ { i'}(x ) = 0 , \ \ \ ( x\in \xi ) , \label{e01_sum}\end{aligned}\ ] ] where @xmath15 are the matrix elements of the representation . when the matrix of this set of equations is non - degenerate , the set @xmath16 must be zero in @xmath9 . in this case , @xmath9 becomes a prohibited zone ( pz ) and the wave function appears as an inherent nodal surface ( ins).@xcite eq . ( [ e01_sum ] ) demonstrates that , for each pair of equivalent operations , a constraint will be imposed on the eigenstates . since the matrixes of representations are absolutely irrelevant to dynamics , the constraint is universal disregarding the kind of systems ( nuclear , atomic , or molecular ) and the details of dynamic parameters . it implies that the states of different systems but belonging to the same set of representations will have exactly the same pz , and their wave functions will have exactly the same ins . on the other hand , some zones are important to binding ( say , for 4-body systems , the zones of equilateral tetrahedron ( et ) and sq ) . whether these zones are prohibited is crucial to the binding energy and the geometric character of a state . furthermore , the number and the locations of the nodal surfaces in a wave function in general determine the strength and mode of oscillation . in particular , the existence of the ins implies an inherent mode . thus , the eigenstates would be seriously affected by the universal symmetry constraint . the decisive effect of the symmetry constraint on the triply and quadruple excited intrashell states of atoms has been revealed previously.@xcite accordingly , these states can be naturally classified according to their inherent nodal structures.@xcite for 4-boson systems , a number of predictions on the structures and internal modes of oscillation have been made previously.@xcite however , these predictions have not yet been confirmed numerically . the present paper generalizes the work of [ 18,19 ] in the following aspects : ( i ) instead of free 4-boson systems , trapped 4-boson systems are considered . thereby a number of tightly bound states can be obtained which are necessary for a systematic analysis . ( ii ) in addition to theoretical analysis , numerical calculations have been performed so that the effect of symmetry constraint can be appraised quantitatively . ( iii ) three types of interactions have been adopted . the aim is to demonstrate the similarity among different kinds of systems . in the next section , the symmetry constraints imposing on 4-boson systems are studied theoretically . we have chosen appropriate sets of coordinates so that eq . ( [ e01_sum ] ) appears in very simple forms and the analysis becomes transparent . then , an isotropic trap together with three types of interaction are introduced , and numerical calculations are performed to diagonalize the hamiltonian . under the trap the total orbital angular momentum @xmath17 , its @xmath18-component @xmath19 , and the parity @xmath20 are good quantum numbers . accordingly , an eigenstate with the c.m . motion removed can be denoted as @xmath21 , where @xmath5 denotes the @xmath5-th state of a @xmath22-series . mostly the @xmath23 states ( the lowest one ) are concerned . therefore , the label @xmath5 is dropped hereafter ( except specified ) . after obtaining the eigenenergies and the eigenstates , a number of quantities ( the root - mean - square radius , the one - body densities for the particle distribution , and the shape - densities ) are further calculated . thereby , inherent physics can be extracted , and a clear comparison among different kinds of system can be made . the emphasis is placed to demonstrate the universality of the pz and the similarity among different systems . a short discussion on 4-fermion systems is also given at the end . the equilateral tetrahedron ( et ) and the square ( sq ) are the two most important geometries . we shall study the symmetry constraint taking place at the et , sq , and their neighborhoods . the zone associated with the extension - contraction of an et along one of its two - fold ( three - fold ) axis is defined as @xmath24 ( @xmath25 ) . @xmath24 and @xmath25 are related to the h - type and k - type of oscillations , respectively , that exist in tetramers.@xcite when the tetrahedrons of @xmath24 are further twisted about the two - fold axis , the extended zone is defined as @xmath26 . when appropriate degrees of freedom are chosen , the general expression eq . ( [ e01_sum ] ) could have very simple form . for @xmath24 , let the jacobi coordinates be @xmath27 , @xmath28 , @xmath29 , then @xmath30 and @xmath31 are required . @xmath32 and @xmath33 are considered as variable . we introduce a body frame @xmath34 with its origin at the c.m . , its @xmath35 lying along @xmath36 , @xmath37 lying along @xmath38 , and @xmath39 lying along @xmath40 . @xmath24 is shown in fig . this zone is 2-dimensional in @xmath34 . in general , an eigenstate can be expressed by using the arguments of a body frame as @xmath41 where @xmath4 denotes the set of coordinates relative to a fixed frame @xmath42 , and @xmath43 denotes the set relative to @xmath44 , @xmath45 denotes the set of euler angles responsible for the transformation from @xmath34 to @xmath42.@xcite let @xmath46 denotes an arbitrary particle permutation , @xmath47 a space inversion with respect to the c.m . , and @xmath48 a rotation about the axis @xmath49 by the angle @xmath50 . in @xmath24 we have the equivalence @xmath51 . when @xmath52 act at the left side of eq . ( [ e02_psi ] ) and @xmath53acts at the right side , we obtain the constraint @xmath54 which is a simple form of eq . ( [ e01_sum ] ) . therefore , @xmath55 is nonzero in @xmath24 only if @xmath56 ( when @xmath57 ) , or @xmath58 ( when @xmath59 ) . we have also the equivalence @xmath60 , which leads to the second constraint @xmath61 therefore , the @xmath62 component exists only if @xmath63 . with this in mind , we know that @xmath24 is a pz for @xmath64 and @xmath65 states ( because they have only @xmath66 components ) , and for @xmath67 and @xmath68 states ( because theydo not have @xmath69 components and their @xmath62 component violates the requirement @xmath63 ) . when @xmath70 , the particles form an et . the zone of et is denoted as @xmath71 which is one - dimensional in @xmath34 and is a subspace of @xmath24 . in @xmath71 oa ( refer to fig . [ fig1]a ) becomes a three - fold axis of an et . thus , additionally , we have @xmath72 where @xmath73 is the angle between oa and @xmath35 . this leads to a set of homogeneous linear equations as @xmath74 \psi_{lq'}(\mathrm{et } ) = 0 , \label{e06_sum}\end{aligned}\ ] ] where @xmath75 is the matrix elements of the rotation about @xmath37.[21 ] under the constraints given by eqs . ( [ e03_psi ] ) , ( [ e04_psi ] ) and ( [ e06_sum ] ) , nonzero solutions exist at the et only if @xmath76 , @xmath77 and @xmath78 ( if @xmath79 ) . they are called et - accessible ( et-ac ) states.@xcite when @xmath80 , the shapes in @xmath24 tends to a sq with its plane lying on the @xmath39-@xmath37 plane ( fig . [ fig1]a ) . the zone of sq is denoted as @xmath81which is also one - dimensional in @xmath34 . for @xmath81 we have @xmath82 . thus , the sq can only exist in @xmath83 states . as a subspace of @xmath24 , the sq must be constrained by eqs . ( [ e03_psi ] ) and ( [ e04_psi ] ) . thus the sq-ac states are @xmath84 , @xmath85 and @xmath78 ( if @xmath79 ) . among them , @xmath84 and @xmath78 are both et- and sq-ac. the geometries in @xmath25 are the regular trihedral pyramids . let @xmath86 denotes the height which is lying along the 3-fold axis of the pyramid , and @xmath87 denotes the side length of the base regular triangle . @xmath86 and @xmath87 are variable . let @xmath35 be lying along the 3-fold axis , and @xmath37 along a height of the base triangle . then we have @xmath88 and @xmath89 . accordingly , @xmath25 is a pz for @xmath64 , @xmath67 and @xmath90 states . in @xmath26 ( refer to fig . [ fig1]b where @xmath36 is lying along @xmath35 and can not be seen ) @xmath91 , @xmath92 , @xmath31 are required . @xmath32 , @xmath33 and the angle @xmath50 between @xmath38 and @xmath40 are considered as variable . in this zone @xmath35 is a 2-fold axis , and @xmath39 lying along the bisector of @xmath50 is also a 2-fold axis . we have @xmath93 and @xmath94 . accordingly , @xmath26 is a pz for @xmath95 and @xmath67 states . for coplanar shapes , let @xmath35 be vertical to the plane of the shapes . then , @xmath96 must hold . we considered the following three cases . \(i ) the coplanar shape is invariant under inversion ( e.g. , a parallelogram ) and , accordingly , @xmath35 is a two - fold axis . the associated zone is denoted as @xmath97 , and the equivalence @xmath82 holds . \(ii ) the shape contains at least a two - fold axis on the @xmath39-@xmath37 plane ( say , an isosceles trapezoid ) . the zone is denoted as @xmath98 , and the equivalence @xmath99 holds . \(iii ) the intersection of @xmath97 and @xmath98 , namely , @xmath100 . this zone includes the rectangles and diamonds . specifically , @xmath101 . associated with the above zones , the pz are summarized in table [ tab1 ] . it is emphasized that the existence of the pz is universal disregarding the kinds of 4-boson systems . since the prohibition is absolutely not violable , the effect of the pz is decisive . the features of all the eigenstates can be thereby more or less predicted . for examples , @xmath84 is free from symmetry constraint . thus , it can choose the structure most favorable to binding , and its wave function is expected to surround an et . @xmath85 is not et-ac but is sq-ac , therefore its wave function is expected to surround a sq , and so on . for the eight zones listed in table [ tab1 ] , the most important zones are the @xmath71 and @xmath81 . the accessibility of these two zones is crucial to the energies of the eigenstates as shown below . when a zone is accessible to a state but some subspaces inside the zone are not ( say , @xmath26 is accessible to @xmath64 but its subspace @xmath24 is not ) , the zone is not a stable zone because the pz inside the zone implies the existence of inherent nodal surfaces , and therefore specific oscillations are excited ( examples are given below ) . [ tab1 ] c\|c\|10c & dim & @xmath84 & @xmath67 & @xmath85 & @xmath68 & @xmath78 & @xmath64 & @xmath65 & @xmath90 & @xmath77 & @xmath102 + @xmath26 & 3 & & & & & & & & & & + @xmath24 & 2 & & & & & & & & & & + @xmath71 & 1 & & & & & & & & & & + @xmath81 & 1 & & & & & & & & & & + @xmath25 & 2 & & & & & & & & & & + @xmath97 & 3 & & & & & & & & & & + @xmath98 & 3 & & & & & & & & & & + @xmath100 & 2 & & & & & & & & & & + we are going to present numerical results to evaluate quantitatively the consequence of the pz . it is assumed that the four particles are confined by an isotropic harmonic trap @xmath103 . the trap is introduced just for supporting more bound states for a systematic analysis , and will not at all affect the appearance of the pz found above . @xmath104 and @xmath105 are used as units of energy and length . with these units three types of interactions are assumed : @xmath106 , @xmath107 ( when @xmath108 ) and @xmath109 ( when @xmath110 ) , and @xmath111 ( when @xmath112 ) and @xmath113 ( when @xmath114 ) , respectively , where @xmath115 . @xmath116 has a short - ranged character and was previously used in nuclear physics for the @xmath117-particles , @xmath118 belongs to the van der waals type for atoms , while @xmath119 is just a repulsive hard core potential . in fact , the three interactions are chosen quite arbitrary , just to show the possible similarity among different systems . with the interaction and the trap , the hamiltonian is @xmath120 where @xmath121 , or @xmath122 . a set of basis functions is introduced to diagonalize the hamiltonian to obtain the spectra and the eigenstates . the details are given in the appendix a. note that @xmath116 and @xmath118 both contain a minimum . thus the total interaction energy would be lower if the particle separations are appropriate . therefore the particles will pursue a better geometry . the et ( all six bonds can be optimized ) and the sq ( four bonds can be optimized ) with an appropriate size will be the first and second choices . however , for @xmath119 , the minimum is not contained , therefore the pursuit to a better geometry is less anxious . the spectra with the three types of interaction are plotted in fig . although the interactions are very different , the three spectra have common features . from table [ tab1 ] we know that the @xmath84 , @xmath85 , @xmath78 , and @xmath123 are et-ac * and/or sq-ac. they form the first group . this group of states are explicitly lower than the others , and the @xmath84 state is the lowest as expected . usually , the states with a larger @xmath17 is higher due to having a stronger rotation energy . however , for @xmath116 and @xmath118 , @xmath85 is higher than @xmath77 because the former is simply sq-ac while the latter is et-ac. nonetheless , for @xmath124 , the superiority of et over sq is weaker , thereby the normal sequence in this group is recovered . for all the three spectra , @xmath64 and @xmath67 are the highest because many pz are contained . these two form the third group . although they contain zero or seldom rotation energy , they are still much higher than the @xmath78 . this fact demonstrates the serious effect of the symmetry constraint . the remaining four states ( @xmath65 , @xmath90 , @xmath102 , and @xmath68 ) are in the middle , they form the second group . totally speaking , the two spectra for @xmath116 and @xmath118 are very similar ( except the energy scale ) . the similarity of the spectrum of @xmath124 to the other two is weaker . we calculate the root mean square radius @xmath125 to evaluate the size of each state ( refer to appendix b ) . the ratios @xmath126 are shown in fig . [ fig3 ] where the three dotted lines guiding the eyes go up and down in a synchronous way . it demonstrates once again the similarity among different systems . for all the three types of interactions , the sizes of the et-ac states @xmath84 , @xmath77 , and @xmath78 are relatively smaller , while the sizes of @xmath64 and @xmath67 are the largest ( because their wave functions are expelled from many pz as shown below ) . if dynamic effect played an essential role , the largest would have @xmath127 because the particles are pushing out by a stronger centrifugal force . but in fact not . the particle distributions in the @xmath128 states are shown by the one - body densities @xmath129 , where @xmath130 and @xmath131 are the magnitude and the polar angle of a position vector @xmath132 originating from the c.m . ( refer to appendix b ) . they are shown in figs . [ fig4 ] and [ fig5 ] . in general the distribution depends on @xmath19 . the choice @xmath133 implies that the @xmath134-axis is chosen to be lying along the direction of @xmath17 . in fig . [ fig4 ] , @xmath84 is isotropic as expected . @xmath67 is nearly isotropic . @xmath68 and @xmath78 have a denser distribution when the particle is closer to the equator ( @xmath135 ) . in general , when @xmath17 is larger , the system will more prefer to increase its moment of inertia so as to reduce the rotation energy . this explains why @xmath68 and @xmath78 prefer to be a little flattened . on the contrary , the distribution of @xmath85 prefers to be closer to the @xmath18-axis . this is not due to dynamics but due to symmetry constraint . since @xmath85 is sq-ac as shown in table [ tab1 ] , it will pursue the sq ( this will be further confirmed later ) . when the @xmath34 as shown in fig . [ fig1 ] is adopted , @xmath35 is chosen to be vertical to the sq . however , the sq can only appear in @xmath62 component due to the constraint eq . ( [ e03_psi ] ) . in this component @xmath17 is also nearly vertical to @xmath35 , and therefore @xmath17 and the sq are nearly coplanar . thus , in the choice @xmath133 , sq and the z - axis are nearly coplanar . this explains why the equator is less preferred by @xmath85 . furthermore , 4b is strikingly similar to 4a , 4c is also similar to 4a but in a less extent . thus , the angular distribution of particle is not seriously affected by the details of interaction . in the following , we shall see once and once that the similarity between the cases with @xmath136 and @xmath137 is high , but weaker with @xmath124 . it has been mentioned that , for @xmath124 , the pursuit to a better geometry is less anxious . this cause the difference in the extent of similarity . high similarity between the cases @xmath136 and @xmath137 emerges also in fig . [ fig5 ] for the odd - parity states . the similarity can be more clearly revealed by observing the wave functions directly . for this purpose , we introduce the hyper - radius @xmath138 , where @xmath139 and @xmath140 are the reduced masses . note that @xmath141 would remain invariant under the transformation between two sets of jacobi coordinates . therefore @xmath141 is suitable for measuring the size . from the normality and from eq . ( [ e02_psi ] ) , integrating over the euler angles , we have @xmath142 there are six arguments included in @xmath43 , one is @xmath141 responsible for the size and the others are denoted by @xmath143 responsible for the shape , and @xmath144 . thus the normality can be rewritten as @xmath145 , where @xmath146 is the probability density that the system has a given size and a given shape ( while the orientation of the shape has already been integrated ) . therefore @xmath147 is called the shape - density . we shall observe @xmath147 in some selected zones . since the et and the sq are both contained in @xmath24 , this zone is important . @xmath147 of the even - parity and odd - parity states observed in @xmath24 are given in figs . [ fig6 ] and [ fig7 ] , where @xmath32 and @xmath33 are variables . since @xmath148 , @xmath68 , @xmath64 , and @xmath65 have their wave functions absolutely being zero in this zone , they do not appear in these figures ( refer to table [ tab1 ] , where @xmath24 is a pzfor them ) . thus the states appear in the figures are @xmath84 , @xmath85 , @xmath78 , @xmath90 , @xmath77 , and @xmath102 . note that the @xmath149 state is both et-ac and sq-ac. due to the superiority of the et over the sq , all the peaks in 6a , 6b , and 6c ( for @xmath136 , @xmath137 , and @xmath124 ) are peaked at an et ( where @xmath150 ) but not a sq ( where @xmath151 ) . it has been mentioned that , for @xmath124 , the pursuit to a better geometry is less anxious . accordingly , the distribution in 6c extends considerably toward a sq . @xmath78 is also both et-ac and sq-ac. the sq of the @xmath78 can be nearly vertical to its @xmath17 because @xmath152component is allowed ( refer to eq . ( [ e03_psi ] ) ) . thereby the moment of inertia can be increased and the rotation energy can be reduced . accordingly , for @xmath78 , the sq becomes more competitive . in fact , for @xmath124 , the sq has replaced the et as the most probable shape as shown in 6i . @xmath77 is et-ac but not sq-ac. accordingly , the distribution is more concentrated around the et and away from the sq as shown in figs . [ fig7]d , [ fig7]e , and [ fig7]f . @xmath85 is not et-ac but sq-ac. accordingly , the distribution is more concentrated around the sq as shown in figs . [ fig6]d , [ fig6]e , and [ fig6]f . among the four states of the second group , @xmath90 and @xmath102 are allowed to appear in this zone , and are shown in the first and third rows of fig . they are neither et-ac nor sq-ac. there are two peaks in each panel , the peak with a smaller @xmath33 is the higher peak ( except in fig . [ fig7]i ) . it implies that the most probable shape is a compressed et along one of its two - fold axis . in general , one can define a specific degree of freedom ( accordingly , a specific 1-dimensional curve in the coordinate space ) to describe a specific mode of oscillation . when a wave function is remarkably distributed along the curve then the specific mode is involved . if a node appears at the curve , the oscillation has been excited . if more nodes appear along the curve , the excitation is more vigorous ( the simplest example is referred to the one - dimensional harmonic oscillator states ) . in each panel , let a curve starting from the upper peak ( a prolonged et ) , extending to the lower peak , and then extending to the abscissa ( refer to the solid curve added in fig . [ fig7]g ) . accordingly , @xmath33 decreases along @xmath153 ( refer to fig . [ fig1]a ) and the prolonged et suffers a compression and is transformed to a sq . once @xmath33 becomes zero , this process will continue but in reverse direction ( namely , @xmath33 increases further along @xmath154 . then , the sq is transformed back to a prolonged et , and repeatedly . the mode associated with the transformation is named the et - sq - et mode ( or the h - type oscillation named in @xcite ) . since the wave functions of @xmath90 and @xmath102 appear as nodes at the et and sq , the et - sq - et mode contains three nodes and therefore is an excited mode . this explains why @xmath90 and @xmath102 are considerably higher than @xmath85 and @xmath78 . from table 1 we know that @xmath25 is a pz for @xmath90 but not for @xmath102 . thus the increase of rotation energy in @xmath102 can be partially compensated due to the @xmath25-ac. therefore , the energy difference between @xmath155 and @xmath102 is small . in fact , the wave function of @xmath102 is also remarkably distributed in @xmath25 . say , for @xmath124 , the maximum of @xmath147 of @xmath102 in @xmath25 is equal to 1.52 located at a flattened regular pyramid , whereas the maximum in @xmath24 is equal to 0.71 located at a prolonged et ( shown in fig . [ fig7]i ) . thus , for @xmath124 , the most probable shape of @xmath156 is the flattened pyramid instead of the prolonged et . the other two member @xmath65 and @xmath68 of the second group do not appear in @xmath24 . from the spectra for @xmath136 and @xmath137 , we know that @xmath65 is higher than @xmath90 , and @xmath68 is higher than @xmath102 . this fact implies that the prohibition of @xmath24 will lead to an increase in energy . from table [ tab1 ] we know that both @xmath65 and @xmath68 will be distributed in @xmath25 . this is shown in fig . [ fig8]a-[fig8]c and [ fig8]d-[fig8]f , respectively . in all the panels , there are two peaks . the upper peak has larger @xmath87 ( i.e. , the base triangle is larger ) , and is associated with a flattened regular pyramid , while the lower peak a prolonged pyramid . let a curve be defined connecting the two peaks . this curve is associated with the transformation between the flattened and prolonged pyramid ( or the k - type of oscillation @xcite ) . since a node appears at the curve ( more exactly , at an et ) , this oscillation has been excited . also from table [ tab1 ] we know that coplanar structures might exist in these two states . this is shown in figs.[fig8]g-[fig8]i , where the two peaks are associated with a flattened and a prolonged diamond , and a node appears between them . thus an excited oscillation of a diamond around a sq is contained . for the @xmath64 state of the third group any coplanar structures are prohibited . this is due to the equivalence @xmath157 , where @xmath35 is chosen to be vertical to the plane ( @xmath158 causes no effect on @xmath159 states , while @xmath47 causes a change of sign on odd parity states , this leads to a contradiction ) thus , from table [ tab1 ] , the wave function of @xmath64 would be essentially distributed in @xmath26 . however , many subspaces in @xmath26 are pz of @xmath64 . this is shown in figs . [ fig9]a-[fig9]c . where @xmath160 and @xmath161 , and @xmath151 are three nodal lines because they are associated with coplanar shapes ( refer to fig . [ fig1]b ) . @xmath162 is also a nodal line because the associated shape belongs also to @xmath24 which is prohibited . furthermore , let the coordinates of the particle 2 relative to @xmath34 be denoted as @xmath163 . accordingly , those of the particles 1 , 3 , and 4 are @xmath164 , @xmath165 , and @xmath166 , respectively . one can see that , if @xmath167 the equivalence @xmath168 holds ( where @xmath46 is an interchange of a pair of particles ) , if @xmath169 the equivalence @xmath170 holds , and if x = y the equivalence @xmath171 holds . for @xmath64 , the left side of each of the equivalence will cause a change of sign while the right side does not . therefore @xmath64 must be zero in each of the subspace with @xmath167 , @xmath169 , or @xmath172 . the equality @xmath167 can be rewritten as @xmath173 . note that @xmath33 and @xmath32 are constrained as @xmath174 in figs . [ fig9]a-[fig9]c . thus @xmath175 is a nodal line . similarly , the case z = x leads to the nodal line @xmath176 , while the case x = y leads to the nodal line @xmath162 mentioned above . totally , six nodal lines are contained in figs . [ fig9]a-[fig9]c . these lines expel the wave function from favorable zones and cause strong inherent oscillation . this explains why the energy of @xmath64 is so high . for @xmath67 , due to the prohibition of @xmath26 and @xmath25 , the structure of tetrahedron is unfavorable . furthermore , due to the prohibition of @xmath98 , the coplanar structures are also unfavorable . it was found that this state is a mixture of a parallelogram and a deformed tetrahedron . the shape - density surrounding the parallelogram is shown in fig.s [ fig9]d-[fig9]f , where the vertical line is associated with the diamonds and is a nodal line for @xmath67 . the two peaks ( each peaked at a parallelogram ) separated by the nodal line implies an excited oscillation of the parallelogram . from the shape - densities plotted in figs . [ fig6 ] to [ fig9 ] , the decisive effect of the pz is clearly shown . since the pz is universal , the wave functions are expelled in a universal way , and the room left for dynamic adjustment is limited . this leads to similarity which is clearly shown by comparing the three patterns of any row in the figures for @xmath136 , @xmath137 , and @xmath124 , respectively . the existence of pz is not only for 4-boson systems but for all kinds of few - body systems containing identical particles . for an example of a 4-fermion system , it is assumed that the spin - dependent interaction tends to align the spins ( alternatively , a magnetic field is applied so that the spins are aligned ) . for this case , the spatial wave functions will be fully anti - symmetric . accordingly , the equivalent operation eq . ( [ e03_psi ] ) that hold in @xmath24 leads to the fact that @xmath177 is nonzero only if @xmath56 ( when @xmath178 ) , or @xmath58 ( when @xmath179 ) . in particular , the @xmath62 component exists only if @xmath178 and @xmath17 being even . accordingly , the et-ac states are now @xmath64 , @xmath68 , and @xmath102 . the sq-ac states are now @xmath85 , @xmath68 , and @xmath78 , while the states rejected from @xmath24 are now @xmath84 , @xmath67 , @xmath65 , and @xmath77 . among them @xmath67 and @xmath77 can both access @xmath25 , while @xmath67 can access also coplanar structures . therefore , for the ten states under consideration , we can predict that @xmath64 , @xmath85 , @xmath68 , and @xmath102 would be the lowest , while @xmath65 and @xmath84 would be the highest . we repeat our previous calculation for the case @xmath136but using fully anti - symmetric basis functions . the resultant spectra is shown in table [ tab2 ] . .the energies of the ten states ( in ascending order ) of the fermion system calculated with @xmath116 ( the unit is @xmath104 ) . [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] the result in table [ tab2 ] confirms the prediction . from the point of classical dynamics , it is incredible that there are zones in the coordinate space favorable in energy but absolutely prohibited . however , this paper demonstrates analytically and numerically that this is a fact in microscopic few - body systems . since the prohibition arises from the symmetry constraint which depends only on the inherent symmetry but is irrelevant to dynamics , the existence of the pz is universal disregarding the kinds of systems and the details of interaction . the quantum states of different systems will have exactly the same pz located at the same subspace when these states are governed by the same inherent symmetry ( namely , they belong to the same set of representations of the symmetry groups ) . since some of the pz have a higher geometric symmetry , they are the zones more favorable in energy . therefore , the prohibition causes serious effect and destroys thoroughly the picture painted simply by dynamics . consequently , the structures of the low - lying states are decisively determined by their inherent symmetry , and the room left for the adjustment from dynamics is narrow , in particular for the @xmath23 states although only 4-body systems with [ 1,1,1,1 ] and [ @xmath180 symmetries are concerned in this paper , obviously the existence of the pz is a universal phenomenon for all kinds of few - body systems with identical particles . acknowledgment : the support from the nsfc under the grant no.10874249 is appreciated . when the c.m . coordinates have been removed , the hamiltonian for internal motion @xmath181 can be expressed by a set of jacobi coordinates @xmath182 , @xmath183 , and @xmath184 . let @xmath185 which is the hamiltonian of harmonic oscillation , then @xmath186 let us introduce a variational parameter @xmath187 . the eigenstates of @xmath188 are denoted as @xmath189 , where @xmath190 and @xmath191 are the radial and angular quantum numbers , respectively . from @xmath189 a set of basis functions for the 4-body system is defined as @xmath192_{lm } , \label{e_a3}\end{aligned}\ ] ] where @xmath193 and @xmath194 are coupled to @xmath195 , then @xmath195 and @xmath196 are coupled to the total orbital angular momentum @xmath17 and its @xmath18-component @xmath19 , @xmath197 the parity , @xmath198 denotes the set of quantum numbers . these functions should be further symmetrized . for bosons , @xmath193 and @xmath194 should be even and we define @xmath199 where the right side is a summation over the permutations . making use of the talmi - moshinsky coefficients , @xcite each term at the right can be expanded in terms of @xmath200 . for an example , @xmath201 where @xmath202 is the talmi - moshinsky coefficients which can be obtained by using the method given in the ref.@xcite . note that the set @xmath203 has not yet been orthonormalized . thus a standard procedure is needed to perform to transform @xmath204 to a orthonormalized set @xmath205 , where @xmath206 is a serial number to denote a basis function , and each basis function can be expanded in terms of @xmath207 . finally , the set @xmath205 is used to diagonalize @xmath181 . due to the talmi - moshinsky coefficients , only the set of coordinates @xmath208 , @xmath209 , and @xmath210 is involved . thus the calculation of the matrix elements is straight forward . when @xmath197 and @xmath17 are given , a series of states will be obtained after the diagonalization . the lowest one of the series is called the first state . @xmath187 has to be adjusted so that the energy of the first state is as low as possible . due to having a trap , the bound states will tend to zero quite fast when @xmath130 tends to infinity . for this case , the above basis functions are suitable and will lead to a better convergency . let @xmath211 , and let the number of basis functions be constrained by @xmath212 . then , as an example of the bosonic case , the resultant energy of the lowest @xmath84 state with the interaction @xmath136 is 2.2965 , 2.2961 , and 2.2960 @xmath213 , respectively , when @xmath214 , 18 , and 20 . since the emphasis is placed at the qualitative aspect , the convergency appears to be satisfactory . after the diagonalization , each eigenstate @xmath217 is expanded in terms of @xmath218 , where @xmath5 denotes the @xmath5-th state of the @xmath219-series . for convenience , a transformation from the set @xmath220 to the set @xmath221 is made , and the eigenstate is re - expanded as @xmath222 . in order to extract the information of a particle from @xmath217 , another set of jacobi coordinates @xmath182 , @xmath223 , and @xmath224 is defined . accordingly , we have the basis function @xmath225_{lm}. \label{e_a6}\end{aligned}\ ] ] since the set @xmath221 and the set @xmath226 can be transformed into each other via the talmi - moshinsky coefficients , the eigenstate can once again expanded as @xmath227 where the coefficients @xmath228 can be known from the diagonalization and from the transformation between sets of basis functions . now , from the normality , we have @xmath229 , \label{e_a8}\end{aligned}\ ] ] where @xmath230 is the polar angle of @xmath231 , and @xmath232 is the distance of particle 3 from the c.m . obviously , the quantity inside the bracket is just the one - body density @xmath233 describing the particle distribution ( all the four particles are distributed in the same way ) .	in some special zones of the high - dimensional coordinate space of few - body systems with identical particles , the operation of an element ( or a product of elements ) of the symmetry groups of the hamiltonian on a quantum state might be equivalent to the operation of another element . making use of the matrix representations of the groups , the equivalence leads to a set of homogeneous linear equations imposing on the wave functions . when the matrix of these equations is non - degenerate , the wave functions will appear as nodal surfaces in these zones . in this case , these zones are prohibited . in this paper , tightly bound 4-boson systems with three types of interaction have been studied analytically and numerically . the existence of the universal prohibited zones has been revealed , and their decisive effect on the structures of the eigenstates is demonstrated .
the statistical model of atom developed by e.fermi @xcite and known as thomas - fermi model , although based on a highly simplified theoretical framework , has been proven surprisingly good in predicting basic properties of condensed matter @xcite . a particular feature of such a model is the description of compressed atoms , in this case the predicted properties were confirmed experimentally @xcite . in a preceding publication @xcite , we focused the attention on this particular aspect of the theory and proposed a simple model for describing systems under pressure . the central point of that work was the development of the concept of `` statistical ionization '' . in simple terms , this is a universal semianalytical function which enables one to describe , as a function of the distance from the point - like nucleus , the balancing process between the antibinding and binding contribution to the total energy within the compressed atom . in spite of the approximations done and extensively discussed , we underlined the utility of the proposed model as a tool for investigating at a basic level and at low computational cost , properties of systems under pressure . however it was also underlined that the properties of semianalycity and universality of the function disappear when higher order of approximation are introduced in the basic tf theory . in the light of what stated before , in this publication we intend to extend the treatment of the previous work to more sophisticated models of the tf approach . that is to say , to include the effects of exchange and correlation into the original tf model and obtain in this framework a `` generalized statistical ionization function '' . the extension of the original tf model is due to p.a.m.dirac ( tfd ) @xcite for the exchange while for the part relative to the correlation , several approaches has been proposed ; in this work the one proposed by h.lewis ( tfdl ) @xcite has been chosen because it is simple and appropriate for the compressed case , since the treatment of the electrons becomes exact in high density limit ( see appendix ) . the paper is organized as follows ; first we obtain the tf , tfd and tfdl equation within a single generalized approach , then we show numerical solutions for different atomic numbers in the neutral uncompressed case , next we illustrate numerical results for the `` generalized statistical ionization function '' in case of @xmath0 ( @xmath1 atomic number ) . finally the equation of state of a compressed system is calculated . comments on the results obtained as well as on advantages and limitations of this model , conclude the work . in this section we derive a unique general form for the tf , tfd and tfdl equation . in a semiclassical approximation the local electronic density of states can be defined as ( see for example @xcite ) : @xmath2=\frac{8\pi}{h^3}p^2({{\bf r } } ) dp({{\bf r}}).\ ] ] the local electron density is therefore given by : @xmath3 which determines the local fermi vector @xmath4 . in the spirit of the ( generalized ) thomas - fermi approach we can express the one - particle energy @xmath5 $ ] through an effective hamiltonian @xmath6 $ ] of free ( interacting ) electrons plus an electrostatic field @xmath7 arising from the direct electron - nucleus and electron - electron interaction ( we must remind that the nucleus is considered a positive point - like charge ) : @xmath8 = h_{\rm el}[p({{\bf r}})]-ev({{\bf r}}),\ ] ] for a system at the equilibrium the chemical potential @xmath9 , defined by the maximum energy of @xmath10\}=e[p_{\rm f}({{\bf r}}),{{\bf r}}]$ ] , has to be a constant @xmath11 independent of @xmath12 . putting for convenience @xmath13 we obtain therefore : @xmath14= e[v({{\bf r } } ) - v_0].\ ] ] the effective field @xmath7 will be determined self - consistently via electrostatic considerations by using the poisson s equation : @xmath15= \frac{1}{r}\frac{d^2}{dr^2}\left\{r\left[v(r)-v_0\right]\right\ } = e\frac{n(r)}{\epsilon_{0 } } , \label{poissonv}\ ] ] where @xmath16 . ( [ poissonv ] ) can be quite simplified by introducing the variables @xmath17 and @xmath18 defined by : @xmath19 @xmath20 eq . ( [ poissonv ] ) becomes thus : @xmath21 and finally , by scaling @xmath22 , we have : @xmath23 this is the `` generalized tf equation '' , where @xmath24 is the atomic bohr radius and @xmath25 a closed system of differential equations is now obtained by expressing back the variable @xmath26 as function of @xmath27 . to this aim the explicit expression of the electronic hamiltonian @xmath28 $ ] is needed . of course this task is in principle quite hard since the solution of a many - body system is necessary , and some kind of approximation is required . the original tf model represents the simplest approximation , i.e. the hamiltonian @xmath29 $ ] is approximated with the only kinetic term : @xmath30=\frac{p^2(r)}{2m}. \label{hptf}\ ] ] the total hamiltonian contains therefore only kinetic and the direct coulomb electron - electron interaction , totally neglecting any quantum contribution . by using again the reduced variables @xmath26 and @xmath27 , we can write eq . ( [ hptf ] ) in the compact form : @xmath31 which , together with eq . ( [ firsteq ] ) , defines the statistical solution of the thomas - fermi model . as stated before higher order of approximation have been introduced in literature to include quantum exchange and correlation contribution to eq . ( [ hptf ] ) . most famous is the so called thomas - fermi - dirac ( tfd ) model which explicitly takes into account the exchange energy ( see @xcite ) . the inclusion of the correlation term has been a more debated issue , and different approaches have been proposed . a particular suitable and simple one is the interpolation formula given by h.w . lewis , which becomes exact in the both the high - density and low - density limits @xcite . we referred and will refer to it as the tfdl model ; in this case the hamiltonian contains kinetic , exchange and correlation terms : @xmath32= \frac{p^2(r)}{2 m } -\frac{e^2p(r)}{4 \pi \epsilon_{0}\pi \hbar } -\frac{me^4(1-\ln2)}{(4 \pi \epsilon_{0 } \pi \hbar)^2 } \ln\left[1 + \frac{(0.89 \alpha \pi -1)\pi p(r ) a_0}{(1-\ln 2)\hbar}\right ] , \label{eqlewis}\ ] ] where @xmath33 introducing the variables @xmath26 and @xmath27 , eq . ( [ eqlewis ] ) reads : @xmath34 , \label{tfdlphi}\ ] ] where @xmath35 @xmath36 and @xmath37 eq . ( [ tfdlphi ] ) can be considered a generalization of the tf model [ eq . ( [ tfphi ] ) ] to include exchange and correlation energies . we find lewis s formula particularly appealing since we are mainly interested in the high - density regime of compressed atoms where the interpolation formula provided by eq . ( [ eqlewis ] ) is expected to work quite well . in spite of the simplicity of the model , to our knowledge , no explicit numerical or analytical treatment of this equation has been published , although several authors @xcite cite this approach in their work . the main reason why this happens is due to the fact that since the tfd model does not produce evident changes with respect to the simple tf solutions , the tfdl is expected to not be much different . however in our paper we derive for the first time explicit numerical solutions of the tfdl model and we use it as basic starting point of our analysis ; the results will be compared with the simple tf model . although we do not expect large differences with respect to the tf , we carry on this study to confirm a qualitative and quantitative equivalence between tf and tfdl and later on the consequent advantages of such a conclusion will be underlined . in this section we illustrate some tf , tfd and tfdl solutions for the neutral uncompressed case , the trend shown for this case is preserved for the compressed case . then we show within a generalized tf approach , the procedure to obtain the `` statistical ionization function '' via the total energy as a function of the distance from the nucleus . ( [ firsteq ] ) and ( [ tfdlphi ] ) , or eqs . ( [ firsteq ] ) and ( [ tfphi ] ) , represent two sets of coupled differential equations which can be solved to obtain @xmath26 or , equivalently , @xmath27 , and as a consequence the local electron density @xmath38 . the boundary conditions required to determine a unique solution are provided by the asymptotic limit : @xmath39 and by fixing the radius of a neutral atom : @xmath40 although it has been widely pointed out in literature , it is important to remind that the inclusion of exchange and correlation terms breaks down the universality of the solution . namely , it is not possible to obtain local electron density @xmath38 for a given atomic number @xmath1 just by a proper scaling of the physical quantities ( length , energy etc ) . in fig . [ fi - x ] we show the radial distribution of @xmath26 for a neutral uncompressed atom for the thomas - fermi model and for the thomas - fermi - dirac - lewis model with statistically relevant values of atomic number : @xmath41 . only a slight discrepancy is found as result of the inclusion of the exchange and correlation terms , pointing out the almost quantitative accuracy of the tf model . as expected , such a discrepancy vanishes as the atomic number @xmath1 is increased . this result show that tf and tfdl are quantitatively as well as qualitatively equivalent ( the trend as stated before holds in the compressed case as well ) . by the knowledge of the solution @xmath26 we can now evaluate the semiclassical expression of the local density of energy @xmath42 which is related to the one electron energy @xmath43 $ ] through the relation @xmath44 e[p(r),r]$ ] . we can identify three different contributions : @xmath45 where @xmath46 \frac{p^2(r)}{2m},\ ] ] @xmath47 @xmath48 where of course @xmath49.\ ] ] a particularly meaningful quantity is also the integrated energy defined @xmath50 by : @xmath51 where @xmath52 is a generic distance @xmath53 . @xmath50 represents therefore the total energy contained in the atomic volume enclosed within the distance @xmath52 from the nucleus . just as @xmath42 , we can think @xmath50 as composed by a kinetic , a coulomb electron - nucleus and a coulomb electron - electron term . after some straightforward manipulations ( see also @xcite ) , by using once more the reduced variables , we can write each of these contributions as : @xmath54 \int_0^x dx v^5(x ) x^{-1/2},\ ] ] @xmath55 \int_0^x dx v^3(x ) x^{-1/2},\ ] ] @xmath56 \int_0^x dx v^3(x ) x^{1/2 } \left[\frac{1}{x}\int_0^x v^3(x ' ) { x'}^{1/2 } + \int_0^{x_0 } v^3(x ' ) { x'}^{-1/2 } \right],\ ] ] where @xmath57 is the scaled variable related to @xmath52 through the relation @xmath58 and where we made use of the relation : @xmath59,\ ] ] valid for an isotropic @xmath60 . as it was done in @xcite , we consider @xmath52 a variable distance within the atom , and @xmath50 , as stated before , is the total energy in classical terms , of the sphere of radius @xmath52 inside the spherical atoms . in simple terms , we do divide the atom into infinitesimally thin concentric shells ; the energy at @xmath52 is the sum of the contributions of all of the shells inside @xmath52 . as a consequence the distance at which @xmath50 has its minimum can be interpreted as the distance at which the binding ( nucleus - electrons ) and antibinding ( electron - electron and kinetic energy ) contributions to the total energy are in exact balance . this allows us to define a sort of electron ionization criterion , where the term `` ionization '' stands for electrons with zero or positive energy ; thus we can address to @xmath61 as the `` generalized statistical ionization function '' . the exact meaning of statistical ionization with the related limitations has been extensively discussed in our previous work , for such a reason we do not spend more discussion about it and focus the attention onto the numerical results . the number of ionized electron is therefore given by @xmath62 which , written in reduced variables , reads : @xmath63 differently from the procedure adopted in our previous work , in this case @xmath61 can be studied only numerically , since it is not possible to solve analytically the integrals and so obtain a semianalytical expression . moreover the dependence of @xmath26 on @xmath1 does not allow to obtain a universal function where to pass from one atom to another is possible by simply scaling in @xmath1 . next we show results obtained by studying numerically the `` generalized statistical ionization function '' for the case @xmath0 ; however since we did find quantitative agreement between tf and tfdl , the results shown are valid for any atomic number by opportunely scaling them . in this section we show results obtained for the `` generalized statistical ionization function '' for the case @xmath0 . as it was already underlined , we expect that the results are quantitatively and qualitatively equivalent in case tf or tfdl model is used , since the solution of the tf and tfdl equation do not differ , as it was shown in the previous section . this means that the generalized ionization function is the numerical equivalent of the semianalytical and universal statistical ionization function of our previous work ; thus the results ( and the equation of state ) shown here for @xmath0 , can be easily generalized to any @xmath1 by simply scaling . indeed the function @xmath61 does not show a different behavior when tf or tfdl are used ; in figure [ etot_s - s ] the plots obtained using tf and using tfdl coincide . as it is possible to see in figure [ etot_s - s ] a minimum is always obtained for any compression , and in figure [ etot_s = n_x ] a particular compression has been chosen and the determination on the number of ionized electrons is pictorially illustrated . accordingly to the procedure of our previous work , at this point we can therefore model the compressed atom as a core with radius @xmath64 plus a number of ionized electrons @xmath65 which in principle is free and is spread over all the atomic volume . the pressure is therefore determined by the density of ` ` free '' ( ionized ) electrons as we proposed in the previous work ( again limitations are extensively discussed there ) : @xmath66 where @xmath65 is given by eq . ( [ nions ] ) and the atomic volume is simply @xmath67 . first we study the ionization as a function of the compression for the single atom ( see fig.[nion - xcore - x0 ] ) and then we extend the procedure to a macroscopic level so that the resulting equation of state is shown in fig . [ p - vf ] , where the pressure in units of rydberg over the bohr atomic volume @xmath68 is plotted as function of the volume . we generalized the concept of `` statistical ionization function '' for electrons in a compressed atom , obtained in a previous work , within the tf approach . the generalization was developed by extending the previous analysis to more sophisticated tf models where exchange and correlation are considered . we found that there is no qualitative and quantitative difference between the original and the generalized approach . although one could have expected this result , we explicitly proved such an equivalence , and for the first time , solutions of the tfdl equation where shown . the important conclusion of this work concerns the fact that because of the equivalence shown , the results reported for the particular case of @xmath0 are valid for any atomic number , provided that a rescaling @xmath69 , where @xmath70 is the atomic number wished , is applied ; above all the results legitimate the use of the semianalytical and universal ionization function obtained in our previous work with the evident advantages of the low computational cost and its extreme simplicity and immediacy . moreover the fact that lewis formula is exact for the high density limit , means that in our case the effects of correlation were well described , as a consequence one can conclude that such effects are not relevant for describing atoms under pressure , at least in first approximation ; this result it is not obvious in an _ a priori _ analysis . as it was discussed in the previous work , the equation of state is a simplification and as a consequence far from being rigorous ; for instance an open problem of our model is the distribution of the `` ionized '' electrons which in the present work is considered simply uniform . however , due to its simplicity and feasibility , the model can be applied for a basic study of compressed matter not only via the determination of the equation of state , but also as a basis for developing more efficient analysis within more sophisticated theories . the example shown in our previous publication , was the determination of the `` ionized '' electrons for a certain compression and the consequent description , in an _ ab initio _ method , of the ionized electrons via plane - waves wavefunctions while the other electrons can be represented as a core or as localized orbitals ; this would speed up the convergence for self - consistent calculations of compressed matter . here we can say more , as it is well known , the _ ab initio _ calculations are based on the concept of pseudopotential ; only the valence electrons are explicitly taken into account while the rest are placed in a core described by an opportune potential which interacts with the valence electrons . valence electrons and core are known for the uncompressed atoms but one may ask what happens in case the system is under high pressure , in this case the valence electrons and the core should be redefined according to the degree of compression . in this case our model , at basic level would be very helpful for estimating the number of ionized ( valence ) electrons and define what is the core . the evident advantage of such a procedure stays in the simplicity of such an estimate and in the very low computational cost . in conclusion we think that the study performed in this work furnishes important information and tools for a computational inexpensive and well founded basic analysis of compressed matter . we briefly illustrate the procedure followed by lewis to develop a suitable formula for the electron correlation , for more details see @xcite and references therein . consider an electronic fermi gas whose maximum momentum is the fermi momentum @xmath71 and density @xmath72 ; the correlation energy at high density can be calculated via the gell - mann s scheme and leads to the following expression @xmath73 @xmath74 and @xmath75 are respectively the electron mass and the electron charge . this expression must be modified in such a way that its validity could be reasonably extended to any density and in particular must be exact at low density . at this point lewis invokes wigner procedure for the calculation of electron correlation for a dilute gas ; i.e. one simply needs to note that a very dilute electron gas in the ground state crystallize into a body - centered cubic lattice and at this point the correlation energy can be determined exactly via a madelung type technique . the expression for low density obtained is @xmath76 where @xmath77 . finally the procedure leads to what lewis defines as `` a suitable interpolation formula '' for the correlation energy valid for any density : @xmath78 \label{formula3}\ ] ] where @xmath79 . this is the expression used by lewis to incorporate the correlation effects into the tfd model . e.fermi , rend.accad.naz.lincei 6 ( 1927 ) 602 . r.p.feynmann , n.metropolis , e.teller , phys.rev . 75 ( 1949 ) 1561 r.f.trunin , m.a.podurets , g.v.simakov,l.v.popov , b.n.moiseev , sov.phys . jetp 35 ( 1972 ) 550 . l.delle site , physica a , 293 ( 2001 ) 71 ; + l.delle site , phisica a , 295 ( 2001 ) 562 . p.a.m.dirac , proc.cambridge phil.soc . 26 ( 1930 ) 376 . h.lewis , phys.rev . 111 ( 1958 ) 1554 . l.landau , m.j.kearsley , e.m.lifshits , statistical physics : part 1 , buttenworth - heinemann , oxford , 1980 . j.f.barnes , phys.rev . 140 ( 1965 ) j.f.barnes , phys.rev . 153 ( 1967 ) 269 . e.e.sapleter , h.s.zapolsky , phys.rev . 158 ( 1967 ) 876 . k.ebina , t.nakamura , j.phys.soc . of jap . 52 ( 1983 ) 1658 . s.l.shapiro , s.a.teukolsky , black holes , white dwarfs and neutron star , wiley , new york , 1983 .	in a previous work one of the authors proposed a simple model for studying systems under pressure based on the thomas - fermi ( tf ) model of single atom . in this work we intend to extend the previous work to more general thomas - fermi models where electronic exchange and correlation are introduced . to do so , we first study numerically the equation obtained by h.w.lewis ( tfdl ) which introduces the effects of exchange and correlation into the original tf equation ; next the procedure followed in the previous work is extended to the new approach and a specific example is illustrated . although one could expect that no big differences were produced by the generalized tf model , we show the qualitative as well as quantitative equivalence with detailed numerical results . these results support the robustness of our conclusions with regards to the model proposed in the previous work and give the character of universality ( i.e. to pass from one atom to another , the quantities calculated must be simply scaled by a numerical factor ) to the properties of compressed systems shown in this work .
we consider a general system of @xmath0 chemical species @xcite inside a fixed volume with @xmath1 denoting the number of molecules . the stoichiometric matrix @xmath2 describes changes in the population size due to @xmath3 different reaction channels , where each @xmath4 describes the change in the number of molecules of type @xmath5 from @xmath6 to @xmath7 caused by an event of type @xmath8 . the probability that event @xmath8 occurs during the time interval @xmath9 equals @xmath10 , where the @xmath11 are called the transition rates . this specification leads to a poisson birth and death process described by the following stochastic equations @xcite @xmath12 where the @xmath13 are independent poisson processes with rate 1 . in order to define the contribution of the @xmath8th reaction @xmath14 to the variability of @xmath15 we first define @xmath16 as the expectation of @xmath17 conditioned on the processes @xmath18 so that @xmath19 is a random variable where timings of reaction @xmath8 have been averaged over all possible times , keeping all other reactions fixed . therefore @xmath20 is a random variable representing the difference between the native process @xmath17 and a time - averaged process of the @xmath8th reaction . now the contribution of the @xmath8th reaction to the total variability of @xmath15 is @xmath21 where @xmath22 denotes the temporal average over all @xmath3 reactions . this definition is similar to the one proposed in @xcite to quantify the contributions of promoter states and mrna fluctuations , respectively , to the protein level variability . in general , it is difficult to calculate or study properties of equation ( [ contrib_def ] ) using a poisson birth and death process framework ( [ kurtz_x ] ) . here instead we use the linear noise approximation ( lna ) , which allows us to model stochastic systems using wiener processes driven @xmath23 instead of poisson processes driven @xmath24 stochastic differential equations @xcite . the lna is valid if the number of interacting molecules is sufficiently large @xcite and decomposes the system s state , @xmath25 , into a deterministic part @xmath26 and a stochastic part @xmath27 @xmath28 here @xmath26 and @xmath27 are described by the deterministic and stochastic differential equations @xmath29 respectively , and their coefficients are given by the following formulae @xmath30 the lna presents a simple way to compute contributions @xmath31 , and here we demonstrate how the total variance can be decomposed into the sum of individual contributions . we first write the explicit solution for the process @xmath32 as @xmath33 and @xmath34 , where @xmath35 is the fundamental matrix of the non - autonomous system of ordinary differential equations and @xmath36 instead of @xmath37 and @xmath38 , respectively . ] @xmath39 now it is straightforward to verify that @xmath40 where @xmath41 , @xmath42 , @xmath43 , @xmath44 and @xmath45 from ( [ sol_xi ] ) and ( [ xi_j ] ) we have @xmath46 with @xmath47 and the time derivative of @xmath48 we obtain for @xmath31 , @xmath49 with @xmath50 this is , of course , analogous to the fluctuation dissipation theorem , with the exception that the diffusion matrix contains zeros at all entries not corresponding to the @xmath8th reaction . now the fact that the total variance @xmath51 can be represented as the sum of individual contributions @xmath52 results directly from the decomposition of the diffusion matrix @xmath53 and the linearity of the equation for @xmath54 , given by the standard fluctuation dissipation theorem @xmath55 with the decomposition ( [ sigmaj_sum ] ) it is in principle possible to detect reactions that make large contributions to the output variability of biochemical reactions . but even simple systems , for which analytic formulae exist , usually have complicated noise structures ; we can nevertheless prove two general propositions that assign surprisingly substantial contributions to the degradation of an output signal . we formulate them in this section ( proofs are in the _ appendix _ ) and illustrate them further below . consider a general system such as described at the beginning of the _ noise decomposition _ section . in addition , assume that the deterministic component of the system @xmath56 has a unique and stable stationary state ( all eigenvalues of matrix @xmath57 have negative real parts ) . if @xmath58 is an output of this system being produced at rate @xmath59 and degraded in the @xmath3th reaction at rate @xmath60 then the contribution of the output s degradation is equal to the half of its mean ; more specifically , @xmath61_{nn}=\frac{1}{2 } \langle x_n \rangle.\ ] ] now consider again a general system but assume that reaction rates are of mass action form and that only three types of reactions are allowed : production from source : : : @xmath62 degradation : : : @xmath63 conversion : : : @xmath64 . to satisfy the openness assumption each species can be created and degraded either directly or indirectly ( via a series of preceding conversion reactions ) . as in _ proposition 1 _ let @xmath58 be the output of the system . under this assumption the degradation of the output contributes exactly half of the total variance of the system s output , @xmath65_{nn}=\frac{1}{2 } \left[\sigma\right]_{nn},\ ] ] where @xmath3 is again the index of the output s degradation reaction . _ proposition 2 _ can be understood as a balance between production and degradation reactions . if we consider all reactions except @xmath3th as producing species @xmath58 , then production and degradation contribute the same amount of noise . usually , however , there is more than one production reaction , and therefore it is more convenient to interpret this result as the contribution of a single reaction . both propositions indicate that a substantial part of noise is a result of the signal s degradation reaction . this observation is particularly interesting as open conversion systems are often good approximations to important biochemical reactions such as michaelis - menten enzymatic conversions or linear signal transduction pathways , which we discuss below ; it suggests also that controlled degradation is an effective mechanism to decrease the overall contribution to variation in protein levels . in this section we demonstrate how our methods can be used to study origins of noise in biochemical systems and present consequences of the above propositions . for illustrative purposes we start with a basic birth and death process , where molecules @xmath66 arrive at rate @xmath67 and degrade at rate @xmath68 . in the lna this process can be expressed by the following stochastic differential equation @xmath69 the stationary distribution of this system is poisson with the mean @xmath70 . therefore in the stationary state death events occur at rate @xmath71 which must be equal the birth rate @xmath67 . the noise terms in the above equation are equal at stationarity indicating that contributions of birth and death reactions are equal . using formula ( [ sigma_j ] ) it is straightforward to verify that the decomposition @xmath72 holds indeed . here we first focus on conversion pathways , where molecules @xmath73 are born at a certain rate @xmath74 and information is transmitted to molecules @xmath58 using ( possibly reversible ) conversion reactions . all molecules can degrade with arbitrary first order rates , and we assume that the final product @xmath58 is not the substrate of any conversion reaction . under these assumptions , regardless of parameter values , length of the pathway , and degradation of intermediates , the degradation of the output @xmath58 , rather unexpectedly , contributes exactly half of the variance of the signalling output , @xmath58 . on the other hand , if information is transmitted not using conversion reactions , but each species catalyses the creation of the subsequent species , i.e. @xmath75 , then according to the _ proposition 1 _ , the contribution of the degradation of @xmath58 to the output s variance equals half of its mean regardless of the cascade length and parameter values . in order to study contributions of intermediate reactions , we considered cascades of the length three ( see figure [ cascades_arrows ] ) and decomposed variances numerically for both conversion and catalytic cases . decompositions are presented in figure [ cascades ] . for conversion cascades we used two parameter sets corresponding to fast and slow conversions to demonstrate that increasing the rates of intermediate steps decreases their contribution as might intuitively be predicted ( top panel of figure [ cascades ] ) . for catalytic cascades increasing the rates of reactions at step two and three ( going from slow to fast ) increases the contribution of reactions at the bottom of the cascades . for slow dynamics during steps two and three , the relatively fast fluctuations at the start are filtered out ( low pass filtering ) . if the dynamics of all species occur at similar times scales then these fluctuation can efficiently propagate downstream ( bottom panel of figure [ cascades ] ) . michaelis - menten enzyme kinetics are fundamental examples of biochemical reactions ; substrate molecules ( @xmath76 ) bind reversibly to enzyme molecules ( @xmath77 ) with the forward rate constant @xmath78 and the backward rate constant @xmath79 to form a complex ( @xmath80 ) , which then falls apart into the enzyme and a product ( @xmath81 ) at rate @xmath82 : @xmath83 . to ensure existence of the steady state , we assume that substrate molecules arrive at rate @xmath84 and are degraded at rate @xmath85 . at the unique steady state , if the concentration of an enzyme is large compared to the substrate , the system is well approximated by a set of mono - molecular reactions : @xmath86 our theory therefore predicts that half of the noise in michaelis - menten kinetics operating in its linear range ( abundant enzyme ) is generated by degradation of the product molecules . in figure [ michaelis ] we have calculated the contributions of each of the four reactions to the variance of all four species for the full model ( without themonomolecular approximation ) . the contribution of the output degradation is as predicted by the monomolecular approximation . the canonical example of a linear catalytic pathway is the gene expression process , which can simply be viewed as the production of rna ( @xmath73 ) from source at the rate @xmath87 , and production of protein ( @xmath88 ) in a catalytic reaction at rate @xmath89 together with first order degradation of both species at rates @xmath90 , @xmath91._proposition 1 _ states that the part of the variance resulting from the protein degradation equals half of the protein mean . moreover , formula ( [ sigma_j ] ) allows us to derive the complete decomposition @xmath92 a similar decomposition was first presented by @xcite , however , only into contributions resulting from fluctuating species and without theoretical explanation ; the latter was provided later by @xcite . below we investigate two extensions of the above model . first , we assume that the promoter can fluctuate between `` on '' and `` off '' states ( similarly to @xcite ) and calculate contributions for different time - scales of these fluctuations . second , we assume that the protein is a fluorophore that undergoes a two - step maturation process before it becomes visible ( folding and joint cyclization with oxidation ) . figure [ promoter ] presents contributions for fast and slow promoter kinetics , showing that fast fluctuations are effectively filtered out ( contributing 10 % ) but remain a substantial contributor when they are slow ( contributing 40 % ) . variability in gene expression is often measured by means of fluorescent proteins that undergo maturation before becoming visible for detection techniques ; but the process of maturation @xcite itself is subject to stochastic effects , and can thus contribute significantly to the observed variability . we used typical parameters @xcite for fast and slow maturing fluorescent proteins and found that maturation contributed 4 and 25% , respectively ( figure [ maturation ] ) to the overall variability ; here our method allows for the rigorous quantification compared to the previous qualitative observations of @xcite . in order to demonstrate that our methodology and predictions are valid also for more general out - of - steady state models we next focus on the p53 regulatory system which incorporates a feedback loop between the tumour suppressor p53 and the oncogene mdm2 , and is involved in the regulation of cell cycle and the response to dna damage . we consider the model introduced in @xcite and later analysed in @xcite that reduces the system to three molecular species , p53 , mdm2 precursor and mdm2 , denoted here by @xmath93 and @xmath94 . the model incorporates six reactions ( @xmath95 ) , and its deterministic version can be written in the form of the following ordinary differential equations @xmath96 using equation ( [ sigma_j ] ) we calculated the contributions of all of the six reactions present in the model . results are presented in figure [ p53_contrib ] . the contributions of each reaction oscillates over an approximately constant range over the course considered here , except for the reaction of @xmath97 degradation : the contribution of the @xmath97 degradation accumulates over time . this observation is consistent with our propositions which predict for the steady - state models that the contribution of output signal degradation is significant , and generally overrides the contributions of other reactions . in @xcite the authors studied the applicability of the lna for the analysis of oscillatory models and found that the approximation fails in this type of models when long time periods are considered . in these cases the total variance diverges to infinity with time . they also decomposed the total variance into an oscillatory and a diverging part ; the diverging part corresponds to the variability in the oscillation s period . the diverging part in figure [ p53_contrib ] results from the p53 degradation , indicating that stochasticity of this reaction is responsible for the variability of the period in this system . the propositions exemplified by the above models demonstrate that the noise resulting from degradation of an output is a substantial source of variability in biochemical systems . in a general system it contributes a fraction of the variance equal to the half of the mean output . here we show how controlling the degradation can reduce this contribution . we consider a simple birth and death process , but the same mechanism is valid for general system considered in the _ proposition 1_. if births and deaths occur at the state dependent rates @xmath98 and @xmath99 , respectively , such a system is described by @xmath100 at equilibrium , we have @xmath101 , and using formula ( [ sigma_j ] ) we can calculate that contribution of degradation is @xmath102 for @xmath103 , @xmath104 this reduces , of course , to the previously discussed case @xmath105 . nevertheless , if functions @xmath106 and @xmath107 are of hill and michaelis - menten type , respectively , then the contribution is no longer directly related to the mean and can be reduced to an arbitrary low level according to the above formula . any decrease in the contribution compared to the mean is achieved by the reduction of the variance resulting from the reduced flux through both reactions and autoregulatory control effects . recent experimental work on protein degradation _ in vitro _ provides the evidence that degradation indeed exhibits michaelis - menten type kinetics @xcite , instead of the linear first order kinetics that are usually used to model degradation . the effect of autoregulation is depicted in figure [ figure1 ] . the noise decomposition method introduced here allows us to investigate in detail where and how noise enters biochemical processes , is propagated through reaction systems , and affects cellular decision making processes . we have shown analytically that in a wide class of systems the signal degradation reaction contributes half of the noise in a system s output regardless of parameter values . we have also carried out numerical study for a system that never reaches steady - state and confirmed the surprisingly important role of degradation in controlling stochasticity of the p53 system , as well as in a range of generic systems biology models . quite generally , the ability to dissect noise propagation through biological systems does enable researchers better to understand the role of noise in function ( and evolution ) , and will also enable synthetic biologists to either harness or dampen the effects of noise in molecular signalling and response networks . one of the central results that we found and report here is the crucial role of degradation of the signal on the overall noise levels . the relevance of degradation has not been studied before in the context of stochastic biochemical dynamics and to our knowledge our study is the first that draws attention to the importance of degradation . this is particularly important as this may indicate new therapeutic targets : in humans and other other sequenced organisms , certainly , the repertoire of proteins involved in protein degradation , in particular ubiquitin - ligases , is as rich and diverse as the repertoire of proteins regulating their activation , the kinases ( and phosphatases ) . thus targeting the degradation of proteins appears as important to biological systems as protein activation and offers an attractive and broad range of new potential therapeutic targets . mk and mphs acknowledge support from the bbsrc ( bb / g020434/1 ) . jm would like to thank the polish ministry of science and higher education for a financial support under the grant n201 362536 . mphs is a royal society wolfson research merit award holder . * 1 ) * interactions of @xmath58 with other species imply that @xmath108 for @xmath109 and @xmath110 . thus matrix @xmath57 can be written as @xmath111 * 2 ) * formula ( [ d_j ] ) implies that all elements of matrix @xmath112 are equal to @xmath113 except @xmath114 , therefore @xmath112 has the form@xmath115 * 3 ) * it is straightforward to verify that the @xmath116 matrix @xmath117 satisfies the equation @xmath118 in an open conversion system @xmath119 ( see e.g. @xcite ) , therefore from _ proposition 1 _ we have @xmath120 , and @xmath127 being the stationary solution of @xmath128 . the following parameters were used : @xmath129 , @xmath130 , @xmath131 @xmath132 , @xmath133 . for fast fluctuations we used @xmath134,@xmath135 , and for slow @xmath136 , @xmath137 . all rates are per hour . [ promoter ] ] . for slow maturation ( average maturation time approx . 5 h. ) we assumed folding and maturation rates to be @xmath138 and @xmath139 , respectively . for fast maturation ( average maturation time approx . 0.5 h ) we set @xmath140 , @xmath141 . all rates are per hour . [ maturation ] ] trajectories ( top left ) and decomposition of variance into the contributions corresponding to each of the @xmath142 reactions for mdm2 ( top right ) , mdm2 rna ( bottom left ) and p53 ( bottom right ) . we used the index @xmath5 of the parameters @xmath143 in equations ( [ p53_1]-[p53_3 ] ) to denote reaction they describe . we used model and parameters published in @xcite . figure demonstrates accumulation of the noise contributed by degradation of p53 ( @xmath142th reaction ) . the following parameters were used : @xmath144 all rates are per hour . ] the effect of noise reduction resulting from regulated degradation . each colour describes production rate @xmath98 ( dotted line ) , degradation rate ( dashed line ) and density of stationary distribution of @xmath66 ( solid line ) . for blue : @xmath145 , @xmath146 ; for black : @xmath147 , @xmath148 ; for green : @xmath149 , @xmath150 . plotted densities are kernel density estimates based on 10000 independent stationary samples generated using gillespie s algorithm @xcite . ]	the phenomena of stochasticity in biochemical processes have been intriguing life scientists for the past few decades . we now know that living cells take advantage of stochasticity in some cases and counteract stochastic effects in others . the source of intrinsic stochasticity in biomolecular systems are random timings of individual reactions , which cumulatively drive the variability in outputs of such systems . despite the acknowledged relevance of stochasticity in the functioning of living cells no rigorous method have been proposed to precisely identify sources of variability . in this paper we propose a novel methodology that allows us to calculate contributions of individual reactions into the variability of a system s output . we demonstrate that some reactions have dramatically different effects on noise than others . surprisingly , in the class of open conversion systems that serve as an approximate model of signal transduction , the degradation of an output contributes half of the total noise . we also demonstrate the importance of degradation in other relevant systems and propose a degradation feedback control mechanism that has the capability of an effective noise suppression . application of our method to some well studied biochemical systems such as : gene expression , michaelis - menten enzyme kinetics , and the p53 system indicates that our methodology reveals an unprecedented insight into the origins of variability in biochemical systems . for many systems an analytical decomposition is not available ; therefore the method has been implemented as a matlab package and is available from the authors upon request . + living cells need to constantly adapt to their changing environment . they achieve this through finely honed decision making and stress response machineries that regulate and orchestrate the physiological adaptation to new conditions . in all studied genomes a large number of proteins have as their primary function the transfer and processing of such information . such proteins are linked through a host of different mechanisms into biochemical circuits that perform a variety of information processing tasks including storage , amplification , integration of and marshalling the response to environmental and physiological signals @xcite . the functioning of these information processing networks depends on thermal or probabilistic encounters between molecules , resulting in a distortion of a transferred information that is best understood as noise . each reaction in the information processing machinery leads to an inevitable loss of information @xcite . therefore , cell functions do not only rely on the necessity to make good " decisions , but also on appropriate ways to cope with the uncertainties arising from the noisy " signal transmission . to deal with the latter type of difficulty , we believe , evolution equipped cells with reliable signal transduction systems by using less noisy reactions or reaction configurations , where needed @xcite . the question , however , which reactions , molecular species or parts of a network contribute most of the variability of a system or are responsible for most of the information loss has not gained much attention , except for some models of gene expression @xcite . origins of stochasticity in biochemical systems have been discussed and argued over at length , but a unified and robust mathematical framework to study this problem has been lacking . here we present a novel and general method to calculate contributions of individual reactions to the total variability of outputs of biochemical systems . this enables us to identify the origins of cell - to - cell variability in dynamical biochemical systems . we derive a modified fluctuation - dissipation theorem which enables us to determine how much of the total variance results from each of the system s reactions . we then obtain some unexpected but general rules governing the noise characteristics of biochemical systems . in particular , we shall show that in an arbitrary system with a sufficiently large number of molecules , degradation of the output ( e.g. a reporter protein or a transcription factor ) contributes to the total variance of the system half the of the output s mean ; for the important class of open conversion systems exactly half of the variance derives from the degradation step of the output signal . these results demonstrate that some reactions may be responsible for higher information loss than others ; but our results also reveal that cells have the option of optimising biochemical network structures in order to avoid the most noisy reactions if necessary . based on these results we propose a mechanism of controlled protein degradation based on a positive feedback that allows an arbitrary noise reduction resulting from the protein degradation . below we first introduce the general framework for modelling chemical reactions and derive a new method to decompose the noise in a biochemical system into contributions from different individual reactions . furthermore , two general properties governing noise are presented . finally , we use biological examples of signal transduction systems to provide novel insights into the origins of variability . in particular , we decompose the variance of the outputs of linear transduction cascades and michaelis - menten enzyme kinetics . in addition , for the oscillatory p53 system we show that stochasticity in p53 protein degradation is responsible for the variability in the oscillations periodicity .
if elongated active , _ i.e. _ self - propelled , objects interact by pushing each other in a dissipative medium or substrate , the objects will tend to locally align @xcite as shown in fig . [ fig - model ] . since these object are self - propelled , once aligned , they will move together in the same direction for a given time . this simple effective alignment mechanism among active objects lead to interesting collective effects @xcite , as the formation of moving cluster as illustrated in fig . [ fig : myxo ] with experiments of myxobacteria . there is a broad range of real - world active systems that consist of active elongated object where this mechanism is at work : gliding bacteria @xcite , dried self - propelled rods @xcite , chemically - driven rods @xcite , and it has been recently argued that also neglecting hydrodynamic effects over steric effects in swimming bacteria @xcite and motility assays @xcite . [ b ] here , we review the large - scale properties of collections of active brownian elongated objects , in particular rods , moving in a dissipative medium / substrate . we address the problem by presenting three different models of decreasing complexity , which we refer to as model i , ii , and iii , respectively . model i is the full physical active brownian rod model introduced in @xcite where particles exhibit a well - defined shape , possess an active force acting along the longest axis of the rod , and interact via volume exclusion effects by pushing each other . in model i there exists a coupling of local density , orientational order , and speed , known to lead to density instabilities and collective phenomena in other active models @xcite . more importantly , in model i active stresses coexist with an an effective local alignment mechanism . due to the combined effect of these two elements , model i displays exciting new physics unseen in other active models , such as the formation of highly dynamical aggregates that constantly eject giant polar cluster containing thousands of active rods @xcite . if we remove from model i the active force , we end up with an equilibrium system ( if noise terms have been adequately chosen ) . with the elongated rods interacting through steric repulsive forces , onsager s argument on thin rods applies @xcite and the system exhibits local nematic order above a given critical density . we discuss the possibility of local nematic order and quasi - long - ranged order ( qlro ) in two - dimensions by introducing model ii , which is a simplified version of model i without anactive force . model ii allows us to argue that the symmetry of the interaction potential in model i is nematic . we introduce model iii to show that the peculiar large - scale properties displayed by model i do not result , as has been argued , from the combined effect of self - propulsion and an effective nematic velocity alignment mechanism . model iii is an active version of model ii and a simplified version of model i without volume exclusion interactions . let us recall that hat most flocking models assume a velocity alignment mechanism whose symmetry is ferromagnetic @xcite . from model iii , we learn that active particles with a nematic velocity alignment exhibit macroscopic nematic structures @xcite , which are not present in model i , which displays polar order at short scales and highly dynamical , highly fluctuating phase - separated phase @xcite . comparing model i , ii , and iii we disentangle the role of activity and interactions and identify the contribution of every modeling element . in particular , we find that by ignoring volume exclusion effects , local and global nematic order seems to be possible , while by adding steric interactions the system is dominated by the interplay of active stresses and local alignment , which prevents the formation of orientational order at large scales in two - dimensions . the physics of active elongated objects , despite its ubiquity in experimental systems , remains still poorly understood . here , we present a detailed state of the art of the unique collective properties of this fascinating physical system . let us consider @xmath0 active brownian rods ( abr ) moving in a two - dimensional space of linear size @xmath1 with periodic boundary conditions . each rod is driven by an active stress / force @xmath2 that is applied along the long axis of the particle . interactions among rods are modeled through a repulsive potential , which we denote , for the @xmath3-th particle , by @xmath4 . the substrate where the rods move acts as a momentum sink . there are three friction drag coefficients , @xmath5 , @xmath6 , and @xmath7 , which correspond to the drags experienced by the rods as the rod moves along the long axis , perpendicular to it , or as it rotates , respectively . in the over - damped limit , the equations of motion of the @xmath3-th rod are given , as in @xcite , by : @xmath8 \\ \label{eq : evol_theta } \dot{\theta}_i & = & \frac{1}{\zeta_{\theta } } \left [ - \frac{\partial u_i}{\partial \theta_i } + \xi_{i}(t ) \right ] \ , , \end{aligned}\ ] ] where the dot denotes a temporal derivative , @xmath9 corresponds to the position of the center of mass and @xmath10 the orientation of the long axis of the rod . the term @xmath4 models the interactions with other rods and @xmath2 is the self - propelling force . the symbol @xmath11 in eq . ( [ eq : evol_x ] ) is the mobility tensor defined as @xmath12 , with @xmath13 and @xmath14 such that @xmath15 . drag friction coefficients can be computed assuming that the rods are surrounded by a liquid @xcite , move on a dried surface as in experiments with granular rods @xcite , or by assuming that eqs . ( [ eq : evol_x ] ) and ( [ eq : evol_theta ] ) represent gliding bacteria , in which case the friction coefficients are arguably connected to presence of the so - called focal adhesion points @xcite . in short , the friction coefficients depend on the specific rod system eqs . ( [ eq : evol_x ] ) and ( [ eq : evol_theta ] ) are supposed to model . here , we use @xmath16 , @xmath17 , @xmath18 , and @xmath19 . ( [ eq : evol_theta ] ) represents the temporal evolution of the orientation of the rod , which is assumed to result from the torque generated by the interactions with the others rods , modeled by @xmath4 , and thus we express this torque as @xmath20 . note that eqs . ( [ eq : evol_x ] ) and ( [ eq : evol_theta ] ) are subject to fluctuations through the terms @xmath21 and @xmath22 , which correspond to delta - correlated vectorial and scalar noise , respectively . for simplicity , we neglect @xmath21 and specify in eq . ( [ eq : evol_theta ] ) @xmath23 and @xmath24 , with @xmath25 ( for more details see @xcite ) . the interactions among the rods are modeled by a soft - core potential that penalizes particle overlapping . for the @xmath3-th rod , the potential takes the form : @xmath26 , where @xmath27 denotes the repulsive potential interaction between the @xmath3-th and @xmath28-th rod , both of length @xmath29 and width @xmath30 , such that @xmath31 . the rods ( in two - dimensions ) can be represented as soft rectangles or spherocylinders as in @xcite or equivalently as straight chains of @xmath32 disks of diameter @xmath30 , as implemented in @xcite , whose centers are separated a given distance @xmath33 ; here @xmath34 . we notice that results obtained with active brownian rods represented by disk - chains are qualitatively identical to those produced with the original abr model introduced in @xcite if and only if @xmath35 , _ i.e. _ as long as the border of the rods is smooth . using this implementation , @xmath27 can be expressed as @xmath36 , where @xmath37 is the potential between disk @xmath38 of the @xmath3-th rod and disk @xmath39 of the @xmath28-th rod , which here we assume to be given by a harmonic repulsive potential : @xmath40 , for @xmath41 and zero otherwise , where @xmath42 is the distance between the centers of the disks and @xmath43 . if we remove the active force and make @xmath44 in eq . ( [ eq : evol_x ] ) , while requiring that @xmath21 and @xmath22 allow us to define a temperature , the resulting system is the equilibrium system of passive rods studied by onsager @xcite to describe , in a simple way , rod - like liquid crystals in two - dimensions . let us recall that according to the so - called mermin - wagner theorem @xcite , this equilibrium system in two - dimensions can not exhibit long - ranged order ( lro ) . furthermore , it is believed that this system exhibits ( in two - dimensions ) a defect - mediated phase transition analogous to the berezinskii - kosterlitz - thouless ( bkt ) transition @xcite . this means that above a given critical point , the system exhibits quasi - long - ranged order ( qlro ) , implying that by increasing the system size , while keeping all intensive parameter fixed , the order parameter ( associated to orientational order ) decreases algebraically . at the mean - field level , _ i.e. _ by neglecting fluctuations , the equilibrium model ii displays an isotropic - nematic transition , which can be understood by considering two rods separated ( their center of mass ) a given distance @xmath45 . to simplify the reasoning , let us neglect the dynamics of the center of mass of the rods and consider them fixed . in this scenario , it is clear that if the interaction between the rods is modeled by an interaction potential that penalizes the overlapping of the rods ( as in model i ) , by making the orientation of the rods parallel to each other , we minimize the interaction potential ( moreover , particles may not even interact ) . in addition , since the interaction potential does not distinguish head and tail of the particles , it should be such that we obtain the same by flipping the orientation of one of the particles by @xmath46 . in short , the interaction potential has to be a function of overlapping area of the two rods , which is , as observed by onsager @xcite , proportional to @xmath47 , where @xmath3 and @xmath28 refer to the label of the particles we are looking at . since @xmath48 , for simplicity we express the potential directly as : @xmath49 where the sum runs over all @xmath28 particles such that @xmath50 , with @xmath51 defining the interaction range , which we could assume to be @xmath52 , and @xmath53 a constant . this defines a simplified dynamics where the equation of motion of the @xmath3-th particle is given by : @xmath54 where @xmath55 and where we have simplified further the model by assuming that @xmath56 is an isotropic delta - correlated vectorial noise . it is important to stress that eqs . ( [ eq : evol_x_model_ii ] ) and ( [ eq : evol_theta_model_ii ] ) do not correspond ( exactly ) to eqs . ( [ eq : evol_x ] ) and ( [ eq : evol_theta ] ) with @xmath44 , but to a simplified version of model i for @xmath44 that shares the same symmetry . the fundamental difference between model i with @xmath44 and model ii is the absence of a term proportional to @xmath57 in eq . ( [ eq : evol_x_model_ii ] ) . despite of this difference , we expect that on large scales eqs . ( [ eq : evol_x_model_ii ] ) and ( [ eq : evol_theta_model_ii ] ) to display a behavior analogous to that of model i with @xmath44 . we know that a system evolving according to eqs . ( [ eq : evol_x_model_ii ] ) and ( [ eq : evol_theta_model_ii ] ) do not exhibit lro , and that the observed transition is a defect - mediated transition with order being qlro as expected for model i with @xmath44 . in addition , and very important for us , at the mean - field level eqs . ( [ eq : evol_x_model_ii ] ) and ( [ eq : evol_theta_model_ii ] ) exhibit an isotropic - nematic transition , characterized by the absence of local and global polar order . we will use model ii as a reference model . the message is that in the absence of activity ( meaning for @xmath44 ) and in two - dimesions , we do not expect the system to exhibit lro but rather qlro . furthermore , from model ii we could imagine that if activity induces order at large scales ( _ i.e. _ lro ) such order should be nematic . we will see that this is the case in model iii , which we introduce next , but counterintuitive not for model i. the third and final model we introduce here is an active version of model ii @xcite . the equations of motion of the @xmath3-th particle are given by : @xmath58 where @xmath59 . despite the fact that the interaction potentials in model i and model iii are short - ranged and share the same symmetry , there is a fundamental difference between both models : the equation for @xmath60 in model iii does not possess a term @xmath57 . the absence of this term implies that in model iii there is no volume exclusion , and thus particles are point - like . as it will become clear below , this leads to the absence of active stresses in model iii , which are present in model i. due to this difference , which may appear at first glance minor , model i and iii exhibit qualitatively different large - scale properties . another difference between model i and iii is the absence of a noise terms in eq . ( [ eq : evol_x_model_iii ] ) , while such terms are present in eq . ( [ eq : evol_x ] ) . such noise terms do not have an impact on the ( qualitative ) large - scale properties of the systems . finally , the difference between model ii and iii is given by the so - called active term , @xmath61 , in eq . ( [ eq : evol_x_model_iii ] ) , replacing the noise term in eq . ( [ eq : evol_x_model_ii ] ) . note that in model iii the equations for @xmath60 and @xmath62 are thus coupled through @xmath63 , while this does not occur in model ii . we will see below that the presence of the active term in model iii has a strong impact on the large - scale properties , which turn to be qualitatively different from those of model ii . we characterize the resulting macroscopic patterns by their level of orientational order through @xmath64 where the averages run over the number of particles and time . polar order corresponds to @xmath65 and nematic order to @xmath66 . the @xmath67 and @xmath68 correspond to global order parameters , _ i.e. _ obtained by performing an average over the entire system . local order parameters can be also defined and should be noticed that a system can exhibit no global order , while still displaying local order . the distribution of particles in space as well as information on the interaction structure can be studied by looking at the cluster size distribution ( csd ) . we use the ( weighted steady state ) csd @xmath69 defined as the time average ( after an initial transient ) of the instantaneous csd : @xmath70 where @xmath71 is the number of clusters of mass @xmath72 present in the system at time @xmath73 . notice that the normalization of this distribution is ensured since @xmath74 ( with @xmath0 the number of particles in the system as defined above ) . clusters are collections of interconnected particles , where any two particles are considered as ( directly ) connected if they interact to each other . for simplicity and to speed up the numerics , the criterion is usually relaxed and particles are considered as connected if they are separated a given distance which is typically of the order of the interaction radius ( the details vary from model to model ) . the functional form of @xmath69 indicates whether the system is `` homogeneous '' , when the csd displays an exponential tail , or phase - separated , when the csd exhibits a peak at large cluster sizes . particles can also self - organized in a kind of `` critical '' state where cluster of all sizes can be observed . this corresponds to a heavy - tailed csd . a detailed analysis of the csd of these models can be found in @xcite . an alternative way to monitor phase separation is by studying the ratio between average cluster size @xmath75 and system size @xmath0 , where @xmath76 , with the advantage of dealing simply with an scalar , @xmath77 . ( a - c ) stationary state typical snapshots at different noise values @xmath78 ( @xmath79 @xmath80 and @xmath81 ) the arrows indicate particles direction of motion ; only a fraction of total particles are shown for clarity reasons . ( a ) @xmath82 , ( b ) @xmath83 , ( c ) @xmath84 , ( d ) @xmath85 ( for clarity reasons particles are represented only by points ) , ( e ) @xmath86 . roman numbers refers to the four phases observed in finite size systems . figure from @xcite.,width=604 ] in order to study the thermodynamical behavior of the models , we analyze the scaling of @xmath87 and @xmath88 with the system size @xmath0 , while keeping all intensive parameter fixed . if @xmath89 , with @xmath90 , the system exhibits lro . qlro , on the other hand , corresponds to @xmath91 , where @xmath92 with @xmath93 in two - dimensions for the xy model . it is worth recalling that for a fully disorder system , we expect @xmath94 , which means that observing an algebraic decay does not necessarily imply qlro . finally , @xmath95 for homogeneous system , indicating that there is a finite value of @xmath75 with @xmath0 . a departure from this behavior is an indication of either phase - separation or the organization of particles in a critical " csd . here , we review results for model i and iii . the large - scale properties of model ii have been already summarized in the model section . as mentioned before , model ii , _ i.e. _ the brownian rod model , is used as a reference model in order to understand the role of the active term . the comparison between model i and iii will allow us to disentangle the role of volume - exclusion and nematic alignment . we start in inverse order , that is , we first revise results for model iii , since the model is simpler and results are , somehow , more intuitive . nematic order parameter @xmath68 as function of the noise amplitude @xmath78 for a density @xmath79 and a linear system size @xmath81 . the figure illustrates the four different regimes observed for a given density and system size . , width=377 ] the numerical results that we are going to review @xcite here correspond to an implementation of model iii _ la vicsek _ , _ i.e. _ using a numerical scheme as in @xcite . it is important to stress that direct numerical interaction of the continuum - time eqs . ( [ eq : evol_x_model_iii ] ) and ( [ eq : evol_theta_model_iii ] ) lead to the same qualitative behavior as the one reported in @xcite ( data not shown ) . results where generated by using the following specific scheme ) is replaced with @xmath96 , which is the actual implementation used in @xcite . ] : @xmath97 \mathbf{v}(\theta_i^t ) \right ] + \gamma \psi_{j}^{t } \ , , \end{aligned}\ ] ] where @xmath98 $ ] refers to the angle associated to the vector @xmath99 if this is expressed in polar coordinates , the sum is taken over all particles @xmath100 within distance @xmath101 of @xmath28 ( including @xmath28 itself ) and @xmath102 is a white noise uniformly distributed in @xmath103 $ ] . the term @xmath78 is the noise amplitude , which means that @xmath104 . simulations were performed with @xmath105 and particle density @xmath106 . a fluctuating ordered phase exists at low noise ( or high density if @xmath107 is used as control parameter ) , and an order / disorder transition line lies in the main parameter plane @xmath108 . varying @xmath78 , we observe in a square domain of linear size @xmath81 , as shown in fig . [ fig:2 ] and [ fig:3 ] , that only _ orientational order arises . despite the polar nature of the particles , the polar order parameter remains near zero for all noise strengths ( fig . [ fig:3 ] ) . both the ordered and the disordered phases are divided in two by the spontaneous segregation , at intermediate @xmath78 values , of the system into high - density , ordered regions and sparse , disordered ones ( fig . [ fig:2]b - d ) . a total of four phases thus seems to exist , labeled i to iv by increasing noise strength hereafter . phases i and ii are nematically ordered , phases iii and iv are disordered . phase i , present at the lowest @xmath78 values , is ordered and spatially homogeneous ( fig . [ fig:2]a ) . its nematic order , which arises quickly from any initial condition , is the superposition , at any time , of two polarly aligned opposite subpopulations of statistically equal size ( fig . [ fig:4]a ) . these subpopulations constantly exchange particles , those which `` turn around '' , an event that occurs on exponentially - distributed times @xmath109 ( fig . [ fig:4]b ) . therefore it is natural to define a typical persistance time @xmath110 and its corresponding persistence length @xmath111 . they are found to decrease as the noise amplitude is increased . while the breaking of a continuous symmetry in a two dimensional equilibrium system can only lead to quasi - long - range - order ( qlro ) as discussed for model ii , in model iii the numerical evidence points towards lro . let us recall that what has been called as active nematics " @xcite , which is equivalent to model ii , where we have to explicitly require the use of a local diffusion tensor depending on @xmath112 , displays qlro . here nematic _ order parameter @xmath68 , for the tested system sizes , decays slower than a power law . a good fit of this decay is given by an algebraic approach to a constant asymptotic value @xmath113 ( fig . [ fig:4]c ) . this phase is characterized by the presence of giant number fluctuations , fig . [ fig:4]c ( for details see @xcite ) . in short , the numerical data suggests the existence of true long - range nematic order . ordered homogeneous phase ( @xmath114 , @xmath115 , @xmath116 ) . ( a ) moving direction ( given by @xmath112 ) distribution in a system of linear system size @xmath81 . ( b ) distribution of particles transition times @xmath109 between the two peaks of panel ( a ) , i.e. particles which `` turn around '' their orientation by @xmath117 . ( c ) nematic order parameter @xmath68 as a function of system size @xmath1 . the vertical red dashed line marks the persistence length @xmath118 . in the inset : a constant value @xmath119 has been subtracted from the nematic order parameter to highlight power low decay to a nonzero constant . the red dashed line decays as @xmath120 . ( d ) number fluctuations @xmath121 as a function of average particle number @xmath122 in a system with @xmath123 . the dashed line marks the power law growth @xmath124 . figure taken from @xcite.,width=302 ] phase ii differs from phase i by the presence , in the steady - state , of a low - density disordered region . in large - enough systems , a narrow , low density channel emerges ( fig . [ fig:2]b ) when increasing @xmath78 . it becomes wider and wider at larger @xmath78 values , so that one can then speak of a high - density nematically ordered band amidst a disordered sea ( fig . [ fig:2]c ) . inside the band , nematic order similar to phase i is found , with particles traveling in roughly equal number along each direction , and turning around or leaving the band at exponentially - distributed times , albeit with a shorter typical persistence time . giant number fluctuations similar to those reported in fig . [ fig:4]d occur . in rectangular domains , the band is typically oriented along the small dimension of the box . bands along the longer dimension can be artificially created , but in sufficiently long and narrow boxes they become wavy under what looks like a finite - wavelength instability and are eventually destroyed , leaving a thicker band along the small dimension . for large - enough ( square ) domains , the band possesses a well - defined profile with sharper and sharper edges as @xmath1 increases . in phase iii , spontaneous segregation into bands still occurs ( for large - enough domains ) , but these now thin structures constantly bend and elongate , getting thinner and vanishing , or merging with others . they never reach a static steady state shape . correspondingly , the nematic order parameter fluctuates on very large time scales and @xmath68 decrease like @xmath125 providing a clear indication that the order in these thin highly dynamical bands self - averages , making phase iii a disordered phase albeit one with huge correlation lengths and times . finally , phase iv , observed for the highest noise strengths , is disordered on small length- and time - scales , without any particular emerging structure ( fig . [ fig:2]e ) . here , we study the long - scale properties of model i by direct integration of eqs . ( [ eq : evol_x ] ) and ( [ eq : evol_theta ] ) . we use as control parameter the aspect ratio @xmath126 , but we could use instead the packing fraction @xmath127 or the noise amplitude @xmath128 . it is important to clarify that when we vary the aspect ratio @xmath126 , we keep fixed the particle area @xmath129 . for a fixed system size @xmath0 , we observe @xmath67 , @xmath68 , and @xmath130 take off above a critical @xmath131 value as shown in fig . [ fig : pt ] . we notice that @xmath131 decreases when @xmath132 is increased . below @xmath131 , we observe a gas phase characterized by the absence of orientational order and an exponential cluster size distribution such that @xmath133 . for @xmath134 the system undergoes a symmetry breaking as observed previously in @xcite . the emerging order is , however , polar as evidenced by the behavior of @xmath67 . we recall that in the presence of polar order , @xmath68 is slaved to @xmath67 . the behavior of @xmath130 , right panel in fig . [ fig : pt ] , indicates that the system starts to spontaneously self - segregate for @xmath134 . here , we find that the onsets of orientational order and phase separation coincide and share the same critical point , as predicted using a simple kinetic model for the clustering process @xcite : due to the effective velocity alignment large polar clusters emerge , which in turn lead to macroscopic polar order . notice that in an equilibrium system of ( hard ) rods ( i.e. @xmath44 ) for @xmath135 and @xmath136 , according to de las heras et al . @xcite , we should observe only an homogeneous disordered phase for this range of parameters . this indicates that the observed phase transition requires @xmath137 , i.e. the active motion of the rods . we performed a finite size study , by increasing simultaneously @xmath0 and @xmath1 while keeping the packing fraction @xmath132 and all other parameters constant . [ fig : fss ] shows the scaling of the ( global ) polar order parameter @xmath67 and average cluster size with respect to system size @xmath130 for aspect ratio @xmath138 and several packing fractions @xmath132 . at low @xmath132 values , i.e. for @xmath139 , we are in the situation @xmath140 ( we recall that the critical @xmath131 value depends on @xmath132 ) . for @xmath140 the system is not phase - separated and does not exhibit orientational order . as expected , the scaling of @xmath67 with @xmath0 shows that @xmath141 , with @xmath142 , which means that the system is fully disordered . in addition , we observe that @xmath143 , with @xmath144 , which indicates that there is a well - defined characteristic cluster size for the system ( that is independent of @xmath0 ) and consequently the system is not phase - separated . at large @xmath132 values , i.e. for @xmath145 and @xmath146 , we observe phase separation and ( global ) polar order for small finite systems ( @xmath147 ) . the finite size study shows that , for @xmath146 , @xmath148 does not decrease ( asymptotically ) with @xmath0 . moreover , for large values of @xmath0 , we even observe an increase . this indicates that @xmath75 is at least proportional to @xmath0 , which implies that the system is phase - separated in the thermodynamical limit as well as in finite systems . at the level of the orientational order parameter @xmath67 , we observe an abrupt change in scaling of @xmath67 with @xmath0 . for @xmath149 ( e.g. , @xmath150 ) for @xmath151 and @xmath152 , @xmath67 is high _ i.e. _ the system displays global polar order . on the other hand , for @xmath153 , while the system remains phase - separated , @xmath67 sharply decreases with @xmath0 . in short , the finite size study reveals that , although phase separation does take place in the thermodynamical limit as well as in finite systems , the phase transition to an orientationally ordered phase is observed only for small finite systems . global order patterns are not present in the thermodynamical limit , with the phase transition occurring , in this limit , between a disordered gas and a phase - separated state with no global orientational order . the reason for observing non - vanishing global order in small systems is the presence of few giant polar clusters as illustrated by the simulation snapshot in fig . [ fig : pt ] . such giant polar clusters can become so big and elongated that they can can even percolate the system , as shown in panel i ) of fig . [ fig : band ] . we refer to such polar percolating structures as bands . inside bands , rods are densely packed , point into the same direction , and exhibit positional order . notice that these bands are distinct from the bands observed in the vicsek model , which are elongated in the direction orthogonal to the moving direction of the particles @xcite . more importantly , the observed polar bands are also fundamentally different from those observed in model iii , where we saw that the point - like self - propelled particles form nematic bands , inside which 50% of the particles move in one direction and 50% in the opposite one . more importantly , our finite size study indicates that the polar patterns observed in abr are a finite size effect that disappear for large enough systems . in short , several of the phases reported for @xmath135 in previous abr works @xcite such as the so - called swarming phase and the bio - turbulence phase vanish in the thermodynamical limit . the abrupt change in scaling of @xmath67 with @xmath0 in fig . [ fig : fss ] suggests that above the crossover system size @xmath154 the polar structures are no longer stable . arguably , the decay in @xmath67 with @xmath0 is due to the fact that rods inside polar clusters are densely packed and hold fixed positions , not being able to exchange neighbors in contrast to other active systems @xcite . [ fig : band ] shows that @xmath155 , _ i.e. _ the total time @xmath156 the system spends in the aggregate phase with respect to the total simulation time @xmath157 . we observe that @xmath156 increases with @xmath0 , in such a way that @xmath158as @xmath159 . this means that the probability of observing the system in an aggregate phase also increases with @xmath0 . for small system sizes @xmath160 we observe moving clusters and bands . large polar structures such as bands form , remain in the system for quite some time , and then quickly break and reform , typically adopting a new orientation . as @xmath161 , bands survive for relatively short periods of time , and quickly bend and break . interestingly , at such large system sizes other macroscopic structures start to frequently emerge . these new macroscopic structures which we refer to as _ aggregates _ are formed by polar clusters of rods that exert stresses on each other and exhibit vanishing polar order , see panel ii ) of fig . [ fig : band ] . in summary , for @xmath162 , the system continuously transitions between highly ordered phases e.g. phases with either a few giant polar clusters or a band and aggregates , as illustrated in fig . [ fig : band ] . for system size larger than @xmath154 bands and polar phases disappear in the thermodynamical limit , while the aggregate phase survives . the transitions between aggregates and bands ( or highly ordered phases ) for @xmath162 results from the competition between elastic energy and the impossibility of the system to sustain long - range polar order . for not too large system sizes , _ i.e. _ for @xmath163 , the shape of the aggregates is roughly circular ( fig . [ fig : band ] , panel ii ) ) and at the center of the aggregate we find one single topological defect : i.e. at the mesoscale , at the center of the aggregate we can not define an average orientation for the rods . due to the active forces , at the center of the aggregate rods are strongly compressed , which implies that the potentials @xmath4 adopts high values . this implies that when one of these aggregates is formed , the total elastic energy of the system @xmath164 increases . on the contrary , in large polar structures such as bands , rods are roughly parallel to each other and therefore are much less compressed by their neighbors , and the total elastic energy is low . the dynamics at @xmath162 can be summarized as follows . large polar clusters form and eventually a band emerges , but since the system is too big for the band to remain stable , at some point the band breaks . the collapse of the band gives rise to the formation of new giant polar clusters which eventually collide head on leading to a large aggregate : a process reminiscent of a traffic jam . the formation of the aggregate leads to a sharp increase of the total elastic energy . let us recall that forces and torques act in such a way that they tend to minimize @xmath4 . in short , the system relaxes by destroying the new formed aggregate , which give rise to the formation of new polar clusters and the cycle starts again . in larger system sizes , i.e. for @xmath165 , aggregates are more complex . inside aggregates , the competition between active forces and local polar alignment leads to new physics . this is particularly evident for very large system sizes , i.e. for @xmath166 , that is when aggregates are big enough to exhibit multiple topological defects of the local orientation of the rods . let us recall that aggregates are formed by polar clusters of rods that are trapped inside these structures . topological defects are areas where , at the mesoscale , as mentioned above , we can not define an average orientation of the rods as illustrated in inset of fig . [ fig : aggregate ] ( areas where the arrows meet ) . in such areas , due to the active forces , rods are strongly compressed by the active push of all surrounding abr ( see inset in fig . [ fig : aggregate ] ) . since the compression is due to the presence of active forces , we refer to this phenomenon as _ active stresses_. for @xmath166 , we observe the emergence of multiple defects that lead to an increase of the elastic energy and the build - up of stresses . notice that more topological defects imply larger values of the elastic energy . there are two clear consequence of the presence of multiple topological defects . on the one hand , aggregates are no longer roundish but rather irregular as illustrated in fig . [ fig : aggregate ] . on the other hand , now the system can relax the elastic energy by reducing the number of topological defects . notice that for @xmath162 , aggregates are relatively small and exhibit one topological defect , and thus , the only way to eliminate the topological defects is by destroying the aggregate . for @xmath166 , given the presence of multiple topological defects , eliminating one topological defect does not require to eliminate the aggregate . as a matter of fact , for very large system sizes , the interplay between topological defects and active stresses lead to large fluctuations of the aggregate boundary and aggregate mass ( i.e. aggregate size ) as indicated in fig . [ fig : aggregate ] . the most distinctive feature of the observed phenomenon is the large fluctuations experienced by the aggregate mass correspond to ejections of remarkably large macroscopic polar clusters from the aggregate , that can be as large as 10% of the system size ( i.e. involving more than @xmath167 rods ) , fig . [ fig : aggregate ] . by this process , i.e. the ejection of large polar clusters , the aggregate manages to decrease its elastic energy . the ejected polar clusters typically dissolve while moving through the gas phase outside the aggregate , leading to a sudden increase of the gas density , top panel in fig . [ fig : aggregate ] . this results in a higher absorption rate of abr by the aggregate that starts again to increase its mass . the comparison of models i , ii , and iii is insightful . model ii and iii share the same interaction potential , which we refer here to as alignment mechanism : eq . ( [ eq : evol_theta_model_ii ] ) is identical to ( [ eq : evol_theta_model_iii ] ) . despite of this , on large scales , model ii and model iii display qualitatively different features . while model ii exhibits for low noise values qlro , as expected for an equilibrium system with continuum symmetry in two - dimensions , model iii at low noise displays nematic lro . the active term in eq . ( [ eq : evol_theta_model_iii ] ) , @xmath168 , makes possible the emergence of stable nematic bands and an homogeneous nematically ordered phase ( phase i ) . in summary , in model ii we observe local nematic order and qlro at large scales , while in model iii we find again local nematic order and lro at large scales . importantly , at the meso and macroscopic sale model ii and iii do not exhibit polar order . in model i , on the other hand , nematic order is never observed ( at least up to packing fractions @xmath169 ) . the difference between model i and model iii is due to the presence of the repulsive term @xmath57 in the equation for @xmath60 . we have learned that the addition of volume exclusion interactions in the evolution of the position of the particle has a dramatic effect on the large - scale properties of the system . at small system sizes and for large enough aspect ratios ( or equivalently packing fractions ) particles self - organize into large moving polar clusters and the system displays global polar order . moreover , the observed bands are polar instead of being nematic as the ones observed in model iii . however , the performed finite size study reveals that polar order ( or any orientational order ) vanishes in the thermodynamical limit ( at least up to packing fractions @xmath169 ) . thus , from model i we learn that volume exclusion effects induce local polar and absence of orientational order at large scale . despite the lack of orientation order , model i undergoes a genuine phase - separation . the phase - separation process , as well as the phase - separated phase are qualitatively different from the ones observed in model iii , where the system spontaneous phase separates in the form of nematically ordered , high - density bands ( referred above as phase ii and iii ) . in model i , phase separation starts with the formation if giant polar clusters @xcite that jam to form what we have called aggregates . the most distinctive feature of these aggregates is the constant ejection from the aggregate of thousands of particles in the form of densely packed and polarly ordered clusters , which leads to large fluctuation of the aggregate size and its boundary . this occurs due to the combined effect of an effective alignment mechanism ( also present in model iii ) and the active pushing ( or stresses ) acting among the particles that requires not only an active force but also a term @xmath57 in the evolution of the position of the particles . the combination of these elements ( active stresses and alignment ) is not present in other active system , making the physics of active brownian rods unique . for instance , self - propelled disks ( spd ) exhibit active stresses but no alignment among the spd @xcite . thus , phase separation in spd resembles a classical coarsening process @xcite which can be described by an effective cahn - hilliard equation @xcite . finally , it important to stress that the phase separation in model i is different from the one observed in active system with an alignment mechanism and a density - dependent speed @xcite . here , while alignment among particles is present , there are no active stresses . in summary , we have characterized the large - scale properties of abr for packing fractions smaller than or equal to @xmath169 . we have shown that large - scale ( orientational ) order patterns can not exist in the thermodynamic limit for ( physical ) abr ( _ i.e. _ in model i ) due to combined effect of active stresses and alignment , which leads to exciting new physics . in particular , we have seen that the interplay of these two elements ( active stresses and alignment ) gives rise to a novel highly fluctuating phase - separated phase characterized by the ejection of giant polar clusters . we point out that the presented evidence can not be used to preclude the emergence of orientational order at higher packing fractions . let us recall that even for brownian rods we expect for large enough packing fractions some kind of isotropic - nematic transition . in simulations with abr at high packing a transition to laning has been reported @xcite . the observed laning phase may remain even in the thermodynamic limit , though its existence should be confirmed by a careful finite size study . finally , the model presented in @xcite for penetrable abr suggested a possible crossover between the large - scale properties exhibited by model i and iii . while this scenario can not be _ a priori _ excluded in @xcite , a finite size study may reveal some of the reported phases vanish in the thermodynamical limit .	lecture notes for the summer school `` microswimmers from single particle motion to collective behaviour '' at forschungszentrum jlich , 2015 .
the magnetism of fermi gases has always received considerable attention in solid - state physics , such as localized and itinerant electrons . while the magnetic properties of bose gases has been less studied . but since the realization of bose - einstein condensation ( bec ) in ultracold atomic gases @xcite , more interests have been cast to this system . the bose gases plays an important role in understanding some exotic quantum phenomena , such as superconductivity and superfluid . the ideal charged bosons were used originally to describe the superconductivity . it has been shown by schafroth @xcite , blatt and butler @xcite that an ideal gas of charged bosons exhibits the essential equilibrium features of superconductor . although the bardeen - cooper - schrieffer ( bcs ) theory @xcite explained the microscopic nature of conventional superconductivity , the charged bose gas exhibits strong diamagnetism at low temperature , which can be attributed to meissner effect . in recent years , the normal - state diamagnetism of high - temperature cuprate superconductors has been explained by real - space charged bosons @xcite . this also recasts new research interest in charged bose gases . experimentally , since the realization of _ spinor _ bec in optical traps @xcite the magnetic properties of _ spinor _ bose gases has received considerable attention . moreover , an ultracold plasma can be created by photoionization of laser - cooled neutral atoms @xcite . the temperatures of electrons and ions can reach as low as 100 mk and 10 @xmath5k , respectively . the ions can be regarded as charged bosons if their spins are integers . the lande - factor for different magnetic ions could also be different . it is known that paramagnetism is from the spin degree of freedom of particles . while charged spinless bose gases can exhibit strong diamagnetism , similar to meissner effect , which comes from the orbital motion of charge degree of freedom in magnetic field . theoretically , both the paramagnetism @xcite in neutral spin-1 bose gases and the diamagnetism of the charged spinless bose gases @xcite have been studied . moreover , we @xcite have discussed the competition of paramagnetism and diamagnetism in charged spin-1 bose gases in external magnetic field , using the lande - factor @xmath1 to evaluate the strength of paramagnetic ( pm ) effect . it is shown that the gas exhibits a shift from diamagnetism to paramagnetism as @xmath1 increases . the ferromagnetism and superconductivity are not compatible in conventional physical models . the meissner - ochsenfeld effect shows the conventional superconductor cancels all magnetic field inside when the temperature below the superconducting transition temperature , which means they become perfectly diamagnetic . the discovery of several ferromagnetic ( fm ) superconductors in experiments @xcite stimulates the research interest in the exotic magnetic properties of fm superconductors . the state of the cooper pairs in the fm superconductors has been wildly studied @xcite . a stronger spin - orbit interaction in uge@xmath6 results in an abnormal huge magnetocrystalline anisotropy @xcite . monthoux et al.@xcite indicates that the favorite superconducting pairing type of this anisotropy is triplet . although the exact symmetry of the paired state has not yet been identified , a spin - triplet pairing is more likely than the spin - singlet pairing in these superconductors @xcite . these behaviors are somewhat like charged spin-1 bosons . thus the charged spin-1 boson model helps to understand the exotic magnetic properties observed in such materials . although the ferromagnetism @xcite in a chargeless spinor bose gas has also been involved in theory , it is little discussed when fm interaction exists in a charged spin system . accordingly the magnetic behavior will become more complex in charged spin systems with fm interactions , where diamagnetism , paramagnetism and ferromagnetism compete with each other in such case . in this paper , the magnetic properties of a charged spin-1 bose gas with fm interactions are studied via mean - field theory . alexandrov et al . found that the coulomb or any other scattering may make charged bose gases superconducting below a critical field @xcite with a specific vortex matter @xcite . superconducting is not obtained in our paper , probably because we used the mean - field approximation to deal with the fm interaction . in despite of this , mean - field theory is still effective to point out the main physics of the magnetism , especially the ferromagnetic transition @xcite . the remainder of this paper is structured as follows . in section 2 , we construct a model including landau diamagnetism , pauli paramagnetism and fm effect . the magnetization density is obtained through the analytical derivation . in section 3 , the results is obtained and the discussions of our results is presented . a summary is given in section 4 . the spin-1 bose gas with fm couplings is described by the following hamiltonian : @xmath7 where @xmath5 is the chemical potential and the landau levels of bosons with charge @xmath8 and mass @xmath9 in the effective magnetic field @xmath10 is @xmath11 where @xmath12 labels different landau levels and @xmath13 is the gyromagnetic frequency . the energy level is degenerate with degeneracy @xmath14 where @xmath15 and @xmath16 are the length in @xmath17 and @xmath18 directions of the system , respectively . the intrinsic magnetic moment associated with the spin degree of freedom leads to the zeeman energy levels split in the magnetic field , @xmath19 where @xmath1 is the lande - factor and @xmath20 denotes the spin - z index of zeeman state @xmath21 ( @xmath22 ) . the contribution to the effective hamiltonian from the fm couplings is @xmath23 where @xmath24 denotes fm coupling and spin polarization @xmath25 . the grand thermodynamic potential is expressed as @xmath26,\end{aligned}\ ] ] where @xmath27 . through converting the sum over @xmath28 to continuum integral , we obtain @xmath29}\},\nonumber\\\end{aligned}\ ] ] where @xmath30 is the volume of the system . ( [ t3 ] ) can be evaluated by taylor expansion , and then performing the integral over @xmath31 . we get @xmath32}}{1-e^{-l\beta\hbar \omega}}.\nonumber\\\end{aligned}\ ] ] for convenience s sake , we introduce some compact notation for the class of sums . it can be defined as @xmath33=\sum_{l=1}^{\infty}\frac{l^{\alpha/2}e^{-l x(\varepsilon+\delta)}}{(1-e^{-l x})^{\kappa}},\end{aligned}\ ] ] where @xmath34 and @xmath35 . within this notation , eq . ( [ t2 ] ) can be rewritten as @xmath36.\end{aligned}\ ] ] with @xmath37 . the particle density @xmath38 can be expressed as @xmath39.\end{aligned}\ ] ] the magnetization density @xmath40 can be obtained from the grand thermodynamic potential , @xmath41 \nonumber\\ & + & x(g\sigma-\frac{1}{2 } ) \sigma_{1\sigma}[2-d,0]- x\sigma_{2\sigma}[2-d,1]\biggr\}.\nonumber\\\end{aligned}\ ] ] the relation among effective magnetic field @xmath10 , external magnetic field @xmath42 and magnetization density @xmath43 is formally expressed as @xmath44 for computational convenience , some dimensionless parameters are introduced below . @xmath45 , @xmath46 , @xmath47 , @xmath48,@xmath49 , @xmath50 , @xmath51 and @xmath52 , and then @xmath53 , where @xmath54 is the characteristic temperature of the system , which is given by @xmath55 . the mean - field self - consistent equations are derived , @xmath56,\\ \label{eq1 } 1&=&\overline \omega t^{1/2}\sum_{\sigma=1,0,-1}\sigma_{1\sigma}^{\prime}[2-d,0],\\ \label{eq2 } \overline m_{t\neq 0}&=&t^{3/2}\sum_{\sigma}\biggl\{\sigma_{1\sigma}^{\prime}[-d,0 ] + x(g\sigma-\frac{1}{2 } ) \sigma_{1\sigma}^{\prime}[2-d,0]\nonumber\\ & -&x\sigma_{2\sigma}^{\prime}[2-d,1]\biggr\},\\ \label{eq3 } \overline \omega&=&h+4\pi\gamma \overline m , \label{eq4}\end{aligned}\ ] ] where @xmath57 , and @xmath58=\sum_{l=1}^{\infty}\frac{l^{\alpha/2}e^{-lx(\overline \varepsilon+\delta)}}{(1-e^{-lx})^{\kappa}},\end{aligned}\ ] ] with @xmath59 . similar method has been used to study the diamagnetism of the charged spinless bose gas @xcite . furthermore , we have extended it to investigate the magnetic properties of charged spin-1 bose gas @xcite . , ( b)@xmath60 versus @xmath61 at reduced temperature @xmath62 and magnetic field @xmath63 . the lande - factor g is chosen as : @xmath64(solid line ) , 0.3(dashed line ) , 0.5(dotted line ) . , scaledwidth=45.0% ] in the following calculations from fig . 1 to fig . 6 , the characteristic parameter @xmath65 has been set as @xmath66 , which is estimated for a system with the charge and mass of @xmath67he , and the particle density being set as @xmath68 . 1 is plotted in a very weak magnetic field @xmath63 . as shown in fig . 1(a ) , the value of total magnetization density @xmath2 presents a turning point from zero to nonzero . it is shown that the zero - field spontaneous magnetization exists in this system with increasing @xmath3 , where @xmath3 is the reduced fm coupling of charged spin-1 bose gases . the curves of @xmath69 versus @xmath3 in fig . 1(b ) are superposed for different lande - factors ( @xmath64 , 0.3 and 0.5 ) . it suggests that @xmath70 is independent with the lande - factor , so @xmath0 at a certain temperature are equal for any lande - factor . here @xmath0 is the critical value of reduced fm coupling of pm phase to fm phase transition . @xmath71 in this situation . when @xmath72 @xmath69 equals 0 , and the value of @xmath69 increases with increasing @xmath3 while @xmath4 until saturate . in the region of @xmath4 , the magnetization density @xmath73 increases with lande - factor for fixed @xmath3 , which is attributed to the pm effect @xcite . diamagnetism , paramagnetism and ferromagnetism compete with each other in such system . the diamagnetism of charged bose gases , which is due to the internal field induced by the spontaneous magnetization , can not overcome ferromagnetism in very weak magnetic field . while the competition between paramagnetism and diamagnetism has been discussed in ref . vs reduced temperature @xmath74 phase diagram of charged spin-1 bose gases at magnetic field @xmath63.,scaledwidth=45.0% ] fig . 2 plots the @xmath0 dependence of temperature at magnetic field @xmath63 . the region below @xmath0 is pm phase , while the region above it is fm phase . as the temperature increases , @xmath0 increases monotonically . it is shown that spontaneous magnetization is hard to occur at high temperature , when the bose statistics reduces to boltzmann statistics . , ( b)@xmath60 as a function of lande - factor @xmath1 of charged spin-1 bose gases at reduced temperature @xmath62 and magnetic field @xmath63 . the reduced fm coupling @xmath3 is chosen as : @xmath75(solid line ) , 0.1(dashed line ) , 0.2(dotted line ) , 0.3(dash dotted line ) , and 0.5(dash dot dotted line).,scaledwidth=45.0% ] it is supposed that @xmath69 will reach to a nonzero equivalence at @xmath76 for arbitrary value of lande - factor for the situation of fig . 1 . to further study the influence of fm coupling to spontaneous magnetization , fig . 3 is plotted . it is shown when @xmath77 , the value of @xmath69 will be zero for any lande - factor values . so the evolution of @xmath69 with lande - factor @xmath1 are superposed and keeps zero for @xmath75 and @xmath78 . for fixed @xmath3 when @xmath79 , the magnetization density @xmath2 increases monotonically with @xmath1 . while @xmath69 maintains a constant in despite of @xmath1 . our results also show that diamagnetism gives little contribution to the magnetism in the weak magnetic field , while paramagnetism and ferromagnetism play significant roles in the magnetization density in the region for @xmath4 . the interaction between paramagnetism and ferromagnetism is intricate . the increase of @xmath69 due to increasing the reduced fm coupling @xmath3 will contribute to the paramagnetism . , ( b)@xmath60 as a function of lande - factor @xmath1 of charged spin-1 bose gases at reduced temperature @xmath80 and magnetic field @xmath81 . the reduced fm coupling @xmath3 is chosen as : @xmath75(solid line ) , 0.01(dashed line ) , 0.1(dotted line ) , 0.3(dash dotted line ) , and 0.5(dash dot dotted line).,scaledwidth=45.0% ] above we have discussed the very weak magnetic field situation , now we turn to investigate the magnetic properties of charged spin-1 bose gases at finite magnetic field , where diamagnetism will emerges clearly . the result of the dependence of the total magnetization density @xmath2 and @xmath82 with lande - factor @xmath1 at a definite magnetic field @xmath81 at reduced temperature @xmath80 is shown in fig . 4 . at low temperature in the definite magnetic field , there is a competition among the paramagnetism , diamagnetism and ferromagnetism . it is shown that diamagnetism dominates in the small @xmath1 region , and therefore the magnetization density exhibits negative value . when @xmath83 , the system presents paramagnetism which is independent of reduced fm coupling @xmath3 . as seen from fig 4 , the curves of @xmath78 , @xmath84 and @xmath85 match together . it means that @xmath69 tends to saturate if @xmath3 is greater than a critical value . the increase of @xmath3 after this critical value does not contribute to the magnetization density . then the system exhibits similar magnetization density at @xmath78 , @xmath84 and @xmath85 . , ( b)@xmath60 as a function of magnetic field @xmath86 of charged spin-1 bose gases at reduced temperature @xmath62 with lande - factor @xmath87 . the reduced fm coupling @xmath3 is chosen as : @xmath75(solid line ) , 0.1(dashed line ) , 0.3(dotted line ) , and 0.5(dash dotted line).,scaledwidth=45.0% ] the discussions above all focused on fixed magnetic field . next we study the influence of magnetic field on magnetism . the evolution of the total magnetization density @xmath2 and @xmath70 with magnetic field at reduced temperature @xmath62 with @xmath87 is shown in fig . the gas always manifests paramagnetism no matter what the values of @xmath3 are . it indicates that in the case of @xmath87 , diamagnetism can not overcome paramagnetism no matter how strong the magnetic field is . this behavior is qualitatively consistent with the result of charged spin-1 bose gases @xcite . in this region , the stronger ferromagnetism induce larger @xmath69 , which will enhance paramagnetism . with increasing the magnetic field , diamagnetism also increases . while this will not change the paramagnetism of this system . whether diamagnetism can increase infinitely with magnetic field is an important issue . , ( b ) the paramagnetization density @xmath88 , and ( c ) the diamagnetization density @xmath89 as a function of magnetic field @xmath86 of charged spin-1 bose gases with @xmath87 and @xmath85 , at reduced temperature @xmath62.,scaledwidth=45.0% ] in order to manifest the paramagnetism and diamagnetism in detail , in fig . 6 we study the dependence of the total magnetization density @xmath2 , the paramagnetization density @xmath88 and the diamagnetization density @xmath89 with magnetic field in reduced temperature @xmath62 with @xmath87 and @xmath85 . @xmath88 holds a constant since fm coupling is larger . @xmath89 tends to saturate with magnetic field . it indicates that diamagnetism will not increase infinitely with magnetic field . this is why in fig . 5 the gas preserves paramagnetism even though the magnetic field is large . it is significant to evaluate the diamagnetic behavior at high magnetic field limit . without consideration of spin , the diamagnetization density , @xmath90,\end{aligned}\ ] ] when @xmath91 , @xmath92 can be reduced to , @xmath93 from equation ( 14b ) , we can obtain , @xmath94 substituting equation ( [ lim3 ] ) into ( [ lim2 ] ) , @xmath95 can be obtained . this analytical result illustrate the diamagnetization density @xmath89 tends to a finite value at high magnetic field . , ( b)@xmath60 versus reduced temperature @xmath74 of charged spin-1 bose gases with @xmath96 and @xmath97 , at magnetic field @xmath63 . the reduced fm coupling @xmath3 is chosen as : @xmath75(solid line ) , 0.1(dashed line ) , 0.3(dotted line ) , 0.5(dash dotted line).,scaledwidth=45.0% ] in order to investigate the magnetic properties of the charged spin-1 bose gas in low temperature , we suppose @xmath96 . the evolution of the total magnetization density @xmath2 and @xmath60 with reduced temperature at @xmath63 and @xmath97 is shown in fig . it is shown that @xmath2 increases with increasing temperature , and reaches a maximum , then decreases at high temperature region . the upward trend at low temperature reflects the diamagnetism , comparing with our results in ref . 21 , which shows a flat trend at the same temperature region . a sharp decline can be seen when @xmath2 close to zero . this suggests that there is a pseudo - condensate temperature in the transition from ferromagnetism to paramagnetism . although condensation has not been considered , the magnetic field is faint in such a case . it is reasonable that the pseudo - critical temperature increases with increasing reduced fm coupling @xmath61 therefore , the temperature region of ferromagnetism enlarges from @xmath75 to @xmath85 in turn . in summary , we study the interplay among paramagnetism , diamagnetism and ferromagnetism of charged spin-1 bose gas with fm coupling within the mean - field theory . in very weak magnetic field , it is shown that the ferromagnetism is stronger than the diamagnetism , where the diamagnetism is related with spontaneous magnetization . the critical value of reduced fm coupling @xmath0 of pm phase to fm phase transition increases with increasing temperature . the lande - factor @xmath1 is supposed as a variable to evaluate the strength of the pm effect . the gas exhibits a shift from diamagnetism to paramagnetism as @xmath1 increases at a finite magnetic field . ferromagnetism plays an important role in the magnetization density in the weak magnetic field . diamagnetism can not increase infinitely with magnetic field at high magnetic field . condensation is predicted to occur through studying the low - temperature magnetic properties in a weak magnetic field . jq would like to thank professor huaiming guo for the helpful discussions . this work was supported by the national natural science foundation of china ( grant no . 11004006 ) , and the fundamental research funds for the central universities of china . 10 anderson m h , ensher j r , matthews m r , wieman c e and cornell e a 1995 _ science _ * 269 * 198 ; davis k b , mewes m - o , andrews m r , druten n j v , durfee d s , kurn d m and ketterle w 1995 _ phys . lett . _ * 75 * 3969 ; bradley c c , sackett c a , tollett j j and hulet r g 1995 _ phys . rev . lett . _ * 75 * 1687 daicic j , frankel n e and kowalenko v 1993 _ phys . * 71 * 1779 ; daicic j , frankel n e , gailis r m and kowalenko v 1994 _ phys . rep . _ * 237 * 63 ; daicic j and frankel n e 1996 _ phys . _ d * 53 * 5745 saxena s s , agarwal p , ahilan k , grosche f m , haselwimmer r k w , steiner m j , pugh e , walker i r , julian s r , monthoux p , lonzarich g g , huxley a , sheikin i , braithwaite d and flouquet j 2000 _ nature _ * 406 * 587	magnetic properties of a charged spin-1 bose gas with ferromagnetic interactions is investigated within mean - field theory . it is shown that a competition between paramagnetism , diamagnetism and ferromagnetism exists in this system . it is shown that diamagnetism , being concerned with spontaneous magnetization , can not exceed ferromagnetism in very weak magnetic field . the critical value of reduced ferromagnetic coupling of paramagnetic phase to ferromagnetic phase transition @xmath0 increases with increasing temperature . the lande - factor @xmath1 is introduced to describe the strength of paramagnetic effect which comes from the spin degree of freedom . the magnetization density @xmath2 increases monotonically with @xmath1 for fixed reduced ferromagnetic coupling @xmath3 as @xmath4 . in a weak magnetic field , ferromagnetism makes immense contribution to the magnetization density . while at a high magnetic field , the diamagnetism inclines to saturate . evidence for condensation can be seen in the magnetization density at weak magnetic field .
in the last decade we entered the data - intensive era of astrophysics , where the size of data has rapidly increased , reaching in many cases dimensions overcoming the human possibility to handle them in an efficient and comprehensible way . in a very close future petabytes of data will be the standard and , to deal with such amount of information , also the data analysis techniques and facilities must quickly evolve . for example the current exploration of petabyte - scale , multi - disciplinary astronomy and earth observation synergy , by taking the advantage from their similarities in data analytics , has issued the urgency to find and develop common strategies able to achieve solutions in the data mining algorithms , computer technologies , large scale distributed database management systems as well as parallel processing frameworks @xcite . astrophysics is one of the most involved research fields facing with this data explosion , where the data volumes from the ongoing and next generation multi - band and multi - epoch surveys are expected to be so huge that the ability of the astronomers to analyze , cross - correlate and extract knowledge from such data will represent a challenge for scientists and computer engineers . to quote just a few , the esa euclid space mission will acquire and process about 100 gbday@xmath1 over at least 6 years , collecting a minimum amount of about @xmath2 tb of data @xcite ; pan - starrs @xcite is expected to produce more than @xmath3 tb of data ; the gaia space mission will build a @xmath4 map of the milky way galaxy , by collecting about one petabyte of data in five years @xcite ; the large synoptic survey telescope ( @xcite ) will provide about @xmath5tb / night of imaging data for ten years and petabytes / year of radio data products . many other planned instruments and already operative surveys will reach a huge scale during their operational lifetime , such as kids ( kilo - degree survey ; @xcite ) , des ( dark energy survey , @xcite ) , herschel - atlas @xcite , hi - gal @xcite , ska @xcite and e - elt @xcite . the growth and heterogeneity of data availability induce challenges on cross - correlation algorithms and methods . most of the interesting research fields are in fact based on the capability and efficiency to cross - correlate information among different surveys . this poses the consequent problem of transferring large volumes of data from / to data centers , _ de facto _ making almost inoperable any cross - reference analysis , unless to change the perspective , by moving software to the data @xcite . furthermore , observed data coming from different surveys , even if referred to a same sky region , are often archived and reduced by different systems and technologies . this implies that the resulting catalogs , containing billions of sources , may have very different formats , naming schemas , data structures and resolution , making the data analysis to be a not trivial challenge . some past attempts have been explored to propose standard solutions to introduce the uniformity of astronomical data quantities description , such as in the case of the uniform content descriptors of the virtual observatory @xcite . one of the most common techniques used in astrophysics and fundamental prerequisite for combining multi - band data , particularly sensible to the growing of the data sets dimensions , is the cross - match among heterogeneous catalogs , which consists in identifying and comparing sources belonging to different observations , performed at different wavelengths or under different conditions . this makes cross - matching one of the core steps of any standard modern pipeline of data reduction / analysis and one of the central components of the virtual observatory @xcite . the massive multi - band and multi - epoch information , foreseen to be available from the on - going and future surveys , will require efficient techniques and software solutions to be directly integrated into the reduction pipelines , making possible to cross - correlate in real time a large variety of parameters for billions of sky objects . important astrophysical questions , such as the evolution of star forming regions , the galaxy formation , the distribution of dark matter and the nature of dark energy , could be addressed by monitoring and correlating fluxes at different wavelengths , morphological and structural parameters at different epochs , as well as by opportunely determining their cosmological distances and by identifying and classifying peculiar objects . in such context , an efficient , reliable and flexible cross - matching mechanism plays a crucial role . in this work we present @xmath0 ( _ command - line catalog cross - match tool and the user guide are available at the page http://dame.dsf.unina.it/c3.html.]_ , @xcite ) , a tool to perform efficient catalog cross - matching , based on the multi - thread paradigm , which can be easily integrated into an automatic data analysis pipeline and scientifically validated on some real case examples taken from public astronomical data archives . furthermore , one of major features of this tool is the possibility to choose shape , orientation and size of the cross - matching area , respectively , between elliptical and rectangular , clockwise and counterclockwise , fixed and parametric . this makes the @xmath0 tool easily tailored on the specific user needs . the paper is structured as follows : after a preliminary introduction , in sec . [ sec : techniques ] we perform a summary of main available techniques ; in sec . [ sect : c3design ] , the design and architecture of the @xmath0 tool is described ; in sections [ sect : config ] and [ sect : optimization ] , the procedure to correctly use @xmath0 is illustrated with particular reference to the optimization of its parameters ; some tests performed in order to evaluate @xmath0 performance are shown in sec . [ sect : performances ] ; finally , conclusions and future improvements are drawn in sec . [ sect : conclusion ] . cross - match can be used to find detections surrounding a given source or to perform one - to - one matches in order to combine physical properties or to study the temporal evolution of a set of sources . the primary criterion for cross - matching is the approximate coincidence of celestial coordinates ( positional cross - match ) . there are also other kinds of approach , which make use of the positional mechanism supplemented by statistical analysis used to select best candidates , like the bayesian statistics @xcite . in the positional cross - match , the only attributes under consideration are the spatial information . this kind of match is of fundamental importance in astronomy , due to the fact that the same object may have different coordinates in various catalogs , for several reasons : measurement errors , instrument sensitivities , calibration , physical constraints , etc . in principle , at the base of any kind of catalog cross - match , each source of a first catalog should be compared with all counterparts contained in a second catalog . this procedure , if performed in the naive way , is extremely time consuming , due to the huge amount of sources . therefore different solutions to this problem have been proposed , taking advantage of the progress in computer science in the field of multi - processing and high performing techniques of sky partitioning . two different strategies to implement cross - matching tools basically exist : web and stand - alone applications . web applications , like openskyquery @xcite , or cds - xmatch @xcite , offer a portal to the astronomers , allowing to cross - match large astronomical data sets , either mirrored from worldwide distributed data centers or directly uploadable from the user local machine , through an intuitive user interface . the end - user has not the need to know how the data are treated , delegating all the computational choices to the backend software , in particular for what is concerning the data handling for the concurrent parallelization mechanism . other web applications , like arches @xcite , provide dedicated script languages which , on one hand , allow to perform complex cross - correlations while controlling the full process but , on the other hand , make experiment settings quite hard for an astronomer . basically , main limitation of a web - based approach is the impossibility to directly use the cross - matching tool in an automatic pipeline of data reduction / analysis . in other words , with such a tool the user can not design and implement a complete automatic procedure to deal with data . moreover , the management of concurrent jobs and the number of simultaneous users can limit the scalability of the tool . for example , a registered user of cds - xmatch has only @xmath6 mb disk space available to store his own data ( reduced to @xmath3 mb for unregistered users ) and all jobs are aborted if the computation time exceeds 100 minutes @xcite . finally , the choice of parameters and/or functional cases is often limited in order to guarantee a basic use by the end - users through short web forms ( for instance , in cds - xmatch only equatorial coordinate system is allowed ) . stand - alone applications are generally command - line tools that can be run on the end - user machine as well as on a distributed computing environment . a stand - alone application generally makes use of apis ( application programming interfaces ) , a set of routines , protocols and tools integrated in the code . there are several examples of available apis , implementing astronomical facilities , such as stil @xcite , and astroml @xcite , that can be integrated by an astronomer within its own source code . however , this requires the astronomer to be aware of strong programming skills . moreover , when the tools are executed on any local machine , it is evident that such applications may be not able to exploit the power of distributed computing , limiting the performance and requiring the storage of the catalogs on the hosting machine , besides the problem of platform dependency . on the contrary , a ready - to - use stand - alone tool , already conceived and implemented to embed the use of apis in the best way , will result an off - the - shelf product that the end - user has only to run . a local command - line tool can be put in a pipeline through easy system calls , thus giving the possibility to the end - user to create a custom data analysis / reduction procedure without writing or modifying any source code . moreover , being an all - in - one package , i.e including all the required libraries and routines , a stand - alone application can be easily used in a distributed computing environment , by simply uploading the code and the data on the working nodes of the available computing infrastructure . one of the most used stand - alone tools is stilts ( stil tool set , @xcite ) . it is not only a cross - matching software , but also a set of command - line tools based on the stil libraries , to process tabular data . it is written in pure java ( almost platform independent ) and contains a large number of facilities for table analysis , so being a very powerful instrument for the astronomers . on one hand , the general - purpose nature of stilts has the drawback to make hard the syntax for the composition of the command line ; on the other hand , it does not support the full range of cross - matching options provided by @xmath0 . in order to provide a more user - friendly tool to the astronomers , it is also available its graphical counterpart , tool for operations on catalogs and tables ( topcat , @xcite ) , an interactive graphical viewer and editor for tabular data , based on stil apis and implementing the stilts functionalities , but with all the intrinsic limitations of the graphical tools , very similar to the web applications in terms of use . regardless the approach to cross - match the astronomical sources , the main problem is to minimize the computational time exploding with the increasing of the matching catalog size . in principle , the code can be designed according to multi - process and/or multi - thread paradigm , so exploiting the hosting machine features . for instance , @xcite evaluated to use a multi - gpu environment , designing and developing their own xmatch tool , @xcite . other studies are focused to efficiently cross - match large astronomical catalogs on clusters consisting of heterogeneous processors including both multi - core cpus and gpus , ( @xcite , @xcite ) . furthermore , it is possible to reduce the number of sources to be compared among catalogs , by opportunely partitioning the sky through indexing functions and determining only a specific area to be analyzed for each source . cds - xmatch and the tool described in @xcite use hierarchical equal area isolatitude pixelisation ( healpix , @xcite ) , to create such sky partition . @xcite , instead , proposed a combined method to speed up the cross - match by using htm ( hierarchical triangle mesh , @xcite ) , in combination with healpix and by submitting the analysis to a pool of threads . healpix is a genuinely curvilinear partition of the sphere into exactly equal area quadrilaterals of varying shape ( see fig . 3 in @xcite ) the base - resolution comprises twelve pixels in three rings around the poles and equator . each pixel is partitioned into four smaller quadrilaterals in the next level . the strategy of htm is the same of healpix . the difference between the two spatial - indexing functions is that htm partitioning is based on triangles , starting with eight triangles , @xmath7 on the northern and @xmath7 on the southern hemisphere , each one partitioned into four smaller triangles at the next level ( see also fig . 2 in @xcite ) . by using one or both functions combined together , it is possible to reduce the number of comparisons among objects to ones lying in adjacent areas . finally openskyquery uses the _ zones _ indexing algorithm to efficiently support spatial queries on the sphere , @xcite . the basic idea behind the _ zones _ method is to map the sphere into stripes of a certain height @xmath8 , called zones . each object with coordinates ( @xmath9 , @xmath10 ) is assigned to a zone by using the formula : @xmath11 a traditional b - tree index is then used to store objects within a zone , ordered by _ zoneid _ and right ascension . in this way , the spatial cross - matching can be performed by using bounding boxes ( b - tree ranges ) dynamically computed , thus reducing the number of comparisons ( fig . 1 in @xcite ) . finally , an additional and expensive test allows to discard false positives . all the cross - matching algorithms based on a sky partitioning have to deal with the so - called block - edge problem , illustrated in fig . [ fig : block - edge ] : the objects @xmath12 and @xmath13 in different catalogs correspond to the same object but , falling in different pieces of the sky partition , the cross - matching algorithm is not able to identify the match . to solve this issue , it is necessary to add further steps to the pipeline , inevitably increasing the computational time . for example , the zhao s tool , @xcite , expands a healpix block with an opportunely dimensioned border ; instead , the algorithm described by @xcite , combining healpix and htm virtual indexing function shapes , is able to reduce the block - edge problem , because the lost objects in a partition may be different from one to another . and @xmath13 in two catalogs . even if corresponding to the same source , they can be discarded by the algorithm , since they belong to two different blocks of the sky partition.,title="fig:",width=226 ] + @xmath0 is a command - line open - source python script , designed and developed to perform a wide range of cross - matching types among astrophysical catalogs . the tool is able to be easily executed as a stand - alone process or integrated within any generic data reduction / analysis pipeline . based on a specialized sky partitioning function , its high - performance capability is ensured by making use of the multi - core parallel processing paradigm . it is designed to deal with massive catalogs in different formats , with the maximum flexibility given to the end - user , in terms of catalog parameters , file formats , coordinates and cross - matching functions . in @xmath0 different functional cases and matching criteria have been implemented , as well as the most used join function types . it also works with the most common catalog formats , with or without header : flexible image transport system ( fits , version tabular ) , american standard code for information interchange ( ascii , ordinary text , i.e. space separated values ) , comma separated values ( csv ) , virtual observatory table ( votable , xml based ) and with two kinds of coordinate system , equatorial and galactic , by using stilts in combination with some standard python libraries , namely _ _ numpy _ _ @xcite , and _ _ pyfits _ _ ] . + despite the general purpose of the tool , reflected in a variety of possible functional cases , @xmath0 is easy to use and to configure through few lines in a single configuration file . main features of @xmath0 are the following : 1 . _ command line _ : @xmath0 is a command - line tool . it can be used as stand - alone process or integrated within more complex pipelines ; 2 . _ python compatibility _ : compatible with python 2.7.x and 3.4.x ( up to the latest version currently available , @xmath15 ) ; 3 . _ multi - platform _ : @xmath0 has been tested on ubuntu linux @xmath16 , windows @xmath17 and @xmath18 , mac os and fedora ; 4 . _ multi - process _ : the cross - matching process has been developed to run by using a multi - core parallel processing paradigm ; 5 . _ user - friendliness _ : the tool is very simple to configure and to use ; it requires only a configuration file , described in sec . [ sect : config ] . the internal cross - matching mechanism is based on the sky partitioning into cells , whose dimensions are determined by the parameters used to match the catalogs . the sky partitioning procedure is described in [ sect : preproc ] . the fig . [ fig : flowchart ] shows the most relevant features of the @xmath0 processing flow and the user parameters available at each stage . as mentioned before , the user can run @xmath0 to match two input catalogs by choosing among three different functional cases : 1 . _ sky _ : the cross - match is done within sky areas ( elliptical or rectangular ) defined by the celestial coordinates taken from catalog parameters ; 2 . _ exact value _ : two objects are matched if they have the same value for a pair of columns ( one for each catalog ) defined by the user ; 3 . _ row - by - row _ : match done on a same row - id of the two catalogs . the only requirement here is that the input catalogs must have the same number of records . the positional cross - match strategy of the @xmath0 method is based on the same concept of the q - fulltree approach , an our tool introduced in @xcite and @xcite : for each object of the first input catalog , it is possible to define an elliptical , circular or rectangular region centered on its coordinates , whose dimensions are limited by a fixed value or defined by specific catalog parameters . for instance , the two full width at half maximum ( fwhm ) values in the catalog can define the two semi - axes of an ellipse or the couple width and height of a rectangular region . it is also possible to have a circular region , by defining an elliptical area having equal dimensions . once defined the region of interest , the next step is to search for sources of the second catalog within such region , by comparing their distance from the central object and the limits of the area ( for instance , in the elliptical cross - match the limits are defined by the analytical equation of the ellipse ) . + in the _ sky _ functional case , the user can set additional parameters in order to characterize the matching region and the properties of the input catalogs . in particular , the user may define : 1 . the shape ( elliptical or rectangular ) of the matching area , i.e. the region , centered on one of the matching sources , in which to search the objects of the second catalog ; 2 . the dimensions of the searching area . they can be defined by fixed values ( in arcseconds ) or by parametric values coming from the catalog . moreover , the region can be rotated by a position angle ( defined as fixed value or by a specific column present in the catalog ) ; 3 . the coordinate system for each catalog ( galactic , icrs , fk4 , fk5 ) and its units ( degrees , radians , sexagesimal ) , as well as the columns containing information about position and designation of the sources . an example of graphical representation of an elliptical cross - match is shown in fig . [ fig : crossmatch ] . in the _ exact value _ case , the user has to define only which columns ( one for each input catalog ) have to be matched , while in the most simple _ row - by - row _ case no particular configuration is needed . @xmath0 produces a file containing the results of the cross - match , consisting into a series of rows , corresponding to the matching objects . in the case of _ exact value _ and _ sky _ options , the user can define the conditions to be satisfied by the matched rows to be stored in the output . first , it is possible to retrieve , for each source , all the matches or only the best pairs ( in the sense of closest objects , according to the match selection criterion ) ; then , the user can choose different join possibilities ( in fig . [ fig : joins ] the graphical representation of available joins is shown ) : @xmath19 and @xmath20 : : only rows having an entry in both input catalogs , ( fig . [ fig : joins]a ) ; @xmath19 or @xmath20 : : all rows , matched and unmatched , from both input catalogs , ( fig . [ fig : joins]b ) ; all from @xmath19 ( all from @xmath20 ) : : all matched rows from catalog @xmath19 ( or @xmath20 ) , together with the unmatched rows from catalog @xmath19 ( or @xmath20 ) , ( fig . [ fig : joins]c - d ) ; @xmath19 not @xmath20 ( @xmath20 not @xmath19 ) : : all the rows of catalog @xmath19 ( or @xmath20 ) without matches in the catalog @xmath20 ( or @xmath19 ) , ( fig . [ fig : joins]e - f ) ; @xmath19 xor @xmath20 : : the `` exclusive or '' of the match - i.e. only rows from the catalog @xmath19 not having matches in the catalog @xmath20 and viceversa , ( fig . [ fig : joins]g ) . + + any experiment with the @xmath0 tool is based on two main phases ( see fig . [ fig : flowchart ] ) : 1 . _ pre - matching : _ this is the first task performed by @xmath0 during execution . the tool manipulates input catalogs to extract the required information and prepare them to the further analysis ; 2 . _ matching : _ after data preparation , @xmath0 performs the matching according to the criteria defined in the configuration file . finally , the results are stored in a file , according to the match criterion described in sec . [ sect : join ] , and all the temporary data are automatically deleted . this is the preliminary task performed by @xmath0 execution . during the pre - matching phase , @xmath0 performs a series of preparatory manipulations on input data . first of all , a validity check of the configuration parameters and input files . then it is necessary to split the data sets in order to parallelize the matching phase and improve the performance . in the _ exact value _ functional case only the first input catalog will be split , while in the _ sky _ case both data sets will be partitioned in subsets . in the latter case , @xmath0 makes always use of galactic coordinates expressed in degrees , thus converting them accordingly if expressed in different format . when required , the two catalogs are split in the following way : in the first catalog all the entries are divided in groups , whose number depends on the multi - processing settings ( see sec . [ sect : config ] ) , since each process is assigned to one group ; in the second catalog the sky region defined by the data set is divided into square cells , by assigning a cell to each entry , according to its coordinates ( fig . [ fig : partitioning ] ) . we used the python multiprocess module to overcome the gil problem , by devoting particular care to the granularity of data to be handled in parallel . this implies that the concurrent processes do not need to share resources , since each process receives different files in input ( group of object of the 1st catalog and cells ) and produces its own output . finally the results are merged to produce the final output . the partitioning procedure on the second catalog is based on the dimensions of the matching areas : the size of the unit cell is defined by the maximum dimension that the elliptical matching regions can assume . if the `` size type '' is `` parametric '' , then the maximum value of the columns indicated in the configuration is used as cell size ; in the case of `` fixed '' values , the size of the cell will be the maximum of the two values defined in the configuration ( fig . [ fig : partitioning]a ) . in order to optimize the performance , the size of the unit cell can not be less than a threshold value , namely the _ minimum partition cell size _ , which the user has to set through the configuration file . the threshold on the cell size is required in order to avoid the risk to divide the sky in too many small areas ( each one corresponding to a file stored on the disk ) , which could slow down the cross - matching phase performance . in sec . [ sect : optimization ] we illustrated a method to optimize such parameter as well as the number of processes to use , according to the hosting machine properties . once the partitioning is defined , each object of the second catalog is assigned to one cell , according to its coordinates . having defined the cells , the boundaries of an elliptical region associated to an object can fall at maximum in the eight cells surrounding the one including the object , as shown in fig . [ fig : partitioning]b . this prevents the block - edge problem previously introduced . once the data have been properly re - arranged , the cross - match analysis can start . in the _ row - by - row _ case , each row of the first catalog is simply merged with the corresponding row of the second data set through a serial procedure . in the other functional cases , the cross - matching procedure has been designed and implemented to run by using parallel processing , i.e. by assigning to each parallel process one group generated in the previous phase . in the _ exact value _ case , each object of the group is compared with all the records of the second catalog and matched according to the conditions defined in the configuration file . in the _ sky _ functional case , the matching procedure is slightly more complex . as described in sec . [ sect : usecases ] , the cross - match at the basis of the @xmath0 method is based on the relative position of two objects : for each object of the first input catalog , @xmath0 defines the elliptical / rectangular region centered on its coordinates and dimensions . therefore a source of the second catalog is matched if it falls within such region . in practice , as explained in the pre - matching phase , having identified a specific cell for each object of a group , this information is used to define the minimum region around the object used for the matching analysis . the described choice to set the dimensions of the cells ensures that , if a source matches with the object , it must lie in the nine cells surrounding the object ( also known as moore s neighborhood , @xcite , see also fig . [ fig : partitioning]b ) . therefore it is sufficient to cross - match an object of a group only with the sources falling in nine cells . in the _ sky _ functional case , @xmath0 performs a cross - matching of objects lying within an elliptical , circular or rectangular area , centered on the sources of the first input catalog . the matching area is characterized by @xmath21 configuration parameters defining its shape , dimensions and orientation . in fig . [ fig : pa ] is depicted a graphical representation of two matching areas ( elliptical and rectangular ) with the indication of its parameters . in particular , to define the orientation of the matching area , @xmath0 requires two further parameters besides the offset and the value of the position angle , representing its orientation . the position angle , indeed , is referred , by default , to the greatest axis of the matching area with a clockwise orientation . the two additional parameters give the possibility to indicate , respectively , the correct orientation ( clockwise / counterclockwise ) and a shift angle ( in degrees ) . finally , the results of the cross - matching are stored in a file , containing the concatenation of all the columns of the input catalogs referred to the matched rows . in the _ sky _ functional case the column reporting the separation distance between the two matching objects is also included . the tool @xmath0 is interfaced with the user through a single configuration file , to be properly edited just before the execution of any experiment . if the catalogs do not contain the source s designation / id information , @xmath0 will automatically assign an incremental row - id to each entry as object designation . for the _ sky _ functional case , assuming that both input catalogs contain the columns reporting the object coordinates , @xmath0 is able to work with galactic and equatorial ( icrs , fk4 , fk5 ) coordinate systems , expressed in the following units : degrees , radians or sexagesimal . if the user wants to use catalog information to define the matching region ( for instance , the fwhms or a radius defined by the instrumental resolution ) , obviously the first input catalog must contain such data . the position angle value / column is , on the contrary , an optional information ( default is 0@xmath22 , clockwise ) . @xmath0 is conceived for a community as wide as possible , hence it has been designed in order to satisfy the requirement of user - friendliness . therefore , the configuration phase is limited to the editing of a setup file , can also automatically generate a dummy configuration file that could be used as template . ] containing all the information required to run @xmath0 . this file is structured in sections , identified by square brackets : the first two are required , while the others depend on the particular use case . in particular , the user has to provide the following information : 1 . the input files and their format ( fits , ascii , csv or votable ) ; 2 . the name and paths of the temporary , log and output files ; 3 . the match criterion , corresponding to one of the functional cases ( _ sky , exact value , row - by - row _ ) . @xmath0 gives also the possibility to set the number of processes running in parallel , through an optional parameter which has as default the number of cores of the working machine ( minus one left available for system auxiliary tasks ) . the configuration for the _ sky _ functional case foresees the setup of specific parameters of the configuration file : those required to define the shape and dimensions of the matching area , the properties of the input catalogs already mentioned in sec . [ sect : usecases ] , coordinate system , units as well as the column indexes for source coordinates and designation . in addition , a parameter characterizing the sky partitioning has to be set ( see sec . [ sect : preproc ] for further information ) . the parameters useful to characterize the matching area are the following : area shape : : it can be elliptical or rectangular ( circular is a special elliptical case ) ; size type : : the valid entries are _ fixed _ or _ parametric_. in the first case , a fixed value will be used to determine the matching area ; in the second , the dimensions and inclination of the matching area will be calculated by using catalog parameters ; first and second dimensions of matching area : : the axes of the ellipse or width and height of the rectangular area . in case of fixed `` size type '' , they are decimal values ( in arcsec ) , otherwise , they represent the index ( integer ) or name ( string ) of the columns containing the information to be used ; parametric factor : : it is required and used only in the case of parametric `` size type '' . it is a decimal number factor to be multiplied by the values used as dimensions , in order to increase or decrease the matching region , as well as useful to convert their format ; pa column / value : : it is the position angle value ( in the `` fixed '' case , expressed in degrees ) or the name / id of the column containing the position angle information ( in the `` parametric '' case ) ; pa settings : : the position angle , which in @xmath0 is referred , by default , to the main axis of the matching area ( greatest ) with a clockwise orientation . the two parameters defined here give the possibility to indicate the correct orientation ( clockwise / counterclockwise ) and a shift angle ( in degrees ) . the user has also to specify which rows must be included in the output file , by setting the two parameters indicating the match selection and the join type , as described in sec . [ sect : join ] . for the _ exact value _ functional case it is required to set the name or i d of the columns used for the match for both input files . the user has also to specify which rows must be included in the output file , by setting the two parameters indicating the match selection and the join type , as described in sec . [ sect : join ] . for the _ row - by - row _ functional case , no other settings are required . the only constrain is that both catalogs must have the same number of entries . as reflected from the description of @xmath0 , the choice of the best values for its internal parameters ( in particular the number of parallel processes and the minimum cell size , introduced in sec . [ sect : preproc ] ) , is crucial to obtain the best computational efficiency . this section is dedicated to show the importance of this choice , directly depending on the features of the hosting machine . in the following tests we used a computer equipped with an intel(r ) core(tm ) @xmath23 , with one @xmath24 , @xmath25 cpu , @xmath26 gb of ram and hosting ubuntu linux @xmath16 as operative system ( os ) on a standard hard disk drive . we proceeded by performing two different kinds of tests : 1 . a series of tests with a fixed value for the minimum cell size ( @xmath27 ) and different values of the number of parallel processes ; 2 . a second series by using the best value of number of parallel processes found at previous step and different values for the minimum cell size . the configuration parameters used in this set of tests are reported in table [ test1:settings ] . the input data sets are two identical catalogs ( csv format ) consisting of @xmath28 objects extracted from the ukidss gps public data @xcite , in the range of galactic coordinates @xmath29 $ ] , @xmath30 $ ] . each record is composed by @xmath31 columns . the choice to cross - match a catalog with itself represent the worst case in terms of cross - matching computational time , since each object matches at least with itself . by setting `` match selection '' as best and `` join type '' as 1 and 2 ( see table [ test1:settings ] ) , we obtained an output of @xmath28 objects matched with themselves as expected . we also performed all the tests by using a random shuffled version of the same input catalog , obtaining the same results . this demonstrates that the @xmath0 output is not affected by the particular order of data in the catalogs . .@xmath0 settings in the first set of tests performed to evaluate the impact of the number of parallel processes and the minimum cell size configuration parameters on the execution time . the choice of same dimensions for the ellipse axes was due to perform a fair comparison with stilts and cds - xmatch , which allow only circular cross - matching . [ cols="^,^",options="header " , ] the first input catalog has been extracted by the ukidss gps data in the range of galactic coordinates @xmath32 $ ] , @xmath30 $ ] , while the second input catalog has been extracted by the glimpse _ data , ( @xcite and @xcite ) , in the same range of coordinates . from each catalog , different subsets with variable number of objects have been extracted . in particular , data sets with , respectively , @xmath33 , @xmath34 , @xmath28 , @xmath35 and @xmath36 objects have been created from the first catalog , while , from second catalog , data sets with @xmath33 , @xmath34 , @xmath28 and @xmath35 rows have been extracted . then , each subset of first catalog has been cross - matched with all the subsets of the second catalog . for uniformity of comparison , due to the limitations imposed by cds - xmatch in terms of available disk space , it has been necessary to limit to only @xmath37 the number of columns for all the subsets involved in the tests performed to compare c@xmath38 and cds - xmatch ( for instance , i d and galactic coordinates ) . for the same reason , the data set with @xmath39 rows has not been used in the comparison between c@xmath38 and cds - xmatch . the common internal configuration used in these tests is shown in table [ test1:settings ] , except for the match selection there was , in fact , the necessity to set it to _ all _ for uniformity of comparison with the cds - xmatch tool ( which makes available only this option ) . then the _ best _ type has been used to compare @xmath0 with stilts and topcat . furthermore , in all the tests , the number of parallel processes was set to @xmath40 and the minimum cell size to @xmath27 , corresponding to the best conditions found in the optimization process of @xmath0 ( see sec . [ sect : optimization ] ) . finally , we chose same dimensions of the ellipse axes in order to be aligned with other tools , which allow only circular cross - matching areas . concerning the comparison among @xmath0 and the three mentioned tools , in the cases of both _ all _ and _ best _ types of matching selection , all tools provided exactly the same number of matches in the whole set of tests , thus confirming the reliability of @xmath0 with respect to other tools ( table [ tab : matchres ] ) . rows has not been used . ] in terms of computational efficiency , @xmath0 has been evaluated by comparing the computational time of its cross - matching phase with the other tools . the pre - matching and output creation steps have been excluded from the comparison , because strongly dependent on the host computing infrastructure . the other configuration parameters have been left unchanged ( table [ test1:settings ] ) . the complete setup for the described experiments is reported in the appendix . in fig . [ fig : c3vsstrows ] we show the computational time of the cross - matching phase for @xmath0 and stilts , as function of the incremental number of rows ( objects ) in the first catalog , and by varying the size of the second catalog in four cases , spanning from @xmath33 to @xmath35 rows . in all diagrams , it appears evident the difference between the two tools , becoming particularly relevant with increasing amounts of data . in the second set of tests performed on the @xmath0 cross - matching phase and stilts , the computational time has been evaluated as function of the incremental number of columns of the first catalog ( from the minimum required @xmath37 up to @xmath31 , the maximum number of columns of catalog 1 ) , and by fixing the number of columns of the second catalog in five cases , respectively , @xmath37 , @xmath5 , @xmath41 , @xmath42 and @xmath43 , which is the maximum number of columns for catalog 2 . in terms of number of rows , in all cases both catalogs were fixed to @xmath35 of entries . in fig . [ fig : c3vsstcols ] the results only for @xmath37 and @xmath43 columns of catalog 2 are reported , showing that @xmath0 is almost invariant to the increasing of columns , becoming indeed faster than stilts from a certain amount of columns . such trend is confirmed in all the other tests with different number of columns of the second catalog . this behavior appears particularly suitable in the case of massive catalogs . finally , in the case of two fits input files instead of csv files , stilts computational time as function of the number of columns is constant and slightly faster than @xmath14 . in the last series of tests , we compared the computational efficiency between the @xmath0 cross - matching phase and cds - xmatch . in this case , due to the limitation of the catalog size imposed by cds - xmatch , the tests have been performed by varying only the number of rows from @xmath33 to @xmath35 as in the analogous tests with stilts ( except the test with @xmath36 rows ) , fixing the number of columns to @xmath37 . moreover , in this case , the cross - matching phase of @xmath0 has been compared with the duration of the phase _ execution _ of the cds - xmatch experiment , thus ignoring latency time due to the job submission , strongly depending on the network status and the state of the job queue , but taking into account the whole job execution . the results , reported in fig . [ fig : c3vsxmrows ] , show a better performance of @xmath0 , although less evident when both catalogs are highly increasing their dimensions , where the differences due to the different hardware features become more relevant . at the end of the test campaign , two other kinds of tests have been performed : ( i ) the verification of the portability of @xmath0 on different oss and ( ii ) an analysis of the impact of different disk technology on the computing time efficiency of the tool . in the first case , we noted , as expected , a decreasing of @xmath0 overall time performance on the windows versions ( @xmath17 and @xmath18 ) , with respect to same tests executed on linux versions ( ubuntu and fedora ) and mac os . on average @xmath0 execution was @xmath44 times more efficient on linux and mac os than windows . this is most probably due to the different strategy of disk handling among various oss , particularly critical for applications , like cross - matching tools , which make an intensive use of disk accesses . this analysis induced us to compare two disk technologies : hdd ( hard disk drive ) vs ssd ( solid state disk ) . both kinds of disks have been used on a sample of the tests previously described , revealing on average a not negligible increasing of computing time performance in the ssd case of @xmath45 times with respect to hdd . for clarity , all test results presented in the previous sections have been performed on the same hdd . in this paper we have introduced @xmath0 , a new scalable tool to cross - match astronomical data sets . it is a multi - platform command - line python script , designed to provide the maximum flexibility to the end users in terms of choice about catalog properties ( i / o formats and coordinates systems ) , shape and size of matching area and cross - matching type . nevertheless , it is easy to configure , by compiling a single configuration file , and to execute as a stand - alone process or integrated within any generic data reduction / analysis pipeline . in order to ensure the high - performance capability , the tool design has been based on the multi - core parallel processing paradigm and on a basic sky partitioning function to reduce the number of matches to check , thus decreasing the global computational time . moreover , in order to reach the best performance , the user can tune on the specific needs the shape and orientation of the matching region , as well as tailor the tool configuration to the features of the hosting machine , by properly setting the number of concurrent processes and the resolution of sky partitioning . although elliptical cross - match and the parametric handling of angular orientation and offset are known concepts in the astrophysical context , their availability in the presented command - line tool makes @xmath0 competitive in the context of public astronomical tools . a test campaign , done on real public data , has been performed to scientifically validate the @xmath0 tool , showing a perfect agreement with other publicly available tools . the computing time efficiency has been also measured by comparing our tool with other applications , representative of different paradigms , from stand - alone command - line ( stilts ) and graphical user interface ( topcat ) to web applications ( cds - xmatch ) . such tests revealed the full comparable performance , in particular when input catalogs increase their size and dimensions . for the next release of the tool , the work will be mainly focused on the optimization of the pre - matching and output creation phases , by applying the parallel processing paradigm in a more intensive way . moreover , we are evaluating the possibility to improve the sky partitioning efficiency by optimizing the calculation of the minimum cell size , suitable also to avoid the block - edge problem . the @xmath0 tool , @xcite , and the user guide are available at the page http://dame.dsf.unina.it/c3.html . the authors would like to thank the anonymous referee for extremely valuable comments and suggestions . mb and sc acknowledge financial contribution from the agreement asi / inaf i/023/12/1 . mb , am and gr acknowledge financial contribution from the 7th european framework programme for research grant fp7-space-2013 - 1 , _ vialactea - the milky way as a star formation engine_. mb and am acknowledge the prin - inaf 2014 _ glittering kaleidoscopes in the sky : the multifaceted nature and role of galaxy clusters_. 99 agrafioti , i. 2012 , from the geosphere to the cosmos , synergies with astroparticle physics , astroparticle physics for europe ( aspera ) , contributed volume , http://www.aspera-eu.org annis , j. t. , 2013 , in american astronomical society meeting abstracts # 221 , des survey strategy and expectations for early science , 221 , id.335.05 becciani , u. , bandieramonte , m. , brescia , m. , et al . 2015 , in proc . adass xxv conf . , advanced environment for knowledge discovery in the vialactea project , in press . ( arxiv:1511.08619 ) benjamin , r. a. , churchwell , e. , babler , b.l . 2003 , , 115 , 953 , doi : 10.1086/376696 boch , t. , pineau , f. x. , & derriere , s. 2014 , cds xmatch service documentation , http://cdsxmatch.u-strasbg.fr/xmatch/doc/ braun , r. 2015 , in proc . of `` the many facets of extragalactic radio surveys : towards new scientific challenges '' ( extra - radsur2015 ) . 20 - 23 october 2015 . bologna , italy . http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=267 , id.34 budavri , t. , & lee , m. a. 2013 , xmatch : gpu enhanced astronomic catalog cross - matching , astrophysics source code library , record ascl:1303.021 budavri , t. , & szalay , a. s. 2008 , , 679 , 301 cavuoti , s. , brescia , m. , longo , g. 2012 , proc . spie , 8451 , 845103 , doi : 10.1117/12.925321 churchwell , e. , babler , b.l . , meade , m.r . 2009 , , 121 , 213 de jong , j. t. a. , verdoes kleijn , g. a. , boxhoorn , d. r. , et al . 2015 , a&a , 582 , a62 douglas , j. , de bruijne , j. , oflaherty , k. , et al . 2007 , esa bulletin , 132 , 26 du , p. , ren , j. j. , pan , j. c. , luo , a. 2014 , scpma , 57 , 577 gorski , k. m. , hivon , e. , banday , a. j. , et al . 2005 , 622 , 759 gray , j. , nieto - santisteban , m. a. & szalay , a. s. 2006 , the zones algorithm for finding points - near - a - point or cross - matchin spatial datasetes , microsoft tech . : msr - tr-2006 - 52 gray , l. 2003 , not . 50 , 200 ivezic , z. , 2009 , in aps april meeting abstracts , lsst : the physics of the dark universe , 54 , w4.003 , http://adsabs.harvard.edu/abs/2009aps..apr.w4003i ivoa recommendation 2005 , an ivoa standard for unified content decriptors version 1.1 ( http://adsabs.harvard.edu/abs/2005ivoa.spec.0819d ) jia , x. & luo , q. 2016 , in proc . conf . on scientific and statistical database management ( ssdbm 16 ) , ed . p. baumann et al . ( new york , ny , acm ) , 12 , doi : 10.1145/2949689.2949705 jia , x. , luo , q. & fan , d. 2015 , in proc . ieee xxi int . conf . on parallel and distributed systems ( icpads ) , 617 , doi : 10.1109/icpads.2015.83 kaiser , n. , 2004 , proc . spie , 5489 , 11 kunszt , p. z. , szalay , a. s. , thakar , a. r. in proc . mpa / eso / mpe workshop , eds banday , a. j. , zaroubi , s. , bartelmann , m. ( berlin : springer ) , 631 , doi : 10.1007/10849171_83 laureijs , r. , racca , g. , stagnaro , l. , et al . 2014 , in proc . spie , 9143 , 91430h , doi : 10.1117/12.2054883 lee , m. a. , & budavri , t. , 2013 , in asp conf . 475 , proc . astronomical data analysis software and systems xxii conf . , cross - identification of astronomical catalogs on multiple gpus , ed . friedel , d. n. , ( san francisco , ca : asp ) , 235 lucas , p. w , hoare , m. g , longmore , a. , et al . 2008 , , 391 , 136 malkov , o. , dluzhnevskaya , o. , karpov , s. , et al . 2012 , balta , 21 , 319 martins , c. j. a. p. , leite , a. c. o. , pedrosa , p. o. j. , 2014 , in statistical challenges in 21st century cosmology , fundamental cosmology with the e - elt , proc . of the international astronomical union , iau symposium , ed . heavens , a. , starck , j .- l . & krone - martins , a. , 306 , 385 - 387 , doi : 10.1017/s1743921314013441 molinari , s. , schisano , e. , elia , d. , et al . 2016 , a&a 591 , a149 , doi : 10.1051/0004 - 6361/201526380 motch , c. , & arches consortium , 2015 , in astronomical data analysis software and systems xxiv , the arches project , ed . a. r. taylor and e. rosolowsky ( san francisco : astronomical society of the pacific ) , 437 nieto - santisteban , m. a. , thakar , a. r. , szalay , a. s. , 2006 , cross - matching very large datasets ( baltimore , md : johns hopkins university ) pineau , f. x. , boch , t. , & derriere , s. 2011 , in asp conf . 442 , proc astronomical data analysis software and systems xx , ed . evans , i. n. , accomazzi , a. , mink , d. j. , & rots , a. h. ( san francisco , ca : asp ) , 85 riccio , g. , brescia , m. , cavuoti , s. , mercurio , a. 2016 , c3 : command - line catalog crossmatch for modern astronomical surveys , astrophysics source code library , record ascl:1610.006 sciacca , e. , vitello , f. , becciani , u. , et al . , 2016 , milky way analysis through a science gateway : workflows and resource monitoring , proceedings of 8th international workshop on science gateways , june 2016 , rome , italy , submitted to ceur - ws , http://ceur-ws.org , issn : 1613 - 0073 . taylor , m. b. , 2005 , in asp conf . ser . 347 , astronomical data analysis software and systems xiv , ed . p. shopbell , m. britton , & r. ebert ( san francisco , ca : asp ) , 29 taylor , m. b. , 2006 , in asp conf . ser . 351 , astronomical data analysis software and systems xv , ed . c. gabriel et al . ( san francisco , ca : asp ) , 666 valiante , e. 2015 , in iau general assembly , meeting 29 , the herschel - atlas survey : main results and data release , 22 , 2257414 van der walt , s. , colbert , s. c. & varoquaux , g. , 2011 , cse , 13 , 22 vanderplas , j. t. , connolly , a. j. , ivezi , & gray , a. 2012 , in proc . conf . on intelligent data understanding ( cidu ) , introduction to astroml : machine learning for astrophysics , 47 - 54 , doi : 10.1109/cidu.2012.6382200 varga - verebelyi , e. , dobos , l. , budavari , t. , 2016 , in from interstellar clouds to star - forming galaxies : universal processes ? , iau symposium 315 , herschel footprint database and service , eprint arxiv:1602.01050 zhao , q. , sun , j. , yu , c. , et al . , 2009 , in algorithms and architectures for parallel processing , proc . of 9th international conference , ica3pp 2009 , a paralleled large - scale astronomical cross - matching function , eds . arrems , h. , chang , s .,- l . , 604 - 614 , isbn : 978 - 3 - 642 - 03095 - 6 this appendix reports the configuration file as used in the example described in sec . [ sect : comparison ] . the text preceded by the semicolon is a comment . .... \textcolor{red}{[i / o files ] } input catalog 1 : \textcolor{olive}{./input / ukidss.csv } format catalog 1 : \textcolor{olive}{csv } \textcolor{blue}{;csv , fits , votable or ascii } input catalog 2 : \textcolor{olive}{./input / glimpse.csv } format catalog 2 : \textcolor{olive}{csv } \textcolor{blue}{;csv , fits , votable or ascii } output : \textcolor{olive}{./output / out.csv } output format : \textcolor{olive}{csv } \textcolor{blue}{;csv , fits , votable or ascii } log file : \textcolor{olive}{./output / out.log } stilts directory : \textcolor{olive}{./libs } working directory : \textcolor{olive}{./tmp } \textcolor{blue}{;temporary directory , removed when completed } \textcolor{red}{[sky parameters ] } area shape : \textcolor{olive}{ellipse } \textcolor{blue}{;ellipse or rectangle } size type : \textcolor{olive}{fixed } \textcolor{blue}{;parametric or fixed } matching area first dimension : \textcolor{olive}{5 } \textcolor{blue}{;arcsec for fixed type - column name / number for parametric type } matching area second dimension : \textcolor{olive}{5 } \textcolor{blue}{;arcsec for fixed type - column name / number for parametric type } parametric factor : \textcolor{olive}{1 } \textcolor{blue}{;multiplicative factor for dimension columns - required for parametric type } pa column / value : \textcolor{olive}{0 } \textcolor{blue}{;degrees for fixed type - column name / number for parametric type } pa settings : \textcolor{olive}{clock , 0 } \textcolor{blue}{;orientation ( clock , counter ) , shift ( degrees ) -empty or default = clock,0 } catalog 2 minimum partition cell size : \textcolor{olive}{100 } \textcolor{blue}{;arcsec } \textcolor{red}{[catalog 1 properties ] } coordinate system : \textcolor{olive}{galactic } \textcolor{blue}{;galactic , icrs , fk4 , fk5 } coordinate units : \textcolor{olive}{deg } \textcolor{blue}{;degrees ( or deg ) , radians ( or rad ) , sexagesimal ( or sex ) } glon / ra column : \textcolor{olive}{l } \textcolor{blue}{;column number or name - required for sky algorithm } glat / dec column : \textcolor{olive}{b } \textcolor{blue}{;column number or name - required for sky algorithm } designation column : \textcolor{olive}{sourceid } \textcolor{blue}{;column number or name - -1 for none } \textcolor{red}{[catalog 2 properties ] } coordinate system : \textcolor{olive}{galactic } \textcolor{blue}{;galactic , icrs , fk4 , fk5 } coordinate units : \textcolor{olive}{deg } \textcolor{blue}{;degrees ( or deg ) , radians ( or rad ) , sexagesimal ( or sex ) } glon / ra column : \textcolor{olive}{l } \textcolor{blue}{;column number or name - required for sky algorithm } glat / dec column : \textcolor{olive}{b } \textcolor{blue}{;column number or name - required for sky algorithm } designation column : \textcolor{olive}{designation } \textcolor{blue}{;column number or name , -1 for none } \textcolor{red}{[output rows ] } match selection : \textcolor{olive}{all } \textcolor{blue}{;all or best } join type : \textcolor{olive}{1 and 2 } \textcolor{blue}{;1 and 2 , 1 or 2 , all from 1 , all from 2 , 1 not 2 , 2 not 1 , 1 xor 2 } ....	modern astrophysics is based on multi - wavelength data organized into large and heterogeneous catalogs . hence , the need for efficient , reliable and scalable catalog cross - matching methods plays a crucial role in the era of the petabyte scale . furthermore , multi - band data have often very different angular resolution , requiring the highest generality of cross - matching features , mainly in terms of region shape and resolution . in this work we present @xmath0 ( command - line catalog cross - match ) , a multi - platform application designed to efficiently cross - match massive catalogs . it is based on a multi - core parallel processing paradigm and conceived to be executed as a stand - alone command - line process or integrated within any generic data reduction / analysis pipeline , providing the maximum flexibility to the end - user , in terms of portability , parameter configuration , catalog formats , angular resolution , region shapes , coordinate units and cross - matching types . using real data , extracted from public surveys , we discuss the cross - matching capabilities and computing time efficiency also through a direct comparison with some publicly available tools , chosen among the most used within the community , and representative of different interface paradigms . we verified that the @xmath0 tool has excellent capabilities to perform an efficient and reliable cross - matching between large data sets . although the elliptical cross - match and the parametric handling of angular orientation and offset are known concepts in the astrophysical context , their availability in the presented command - line tool makes @xmath0 competitive in the context of public astronomical tools .
manipulating the molecular rotational degrees of freedom in gas phase by means of laser fields remains a very attractive topic in quantum control @xcite with a wide range of applications in photochemistry extending from chemical reactivity @xcite to nanoscale design @xcite , stereochemistry @xcite , surface processing @xcite , catalysis @xcite , and attosecond molecular dynamics @xcite . such phenomena play also a role in quantum computing @xcite and high - order harmonic generation @xcite . in this setting , molecular alignment and orientation can be identified as crucial prerequisites before exploring more complex control scenarios @xcite . the alignment process @xcite is by now a well - established concept from both the experimental and theoretical points of view with recent extensions ranging from the deflection of aligned molecules @xcite , the introduction of planar alignment @xcite , the control of molecular unidirectional motion @xcite , the study of molecular superrotors @xcite or the analysis of dissipation effects due to molecular collisions @xcite . the description and the control of molecular orientation are not currently at the same degree of improvement , in particular from the experimental point of view . molecular orientation was achieved in the adiabatic regime @xcite . if no static field is used , a rapid turn - off of the laser field allows to get orientation under field - free conditions . at low temperature , a very high degree of orientation can be obtained using such control strategies and a molecular quantum - state selection @xcite . control schemes in the sudden regime , where the duration of the control field is short with respect to the rotational period , have been also developed @xcite . among other techniques , we can mention the interaction of the molecule with a terahertz ( thz ) laser pulse and the @xmath0- scheme . note that a larger degree of orientation can be achieved with the two - color mechanisms through ionization depletion @xcite . in this second regime , molecular orientation has been recently addressed experimentally for linear molecules using only a thz field @xcite or its combination with a laser field @xcite , and in @xcite for an excitation process with two - color laser fields . in this work , we complement the previous experimental and theoretical works on the orientation dynamics produced by thz fields by exploring the orientation at high temperature ( typically room temperature ) of a symmetric top molecule , ch@xmath1i . this molecule is a good candidate to achieve a high degree of orientation at room temperature due to its large permanent dipole moment and its relatively small rotational constant @xcite . the thz pulses are obtained from the excitation of a plasma by a two - color femtosecond laser field @xcite , while the detection process is based on the free - induction decay ( fid ) emitted by the molecular sample after the thz excitation @xcite . we extend the previous studies on the subject by considering the case of a symmetric top molecule at room temperature . we show that a noticeable degree of orientation can be reached . a theoretical description of the propagation of a thz field in the sample shows that the fid is not proportional to the degree of orientation but to its time derivative . a complete analytical derivation of this result is given in this paper . note that this dependency has been already mentioned in @xcite . the relation between the fid and the degree of orientation allows us to quantitatively compare the experimental observations with the numerical simulations . a very good match is found for the first two orientation revivals . this agreement has been improved by accounting for the centrifugal distorsion and the relaxation effects in the computations @xcite . we also use this theoretical description to explore the influence of the laser parameters on the orientation dynamics . the paper is organized as follows . the experimental setup is described in sec . [ sec3 ] , with a special emphasis on the generation of thz pulses and on the detection process . we show in sec . [ sec2p ] that the fid is given at first order by the time derivative of the degree of orientation . the model system is introduced in sec . the numerical and the experimental results are discussed in sec . conclusions and prospective views are given in sec . the experimental set - up for producing and measuring molecular orientation is shown in fig.[set_up ] . it is based on a thz pulsed source and an electro - optical sampling device for the detection . the thz pulses are produced through plasma generation in gases with two - color femtosecond pulses @xcite . a fraction of the output of a chirped pulse amplifier ( cpa , 800 nm wavelength , 100 fs pulse duration , 7.5 mj energy , 100 hz repetition rate ) is focused in dry nitrogen . a type i phase matching @xmath2-barium borate ( bbo ) crystal is inserted on the beam path at a given distance from the focusing point to produce the second harmonic at 400 nm . the generation of thz pulses is optimized by adjusting the phase matching angle of the frequency doubling crystal and its longitudinal position using a micrometric stage . the thz source provides pulses of typically few hundred femtoseconds pulse duration with only one cycle ( see fig . [ fig2 ] below for details ) and covers a spectral range from 0 to 4 thz , with a maximum at about 1.5 thz . after collimation by an off - axis parabolic mirror , the two incoming beams at 800 and 400 nm are filtered out by means of a 2-mm thick plate from polytetrafluoroethylene and a thin black polyethylene sheet . the thz pulse is then focused in a cell ( 40 mm optical pathway ) containing the sample at room temperature . the ch@xmath3i sample ( _ sigma aldrich _ , batch reference _ bcbk 1300 v _ , reagent grade chemicals ) is initially stored in the liquid phase and vaporized by expansion into the cell under vacuum , just before the experiment so as to avoid water pollution of the sample . the thz electric field in the gas sample is less than 100 kv@xmath4 . the typical pressure used in this work is around 0.2 - 0.3 bar ( _ e.g. _ below the saturation vapour pressure of ch@xmath3i ) . for a good transmission of the thz beam through the gas cell , the windows are in polymethylpentene polymer ( tpx , 38.1 mm diameter , 2.0 mm thickness ) . the transmitted thz radiation is collected from the sample by another off - axis parabolic mirror and sent to the detection device . this latter is based on an electro - optical sampling of the thz pulse shape in the time domain @xcite . the weak probe beam ( typically @xmath5 nj ) derived from the output of the cpa is focused and spatially overlapped with the thz radiation in a znte ( 110 ) crystal . the polarization of the probe beam is modified through the pockels effect induced by the thz beam ( electro - optic detection process ) . this allows us to sample the thz electric field by changing the time delay between the two pulses with a motorized delay line stage . the change of the polarization state of the probe beam is measured from the combination of a quarter wave plate ( qwp ) , a wollaston prism ( wp ) , and two head - to - tail connected photodiodes so that the difference of their signals is directly obtained ( see fig . [ set_up ] ) . the difference signal is then amplified and sent to a lock - in amplifier synchronized with the laser repetition rate . the quarter wave plate is oriented so as to get a circular polarization without thz pulse and equivalent signals for the two photodiodes . the thz pathway is included in a box continuously purged ( relative humidity @xmath6 7 @xmath7 ) with dry nitrogen to avoid absorption by water vapor . [ ht ] the signal @xmath8 recorded by the lock - in amplifier is proportional to @xmath9 , the difference between the intensities measured by the two head - to - tail connected photodiodes . @xmath9 is given by @xcite : @xmath10 where @xmath11 is the probe intensity , @xmath12 the probe angular frequency , @xmath13 the refraction index at the probe frequency , @xmath14 the thz electric field , @xmath15 the electro - optic coefficient of the znte crystal ( pockels effect , @xmath15=4 pm v@xmath16 ) , @xmath17 the crystal length ( 200 @xmath18 m ) , and @xmath19 the speed of light in vacuum . the typical ratio @xmath20 measured in the experiment is @xmath21 1 - 5 @xmath7 , @xmath22 being the signal produced by the thz pulse and @xmath23 the signal delivered by each photodiode . the electric field @xmath14 can be directly evaluated by using eq . ( [ firsteq ] ) and taking into account the transmission and reflection coefficients of the different optical components . the estimated electric field in the gas sample is typically within the range 6 - 30 kv @xmath4 . note that @xmath14 in eq . ( [ firsteq ] ) is the total electric field including the transmitted thz pulse @xmath24 and the fid electric field , as discussed below . a typical recording of the electro - optical sampling signal is depicted in fig . [ typical experimental signal ] . it exhibits the transmitted thz pulse at zero delay and the two first orientational revivals at 67 and 134 ps . the goal of sec . [ sec2p ] is to interpret this experimental trace in terms of orientation efficiency . [ ht ] i. the pressure in the gas cell was 0.26 bar . the transmitted thz pulse around zero delay and two first orientational revivals at 67 and 134 ps are shown.,title="fig:",width=377 ] this section is aimed at describing the production of fid and its propagation in a gaseous sample . we follow here the formalism and the approximations used in @xcite . to be more concrete , we consider a thz pulse linearly polarized along the @xmath25- direction and propagating along the @xmath26- one . this field is of the form @xmath27 at @xmath28 . this thz field experiences an instantaneous dipole @xmath29 along the @xmath25- direction which satisfies : @xmath30 where @xmath31 is the permanent dipole moment , @xmath32 d for the ch@xmath3i molecule @xcite , and @xmath33 stands for the angle between the molecular axis defined by the c - i bond and the field polarization direction . the induced dipole @xmath29 generates a contribution to the thz field which also propagates within the sample . we introduce the polarization @xmath34 of the sample given in the frequency domain by : @xmath35 where @xmath36 is the number of molecules and @xmath37 the corresponding volume . this polarization is also related to the electric field through the susceptibility parameter @xmath38 : @xmath39 where @xmath40 is the vacuum permittivity . the spectral distribution of the field @xmath41 is defined as : @xmath42 which leads to : @xmath43 in a linear approximation framework where @xmath44 does not depend on the amplitude of the field , the resolution of the maxwell equations gives that the propagation of the field can be written as follows @xcite : @xmath45}d\omega,\ ] ] with @xmath46 , @xmath47 . in the case of a dilute medium , the complex refractive index is given by @xmath48 . expanding the exponential term @xmath49 $ ] in taylor series up to the second order , we arrive at : @xmath50 which leads to : @xmath51 where @xmath52 is the initial thz pulse . we then get : @xmath53 where we have used the fact that the time derivative of a fourier transform ( @xmath54 ) of a function @xmath55 is given by : @xmath56 = -i\omega ft[f(t)]$ ] . the thz electric field can finally be written as follows @xmath57 where @xmath58 is a positive scalar factor depending upon the propagation coordinate @xmath26 . the second term of the right hand side of eq . ( [ eq : a10 ] ) is the fid electric field emitted by the molecules of the sample . this contribution is used in the detection process as described above . from the experimental point of view , we point out that the fid manifests itself as recurrent thz echos launched by the molecular sample after its interaction with the thz pulse . it originates from transient orientation revivals of molecules inducing , under field - free conditions , a non - zero dipole . another way of interpreting this phenomenon is to consider that the thz pulse experiences spectral shaping during its propagation . absorption related to @xmath59 transitions produces periodic holes in the spectrum with a spectral separation in angular frequency @xmath60 , with @xmath61 the rotational constant ( neglecting the centrifugal distortion ) . a periodic modulation in the frequency domain leads to periodic replica in the time domain every @xmath62 . we emphasize that the first term of eq . ( [ eq : a10 ] ) does not reflect the absorption experienced by the incident pulse because of the use of first order expansion . we describe in this section the model used in the numerical computations to study the control of molecular orientation of the symmetric top molecule of methyl - iodide ch@xmath3i . we consider the rotational dynamics of this molecular system , which is assumed to be in its ground vibronic state , in interaction with a linear polarized thz field . the hamiltonian of the system can be written as : @xmath63 where @xmath64 and @xmath65 describe respectively the field - free hamiltonian and the interaction with the laser field . the hamiltonian of the molecular system is given by @xcite : @xmath66 where @xmath67 is the angular momentum operator and @xmath68 the component of @xmath67 along the body - fixed @xmath69- axis defined by the c - i bond . the energy eigenvalues @xmath70 of the operator @xmath71 in the wigner basis @xmath72 for a prolate symmetric top can be expressed as follows : @xmath73 where @xmath74 and @xmath75 are the rotational constants . the states @xmath72 are the eigenstates of the square of the angular momentum operator @xmath76 and of its projections , @xmath77 and @xmath78 on the body - fixed @xmath69- axis and space - fixed @xmath25- axis , respectively @xcite . the molecular parameters of the ch@xmath3i molecule used in the numerical computations are given in tab . [ tab:01 ] @xcite . .values of the rotational and of the centrifugal constants of the ch@xmath3i molecule used in the numerical computations . [ cols="<,<",options="header " , ] the interaction between the molecular system and the external electromagnetic field reads : @xmath79 where the function @xmath14 represents here the amplitude of the thz electric field . we neglect in this paper the effect of the polarizability components since the maximum intensity of the electric field remains moderate . the units used are atomic units unless otherwise specified . at room temperature , the system is described by a density matrix @xmath80 whose dynamics is governed by the liouville - von neumann equation @xcite : @xmath81 , \label{eq:05}\ ] ] where the initial condition @xmath82 is given by the canonical density operator at thermal equilibrium @xmath83 where @xmath84 is the partition function , with @xmath85 the temperature fixed to @xmath86 k and @xmath87 the boltzmann constant . the degree of orientation of the molecular system is given by the expectation value : @xmath88 . \label{eq:07}\ ] ] furthermore , in order to simulate more realistic experimental conditions ( see sec . [ sec4 ] ) , we add to the model system the dissipative effects due to molecular collisions @xcite . to limit the complexity of the numerical computations , we consider in this paper the effective approach proposed in @xcite to account for coherence relaxation described by the time @xmath89 , where @xmath90 is the pressure of the sample . we approximate the decoherence by an exponential decay such that the final degree of orientation is given by : @xmath91 where the non - dissipative orientation , @xmath92 , is computed from the liouville - von neumann equation ( [ eq:05 ] ) . in this paragraph , we first investigate theoretically the degree of orientation that can be achieved with a thz laser pulse in the experimental conditions of the set - up . the hilbert space is spanned by the wigner s functions @xmath93 , with @xmath94 , @xmath95 and @xmath96 . @xmath97 and @xmath98 being good quantum numbers , the hamiltonian of the system only depends on the angle @xmath33 . numerically , we consider a finite dimensional hilbert space with @xmath99 , @xmath100 . from a physical point of view , this reduction can be justified by the fact that the thz excitation only transfers a finite amount of energy to the system , which thus stays in a finite dimensional subspace . the control pulse can be approximated by a set of hermite polynomials as follows : @xmath101\ ] ] where @xmath102 , and @xmath103 stands for the hermite polynomials of order @xmath13 , @xmath104 and @xmath105 . the parameter @xmath106 is given by the relation @xmath107 with @xmath108 ps . the reasonable match between the theoretical thz pulse and the experimental one shown in fig . [ fig2 ] justifies the choice made for the electric field in eq . ( [ eqherm ] ) . note that the peak - to - peak amplitude is experimentally of 9.4 kv @xmath4 in this case . [ ht ] ) , thz pulses at delay zero . the amplitude @xmath109 has been adjusted to get the best match between the two pulses.,title="fig:",width=302 ] we start the analysis of the dynamics by giving a global picture of the time evolution of the molecular orientation as displayed in fig . [ fig3 ] , where the relaxation effects are taken into account with @xmath110 ps atm @xcite . the pressure of the cell is @xmath111 . here , we fix the amplitude @xmath109 to 100 kv @xmath4 , which is the maximum experimental available amplitude . revivals are observed at times multiples of the rotational period . the maximum of orientation is respectively of the order of @xmath112 and @xmath113 for the first and second revivals . note that this maximum is larger than @xmath114 in the non - dissipative case . [ ht ] i molecule after an excitation with a @xmath115 pulse centered at @xmath116 . the small insert displays the numerical evolution of the maximum degree of orientation at delay zero ( blue , circle ) and for the first ( red , rectangle ) and second ( yellow , diamond ) revivals as a function of the amplitude @xmath109 of the thz field.,title="fig:",width=302 ] in fig . [ fig3 ] , we also study how the amplitude of the initial thz field affects the behavior of the orientation dynamics . due to the low intensity of the field , we observe that the maximum values of each transient evolve linearly with respect to the amplitude . the effect of the pulse duration on the degree of orientation is displayed in fig . [ fig4new ] . more precisely , we consider the role of the parameter @xmath117 as defined in eq . ( [ eqherm ] ) . note that the overall structure of the field is not modified by this parameter , the field is only compressed or extended in time . this modification corresponds to the currently available shaping techniques of thz pulses . in fig . [ fig4new ] , we observe a nonlinear behavior of the degree of orientation , which is maximum for the two revivals for a value of @xmath117 of the order of 2 ps . [ ht ] for the delay zero ( blue - circle ) and the first ( orange - square ) and the second ( yellow - diamond ) revivals as a function of the parameter @xmath117 of the pulse given by eq . ( [ eqherm ] ) . in the experiment , the parameter @xmath117 is of the order of 1 ps.,title="fig:",width=302 ] as shown in sec . [ sec2p ] , the detection process is sensitive to the time derivative of the degree of orientation , i.e. @xmath118}{dt}$ ] and not directly to the orientation . in fig . [ var_1tr ] , we study the effect of the amplitude of the field on the first revival . we observe that the shape of the transient does not change when the amplitude is varied . similar results are obtained for the second revival . [ ht ] of the first revival as a function of the peak amplitude of the electric field which is set to 20 , 40 , 60 , 80 and 100 kv @xmath4 . the @xmath119 parameter is set to 1 kv @xmath4 ps.,title="fig:",width=302 ] after this complete theoretical description of the orientation dynamics , the goal is now to make a full comparison of the experimental and theoretical results . a scaling factor and a shift parameter along the vertical axis are determined to get the best match between the two sets of data . as shown in fig . [ fig : figs ] , the numerical simulation reproduces quite well the experimental signal for the two first transients . the signal is due to a large number of molecules within the volume of the sample , each molecule being excited by a field of different amplitude . the fact that the shape of the revivals does not depend on the intensity of the electric field explains the good agreement observed in fig . [ fig : figs ] . the signal is too weak after the second revival to pursue this comparison . is set to 5.75 kv @xmath4 ps . ] in this article we have investigated the orientation of a symmetric top molecule , namely ch@xmath3i . we have shown the efficiency of the full - optical ultrafast set - up resolved in time to generate thz pulses and to produce molecular orientation . we provide a detailed description of the experimental set - up used and a complete numerical study of the corresponding dynamics . the analysis of the detection process shows that the degree of orientation is indirectly measured via the time derivative of the expectation value of @xmath120 . the theoretical model reproduces accurately the experimental results up to the second orientation revival . additional numerical simulations reveal that the orientation dynamics induced by this thz pulse is qualitatively similar for the linear molecule , ocs . it will be interesting to consider also asymmetric top molecules which have a more complex and non - periodic field - free evolution than linear or symmetric top molecules . the results of this work can be viewed as an important step forward for the control of molecular orientation . the good match between theory and experiment will allow us to explore the efficiency of more complex strategies using for instance a pre - alignment by a laser field . such approaches are necessary to increase the degree of orientation and reach efficiencies where molecular orientation could be useful in practice . + + * acknowledgment * + d. sugny acknowledges the support from the anr - dfg research programs explosys ( anr-14-ce35 - 0013 - 01 ) and coqs ( anr-15-ce30 - 0023 - 01 ) . this work was supported by the conseil rgional de bourgogne under the photcom pari program as well as the labex action program ( anr-11-labx-01 - 01 ) and the coconics program ( contract no . anr-13-bs08 - 0013 ) . this work has been done with the support of the technical university of munich , institute for advanced study , funded by the german excellence initiative and the european union seventh framework programme under grant agreement 291763 .	we investigate experimentally and numerically the field - free orientation of the symmetric top molecule of methyl - iodide at high temperature using a terahertz radiation generated by a plasma induced by a two - color laser beam . the degree of orientation is measured from the free - induction decay emitted by the sample . the observed experimental signal is reproduced with a good accuracy by numerical simulations .
gaussian process classifiers are a very effective family of non - parametric methods for supervised classification @xcite . in the binary case , the class label @xmath0 associated to each data instance @xmath1 is assumed to depend on the sign of a function @xmath2 which is modeled using a gaussian process prior . given some data @xmath3 , learning is performed by computing a posterior distribution for @xmath2 . nevertheless , the computation of such a posterior distribution is intractable and it must be approximated using methods for approximate inference @xcite . a practical disadvantage is that the cost of most of these methods scales like @xmath4 , where @xmath5 is the number of training instances . this limits the applicability of gaussian process classifiers to small datasets with a few data instances at most . recent advances on gaussian process classification have led to sparse methods of approximate inference that reduce the training cost of these classifiers . sparse methods introduce @xmath6 inducing points or pseudoinputs , whose location is determined during the training process , leading to a training cost that is @xmath7 @xcite . a notable approach combines in @xcite the sparse approximation suggested in @xcite with stochastic variational inference @xcite . this allows to learn the posterior for @xmath2 and the hyper - parameters ( inducing points , length - scales , amplitudes and noise ) using stochastic gradient ascent . the consequence is that the training cost is @xmath8 , which does not depend on the number of instances @xmath5 . similarly , in a recent work , expectation propagation ( ep ) @xcite is considered as an alternative to stochastic variational inference for training these classifiers @xcite . that work shows ( i ) that stochastic gradients can also be used to learn the hyper - parameters in ep , and ( ii ) that ep performs similarly to the variational approach , but does not require one - dimensional quadratures . a disadvantage of the approach described in @xcite is that the memory requirements scale like @xmath9 since ep stores in memory @xmath10 parameters for each data instance . this is a severe limitation when dealing with very large datasets with millions of instances and complex models with many inducing points . to reduce the memory cost , we investigate in this extended abstract , as an alternative to ep , the use of stochastic propagation ( sep ) @xcite . unlike ep , sep only stores a single global approximate factor for the complete likelihood of the model , leading to a memory cost that scales like @xmath11 . we now explain the method for gaussian process classification described in @xcite . consider @xmath12 the observed labels . let @xmath13 be a matrix with the observed data . the assumed labeling rule is @xmath14 , where @xmath15 is a non - linear function following a zero mean gaussian process with covariance function @xmath16 , and @xmath17 is standard normal noise that accounts for mislabeled data . let @xmath18 be the matrix of inducing points ( _ i.e. _ , virtual data that specify how @xmath2 varies ) . let @xmath19 and @xmath20 be the vectors of @xmath2 values associated to @xmath21 and @xmath22 , respectively . the posterior of @xmath23 is approximated as @xmath24 , with @xmath25 a gaussian that approximates @xmath26 , _ i.e. _ , the posterior of the values associated to @xmath22 . to get @xmath25 , first the full independent training conditional approximation ( fitc ) @xcite of @xmath27 is employed to approximate @xmath26 and to reduce the training cost from @xmath4 to @xmath7 : @xmath28 where @xmath29 , @xmath30 and @xmath31 , with @xmath32 , @xmath33 , @xmath34 and @xmath35 is the marginal likelihood . furthermore , @xmath36 is a matrix with the prior covariances among the entries in @xmath37 , @xmath38 is a row vector with the prior covariances between @xmath39 and @xmath37 and @xmath40 is the prior variance of @xmath39 . finally , @xmath41 denotes the p.d.f of a gaussian distribution with mean vector equal to @xmath42 and covariance matrix equal to @xmath43 . next , the r.h.s . of ( [ eq : posterior ] ) is approximated in @xcite via expectation propagation ( ep ) to obtain @xmath25 . for this , each non - gaussian factor @xmath44 is replaced by a corresponding un - normalized gaussian approximate factor @xmath45 . that is , @xmath46 , where @xmath47 is a @xmath48 dimensional vector , and @xmath49 , @xmath50 and @xmath51 are parameters estimated by ep so that @xmath44 is similar to @xmath45 in regions of high posterior probability as estimated by @xmath52 . namely , @xmath53 , where @xmath54 is the kullback leibler divergence . we note that each @xmath45 has a one - rank precision matrix and hence only @xmath10 parameters need to be stored per each @xmath45 . the posterior approximation @xmath25 is obtained by replacing in the r.h.s . of ( [ eq : posterior ] ) each exact factor @xmath44 with the corresponding @xmath45 . namely , @xmath55 , where @xmath56 is a constant that approximates @xmath35 , which can be maximized for finding good hyper - parameters via type - ii maximum likelihood @xcite . finally , since all factors in @xmath25 are gaussian , @xmath25 is a multivariate gaussian . in order for gaussian process classification to work well , hyper - parameters and inducing points must be learned from the data . previously , this was infeasible on big datasets using ep . in @xcite the gradient of @xmath57 w.r.t @xmath58 ( _ i.e. _ , a parameter of the covariance function @xmath59 or a component of @xmath22 ) is : @xmath60 where @xmath61 and @xmath62 are the expected sufficient statistics under @xmath25 and @xmath63 , respectively , @xmath64 are the natural parameters of @xmath63 , and @xmath65 is the normalization constant of @xmath66 . we note that ( [ eq : gradient ] ) has a sum across the data . this enables using stochastic gradient ascent for hyper - parameter learning . a batch iteration of ep updates in parallel each @xmath45 . after this , @xmath25 is recomputed and the gradients of @xmath57 with respect to each hyper - parameter are used to update the model hyper - parameters . the ep algorithm in @xcite can also process data using minibatches of size @xmath67 . in this case , the update of the hyper - parameters and the reconstruction of @xmath25 is done after processing each minibatch . the update of each @xmath45 corresponding to the data contained in the minibatch is also done in parallel . when computing the gradient of the hyper - parameters , the sum in the r.h.s . of ( [ eq : gradient ] ) is replaced by a stochastic approximation , _ i.e. _ , @xmath68 , with @xmath69 the set of indices of the instances of the current minibatch . when using minibatches and stochastic gradients the training cost is @xmath8 the method described in the previous section has the disadvantage that it requires to store in memory @xmath70 parameters for each approximate factor @xmath45 . this leads to a memory cost that scales like @xmath9 . thus , in very big datasets where @xmath5 is of the order of several millions , and in complex models where the number of inducing points @xmath48 may be in the hundreds , this cost can lead to memory problems . to alleviate this , we consider training via stochastic expectation propagation ( sep ) as an alternative to expectation propagation @xcite . sep reduces the memory requirements by a factor of @xmath5 . r0.5 ll + & for each approximate factor @xmath45 to update : + 1.1 : & + 1.2 : & @xmath71 + 2 : & reconstruct @xmath25 : @xmath72 + ll + & set the new global factor @xmath73 to be uniform . + 2 : & for each exact factor @xmath44 to incorporate : + 2.1 : & + 2.2 : & @xmath71 + 2.3 : & @xmath74 + 3 : & reconstruct @xmath25 : @xmath75 + ll + & set @xmath25 to the prior . for each @xmath44 to process : + 1.1 : & + 1.2 : & @xmath71 + 2 : & update @xmath25 : @xmath76 + in sep the likelihood of the model is approximated by a single global gaussian factor @xmath77 , instead of a product of @xmath5 gaussian factors @xmath45 . the idea is that the natural parameters @xmath78 of @xmath77 approximate the sum of the natural parameters @xmath79 of the ep approximate factors @xmath45 . this approximation reduces by a factor of @xmath5 the memory requirements because only the natural parameters @xmath78 of @xmath77 need to be stored in memory , and the size of @xmath78 is dominated by the precision matrix of @xmath77 , which scales like @xmath11 . when sep is used instead of ep for finding @xmath25 some things change . in particular , the computation of the cavity distribution @xmath80 is now replaced by @xmath81 , @xmath82 . furthermore , in the case of the batch learning method described in the previous section , the corresponding approximate factor @xmath45 for each instance is computed as @xmath53 to then set @xmath83 . this is equivalent to adding natural parameters , _ i.e. _ , @xmath84 . in the case of minibatch training with minibatches of size @xmath85 the update is slightly different to account for the fact that we have only processed a small amount of the total data . in this case , @xmath86 , where @xmath69 is a set with the indices of the instances contained in the current minibatch . finally , in sep the computation of the gradients for updating the hyper - parameters is done exactly as in ep . figure [ fig : fig_ep_vs_sep ] compares among ep , sep and adf @xcite when used to update @xmath25 . in the figure training is done in batch mode and the update of the hyper - parameters has been omitted since it is exactly the same in either ep , sep or adf . in adf the cavity distribution @xmath87 is simply the posterior approximation @xmath25 , and when @xmath25 is recomputed , the natural parameters of the approximate factors are simply added to the natural parameters of @xmath25 . adf is a simple baseline in which each data point is _ seen _ by the model several times and hence it underestimates variance @xcite . we evaluate the performance of the model described before when trained using ep , sep and adf . * performance on datasets from the uci repository : * first , we consider 7 datasets from the uci repository . the experimental protocol followed is the same as the one described in @xcite . in these experiments we consider a different number of inducing points @xmath48 . namely , @xmath88 , @xmath89 and @xmath90 of the total training instances and the training of all methods is done in batch mode for 250 iterations . table [ tab : ll_uci ] shows the average negative test log likelihood of each method ( the lower the better ) on the test set . the best method has been highlighted in boldface . we note that sep obtains similar and sometimes even better results than ep . by contrast , adf performs worse , probably because it underestimating the posterior variance . in terms of the average training time all methods are equal . .average negative test log likelihood for each method and average training time in seconds . [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] * performance on big datasets : * we carry out experiments when the model is trained using minibatches . we follow @xcite and consider the mnist dataset , which has 70,000 instances , and the airline delays dataset , which has 2,127,068 data instances ( see @xcite for more details ) . in both cases the test set has 10,000 instances . training is done using minibatches of size 200 , which is equal to the number of inducing points @xmath48 . in the case of the mnist dataset we also report results for batch training ( in the airline dataset batch training is infeasible ) . figure [ fig : stochastic ] shows the avg . negative log likelihood obtained on the test set as a function of training time . in the mnist dataset training using minibatches is much more efficient . furthermore , in both datasets sep performs very similar to ep . however , in these experiments adf provides equivalent results to both sep and ep . furthermore , in the airline dataset both sep and adf provide better results than ep at the early iterations , and improve a simple linear model after just a few seconds . the reason is that , unlike ep , sep and adf do not initialize the approximate factors to be uniform , which has a significant cost in this dataset . r0.5 @xmath91 the results obtained in the large datasets contradict the results obtained in the uci datasets in the sense that adf performs similar to ep . we believe the reason for this is that adf may perform similar to ep only when the model is simple ( small @xmath48 ) and/or when the number of training instances is very large ( large @xmath5 ) . to check that this is the case , we repeat the experiments with the mnist dataset with an increasing number of training instances @xmath5 ( from @xmath92 to @xmath93 ) and with an increasing number of inducing points @xmath48 ( from @xmath94 to @xmath95 ) . the results obtained are shown in figure [ fig : n_vsm ] , which confirms that adf only performs similar to ep in the scenario described . by contrast , sep seems to always perform similar to ep . finally , increasing the model complexity ( @xmath48 ) seems beneficial . stochastic expectation propagation ( sep ) @xcite can reduce the memory cost of the method recently proposed in @xcite to address large scale gaussian process classification . such a method uses expectation propagation ( ep ) for training , which stores @xmath9 parameters in memory , where @xmath6 is some small constant and @xmath5 is the training set size . this cost may be too expensive in the case of very large datasets or complex models . sep reduces the storage resources needed by a factor of @xmath5 , leading to a memory cost that is @xmath11 . furthermore , several experiments show that sep provides similar performance results to those of ep . a simple baseline known as adf may also provide similar results to sep , but only when the number of instances is very large and/or the chosen model is very simple . finally , we note that applying bayesian learning methods at scale makes most sense with large models , and this is precisely the aim of the method described in this extended abstract . * acknowledgments : * yl thanks the schlumberger foundation for her faculty for the future phd fellowship . jmhl acknowledges support from the rafael del pino foundation . ret thanks epsrc grant # s ep / g050821/1 and ep / l000776/1 . tb thanks google for funding his european doctoral fellowship . dhl and jmhl acknowledge support from plan nacional i+d+i , grant tin2013 - 42351-p , and from comunidad autnoma de madrid , grant s2013/ice-2845 casi - cam - cm . dhl is grateful for using the computational resources of _ centro de computacin cientfica _ at universidad autnoma de madrid . j. hensman , a. matthews , and z. ghahramani . scalable variational gaussian process classification . in _ proceedings of the eighteenth international conference on artificial intelligence and statistics _ , 2015 .	a method for large scale gaussian process classification has been recently proposed based on expectation propagation ( ep ) . such a method allows gaussian process classifiers to be trained on very large datasets that were out of the reach of previous deployments of ep and has been shown to be competitive with related techniques based on stochastic variational inference . nevertheless , the memory resources required scale linearly with the dataset size , unlike in variational methods . this is a severe limitation when the number of instances is very large . here we show that this problem is avoided when stochastic ep is used to train the model .
the successful bose - einstein condensation of @xmath0cr atoms @xcite realizes for the first time bose - einstein condensates ( becs ) with significant dipole - dipole interactions . these long - range and anisotropic interactions introduce rich physical effects , as well as new opportunities to control becs . a basic example is how dipole - dipole interactions modify the shape of a trapped bec . in a prolate ( elongated ) dipolar gas with the dipoles polarised along the long axis the net dipolar interaction is attractive , whereas for an oblate ( flattened ) configuration with the dipoles aligned along the short axis the net dipolar interaction is repulsive . as a result , in comparison to _ s_-wave becs ( which we define as systems in which atom - atom scattering is dominated by the _ s_-wave channel ) , a dipolar bec elongates along the direction of an applied polarizing field @xcite . a full theoretical treatment of a trapped bec involves solving the gross - pitaevskii equation ( gpe ) for the condensate wave function . the non - local nature of the mean - field potential describing dipole - dipole interactions means that this task is significantly harder for dipolar becs than for _ s_-wave ones . however , in the limit where the bec contains a large number of atoms the problem of finding the ground state density profile and low - energy dynamics simplifies . in a harmonic trap with oscillator length @xmath1 , a bec containing @xmath2 atoms of mass @xmath3 which have repulsive _ s_-wave interactions characterized by scattering length @xmath4 enters the thomas - fermi ( tf ) regime for large values of the parameter @xmath5 . in the tf regime the zero - point kinetic energy can be ignored in comparison to the interaction and trapping energies and the gross - pitaevskii equation reduces to the equations of superfluid hydrodynamics at @xmath6 . when applied to a trapped @xmath7-wave bec these equations are known to admit a large class of exact analytic solutions @xcite . the tf approximation can also be applied to dipolar becs @xcite . although the resulting superfluid hydrodynamic equations for a dipolar bec contain the non - local dipolar potential , exact solutions can still be found @xcite and we make extensive use of them here . the calculations in this paper are all made within the tf regime . condensates are quantum fluids described by a macroscopic wave function @xmath8 $ ] , where @xmath9 is the condensate density and @xmath10 is the condensate phase . this constrains the velocity field @xmath11 to be curl - free @xmath12 . in an experiment rotation of the condensate can be accomplished by applying a rotating elliptical deformation to the trapping potential @xcite . at low rotation frequencies the elliptical deformation excites low - lying collective modes ( quadrupole etc . ) with quantized angular momentum which may be viewed as surface waves ( and which obey @xmath12 ) . above a certain critical rotation frequency vortices are seen to enter the condensate and these satisfy the @xmath12 condition by having quantized circulation . the hydrodynamic equations for a bec provide a simple and accurate description of the low - lying collective modes . furthermore , they predict these modes become unstable for certain ranges of rotation frequency @xcite . comparison with experiments @xcite and full numerical simulations of the gpe @xcite have clearly shown that the instabilities are the first step in the entry of vortices into the condensate and the formation of a vortex lattice . crucially , the hydrodynamic equations give a clear explanation of why vortex lattice formation in @xmath7-wave becs was only observed to occur at a much greater rotation frequency than that at which they become energetically favorable . it is only at these higher frequencies that the vortex - free condensate becomes dynamically unstable . individual vortices @xcite and vortex lattices @xcite in dipolar condensates have already been studied theoretically . however , a key question that remains is how to make such states in the first place . in this paper we extend the tf approximation for rotating trapped condensates to include dipolar interactions , building on our previous work @xcite . specifically , starting from the hydrodynamic equations of motion we obtain the stationary solutions for a condensate in a rotating elliptical trap and find when they become dynamically unstable to perturbations . this enables us to predict the regimes of stable and unstable motion of a rotating dipolar condensate . for a non - dipolar bec ( in the tf limit ) the transition between stable and unstable motion is independent of the interaction strength , and depends only on the rotation frequency and trap ellipticity in the plane perpendicular to the rotation vector @xcite . we show that for a dipolar bec it is additionally dependent on the strength of the dipolar interactions and also the axial trapping strength . all of these quantities are experimentally tunable and this extends the routes that can be employed to induce instability . meanwhile , the critical rotation frequency at which vortices become energetically favorable @xmath13 is also sensitive to the trap geometry and dipolar interactions @xcite , and means that the formation of a vortex lattice following the instability can not be assumed . using a simple prediction for this frequency , we indicate the regimes in which we expect vortex lattice formation to occur . by considering all of the key and experimentally tunable quantities in the system we outline several accessible routes to generate instability and vortex lattices in dipolar condensates . this paper is structured as follows . in section ii we introduce the mean - field theory and the tf approximation for dipolar becs , in section iii we derive the hydrodynamic equations for a trapped dipolar bec in the rotating frame , and in section iv we obtain the corresponding stationary states and discuss their behaviour . in section v we show how to obtain the dynamical stability of these states to perturbations , and in section vi we employ the results of the previous sections to discuss possible pathways to induce instability in the motion of the bec and discuss the possibility that such instability leads to the formation of a vortex lattice . finally in section vii we conclude our findings and suggest directions for future work . we consider a bec with long - range dipolar atomic interactions , with the dipoles aligned in the @xmath14 direction by an external field . the condensate wave function ( mean - field order parameter ) for the condensate @xmath15 satisfies the gpe which is given by @xcite , @xmath16\psi \label{gpe } , \nonumber \\\end{aligned}\ ] ] where @xmath3 is the atomic mass . the @xmath17-term arises from kinetic energy and @xmath18 is the external confining potential . becs typically feature _ s_-wave atomic interactions which gives rise to a local cubic nonlinearity with coefficient @xmath19 , where @xmath4 is the _ s_-wave scattering length . note that @xmath4 , and therefore @xmath20 , can be experimentally tuned between positive values ( repulsive interactions ) and negative values ( attractive interactions ) by means of a feshbach resonance @xcite . the dipolar interactions lead to a non - local mean - field potential @xmath21 which is given by @xcite , @xmath22 where @xmath23 is the condensate density and @xmath24 is the interaction potential of two dipoles separated by a vector @xmath25 , where @xmath26 is the angle between @xmath25 and the polarization direction , which we take to be the @xmath14-axis . the dipolar becs made to date have featured permanent magnetic dipoles . then , assuming the dipoles to have moment @xmath27 and be aligned in an external magnetic field @xmath28 , the dipolar coupling is @xmath29 @xcite , where @xmath30 is the permeability of free space . alternatively , for dipoles induced by a static electric field @xmath31 , the coupling constant @xmath32 @xcite , where @xmath33 is the static polarizability and @xmath34 is the permittivity of free space . in both cases , the sign and magnitude of @xmath35 can be tuned through the application of a fast - rotating external field @xcite . we will specify the interaction strengths through the parameter @xmath36 which is the ratio of the dipolar interactions to the s - wave interactions @xcite . we take the _ s_-wave interactions to be repulsive , @xmath37 , and so where we discuss negative values of @xmath38 , this corresponds to @xmath39 . we will also limit our analysis to the regime of @xmath40 , where the thomas - fermi approach predicts that non - rotating stationary solutions are robustly stable @xcite . outside of this regime the situation becomes more complicated since the non - rotating system becomes prone to collapse @xcite . we are concerned with a bec confined by an elliptical harmonic trapping potential of the form , @xmath41.\ ] ] in the @xmath42 plane the trap has mean trap frequency @xmath43 and ellipticity @xmath44 . the trap strength in the axial direction , and indeed the geometry of the trap itself , is specified by the trap ratio @xmath45 . when @xmath46 the bec shape will typically be oblate ( flattened ) while for @xmath47 it will typically be prolate ( elongated ) , although for strong enough dipolar interactions the electrostrictive / magnetostrictive effect can cause a bec in an oblate trap to become prolate itself . the time - dependent gpe ( [ gpe ] ) can be reduced to its time - independent form by making the substitution @xmath48 , where @xmath49 is the chemical potential of the system . we employ the tf approximation whereby the kinetic energy of static solutions is taken to be negligible in comparison to the potential and interaction energies . the validity of this approximation in dipolar becs has been discussed elsewhere @xcite . then , the time - independent gpe reduces to , @xmath50 for ease of calculation the dipolar potential @xmath51 can be expressed as , @xmath52 where @xmath53 is a fictitious ` electrostatic ' potential defined by @xcite , @xmath54 this effectively reduces the problem of calculating the dipolar potential ( [ dipole_a ] ) to the calculation of an electrostatic potential of the form ( [ potential ] ) , for which a much larger theoretical body of literature exists . exact solutions of eq . ( 6 ) for @xmath55 , @xmath53 and hence @xmath56 can be obtained for any general parabolic trap , as proven in appendix a of ref in particular , the solutions of @xmath55 take the form @xmath57 where @xmath58 is the central density . remarkably , this is the general inverted parabola density profile familiar from the tf limit of non - dipolar becs . an important distinction , however , is that for the dipolar bec the aspect ratio of the parabolic solution differs from the trap aspect ratio . having introduced the tf model of a dipolar bec we now extend this to include rotation and derive hydrodynamic equations for the rotating system . we consider the rotation to act about the @xmath14-axis , described by the rotation vector @xmath59 where @xmath60 is the rotation frequency and the hamiltonian in the rotating frame is given by , @xmath61 where @xmath62 is the hamiltonian in absence of the rotation and @xmath63 is the quantum mechanical angular momentum operator . using this result with the hamiltonian @xmath62 from eq . ( [ gpe ] ) we obtain @xcite , @xmath64 \psi(\underline{r},t).\label{gpe_rotating_d}\end{aligned}\ ] ] note that all space coordinates @xmath25 are those of the rotating frame and the time independent trapping potential @xmath65 , given by eq . ( [ harmonicv ] ) , is stationary in this frame . momentum coordinates , however , are expressed in the laboratory frame @xcite . we can express the condensate mean field in terms of a density @xmath66 and phase @xmath67 as @xmath68 $ ] , and so that the condensate velocity is @xmath69 . substuting into the time - dependent gpe ( [ gpe_rotating_d ] ) and equating imaginary and real terms leads to the following equations of motion , @xmath70,\\ m \frac{\partial \underline{v}}{\partial t } & = & -\underline{\nabla } \left ( \frac{1}{2 } m \underline{v } \cdot \underline{v } + v(\underline{r } ) + \phi_{dd}(\underline{r } ) \right . \nonumber \\ & & + \left . g \rho - m \underline{v } \cdot \left [ \underline{\omega } \times \underline{r } \right ] \right).\label{motion_eq_d}\end{aligned}\ ] ] in the absence of dipolar interactions ( @xmath71 ) eqs.([cont_eq_d ] ) and ( [ motion_eq_d ] ) are commonly known as the superfluid hydrodynamic equations since they resemble the equation of continuity and the euler equation of motion from dissipationless fluid dynamics . here we have extended them to include dipolar interactions . note that the form of condensate velocity leads to the relation , @xmath72 which immediately reveals that the condensate is irrotational . the exceptional case is when the velocity potential @xmath73 is singular , which arises when a quantized vortex occurs in the system . we now search for stationary solutions of the hydrodynamic eqs.([cont_eq_d ] ) and ( [ motion_eq_d ] ) . these states satisfy the equilibrium conditions , @xmath74 following the approach of recati _ et al_. @xcite we assume the velocity field ansatz , @xmath75 here @xmath33 is a velocity field amplitude that will provide us with a key parameter to parameterise our rotating solutions . note that this is the velocity field in the laboratory frame expressed in the coordinates of the rotating frame , and also that it satisfies the irrotationality condition ( [ irrotationality ] ) . combining eqs . ( [ motion_eq_d ] ) and ( [ velocity_ansatz ] ) we obtain the relation , @xmath76 where the _ effective _ trap frequencies @xmath77 and @xmath78 are given by , @xmath79 @xmath80 the dipolar potential inside an inverted parabola density profile ( [ tf ] ) has been found in refs . @xcite to be , @xmath81 - \frac{\rho}{3 } \nonumber \\\end{aligned}\ ] ] where we have defined the condensate aspect ratios @xmath82 and @xmath83 , and where the coefficients @xmath84 are given by , @xmath85 where @xmath86 , @xmath87 and @xmath88 are integers . note that for the cylindrically symmetric case , where @xmath89 , the integrals @xmath84 evaluate to @xcite , @xmath90 where @xmath91 denotes the gauss hypergeometric function @xcite . thus we can rearrange eq . ( [ mu ] ) to obtain an expression for the density profile , @xmath92 } { g\left(1-\varepsilon_{dd}\right)}. \nonumber \\ \label{rho}\end{aligned}\ ] ] comparing the @xmath93 , @xmath94 and @xmath95 terms in eq . ( [ tf ] ) and eq . ( [ rho ] ) we find three self - consistency relations that define the size and shape of the condensate : @xmath96 where @xmath97 $ ] . furthemore , by inserting eq . ( [ rho ] ) into eq . ( [ cont_eq_d ] ) we find that stationary solutions satisfy the condition , @xmath98 we can now solve eq . ( [ alpha ] ) to give the velocity field amplitude @xmath33 for a given @xmath38 , @xmath99 and trap geometry . in the limit @xmath100 this amplitude is independent of the _ s_-wave interaction strength @xmath20 and the trap ratio @xmath101 . however , in the presence of dipolar interactions the velocity field amplitude becomes dependent on both @xmath20 and @xmath101 . for fixed @xmath38 and trap geometry , eq . ( [ alpha ] ) leads to branches of @xmath33 as a function of rotation frequency @xmath99 . these branches are significantly different between traps that are circular ( @xmath102 ) or elliptical ( @xmath103 ) in the @xmath42 plane , and so we will consider each case in turn . note that we restrict our analysis to the range @xmath104 : for @xmath105 the static solutions can disappear , with the condensate becoming unstable to a centre - of - mass instability @xcite . we first consider the case of a trap with no ellipticity in the @xmath42 plane ( @xmath102 ) . in fig.[fig1](a ) we plot the solutions of eq . ( [ alpha ] ) as a function of rotation frequency @xmath99 for a spherically - symmetric trap @xmath106 and for various values of @xmath38 . before dicussing the specific cases , let us first point out that for each @xmath38 the solutions have the same qualitative structure . up to some critical rotation frequency only one solution exists corresponding to @xmath107 . at this critical point the solution bifurcates , giving two additional solutions for @xmath108 and @xmath109 on top of the original @xmath107 solution . we term this critical frequency the bifurcation frequency @xmath110 . of the static condensate solutions as a function of the trap rotation frequency @xmath99 in a spherically - symmetric trap ( @xmath106 and @xmath102 ) . various values of @xmath38 are presented : @xmath111 , @xmath112 , @xmath113 and @xmath114 . insets illustrate the geometry of the condensate in the @xmath42 plane . ( b ) the bifurcation frequency @xmath110 ( the point at which the solutions of @xmath33 in ( a ) bifurcate ) according to eq . ( [ eq : bif ] ) versus trap ratio @xmath101 . plotted are the results for @xmath111 , @xmath115 , @xmath116 , @xmath112 , @xmath117 , @xmath118 , @xmath119 , @xmath120 , @xmath121 and @xmath114 . in ( a ) and ( b ) @xmath38 increases in the direction of the arrow . , width=283 ] for @xmath100 we regain the results of refs.@xcite with a bifurcation point at @xmath122 and , for @xmath123 , non - zero solutions given by @xmath124 @xcite . the physical significance of the bifurcation frequency has been established for the non - dipolar case and is related to the fact that the system becomes energetically unstable to the spontaneous excitation of quadrupole modes for @xmath125 . in the tf limit , a general surface excitation with angular momentum @xmath126 , where @xmath127 is the tf radius and @xmath128 is the quantized wave number , obeys the classical dispersion relation @xmath129 involving the local harmonic potential @xmath130 evaluated at @xmath127 . consequently , for the non - rotating and non - dipolar bec @xmath131 . meanwhile , inclusion of the rotational term in the hamiltonian ( [ heff ] ) shifts the mode frequency by @xmath132 . then , in the rotating frame , the frequency of the @xmath133 quadrupole surface excitation becomes @xmath134 . the bifurcation frequency thus coincides with the vanishing of the energy of the quadrupolar mode in the rotating frame , and the two additional solutions arise from excitation of the quadrupole mode for @xmath135 . for the non - dipolar bec it is noteworthy that @xmath110 does not depend on the interactions this feature arises because the mode frequencies @xmath136 themselves are independent of @xmath20 . however , in the case of long - range dipolar interactions the potential @xmath137 of eq . ( [ eq : phidd ] ) gives non - local contributions , breaking the simple dependence of the force @xmath138 upon @xmath127 @xcite . thus we expect the resonant condition for exciting the quadrupolar mode , i.e. @xmath139 ( with @xmath133 ) , to change with @xmath38 . in fig . [ fig1](a ) we see that this is the case : as dipole interactions are introduced , our solutions change and the bifurcation point , @xmath110 , moves to lower ( higher ) frequencies for @xmath140 ( @xmath141 ) . note that the parabolic solution still satisfies the hydrodynamic equations providing @xmath142 . outside of this range the parabolic solution may still exist but it is no longer guaranteed to be stable against perturbations . density profiles for @xmath107 have zero ellipticity in the @xmath42 plane . by contrast , the @xmath143 solutions have an elliptical density profile , even though the trap itself has zero ellipticity . this remarkable feature arises due to a spontaneous breaking of the axial rotational symmetry at the bifurcation point . for @xmath108 the condensate is elongated in @xmath144 while for @xmath109 it is elongated in @xmath145 , as illustrated in the insets in fig . [ fig1](a ) . in the absence of dipolar interactions the @xmath143 solutions can be intepreted solely in terms of the effective trapping frequencies @xmath146 and @xmath147 given by eqs.([omax ] ) and ( [ omay ] ) . the introduction of dipolar interactions considerably complicates this picture , since they also modify the shape of the solutions . notably , for @xmath140 the dipolar interactions make the bec more prolate , i.e. , reduce @xmath148 and @xmath149 , while for @xmath141 they make the bec more oblate , i.e. , increase @xmath148 and @xmath149 . in fig . [ fig1](a ) we see that as the dipole interactions are increased the bifurcation point @xmath110 moves to lower frequencies . the bifurcation point can be calculated analytically as follows . first , we note that for @xmath150 the condensate is cylindrically symmetric and @xmath151 . in this case the aspect ratio @xmath152 is determined by the transcendental equation @xcite @xmath153+\frac{\left(\varepsilon_{dd}-1\right)\left(\kappa^2-\gamma^2\right)}{3\kappa^2\varepsilon_{dd}}=0 \nonumber \\\end{aligned}\ ] ] where @xmath154}{2\left(1-\kappa^2\right)}\end{aligned}\ ] ] with @xmath155 $ ] for the prolate case ( @xmath156 ) , and @xmath157 $ ] for the oblate case ( @xmath158 ) . for small @xmath159 , we can calculate the first order corrections to @xmath148 and @xmath149 with respect to @xmath152 from eqs . ( [ kx],[ky ] ) . we can then insert these values in eq . ( [ alpha ] ) and solve for @xmath99 , noting that in the limit @xmath160 we have @xmath161 . thus , we find @xmath162 in fig . [ fig1](b ) we plot @xmath110 [ eq.([omega_b ] ) ] as a function of @xmath101 for various values of @xmath38 . for @xmath100 we find that the bifurcation point remains unaltered at @xmath163 as @xmath164 is changed @xcite . as @xmath38 is increased the value of @xmath101 for which @xmath110 is a minimum changes from a trap shape which is oblate ( @xmath165 ) to prolate ( @xmath166 ) . note that for @xmath167 the minimum bifurcation frequency occurs at @xmath168 , which is over a @xmath169 deviation from the non - dipolar value . for more extreme values of @xmath38 we can expect @xmath110 to deviate even further , although the validity of the inverted parabola tf solution does not necessary hold . for a fixed @xmath101 we also find that as @xmath38 increases the bifurcation frequency decreases monotonically . , as a function of the trap rotation frequency , @xmath99 , for a trap ratio @xmath106 and ellipticity @xmath170 . various values of @xmath38 are presented , @xmath111 , @xmath112 , @xmath113 and @xmath114 , with @xmath38 increasing in the direction of the arrow . insets illustrate the geometry of the condensate in the @xmath42 plane . ( b ) backbending point @xmath110 versus @xmath38 for @xmath170 and @xmath171 ( solid curve ) , @xmath172 ( long dashed curve ) and @xmath173 ( short dashed curve).[fig2],width=283 ] consider now the effect of finite ellipticity in the @xmath42 plane ( @xmath174 ) . rotating elliptical traps have been created experimentally with laser and magnetic fields @xcite . following the experiment of madison _ et al . _ @xcite we will employ a weak trap ellipticity of @xmath170 . in fig . [ fig2](a ) we have plotted the solutions to eq . ( [ alpha ] ) for various values of @xmath38 in a @xmath106 trap . as predicted for non - dipolar interactions @xcite the solutions become heavily modified for @xmath103 . there exists an upper branch of @xmath108 solutions which exists over the whole range of @xmath99 , and a lower branch of @xmath109 solutions which back - bends and is double - valued . we term the frequency at which the lower branch back - bends to be the back - bending frequency @xmath110 . the bifurcation frequency in non - elliptical traps can be regarded as the limiting case of the back - bending frequency , with the differing nonclamenture employed to emphasise the different structure of the solutions at this point . however , for convenience we will employ the same parameter for both , @xmath110 . no @xmath107 solution exists ( for any non - zero @xmath99 ) . in the absence of dipolar interactions the effect of increasing the trap ellipticity is to _ increase _ the back - bending frequency @xmath110 . turning on the dipolar interactions , as in the case of @xmath102 , _ reduces _ @xmath110 for @xmath175 , and _ increases _ @xmath110 for @xmath176 . this is more clearly seen in fig . [ fig2](b ) where @xmath110 is plotted versus @xmath38 for various values of the trap ratio @xmath101 . also , as in the @xmath102 case , increasing @xmath38 decreases both @xmath148 and @xmath149 , i.e. the bec becomes more prolate . importantly , the back - bending of the lower branch can introduce an instability . consider the bec to be on the lower branch at some fixed rotation frequency @xmath99 . now consider decreasing @xmath38 . the back - bending frequency @xmath110 increases and at some point can exceed @xmath99 . in other words , the static solution of the bec suddenly disappears and the bec finds itself in an unstable state . we will see in section vi that this type of instability can also be induced by variations in @xmath101 and @xmath44 . although the solutions derived above are static solutions in the rotating frame they are not necessarily stable , and so in this section we analyze their dynamical stability . consider small perturbations in the bec density and phase of the form @xmath177 and @xmath178 . then , by linearizing the hydrodynamic equations eqs . ( [ cont_eq_d ] , [ motion_eq_d ] ) , the dynamics of such perturbations can be described as , @xmath179 = -\left[\begin{array}{cc } \underline{v}_c \cdot \nabla & g\left(1+\varepsilon_{dd}k\right)/m \\ \nabla \cdot \rho_0 \nabla & \left[\left(\nabla \cdot \underline{v}\right)+\underline{v}_c \cdot \nabla \right ] \\ \end{array } \right ] \left[\begin{array}{c } \delta s \\ \delta \rho \\ \end{array } \right ] \end{aligned}\ ] ] where @xmath180 and the integral operator @xmath181 is defined as @xmath182 the integral in the above expression is carried out over the domain where @xmath183 , that is , the general ellipsoidal domain with radii @xmath184 of the unperturbed condensate . extending the integration domain to the region where @xmath185 adds higher order effects since it is exactly in this domain that @xmath186 . to investigate the stability of the bec we look for eigenfunctions and eigenvalues of the operator ( [ stability ] ) : dynamical instability arises when one or more eigenvalues @xmath187 possess a positive real part . the size of the real eigenvalues dictates the rate at which the instability grows . note that the imaginary eigenvalues of eq . ( [ stability ] ) relate to stable collective modes of the system @xcite , e.g. sloshing and breathing , and have been analysed elsewhere for dipolar becs @xcite . in order to find such eigenfunctions we follow refs . @xcite and consider a polynomial ansatz for the perturbations in the coordinates @xmath188 and @xmath14 , of total degree @xmath2 . all operators in ( [ stability ] ) , acting on polynomials of degree @xmath2 , result in polynomials of ( at most ) the same degree , including the operator @xmath181 . this latter fact was known to 19th century astrophysicists who calculated the gravitational potential of a heterogeneous ellipsoid with polynomial density @xcite . the integral appearing in eq . ( [ koperator ] ) is exactly equivalent to such a potential . a more recent paper by levin and muratov summarises these results and presents a more manageable expression for the resulting potential @xcite . hence , using these results the operator @xmath181 can be evaluated for a general polynomial density perturbation @xmath189 , with @xmath190 and @xmath191 being non - negative integers and @xmath192 . therefore , the perturbation evolution operator ( [ stability ] ) can be rewritten as a scalar matrix operator , acting on vectors of polynomial coefficients , for which finding eigenvectors and eigenvalues is a trivial computational task . ) ( solid curves ) for the upper - branch solutions of @xmath33 as a function of @xmath99 . we assume @xmath170 , @xmath106 and @xmath193 , and present various dipolar strengths @xmath111 , @xmath112 , @xmath113 and @xmath114 , with @xmath38 increasing in the direction of the arrow . the inset shows the full region of dynamical instability in the @xmath194 plane for @xmath100 . the narrow regions have negligible effect and so we only consider the main instability region ( bounded by the dashed line ) . [ figeigs],width=283 ] using the above approach we determine the real positive eigenvalues of eq . ( [ stability ] ) and thereby predict the regions of dynamical instability of the static solutions . we focus on the case of an elliptical trap since this is the experimentally relevant case . recall the general form of the branch diagram for this case , i.e. fig . [ fig2](a ) . in the @xmath109 half - plane , the static solutions nearest the @xmath107 axis never become dynamically unstable , except for a small region @xmath195 , due to a centre - of - mass instability of the condensate @xcite . the other lower branch solutions are always dynamically unstable and therefore expected to be irrelevant to experiment . thus , we only consider dynamical instability for the upper branch solutions , i.e. the branch in the upper half plane ( where @xmath196 ) . in fig . [ figeigs ] we plot the maximum positive real eigenvalues of the upper branch solutions as a function of @xmath99 for a fixed ellipticity @xmath170 . the maximum polynomial perturbation was set at @xmath193 , since for this ellipticity it was found that higher order perturbations did not alter the region of instability , and such modes are therefore not displayed . for a given @xmath38 and @xmath101 there exists a dynamically unstable region in the @xmath194 plane . an illustrative example is shown in fig . [ figeigs](inset ) for @xmath100 and @xmath106 . the instability region ( shaded ) consists of a series of crescents @xcite . each crescent corresponds to a single value of the polynomial degree @xmath2 , where higher values of @xmath2 add extra crescents from above . at the high frequency end these crescents merge to form a main region of instability , characterised by large eigenvalues . at the low frequency end the crescents become vanishingly thin and are characterised by very small eigenvalues which are at least one order of magnitude smaller than in the main instability region @xcite . as such these regions will only induce instability in the condensate if they are traversed very slowly . this was confirmed by numerical simulations in ref . @xcite where it was shown that the narrow instability regions have negligible effect when ramping @xmath99 at rates greater than @xmath197 . it is unlikely that an experiment could be sufficiently long - lived for these narrow instability regions to play a role . for this reason we will subsequently ignore the narrow regions of instability and define our instability region to be the main region , as bounded by the dashed line in fig . [ figeigs](inset ) . for the experimentally relevant trap ellipticities @xmath198 the unstable region is defined solely by the @xmath193 perturbations . we define the lower bound of the instability region to be @xmath199 ( this corresponds to the dashed line in the inset ) . this is the key parameter to characterise the dynamical instability . as @xmath38 is increased @xmath199 decreases and , accordingly , the unstable range of @xmath99 widens . note that the upper bound of the instability region is defined by the endpoint of the upper branch at @xmath200 . for a non - dipolar bec the static solutions and their stability in the rotating frame depend only on rotation frequency @xmath99 and trap ellipticity @xmath44 . adiabatic changes in @xmath44 and @xmath99 can be employed to evolve the condensate through the static solutions and reach a point of instability . indeed , this has been realized both experimentally @xcite and numerically @xcite , with excellent agreement to the hydrodynamic predictions . for the case of a dipolar bec we have shown in sections [ secstationarystates ] and [ secstability ] that the static solutions and their instability depend additionally on the trap ratio @xmath101 and the interaction parameter @xmath38 . since all of these parameters can be experimentally tuned in time , one can realistically consider each parameter as a distinct route to traverse the parameter space of solutions and induce instability in the system . examples of these routes are presented in fig . [ figroutes ] . specifically , fig . [ figroutes ] shows the static solutions @xmath33 of eq . ( [ alpha ] ) as a function of @xmath99 [ fig . [ figroutes](a ) ] , @xmath44 [ fig . [ figroutes](b ) ] , @xmath201 [ fig . [ figroutes](c ) ] and @xmath101 [ fig . [ figroutes](d ) ] . in each case the remaining three parameters are fixed at @xmath202 , @xmath203 , @xmath204 , and @xmath205 . dynamically unstable solutions are indicated with red circles . grey arrows mark routes towards instability ( the point of onset of instability being marked by an asterisk ) , where the free parameter @xmath99 , @xmath44 , @xmath201 , or @xmath101 is varied adiabatically until either a dynamical instability is reached , or the solution branch backbends and so ceases to exist . for solutions with @xmath206 , the instability is always due to the system becoming dynamically unstable ( dashed arrows ) , whereas for @xmath207 the instability is always due to the solution branch backbending on itself ( solid arrows ) and so ceasing to exist . numerical studies @xcite indicate that these two types of instability involve different dynamics and possibly have distinct experimental signatures . below we describe adiabatic variation of each parameter in more general detail , beginning with the established routes towards instability in which ( i ) @xmath99 and ( ii ) @xmath44 are varied , and then novel routes based on adiabatic changes in ( iii ) @xmath38 and ( iv ) @xmath101 . in each case it is crucial to consider the behaviour of the points of instability , namely the back - bending point @xmath110 and the onset of dynamical instability of the upper branch @xmath199 . , determined from eq.([alpha ] ) . dynamically unstable solutions are marked with red circles . in each of the figures the trap rotation frequency @xmath99 ( a ) , trap ellipticity @xmath44 ( b ) , dipolar interaction strength @xmath201 ( c ) and axial trapping strength @xmath101 ( d ) are varied adiabatically , whilst the remaining parameters remain fixed at @xmath208 , @xmath209 , @xmath210 , and @xmath203 . the adiabatic pathways to instability ( onset marked by red asterisk ) are schematically shown by the dashed and solid arrows . dashed arrows indicate a route towards dynamical instability , whereas solid arrows indicate an instability due to disappearance of the stationary state.[figroutes],width=283 ] versus @xmath110 ( solid curves ) and @xmath199 ( dashed curves ) for @xmath106 and ( i ) @xmath111 , ( ii ) @xmath113 and ( iii ) @xmath114 . ( b ) phase diagram of @xmath38 versus @xmath110 ( solid curves ) and @xmath199 ( dashed curves ) for @xmath170 and ( i ) @xmath171 , ( ii ) @xmath211 and ( iii ) @xmath212 . ( c ) phase diagram of @xmath101 versus @xmath110 ( solid curves ) and @xmath199 ( dashed curves ) for @xmath170 and ( i ) @xmath111 , ( ii ) @xmath112 , ( iii ) @xmath113 and ( iv ) @xmath114 . in each case the solid ( dashed ) arrows depict the routes to instability shown in fig . 4.,width=283 ] * i ) adiabatic introduction of @xmath99 : * the relevant parameter space of @xmath44 and @xmath99 is presented in fig . [ figoboi](a ) , with the instability frequencies @xmath213 ( solid curves ) and @xmath214 ( dashed curves ) indicated . for a bec initially confined to a non - rotating trap with finite ellipticity @xmath44 , as the rotation frequency @xmath99 is increased adiabatically the bec follows the upper branch solution [ fig . [ figroutes](a ) ( dashed arrow ) ] . this particular route traces out a horizontal path in fig.[figoboi](a ) until it reaches @xmath214 , where the stationary solution becomes dynamically unstable . for the specific parameters of fig . [ figroutes](a ) the system becomes unstable at @xmath215 . more generally , fig . [ figoboi](a ) shows that as @xmath38 is increased , @xmath214 is decreased and as such instabilities in the stationary solutions will occur at lower rotation frequencies . at @xmath216 , the curve for @xmath199 displays a sharp kink , arising from the shape of the dynamically unstable region , as shown in fig . [ figeigs](inset ) . * ii ) adiabatic introduction of @xmath44 : * here we begin with a cylindrically symmetric ( @xmath102 ) trap , rotating at a fixed frequency @xmath99 . the trap ellipticity @xmath44 is then increased adiabatically and in the phase diagram of fig.[figoboi](a ) the bec traces out a vertical path starting at @xmath217 . the ensuing dynamics depend on the trap rotation speed relative to @xmath218 : \(a ) for @xmath219 the condensate follows the upper branch of the static solutions shown in fig . [ figroutes](a ) . this branch moves progressively to larger @xmath33 . for @xmath220 the bec remains stable but as @xmath44 is increased further the condensate eventually becomes dynamically unstable . figure [ figoboi](a ) shows that as @xmath38 is increased @xmath214 is decreased and as such the dynamical instability of the stationary solutions occurs at a lower trap ellipticity . \(b ) for @xmath221 the condensate accesses the lower branch solutions nearest the @xmath107 axis . these solutions are always dynamically stable and the criteria for instability is instead determined by whether the solution exists . as @xmath44 is increased the back - bending frequency @xmath213 increases . therefore , when @xmath44 exceeds some critical value the lower branch solutions disappear for the chosen value of rotation frequency @xmath99 . this occurs when @xmath222 . figure [ figoboi](a ) shows that as @xmath38 is increased @xmath213 is decreased and as such instabilities in the system will occur at a higher trap ellipticity . at this point the parabolic condensate density profile no longer represents a stable solution . the particular route indicated in fig . [ figroutes](b ) is included in fig . [ figoboi](a ) as a vertical , solid grey arrow . * iii ) adiabatic change of @xmath38 : * the relevant parameter space of @xmath38 and @xmath99 is shown in fig . 5(b ) for several different trap ratios . consider that we begin from an initial bec in a trap with finite ellipticity @xmath170 and rotation frequency @xmath99 . ( this can be achieved , for example , by increasing @xmath44 from zero at fixed @xmath99 . ) then , by changing @xmath38 adiabatically an instability can be induced in two ways : \(a ) for @xmath223 the condensate follows the upper branch solutions until they become unstable . this route to instability in shown in fig . 4(c ) by the dashed arrow , with the corresponding path in fig . 5(b ) shown by the vertical dashed arrow . thus for @xmath224 the motion remains stable . however , for @xmath225 the upper branch becomes dynamically unstable . in fig . [ figoboi](b ) @xmath226 ( dashed curves ) is plotted for different trap ratios . as can be seen , the stable region of the upper branch becomes smaller as @xmath38 is increased . \(b ) for @xmath227 the condensate follows the lower branch solutions nearest the @xmath107 axis . these solutions are always stable and hence an instability can only be induced when this solution no longer exists , i.e. @xmath228 . figure [ figoboi](b ) shows @xmath229 ( solid curves ) for various trap aspect ratios . as can be seen the back - bending frequency @xmath110 decreases as @xmath38 is increased . thus if @xmath38 is increased the system will remain stable . however if @xmath38 is decreased then the system will become unstable when @xmath230 . * iv ) adiabatic change of @xmath101 : * figure [ figoboi](c ) shows the parameter space of @xmath101 and @xmath99 . consider , again , an initial stable condensate with finite trap rotation frequency @xmath99 and ellipticity @xmath170 . then through adiabatic changes in @xmath101 the condensate can traverse the parameter space and , depending on the initial conditions , the instability can arise in two ways : \(a ) for @xmath231 the condensate exists on the upper branch . it is then relevant to consider the onset of dynamical instability @xmath232 ( dashed curves in fig . [ figoboi](c ) ) . providing @xmath233 the solution remains dynamically stable . however , once @xmath234 the upper branch solutions become unstable . \(b ) for @xmath235 the condensate exists on the lower branch nearest the @xmath107 axis . these solutions are always dynamically stable and instability can only occur when the motion of the back - bending point causes the solution to disappear . this occurs when @xmath231 , with @xmath236 shown in fig . [ figoboi](c ) by solid curves for various dipolar interaction strengths . these two paths to instability are shown in fig . [ figroutes](d ) and are also indicated in fig . [ figoboi](c ) as vertical grey arrows , where the dashed ( solid ) arrow corresponds to the @xmath237 ( @xmath207 ) path . having revealed the points at which a rotating dipolar condensate becomes unstable we will now address the question of whether this instability leads to a vortex lattice . first , let us review the situation for a non - dipolar bec . the presence of vortices in the system becomes energetically favorable when the rotation frequency exceeds a critical frequency @xmath13 . working in the tf limit , with the background density taking the parabolic form ( [ tf ] ) , @xmath13 can be approximated as @xcite , @xmath238 here the condensate is assumed to be circularly symmetric with radius @xmath127 , and @xmath239 is the healing length that characterises the size of the vortex core . for typical condensate parameters @xmath240 . it is observed experimentally , however , that vortex lattice formation occurs at considerably higher frequencies , typically @xmath241 . this difference arises because above @xmath13 the vortex - free solutions remain remarkably stable . it is only once a hydrodynamic instability occurs ( which occurs in the locality of @xmath242 ) that the condensate has a mechanism to deviate from the vortex - free solution and relax into a vortex lattice . another way of visualising this is as follows . above @xmath13 the vortex - free condensate resides in some local energy minimum , while the global minimum represents a vortex or vortex lattice state . since the vortex is a topological defect , there typically exists a considerable energy barrier for a vortex to enter the system . however , the hydrodynamic instabilities offer a route to navigate the bec out of the vortex - free local energy minimum towards the vortex lattice state . ( long dashed red curve ) and @xmath199 ( short dashed curve ) , and the critical rotation frequency for vorticity @xmath13 ( solid curve ) for ( a ) an oblate trap @xmath243 and ( b ) a prolate trap @xmath244 . the instability frequencies are based on a trap with ellipticity @xmath170 while @xmath13 is obtained from eq . ( [ crit_freq ] ) under the assumption of a @xmath0cr bec with @xmath245 atoms and scattering length @xmath246 nm in a circularly symmetric trap with @xmath247hz.[fig5],width=283 ] note that vortex lattice formation occurs via non - trivial dynamics . the initial hydrodynamic instability in the vortex - free state that we have discussed in this paper is only the first step @xcite . for example , if the condensate is on the upper branch of hydrodynamic solutions ( e.g. under adiabatic introduction of @xmath99 ) and undergoes a dynamical instability , this leads to the exponential growth of surface ripples in the condensate @xcite . alternatively , if the condensate is on the lower branch and the static solutions disappear ( e.g. following the introduction of @xmath44 ) the condensate undergoes large and dramatic shape oscillations . in both cases the destabilisation of the vortex - free condensate leads to the nucleation of vortices into the system . a transient turbulent state of vortices and density perturbations then forms , which subsequently relaxes into a vortex lattice configuration @xcite . in the presence of dipolar interactions , however , the critical frequency for a vortex depends crucially on the trap geometry @xmath101 and the strength of the dipolar interactions @xmath38 . following reference @xcite we will make a simple and approximated extension of eq . ( [ crit_freq ] ) to a dipolar bec . we will consider a circularly - symmetric dipolar condensate with radius @xmath248 that satisfies eqs . ( [ kx])-([size ] ) , and insert this into eq . ( [ crit_freq ] ) for the condensate radius . this method still assumes that the size of the vortex is characterised by the s - wave healing length @xmath249 . although one does expect the dipolar interactions to modify the size of the vortex core , it should be noted that eq . ( [ crit_freq ] ) only has logarithmic accuracy and is relatively insensitive to the choice of vortex core length scale . the dominant effect of the dipolar interactions in eq . ( [ crit_freq ] ) comes from the radial size and is accounted for . note also that this expression is for a circularly symmetric system while we are largely concerned with elliptical traps . however we will employ a very weak ellipticity @xmath170 for which we expect the correction to the critical frequency to be correspondingly small . as an example , we take the parameter space of rotation frequency @xmath99 and dipolar interactions @xmath38 . we first consider the behavior in a quite oblate trap with @xmath243 . in fig . [ fig5](a ) we plot the instability frequencies @xmath199 and @xmath110 for this system as a function of the dipolar interactions @xmath38 . depending on the specifics of how this parameter space is traversed , either by adiabatic changes in @xmath99 ( vertical path ) or @xmath38 ( horizontal path ) , the condensate will become unstable when it reaches one of the instability lines ( short and long dashed lines ) . these points of instability decrease weakly with dipolar interactions and have the approximate value @xmath250 . on the same plot we present the critical rotation frequency @xmath13 according to eq . ( [ crit_freq ] ) . in order to calculate this we have assumed a bec of 150,000 @xmath0cr atoms confined within a trap with @xmath247hz . in this oblate system we see that the dipolar interactions lead to a decrease in @xmath13 , as noted in @xcite . this dependence is very weak at this value of @xmath101 , and throughout the range of @xmath38 presented it maintains the approximate value @xmath251 . importantly these results show that when the condensate becomes unstable a vortex / vortex lattice state is energetically favored . as such , we expect that in an oblate dipolar bec a vortex lattice will ultimately form when these instabilities are reached . in fig . [ fig5](b ) , we make a similar plot but for a prolate trap with @xmath244 . the instability frequencies show a somewhat similar behaviour to the oblate case . however , @xmath13 is drastically different , increasing significantly with @xmath38 . we find that this qualitative behaviour occurs consistently in prolate systems , as noted in @xcite . this introduces two regimes depending on the dipolar interactions . for @xmath252 , @xmath253 , and so we expect a vortex / vortex lattice state to form following the instability . however , for @xmath254 we find an intriguing new regime in which @xmath255 . in other words , while the instability in the vortex - free parabolic density profile still occurs , a vortex state is not energetically favorable . the final state of the system is therefore not clear . given that a prolate dipolar bec is dominated by attractive interactions ( since the dipoles lie predominantly in an attractive end - to - end configuration ) one might expect similar behavior to the case of conventional becs with attractive interactions ( @xmath256 ) where the formation of a vortex lattice can also be energetically unfavorable . suggestions for final state of the condensate in this case include centre - of - mass motion and collective oscillations , such as quadrupole modes or higher angular momentum - carrying shape excitations @xcite . however the nature of the true final state in this case is beyond the scope of this work and warrants further investigation . by calculating the static hydrodynamic solutions of a rotating dipolar bec and studying their stability , we have predicted the regimes of stable and unstable motion . in general we find that the backbending or bifurcation frequency @xmath110 decreases with increasing dipolar interactions . in addition , the onset of dynamical instability in the upper branch solutions , @xmath199 , decreases with increasing dipolar interactions . furthermore these frequencies depend on the aspect ratio of the trap . by utilising the novel features of dipolar condensates we detail several routes to traverse the parameter space of static solutions and reach a point of instability . this can be achieved through adiabatic changes in trap rotation frequency @xmath99 , trap ellipticity @xmath44 , dipolar interactions @xmath38 and trap aspect ratio @xmath101 , all of which are experimentally tunable quantities . while the former two methods have been employed for non - dipolar becs , the latter two methods are unique to dipolar becs . in an experiment the latter instabilities would therefore demonstrate the special role played by dipolar interactions . furthermore , unlike for conventional becs with repulsive interactions , the formation of a vortex lattice following a hydrodynamic instability is not always favored and depends sensitively on the shape of the system . for a prolate bec with strong dipolar interactions there exists a regime in which the rotating spheroidal parabolic thomas - fermi density profile is unstable and yet it is energetically unfavorable to form a lattice . other outcomes may then develop , such as a centre - of - mass motion of the system or collective modes with angular momentum . however , for oblate dipolar condensates , as well as prolate condensates with weak dipolar interactions , the presence of vortices is energetically favored at the point of instability and we expect the instability to lead to the formation of a vortex lattice . 99 a. griesmaier , j . werner , s. hensler , j. stuhler and t. pfau , phys . lett . * 94 * , 160401 ( 2005 ) . t. lahaye , t. koch , b. frohlich , m. fattori , j. metz , a. griesmaier , s. giovanazzi and t. pfau , nature ( london ) * 448 * , 672 ( 2007 ) . t. koch , t. lahaye , j. metz , b. frohlich , a. griesmaier and t. pfau , nat . phys . * 4 * , 218 ( 2008 ) . l. santos , g.v . shlyapnikov , p. zoller , and m. lewenstein , phys . rev . lett . * 85 * , 1791 ( 2000 ) . s. yi and l. you , phys . a * 63 * , 053607 ( 2001 ) . c. j. pethick and h. smith , _ bose - einstein condensation in dilute gases _ ( cambridge university press , 2002 ) . l. pitaevskii and s. stringari , _ bose - einstein condensation _ ( oxford , 2003 ) , p183 . p. nozieres and d. pines , _ theory of quantum liquids _ ( westview , new york , 1999 ) . n.g . parker and d. h. j. odell , phys . a * 78 * , 041601(r ) ( 2008 ) . d. h. j. odell , s. giovanazzi and c. eberlein , phys . * 92 * , 250401 ( 2004 ) . c. eberlein , s. giovanazzi and d.h.j . odell , phys . a * 71 * , 033618 ( 2005 ) . k. w. madison , f. chevy , w. wohlleben and j. dalibard , phys . lett . * 84 * , 806 , ( 2000 ) ; k. w. madison , f. chevy , v. bretin and j. dalibard , _ ibid . _ * 86 * 4443 ( 2001 ) . et al . _ , * 88 * , 010405 ( 2002 ) . a. recati , f. zambelli and s. stringari , phys . lett . * 86 * , 377 ( 2001 ) . s. sinha and y. castin , phys . * 87 * , 190402 ( 2001 ) . e. lundh , j .- martikainen and k .- a . suominen , phys . a * 67 * , 063604 ( 2003 ) . n. g. parker , r. m. w. van bijnen and a. m. martin , phys . a * 73 * , 061603(r ) ( 2006 ) . i. corro , n. g. parker and a. m. martin , j. phys . b * 40 * , 3615 ( 2007 ) . r. m. wilson , s. ronen , j. l. bohn , and h. pu , phys . 100 , 245302 ( 2008 ) . n. r. cooper , e. h. rezayi and s. h. simon , phys . lett . * 95 * 200402 ( 2005 ) . j. zhang and h. zhai , phys . rev . * 95 * , 200403 ( 2005 ) . s. komineas and n. r. cooper , phys . a * 75 * , 023623 ( 2007 ) . r. m. w. van bijnen , d. h. j. odell , n. g. parker and a. m. martin , phys . * 98 * , 150401 ( 2007 ) . a. m. martin , n. g. parker , r. m. w. van bijnen , a. dow and d. h. j. odell , las . phys . * 18 * , 322 ( 2008 ) . k. gral , k. rzaewski , and t. pfau , phys . rev . a * 61 * , 051601 ( 2000 ) . s. yi and l. you , phys . a * 61 * , 041604 ( 2000 ) . m. marinescu and l. you , phys . lett . * 81 * , 4596 ( 1998 ) . a. j. leggett , _ quantum liquids : bose - condensation and cooper pairing in condensed matter systems _ ( oxford university press , 2006 ) . a. j. leggett , in _ bose - einstein condensation : from atomic physics to quantum fluids _ , proceedings of the thirteenth physics summer school , edited by c. m. savage and m. p. das , ( world scientific , 2001 ) , p. 1 - 42 . l. d. landau and e. m. lifshitz , _ course of theoretical physics : mechanics _ ( butterworth - heinemann , third edition , 1982 ) l. s. gradshteyn and i. m. ryzhik , eds . , _ table of integrals , series , and products _ ( academic press , san diego , 2000 ) , sixth ed . m. abramowitz and i. stegun , eds . , _ handbook of mathematical functions _ ( dover , new york , 1974 ) . y. castin , in _ coherent matter waves _ , lecture notes of les houches summer school , edited by r. kaiser , c. westbrook and f. david ( springer - verlag , 2001 ) , p. 1 r. m. w. van bijnen , n. g. parker , a. m. martin and d. h. j. odell , preprint ( 2008 ) . n. m. ferrers , quart . j. pure and appl . math * 14 * , 1 ( 1877 ) . f. w. dyson , quart . j. pure and appl . math , * 25 * , 259 ( 1891 ) . m. l. levin and r. z. muratov , astrophys . j. * 166 * , 441 ( 1971 ) . e. lundh , c.j . pethick and h. smith , phys . a * 55 * , 2126 ( 1997 ) . n. k. wilkin , j.m.f . gunn and r.a . smith , phys . 80 * , 2265 ( 1998 ) . b. mottelson , phys . lett . * 83 * , 2695 ( 2000 ) . c. j. pethick and l. pitaevskii , phys . a * 62 * , 033609 ( 2000 ) . n. g. parker and c. s. adams , phys . lett . * 95 * , 145301 ( 2005 ) ; j. phys . b * 39 * , 43 ( 2006 ) . e. lundh , c. j. pethink and h. smith , phys . a * 55 * , 2126 ( 1997 ) .	we present a theoretical analysis of dilute gas bose - einstein condensates with dipolar atomic interactions under rotation in elliptical traps . working in the thomas - fermi limit , we employ the classical hydrodynamic equations to first derive the rotating condensate solutions and then consider their response to perturbations . we thereby map out the regimes of stability and instability for rotating dipolar bose - einstein condensates and in the latter case , discuss the possibility of vortex lattice formation . we employ our results to propose several novel routes to induce vortex lattice formation in a dipolar condensate .
to understand the evolution of galaxies one may attempt to match the observational data by models which describe the global processes star - formation rate , gas infall , gas loss with suitable formulations . by adjusting free parameters , quite good fits can be achieved . however , the number of these free parameters often is uncomfortably large . moreover , this approach may not lead to a unique identification of the dominant physical process , as the persisting g dwarf-`problem ' ( pagel & patchett 1975 ) and the formation of radial abundance gradients ( gtz & kppen 1992 ) illustrate . our chemodynamical approach ( hensler 1987 , theis et al . 1992 , samland et al . 1997 ) tries to describe as precisely as possible the known physical processes present in the interstellar medium ( ism ) and its interaction with the stars . these local processes are coupled with the global dynamics of the gaseous and stellar components , constituting a physical description of the evolution of a galaxy . since it is unrealistic to include all processes in their full complexity , one has to define a sufficiently simple but accurate network of interactions in the ism . our prescription , based on the three - component ism model of mckee & ostriker ( 1977 ) and on the formulations of habe et al . ( 1981 ) and ikeuchi et al . ( 1984 ) , has successfully been coupled with the global dynamics for models of elliptical ( theis et al . 1992 ) and disk galaxies ( samland et al . 1997 ) . another important aspect is the degree of non - linearity of the network which determines the behaviour of the model . since the dependences of the rate coefficients are not well known , some caution is necessary to avoid the appearance of complex behaviour solely due to the mathematical formulation . we can not yet fully settle these questions , but what is needed is a more complete understanding of the behaviour of this type of model and an identification of the crucial processes . in their chemodynamical models theis et al . ( 1992 ) find that most often the star - formation rate varies slowly with time , but under certain conditions it undergoes strong nonlinear oscillations , involving the condensation and evaporation of the cool clouds embedded in the hot intercloud gas . the phases of slow variation are due to an equilibrium caused by the self - regulation of the star - formation rate ( sfr ) whose efficiency is reduced as the massive stars heat the gas by their ionizing continuum radiation . in a partial network with a single gas phase kppen et al . ( 1995 ) show that this equilibrium results in a quadratic dependence of the sfr on gas density independent of what was assumed for the stellar birth function . under realistic conditions this is quickly reached , and it is unconditionally stable , quite insensitive to the rate coefficients used . the present study extends the network of kppen et al . ( 1995 ) to two gas components , clouds and intercloud gas , described in sect . we investigate its behaviour by numerical solution which allows the extraction of analytical conditions and relations . this permits the identification of the origin of the oscillations of the sfr ( sect . 3 ) , and the formulation of a physically consistent description ( sect . 4 ) which leads to the identification of a second equilibrium , namely that between condensation and evaporation of the clouds . in sect . 5 we extend the prescription to the more realistic one by samland et al . ( 1997 ) , having condensation and evaporation occurring simultaneously in a cloud population . we shall consider a somewhat simplified version of the present cde models which captures their characteristic behaviour . as in the full models , there are four components : the hot intercloud gas ( named hereafter ` gas ' , with a mass density @xmath0 ) , the gas in the form of clouds ( ` clouds ' @xmath1 ) , as well as massive stars ( @xmath2 ) , and low mass stars and remnants ( @xmath3 ) . between the components the following interactions are taken into account : star - formation , gas return from dying stars , evaporation of clouds , condensation of gas onto clouds , ( radiative or mechanical ) heating of the gas by massive stars , radiative cooling of the gas . the full network also includes other processes , such as the formation of clouds by compression in supernova shells , dissipation by cloud - cloud collisions . these will not be included in our investigation , because comparison with the results of the complete network showed that they do not essentially determine the type of the system s behaviour . then the time evolution of the mass densities of the components is described by the following equations : @xmath4 throughout the paper , we shall use the units parsec , @xmath5 years , and solar masses . star - formation is described by the stellar birth function used in the form of kppen et al . ( 1995 ) @xmath6 normally we use a quadratic dependence on density ( @xmath7 and @xmath8 ) . the exponential factor involving the temperature @xmath9 of the cloud gas describes what fraction of a cloud is in the form of star forming molecular clumps . the mass returned to the interstellar gas by dying massive stars ( with a mean life - time @xmath10 myr ) is taken to be the fraction @xmath11 of the stellar mass . of all stars born , the fraction @xmath12 is in the form of massive stars . the remaining terms pertain to evaporation of clouds , whose rate coefficient @xmath13 can be a function of densities and temperatures , and condensation of gas onto clouds ( coefficient @xmath14 ) . in the formulations of hensler & burkert ( 1991 ) and theis et al . ( 1992 ) the cloudy medium is composed of clouds which have identical properties ( radius @xmath15 , mass @xmath16 , density @xmath17 ) and which are embedded in the ( hot ) intercloud gas . one assumes pressure equilibrium ( @xmath18 ) , which gives for the volume filling factor of the cloudy medium : @xmath19 with the mean molecular masses @xmath20 , @xmath21 in the two gas phases . for simplicity , we set @xmath22 equal to one proton mass , in all what follows . the true mean density @xmath17 within a single cloud is given by @xmath23 assuming that the clouds are just unstable to the jeans criterion , the mass is determined by @xmath24 with boltzmann s constant @xmath25 and gravitational constant @xmath26 . hence the radius of the cloud is @xmath27 & = & 298.0 \ , \cdot ( t_c/10 ^ 4\,{\rm k})^{1/2 } \cdot ( \rho_c [ { { \rm m}_\odot}/{\rm pc}^3 ] ) ^{-1/2 } \nonumber \end{aligned}\ ] ] depending on its properties , such a cloud evaporates into the surrounding medium due to thermal conduction from the ambient hot gas or the gas condenses onto it . cowie et al . ( 1981 ) give a criterion for this behaviour : if the quantity defined as @xmath28 } \right)^{-1 } \left({n_g [ { \rm cm^{-3 } } ] } \right)^{-1 } \\ & = & 0.01246 \,\ , \cdot ( t_g/10 ^ 7 { \rm k})^2 \cdot ( r_c [ { \rm pc}])^{-1 } \cdot ( g [ { { \rm m}_\odot}/{\rm pc}^3 ] ) ^{-1 } \nonumber \end{aligned}\ ] ] is smaller than 0.03 , condensation occurs , otherwise the clouds evaporate . note that cowie et al . use a slightly different notation : @xmath29 . since the clouds have identical properties , the whole system switches between condensation and evaporation , depending on the criterion . in a further development , samland et al . ( 1996 ) consider a cloud population with a spectrum of masses , and therefore one has at any time clouds that evaporate as well as those condensing . we discuss the implications in sect . [ s : samland ] . the mass loss rate per cloud due to evaporation is from cowie et al . : @xmath30 & = & 2.75 \cdot 10 ^ 4 \cdot t_g^{2.5 } \cdot r_c [ \rm pc ] \\ \vert { \raisebox{0.25ex}{$\stackrel{.}{m}_c$}}\vert \ , [ { { \rm m}_\odot}/{\rm myr } ] & = & 1.38 \cdot 10 ^ 6 \cdot ( t_g/10 ^ 7 { \rm k})^{2.5 } \cdot r_c [ \rm pc ] \end{aligned}\ ] ] for @xmath31 the mass loss rate decreases with increasing @xmath32 . in what follows , we do not consider this additional detail , as it alters the behaviour of the models only slightly . this applies also to the modifications introduced by mckee & begelman ( 1990 ) . from the number density of the clouds @xmath33 one gets the rates for evaporation and condensation @xmath34 we note that with @xmath35 , on the border between evaporation and condensation behaviour , the rates are equal : @xmath36 the changes in the internal energy densities are @xmath37 with the abbreviation @xmath38 in our units . the first term is the heating of the gas by massive stars . because in our treatment of the stars the rate from continuous heating during the life - time of the stars @xmath39 has the same dependence on @xmath2 as that from an explosive stellar death e.g. @xmath40 , processes of both types are included in the heating term : photoionization ( with all stellar ionizing photons being absorbed by the gas ) , deposition of mechanical energy by stellar winds and supernova explosions . all three processes give rate coefficients of the same order @xmath41 . to permit a direct comparison with the model of kppen et al . ( 1995 ) we keep @xmath42 ( for photoionization with an efficiency of @xmath43 for the conversion into thermal energy ) . a higher value gives a smaller self - regulated sfr and thus a longer star - formation time - scale . the second term is radiative cooling with a general cooling function @xmath44 ( cf . theis et al . 1992 ) . in our units , and by noting the use of mass densities instead of number densities : the third and fourth terms are the energy gained by the gas through the addition of the evaporated cloud material whose temperature shall be @xmath46 and the energy lost by the gas which is condensed onto the clouds . the ( usually negligibly small ) fifth term is that part of the gain if the stellar ejecta had gas temperature . we include this merely for convenience of analytical considerations . likewise , the energy density of the cloudy medium is changed @xmath47 by gains from heating by massive stars and losses by radiative cooling , by the losses due to matter locked up into stars ( usually a minor term ) , and by the losses due to evaporating material or by gains from incorporating the condensing gas . the latter temperature is designated as @xmath48 . for simplicity , we consider @xmath49 . the metal masses in the gas @xmath50 , where @xmath51 is the metallicity , and the other components evolve like : @xmath52 the first term in the equation for @xmath53 describes the supply to the ism of metals from the massive stars . it is composed of two parts : the metals that where incorporated into the stars at their birth and are unchanged , and the metals freshly synthesized in the stars . for a whole stellar generation the latter is @xmath54 , with the fraction @xmath55 of the mass locked up into remnants which is here @xmath56 . since the yield @xmath57 ( e.g. kppen & arimoto 1991 ) refers to the entire stellar mass spectrum , the massive stars alone contribute @xmath58 . for a primary element such as oxygen , the yield is constant , except for the ( negligible ) factor @xmath59 which takes into account that the primordial elements are used up with metal enrichment . for a secondary element ( nitrogen ) one has @xmath60 . the abundances of every metal can be scaled to convenient reference values . in this paper , we divide the abundances by the yields , which are taken to be solar . this set of equations ( eqns . [ e : gas ] to [ e : metal.rem ] ) , which includes the switching according to the criterion of cowie et al . ( 1981 ) , had been incorporated into the chemodynamical models to compute the evolution of a closed - box system ( hensler & burkert 1991 , theis et al . 1992 ) , and coupled with 1-dimensional hydrodynamics to study the evolution of spheroidal systems ( theis et al . 1992 ) . it was found that star - formation occurs in two modes : in a continuous way as the consequence of self - regulation ( cf . kppen et al . 1995 ) , but in a certain range of densities it would fluctuate very strongly . these fluctuations actually are regular , large amplitude , non - linear oscillations of the cloud temperature and hence of the star - formation rate ( cf . fig . 7 in theis et al . the period of the order of 100 myrs is longer than the cooling time scale of the cloud gas . this indicates that the strongly damped oscillations which may occur before star - formation reaches the self - regulated mode ( kppen et al . 1995 ) can not be the origin . further inspection reveals that the oscillations show up most strongly in the cloud temperature and the gas density , rather weakly in the gas temperature , and were almost absent in the cloud density . in fig . [ f : full.swing ] we show the track in the ( @xmath61)-plane of a model entering the oscillatory mode and performing a few cycles . starting from a point marked ` a ' , it quickly and rather directly enters a limit cycle . at point ` b ' the system changes over from evaporation to condensation . while just a little amount of gas is condensed , the clouds heat up rapidly to a maximum , then cool along the dot - dashed curve , while much more gas condenses . this continues to point ` c ' where the system changes over to evaporation . cooling in the clouds follows , with initially little gas return which increases thereafter . passing through a minimum temperature , the clouds reach a temperature marked by the horizontal dashed line , along which the system reaches again point ` b ' and switches back to condensation . as will be shown below , the horizontal dashed line is identified as the self - regulated star - formation mode ; the passage through the temperature minimum is merely the transient while the system settles into equilibrium ( kppen et al . 1995 ) . the two dashed curves sloping down to the right delineate the border where the system changes between evaporation and condensation . one notices that points ` b ' and ` c ' do not lie on the same curve . this is because the gas temperature @xmath62 at point ` b ' is about 10 percent lower than at point ` c ' . though this amplitude is rather modest , the strong dependence of the switch criterion on @xmath62 leads to a strong shift of the switch curve in the diagram . but these oscillations in @xmath62 are not essential for the behaviour of the model . the fact that the numerical simulations show the appearance of the oscillations primarily in cloud temperature and gas density indicates that these two variables form an ` inner ' system whose behaviour is governed by the values of the control parameters @xmath63 . a most convenient way to analyze this is plotting the streamlines of the equations in the ( @xmath64)-plane , with the other variables @xmath1 , @xmath2 , and @xmath62 either changing with time or held constant . we find that for the explanation of the fundamental structures , the exact dependences of the rate coefficients on density or temperature as given by mckee & begelman ( 1990 ) or cowie et al . ( 1981 ) are of secondary importance . it is completely sufficient to assume that the rate coefficients for condensation and evaporation are constant ( @xmath65 ) . in the following , we regard this simplified description ; for the criterion of switching between condensation and evaporation , the prescription to compute @xmath32 is kept as given above , but we shall consider the threshold value of 0.03 as another free constant @xmath29 . the resulting streamlines are shown in fig . [ f : simp.all ] , which corresponds to fig . [ f : full.swing ] except for the slightly different position of the limit cycle . one distinguishes the two different regions of evaporation and condensation . in the evaporation regime , all streamlines are always oriented towards increasing gas density , and bunch together to form a funnel centered at constant @xmath66 k ( corresponding to @xmath67 k in fig . [ f : full.swing ] ) . along this feature , the clouds evaporate at constant temperature , until the gas density exceeds a critical value corresponding to point ` b ' in the full model and condensation commences . setting @xmath68 in eqns . [ e : gas ] , [ e : cloud ] , and [ e : energycloud ] and using @xmath69 , one gets the equation for the streamlines : @xmath70 at @xmath71 all streamlines are horizontal , which is merely a feature of the logarithmic representation and gives the impression of the fan structure on the left as the lines bunch together into the funnel . the important horizontal streamline along the funnel is described by the balance between stellar heating and radiative cooling @xmath72 , i.e. the evolution proceeds with the clouds being essentially in thermal equilibrium . as these are the only features , it is evident that regardless of its initial state , the system will always end up in the funnel , along which it evolves at constant cloud temperature towards higher gas densities . in the condensation regime , all streamlines are oriented towards lower gas densities . here , one finds a funnel , marked by the short - dashed curve sloping down from upper right to lower left . all streamlines , whether coming from lower or higher temperatures , merge into this funnel which meets the switch line at @xmath73 and @xmath74 k , and the system is changed back to evaporation corresponding to point ` c ' in fig . [ f : full.swing ] . setting @xmath75 yields the equation for the streamlines @xmath76 figure [ f : simp.low ] depicts the whole condensation regime in the ( @xmath64)-plane . to show the behaviour at large densities , a lower value of @xmath77 k is assumed . apart from a region of horizontal streamlines at @xmath71 , as for the evaporation region , one finds these structures : * at low gas densities , a horizontal streamline exists where @xmath78 , the same condition as for the evaporation funnel . its position agrees closely with the upper short dashed curve in fig . [ f : simp.low ] , which is the locus where one has thermal balance in the clouds @xmath79 . * at large gas densities , a horizontal streamline appears where @xmath80 . the dot - dashed curve ( labeled ` a ' ) shows where the cloud temperature does not change with time . the clouds are not in thermal equilibrium , because the condensation occurs faster than the cooling time , and the clouds are kept at the condensates temperature . * both two parts are connected with a curve , along which the evolution proceeds towards thermal equilibrium in the clouds . where the transition occurs , is essentially determined by the ratio of @xmath81 and @xmath82 . * there is an additional feature , showing up as a clump of streamlines : the intersection of the condensation funnel with the condition @xmath83 at @xmath84 and @xmath85 k ( in fig . [ f : simp.low ] ) forms an attracting node . the evolution of the clouds proceeds in thermal equilibrium and in balance between stellar gas return and condensation . independent of the initial conditions , the system always reaches the condensation funnel , where the clouds cool and the gas density decreases . however , under certain conditions , the system may enter an equilibrium evolution at constant gas density . the overall evolution can thus be pieced together : the two regimes of evaporation and condensation are separated by the curve where the criterion of cowie et al . ( 1981 ) is met exactly ( @xmath86 ) , depicted in fig . [ f : simp.all ] . one may work out how the position of this switching line depends on the parameters , using the prescriptions ( eqns . [ e : filling ] to [ e : sigma ] ) . for a cloud filling factor much smaller than unity , which is the common case , one obtains : @xmath87 the curve s location is independent of the cloud density @xmath1 , and it is shifted towards lower gas densities , if one raises the threshold @xmath29 or lowers the gas temperature @xmath62 , which is the most sensitive parameter . the existence of the funnels makes it relatively easy to predict the conditions under which oscillations occur . from the pattern of the streamlines it is clear that a condition for a closed track in the plane is the intersection of evaporation funnel and switching line should happen at a lower cloud temperature ( and higher gas density ) than that of condensation funnel and switch curve . the conditions for the funnels and the switch line constitute a set of simultaneous non - linear equations which can be solved numerically to yield the curve in the @xmath88-plane , and the positions of the intersections . the stellar density can be estimated from the low density limit @xmath89 ( cf . kppen et al . the geometry of the funnels thus allows a classification of the behaviour , as a function of the ` outer ' parameters @xmath1 and @xmath62 , as shown in fig . [ f : simp.scan ] . furthermore , the difference and ratio of the densities @xmath0 can be used together with the two forms of eqn . [ e : gas ] to estimate the times spent in each phase , the oscillation period and the keying ratio . because these considerations neglect the changing gas temperature , they predict the types of behaviour of the complete network only in a qualitative way . all these investigations together with numerical solutions of the complete system show that oscillations are restricted to a finite range in cloud density , which may be quite narrow ( less than a decade ) and which is rather sensitive to some of the parameters and constants in the model . during a cycle , the following sequence of physical processes takes place : when condensation starts , the clouds are rapidly heated up by the gas condensates , until the condensation funnel is reached . there , the clouds cool radiatively and evolve towards thermal equilibrium . when the switching criterion is met again , evaporation commences . lacking the heating by the condensates , the clouds cool until the new thermal equilibrium is reached . then the on - going evaporation of the clouds increases the gas density , until the condensation criterion is again fulfilled . the cycle is thus essentially driven by the deposition of thermal energy of the condensates into the clouds , which radiate the energy away in the form of line emission . thus the source which makes the cycle possible is the deposition of sufficiently hot material onto the clouds . the temperature @xmath90 of the condensates is an extremely sensitive parameter : oscillations can only be found if the condensing gas remains hot @xmath91 . assumption of a lower temperature greatly reduces the region for the occurrence of oscillations , or can completely suppress them . this is quite easily understood in terms of the condensation funnel ( fig . [ f : simp.low ] ) : the position of its high - density branch is determined by the condensate temperature . any cooling reduces the range in cloud temperatures that the funnel covers and thus the possibility for having a cyclic path between the two regimes is reduced . thus , the prescription of the transfer of energy during the condensation phase is a most critical part of the network . mckee & cowie ( 1977 ) and mckee & begelman ( 1990 ) give expressions for the mass transfer rate , but not for the rate of energy exchange between gas and clouds . therefore one considered in the previous chemodynamical models ( e.g. theis et al . 1992 ) limiting cases , such as that the condensing gas does not cool or cools down to some fixed temperature . forming only one of the many processes of the ism model network , this rather simple recipe was preferred . of course , it is not a physically entirely satisfactory description , even more so since the oscillations were found to be rather sensitive to the exact recipe . a physically consistent prescription can be formulated from noting that mckee & cowie ( 1977 ) as well as mckee & begelman ( 1990 ) derive their evaporation and condensation rates on the basis of a _ stationary _ solution for the gas flow between cloud and intercloud matter , with a fixed temperature profile . this means that any parcel of condensing gas follows the local temperature from the intercloud gas to the cloud interior , and that it arrives in the cloud with the cloud s temperature : @xmath92 . as discussed above , cooling of the condensating material results in a lowering of the high - density part of the condensation funnel ( cf . figs . [ f : simp.all ] and [ f : simp.low ] ) . if now the condensates cool down to the cloud temperature , the condensation funnel is merely a horizontal line , at the same height of the evaporation funnel . because of the opposite directions of the streamlines in the two regimes , the intersection of the funnels which are characterized by thermal balance in the clouds with the switching curve is an attracting node , in which the system will always end up . here , the system evolves with the cloud gas in thermal equilibrium , and without any oscillations . the switching condition implies that @xmath93 ( eqn . [ e : switchbalance ] ) and so the evaporation and condensation terms in eqns . [ e : gas ] , [ e : cloud ] , [ e : energygas ] , and [ e : energycloud ] vanish . usually one finds that @xmath94 , so that one has @xmath95 , and the overall evolution is described by @xmath96 which is nothing but the single - gas - phase system which has been shown to evolve almost exclusively in an equilibrium between star - formation and its inhibition due to the heating of the ambient gas by the massive stars ( kppen et al . 1995 ) . note that in this more general view , the conversion of gaseous matter into remnants is _ independent of the rate coefficients for evaporation and condensation , and even on the actual prescription for the switching_. but how the gas is divided among clouds and intercloud gas , depends on these specifications . the switching of the system between condensation and evaporation gives rise to non - continuous terms in the differential equations . however , since in both regimes the system evolves towards the switching condition , this apparent mathematical difficulty makes the solution instead more simple : it forces the system to _ evolve along the switch line_. this permits to estimate the timescales of the equations : for conditions appropriate to a galactic disk ( @xmath97 m@xmath98/pc@xmath99 ) one has @xmath74 k from the figures shown . the gas density is typically @xmath100 . the cloud filling factor is usually quite small @xmath101 . from eqn . [ e : switch ] , one gets for the temperature of the intercloud gas @xmath102 which is quite insensitive to the actual gas density . the rate coefficients are obtained from eqns . [ e : filling ] to [ e : krate ] @xmath103 this makes these processes much slower than the cooling in the clouds @xmath104 myr , but faster than star - formation @xmath105 myr . for larger heating coefficients than we used here , the star - formation is even slower ( few gyr ) , and the two equilibria influence each other even less . thus , after some transient phase , the evolution proceeds in a _ hierarchy of equilibria _ : the fast cooling in the clouds will establish the self - regulated mode of star - formation ; on a longer timescale the condensation and evaporation will balance the distribution of the gaseous matter among cloud and intercloud gas ; and even more slowly , all gas is consumed to be turned into stellar remnants . how the gas is distributed among the gas and the cloud component , depends on the condensation and evaporation . as stated above , the system evolves along the switching line , which can be formulated as : @xmath106 thermal equilibrium in the clouds implies constant temperature , so the last term vanishes , and with eqn . [ e : energygas ] one gets @xmath107 the changes of the densities can be written in the low density limit of self - regulated star - formation @xmath108 @xmath109 \nonumber \\ { \raisebox{0.25ex}{$\stackrel{.}{c}$ } } & = & -(1-\xi\eta ) \psi - { \raisebox{0.25ex}{$\stackrel{.}{g}$}}\nonumber \end{aligned}\ ] ] where @xmath62 is taken from eqn . [ e : tgas.form ] , and using the abbreviations @xmath110 one notes that the equations have a critical point which obeys the condition @xmath111 , or @xmath112 thus the ratio @xmath113 due to the balance of condensation and evaporation increases only very slightly as the cloud gas is consumed . it is independent of the actual rate coefficients , and only weakly dependent on the switching threshold . these properties are found in the numerical solutions , shown in fig . [ f : gc.sigma ] . after the transient evolution subsides within 1 gyr , which depends on the rate coefficients the system enters the equilibrium , where the above derived dependences are obeyed . this shows that , apart from transient phases , the results of the chemo - dynamical models are quite insensitive to the details of the recipe of cowie et al . ( 1981 ) , for example the correction factor @xmath114 for the influence of magnetic fields . the stability of the system against perturbations of the gas and cloud densities and temperatures are investigated by numerical experiments , adding or removing a substantial ( 50 percent ) portion of the gas present in the system or changing strongly the temperatures . after a transitory phase lasting some 100 myrs , a new equilibrium solution is reached which has the same character as before , i.e. without oscillations . in the transient there may be a few bumps and wiggles , but the system never breaks into oscillations . this is as expected , since the disturbance moves the system in e.g. the @xmath88-plane to another position , from where it always evolves towards one of the funnels to end up in the intersection with the switch line . since the character of the streamlines can not be altered into the pattern responsible for oscillations , the system is also very robust against inflows , outflows , and external heating . in fig . [ f : simp.primary ] we depict the relation of abundance in the cloud component of a primary element with the gas fraction @xmath115 . in the simple model ( closed - box with instantaneous recycling ) the gas metallicity follows @xmath116 with the true yield @xmath57 , shown as the dashed line . for simplicity , we divide the abundances by the yield . the abundance rises more steeply at early times than the simple model , because we allow for a finite life - time of the massive stars which are responsible for the metal enrichment . for comparison , we also show the model where the condensing gas does not cool : the metallicity remains constant during the evaporative phases , separated by a steep rise when the metal - rich gas is mixed to the clouds at the beginning of the condensation phase . if one allows the gas to cool down to cloud temperature before condensing , after a single evaporative intervall , the evolution proceeds in what seems to be the average evolution of the model with oscillations . it is worth emphasizing that for @xmath117 , the evolution of either model is quite close ( within 0.1 dex ) to that of the simple model , i.e. the effective yield is nearly equal to the true yield . in fig . [ f : simp.secondary ] we show the behaviour of the abundance of a secondary element such as nitrogen , by the n / o vs. o / h plot . the simple model predicts a strictly linear relationship , viz . a straight line with slope 1 for the logarithmic values . apart from a systematically lower yield for nitrogen ( by 0.1 dex ) , the numerical models show a quite similar slope . the oscillations are less noticeable here than in the abundances themselves , and in the overall evolution , they make the slope of the n / o - o / h relation steeper . if one allows for cooling of the condensing gas , the slope is very close to unity . thus , in both aspects , the chemical enrichment of the clouds from which the stars are born occurs in the very much same way as in the simple model . this should not be too surprising , because we do consider a closed - box model . moreover , the time - scales for mixing of the freshly - produced metals into the cloud gas are determined by the condensation / evaporation times , which are still short compared to the overall evolution , viz . the gas consumption . the models that treat the clouds as being identical are certainly a simplification . for a more realistic approach , one would consider a cloud population with a distribution in e.g. mass , as is done in the models of samland et al . with such a description of the cloud phase , the cowie et al . criterion implies that all clouds smaller than a certain radius will evaporate , while those larger will grow by condensation . thus , one has at the same time both processes taking place , and for the overall rates one integrates over the cloud spectrum . how this is done and which rates one gets , depends on any further assumptions on the evolution of the cloud population . samland et al . obtain for the solar neighbourhood , coefficients @xmath118 myr@xmath119 and @xmath120 myr@xmath119 which remain quite constant throughout the evolution . such a system settles quickly , within a few condensation time - scales @xmath121 into the balance between condensation and evaporation @xmath122 which is time - independent . the overall gas consumption takes place much more slowly , with @xmath123 . for a linear sfr , this can easily be shown by solving the equations analytically . since one normally has @xmath124 , most of the gaseous matter is in the cloud component , which is slowly converted into stellar remnants . in fig . [ f : gc.trans ] we show the ratio @xmath113 as a function of cloud density for models with @xmath125 , starting with an initial density @xmath126 . the self - regulated star - formation is quickly established as soon as the star - formation time - scale has become larger than the mean life - time of the massive stars ( @xmath127 or @xmath128 ) . as long as the condensation occurs faster than the stellar gas return ( @xmath129 myr@xmath119 ) , the system settles into the condensation / evaporation equilibrium . in models with very low condensation coefficients , the ejecta from the stars formed in the initial period of rather high star - formation are first accumulated in the hot gas , whose mass may remain constant while the clouds continue to be consumed @xmath130 . after a few condensation time - scales the gas condenses following the steady - state condition ( eqn . [ e : steady ] ) . the peak in the curve for the rather extreme model with @xmath131 at @xmath132 marks where the gas and cloud phases are in equilibrium . in this model , about 10 percent of the initial mass is present in the gas component for about a few gyr . in fig . [ f : oxy.trans ] we show the ratio of the metallicities in gas and cloud components , to study the efficiency of the mixing between the two components . comparing with the evolution of the density ratio , one notices that the metallicities become equal as soon as the model reaches the steady - state solution ( eqn . [ e : steady ] ) . the oxygen abundance in gas and clouds as a function of the gas fraction is depicted in fig . [ f : yield ] , compared to the expectation from the simple model . if condensation and evaporation occur faster than star - formation , gas and clouds remain well - mixed , and the model follows the simple model very closely . the metallicity of the clouds is essentially determined by the amount of mixing due to condensation , it does not depend on the value of @xmath13 . on the other hand , the gas metallicity is determined by how much the stellar ejecta are mixed with the low metal material evaporated from the clouds . for low values of @xmath13 the mixing is poor , and the metallicity is as large as the metal abundance of the fresh stellar ejecta which is as large as @xmath133 ( eqn . [ e : metal.gas ] ) . eventually , all models evolve with the true yield . in systems with a large initial density @xmath134 m@xmath98/pc@xmath99 ( shown in fig . [ f : yield.high ] ) the initial sfr @xmath135 is greater than the condensation rate @xmath136 . as a consequence , metals accumulate in the gas , and because of the small but existing reprocessing of enriched cloud matter the metallicity even rises . the condensation of the metal - rich gas occurs only at a more evolved stage , and then the large gas metallicity causes the clouds to have metallicities larger than expected from the simple model . in a galaxy , the total mass of a primary element such as oxygen is proportional to the total number of type ii supernovae that have exploded until the present time , independent of when and where this happened . this makes the total metal mass an essential indicator for the global state of a galaxy , though it is not easy to obtain , as it requires the determination of not only the mean abundances of gas in the various forms and stars , but their mass fractions as well . this would involve different wavelength regions and abundance analysis techniques with the problems associated combining the data . is it necessary to measure all gaseous components , or would the observation of a single phase , e.g. the cloud components being observable with h ii regions , suffice ? what is the magnitude of the errors involved ? in the following , we use our closed - box chemodynamical models to provide some rough answers . since the freshly produced metals are first injected into the warm or hot intercloud gas , observations of the dense cloud medium might miss an important fraction of the metals . in figs . [ f : omassa.frac ] to [ f : omassc.frac ] we show how the distribution of oxygen mass over gas , clouds , and remnants depends on the gas mass fraction @xmath137 , which indicates the chemical age of a system . the metals in chemically very young systems are still mainly in the hot gas . at more advanced stages , the metal rich gas has condensed into the clouds , which then contain most metal mass . in highly evolved systems , the metals are found in the remnants . the shape of the curves in intermediately evolved systems , i.e. around the peak of the contribution from the clouds , is mainly determined by the ratio of the inverse time - scales for condensation and initial star formation @xmath138 . because in the closed - box model most of the star formation occurs during the initial phase , this time - scale determines how much metals are produced and which are released from the intercloud medium by condensation @xmath14 . for @xmath139 m@xmath98pc@xmath140 myr@xmath119 ( fig . [ f : omassa.frac ] ) the metals are hardly to be found in the gas , because of the rapid condensation into clouds . for values smaller than about 0.1 ( fig . [ f : omassc.frac ] ) the metals remain in the gas until rather late , and because of the fast conversion of clouds into stars , they are swiftly locked up into remnants . as the clouds may contain less than a third of all the metals , measurements of the metallicity of the cloud gas only could be rather misleading for the determination of the total metal mass in a galaxy . such a situation arises for low condensation rates or for high initial gas densities , i.e. a centrally rather concentrated proto - galactic cloud . the shapes of the curves are much less strongly influenced by the ratio @xmath141 which we kept constant in the figures : it determines the ratio of metal mass contained in gas and clouds in the late evolution ; thus because one has @xmath142 , the clouds keep a larger portion of the metal mass than the gas . we investigate the behaviour of one - zone chemical evolution models which utilize the chemodynamical prescription for a two - component interstellar medium . this is done by numerical solution , topological analysis of the equations , and identification of the conditions for equilibrium solutions . the self - regulated star - formation found in models with a single gas component ( kppen et al . 1995 ) is not at all disturbed by the additional condensation and evaporation processes . due to its shorter time - scale , it dominates the evolution . the models follow the self - regulated solution very closely , without any oscillations . on the other hand , it is the distribution of the gas among the cloud and intercloud phases which is determined by the mass exchange due to condensation and evaporation . the evolution also follows an equilibrium which is either ruled by the switching condition @xmath143 or by the relation @xmath144 . the mass of gas in the intercloud phase is usually very small compared to that found in clouds . from realistic rate coefficients for condensation and evaporation in a cloud population ( samland et al . 1997 ) one finds that the evolution of the model follows a hierarchy of nested equilibria , which are in ascending order of their quite separate time - scales : thermal equilibrium in the cloud gas , regulation of the sfr by heating of gas due to radiation from massive stars , equilibrium of mass exchange between gas and clouds . this is further embedded into the overall conversion of gas into stellar remnants . during the evolution in the equilibrium between condensation and evaporation @xmath144 , the relation between metallicity and gas mass fraction follows a simple model quite closely . gas and cloud phases are well - mixed , their metallicities are the same . together with the quadratic dependence of the star - formation rate on the cloud density ( cf . kppen et al . 1995 ) , these aspects make up the global behaviour that is typical for chemo - dynamical evolution models . in models with low condensation coefficients and/or high initial cloud densities , the metal - rich gas ejected by the stars formed in the initial period of intense star - formation may remain in the intercloud medium quite long , before condensation into clouds commences and metal enrichment of the clouds occurs . from these models we compute how the total metal mass in a galaxy is distributed among gas , clouds , and stellar remnants and how this changes with the chemical age of the system . for conditions similar to the solar neighbourhood , unevolved gas - rich systems have their metals in both cloud and intercloud gas . in high - density models ( or those with small condensation rates ) the metals may remain in the intercloud medium even in rather evolved , gas - rich galaxies , and those contained in the cloud phase may be a rather portion . work is under way to investigate the properties of full ` chemodynamical ' models with the complete network of interactions and the coupling to the global dynamics . _ acknowledgements : _ we thank m. samland and r. spurzem for valuable discussions . j.k . thanks the deutsche forschungsgemeinschaft for financial support ( project he 1487/13 - 1 ) and the observatoire de strasbourg for its generous hospitality . cowie l.l . , mckee c.f . , ostriker j.p . , 1981 , gtz m. , kppen j. , 1992 , habe a. , ikeuchi s. , tanaka y.d . , 1981 , hensler g. , burkert a. , 1991 , in : proc . elba workshop `` chemical and dynamical evolution of galaxies '' eds . f.matteucci et al . , p.168 ikeuchi s. , habe a. , tanaka y.d . , 1984 , kppen j. , arimoto n. , 1991 , and * 89 * , 420 kppen j. , theis ch . , hensler g. , 1995 , mckee c.f . , begelman m.c . , 1990 , mckee c.f . , cowie l.l . , 1977 , mckee c.f . , ostriker j.p . , 1977 pagel b.e.j . , patchett b.e . , 1975 , samland m. , hensler g. , theis ch . , 1997 , theis ch . , burkert a. , hensler g. , 1992 ,	the network of interactions between hot gas , cool clouds , massive stars , and stellar remnants used in the chemodynamical modeling of the interstellar medium is investigated for the types of its solutions . in a physically consistent formulation for the energy transfer during condensation , oscillations due to a cyclic switching between condensation and evaporation never occur . a closed - box system evolves in a hierarchy of equilibria : thermal balance in the cloud gas , star - formation self - regulated due to heating of the clouds by massive stars , and the balance of condensation and evaporation . except for an initial transitory phase , the evolution of the metallicity follows that of the simple model quite closely . for galaxies with a high initial density , or if the condensation rate is low , the metals produced by the stars may remain stored in the hot gas phase even in evolved systems with low gas fractions . = 10000 # 1#2 aj * # 1 * , # 2 # 1#2 apj * # 1 * , # 2 # 1#2 apj suppl.ser . * # 1 * , # 2 # 1#2 apss * # 1 * , # 2 # 1#2 a&a * # 1 * , # 2 # 1#2 a&as * # 1 * , # 2 # 1#2 araa * # 1 * , # 2 # 1#2 astr.zh . * # 1 * , # 2 # 1#2 iau symp . * # 1 * , # 2 # 1#2 mnras * # 1 * , # 2 # 1#2 pasj * # 1 * , # 2 # 1#2 pasp * # 1 * , # 2 # 1#2 proc.nat.acad.sci . * # 1 * , # 2 # 1#2 rev.mex.a&a * # 1 * , # 2 _ g _ g _ c _ c _ c _ c _ g _ s _ r
two successful routes that are being actually followed to produce magnetic organic materials @xcite are the addition of magnetic atoms @xcite and the use of polyradicals @xcite . in particular , carbon - based nickel compounds that show spontaneous field - dependent magnetization and hysteresis at room temperature , have been recently synthesized @xcite . moreover , the combination of two radical modules with different spins has allowed the obtaining of organic polymers with ferro- or antiferromagnetic ordering @xcite . research on molecules containing polyradicals goes back to the early nineties @xcite and has produced a variety of results as , for example , the synthesis of high spin organic molecules . in some of these molecules the failure of hund s rule has been demonstrated @xcite . on the other hand , experimental and theoretical evidence has been recently presented indicating that 5-dehydro - m - xylylene or dmx was the first example of an organic tri - radical with an open - shell doublet ground - state@xcite . both methods share a common strategy : the use of ingredients ( either radicals or atoms ) that provide a finite spin . in this work we follow a different approach . specifically , we predict the existence of spin polarized organic molecules derived from non magnetic @xmath0-conjugated polycyclic aromatic hydrocarbons ( pahs ) by selective hydrogenation of their peripheral c atoms . high hydrogenation of pahs has been proposed as a method for hydrogen storage @xcite . more recently , the feasibility of double hydrogenation of those compounds has been investigated theoretically @xcite . our work is inspired upon lieb s theorem for bipartite lattices that shows the appearance of magnetism whenever they are unbalanced @xcite . according to lieb , if a nearest neighbor model with a local on - site interaction is applicable to a bipartite lattice , the spin multiplicity of the ground state is @xmath1 , where @xmath2 and @xmath3 are the number of atoms in each sublattice . most pahs are alternant hydrocarbons where carbon atoms can be separated into two disjoint subsets so that an atom in one set only has neighbors in the other set ( figs . 1 and 2 show a colored version of the partition ) . the same theorem has been used to support the existence of magnetism in graphene ribbons and islands @xcite . all work we know is based on single - determinantal methods , i.e. , on a more or less sophisticated form of self - consistent - field ( scf ) calculation . let us remark that being @xmath0-orbital magnetism a direct result of the strong correlation among @xmath0-electrons , only methods designed explicitly to catch these effects ( like cisd and mcscf , used in our work ) can help to resolve the doubts regarding the appearance of magnetism in graphite - derived systems . according to its symmetry group ) , 1,3,5,7,9,11-hexahydrocoronene ( b , hereafter referred to as @xmath4 ) and planar 1,4,5,6,7,10-hexahydrocorannulene ( c , hereafter referred to as @xmath5 ) . saturated carbon atoms are represented by black symbols while dark gray ( magenta ) and light gray symbols are used to distinguish carbon atoms belonging to different sublattices . corannulene is a non - alternant hydrocarbon , that is , a frustrated cluster of carbon atoms ( note the fully magenta bond between two magenta atoms ) . ] the rest of the paper is organized as follows . the _ ab initio _ methods ( both mono- and multi - determinantal ) used in this work are discussed in some detail in section ii , while the results obtained with those methods are reported and discussed in section iii . section iv in turn is devoted to the analysis of the _ ab initio _ results by means of model hamiltonians , in particular the hubbard and the pariser - parr - pople hamiltonians . finally , the conclusions of our work are gathered in section v. ) . as in fig . 1 , black symbols indicate carbon atoms forming only single bonds while dark gray ( magenta ) and light gray symbols denote each of the two sublattices in which carbon atoms can be separated . ] . total energies ( in hartrees ) for atomic hydrogen , molecular hydrogen , coronene c@xmath6h@xmath7 , corannulene c@xmath8h@xmath9 , two molecules obtained from hexahydrogenation of coronene and one derived from hexahydrogenation of corannulene ( in the latter case the results correspond to the planar geometry shown in fig . the results were obtained using three basis sets ( midi , cc - pvdz and cc - pvtz ) , two scf methods ( rhf and rb3lyp ) and one multi - configurational method cisd . the number of occupied ( m ) and empty ( n ) @xmath0 molecular orbitals included in the cisd calculations as well as the number of electrons that fill them ( n ) is indicated as ( m+n , n ) . small stars emphasize the spin multiplicity of the more stable state . [ cols="^,^,^,^,^ " , ] [ table : energy - md ] total energies for the singlet and the relevant multiplet of hydrogenated coronene @xmath10 , @xmath4 and planar hydrogenated corannulene @xmath5 ( a , b and c in fig . 1 ) are reported in table i. it is first noted that whereas the energies obtained with the small basis set midi and those obtained with the already large cc - pvdz , differ in 4 - 6 hartrees ( approximately 0.6% ) , the difference is reduced to 0.1 - 0.3 hartrees ( approximately 0.02% ) when cc - pvdz is replaced by the largest basis used in this work , namely , the cc - pvtz basis set . this indicates that convergence , as far as the basis set is concerned , is rather acceptable . in the case of hexahydrogenated coronene ( briefly 6h - coronene ) , results clearly show that , no matter the method or the basis set used , the ground state of molecule @xmath10 is a septuplet and that of molecule @xmath4 a singlet . we have checked that other spin states lie between those two . in molecule @xmath10 the largest energy difference between the high spin ground state and the singlet occurs for rhf ( 0.23 - 0.26 hartrees ) . this difference is reduced to approximately 0.1 hartrees for rb3lyp , increasing again using the cisd method . on the other hand , all results for @xmath4 conformation show that the singlet is below the septuplet by more than 0.2 hartrees . similar results are obtained for 6h - corannulene , although energy differences are slightly smaller . table i also reports total energy results for atomic and molecular hydrogen , coronene and corannulene that allow the calculation of fragmentation energies ( table ii analysis ) . these are negative relative to atomic hydrogen but not relative to the molecular form . therefore , actual synthesis of the hydrogenated molecules would need sophisticated reaction paths @xcite . we also note that the singlet ground state of @xmath4 hydrogenated coronene is more stable than that of the molecule having a septuplet ground state ( @xmath10 ) . presumably , other forms of 6h - corannulene would also show deeper ground state energies than that of the studied magnetic conformation . note also that hydrogenation of the curved ( stable ) geometry of corannulene ( see fig . 2 ) is slightly less favorable than that of its planar geometry ( compare results for @xmath5 and @xmath11 in table ii ) . anyhow , as in the planar geometry , the quintuplet has a lower energy than the singlet . 3 depicts the total spin densities of the septuplet states ( s=3 ) of 1,4,5,8,9,12-hexahydrocoronene and 1,3,5,7,9,11-hexahydrocoronene ( a and b ) and the quintuplet ( s=2 ) of planar 1,4,5,6,7,10-hexahydrocorannulene . concerning 1,4,5,8,9,12-hexahydrocoronene , the most appealing result is that the spin density is finite only on the carbon atoms of one sublattice . more precisely , spin density is located in the sublattice to which no additional h atoms were attached . this result is highly illustrative allowing some intuition on the reasons for a magnetic ground state : electron - electron repulsion is minimized because each electron avoids sitting at nearest - neighbors distances from the others . however , in 1,3,5,7,9,11-hexahydrocoronene , a molecule with a singlet ground state , the spin is equally spread over the two sublattices implying larger electronic repulsions at the central hexagon . the case of 1,4,5,6,7,10-hexahydrocorannulene is even more interesting as , being a frustrated molecule , at least one bond between atoms of the same sublattice should be present . this is clearly visible in fig . 3 once a sublattice is identified as the sites showing spin density while the rest belong to the other sublattice ( colors in fig . 1 have anticipated this feature ) . we will show later that the model hamiltonian calculations for 1,4,5,6,7,10-hexahydrocorannulene show frustration at the same bond than _ ab initio _ calculations ( compare figs . 3 and 4 ) . having identified the atoms at each sublattice , it is tempting to use the unbalance in the molecule ( @xmath12=4 ) to predict the total spin of the ground state using lieb s formula . the result ( s=2 ) is in perfect agreement with numerical results . this is particularly interesting as in principle lieb s theorem should only work on non - frustrated systems . spin multiplicity of the ground state of a molecule is usually predicted by means of hund rule applied to mo energies obtained by an appropriate method . we have checked that the spin of the ground states of the molecules here investigated is consistent with the degeneracy of the homo that hckel s method gives for the skeleton of c atoms having an unsaturated @xmath0 orbital . this is true not only for 6h - coronene , but also for 6h - corannulene . although the extended hckel s method used by _ ab initio _ codes to initialize the self - consistency process slightly lifts this degeneracy , the homo still appears as a narrow bunch containing a number of orbitals compatible with the spin of the ground states of the three planar molecules depicted in fig . , as in hund rule , such a distribution of molecular orbitals favors high spin ground states through a winning competition of interaction energy gains against kinetic energy losses . let us critically examine the applicability of lieb s theorem as the predicting tool of the multiplicity of the ground state of hydrogenated pahs . the underlying hubbard model ignores that : ( i ) transfer integrals in any realistic system are not limited to nearest neighbors sites , ( ii ) @xmath13orbitals appear around the homo - lumo gap in the same energy interval as @xmath0orbitals , ( iii ) interaction among electrons is not limited to on - site coulomb repulsion . in our opinion , the success of a theorem or rule based on the simplest interacting model comes from its actual capability of describing the correct antiferromagnetic spin - spin correlations between nearest @xmath0 electrons . strong correlation is the basis for the basic correctness of a simplified image in which up and down spins alternate @xcite . even if the spin multiplicity of the ground state is predicted either by hund rule or lieb s theorem , a deeper understanding of underlying correlations calls for a complete numerical solution of simple interacting models . we have analyzed both pariser - parr - pople ( ppp ) model hamiltonian @xcite and the local version of hubbard hamiltonian @xcite , which actually is a particular case of the former . the ppp hamiltonian contains a non - interacting part @xmath14 and a term that incorporates the electron - electron interactions @xmath15 : @xmath16 the non - interacting term is written as , @xmath17 where the operator @xmath18 creates an electron at site @xmath19 with spin @xmath13 , @xmath20 is the energy of carbon @xmath0orbital , and @xmath21 is the hopping between nearest neighbor pairs ( kinetic energy ) . @xmath22 is the number of unsaturated c atoms . the interacting part is in turn given by , @xmath23 where @xmath24 is the on - site coulomb repulsion and @xmath25 is the inter - site coulomb repulsion , while the density operator is @xmath26 this hamiltonian reduces to the hubbard model for @xmath25=0 we start discussing the fitting of spin state energies by means of hubbard hamiltonian . lanczos algorithm in the whole hilbert occupation space is used to get numerically exact many - body states @xcite . coulomb on - site repulsion has been adjusted to describe spin excitations of some pahs . because the interacting model can not be solved exactly for 6h - coronene ( 18 orbitals or equivalently sites lead to a hilbert occupation space of dimension equal to @xmath27 which is beyond actual computational facilities ) , the case of a coronene molecule with all peripheral c atoms saturated by additional h has been considered . this leaves a molecule with only 12 @xmath0 orbitals , a cluster size that can be easily handled by means of lanczos algorithm . also benzene ( 6 sites ) , anthracene ( 14 sites ) and 6h - corannulene ( 14 sites ) have been fitted . calculations were carried out by taking the hopping integral commonly used to describe graphene sheets , @xmath28 ev , and varying the on - site repulsion @xmath24 . the results depicted in the left panel of fig . 4 indicate that spin states of these molecules can be reasonably fitted with @xmath29 ev ( benzene , for which the fitting is as good as for anthracene , is not shown for the sake of clarity ) . albeit noticeable deviations occur in the three lowest lying states of 6h - corannulene , the state ordering is the correct one . we have checked that the failure to correctly separate the lower excitations of 6h - corannulene is not exclusive of the simplest interacting model : a ppp calculation using ohno s interpolation scheme @xcite shows a similar weakness . let us remark once more that , despite the rather small @xmath24 ( @xmath30 ) resulting from the fittings shown in fig . 4 , anti - ferromagnetic correlations in these molecules , as calculated by means of the interacting model , are significant . particularly attracting is the case of corannulene for which there is a bond at which the spin - spin correlation is significantly smaller than at other bonds of the molecule ( top right panel of fig . 4 ) . interestingly enough , placing frustration ( two adjacent @xmath0 orbitals of the same sublattice as shown in fig . 1c ) at that bond gives a difference between carbon atoms in the two sublattices of four atoms , which , using lieb s formula , predicts a ground state of total spin 2 , in agreement with our numerical results . summarizing , @xmath24 = 3.3 ev works satisfactorily describing the spin states of aromatic molecules and , in particular , the multiplicity of the ground state . the same simple hubbard model fails , however , in describing the charged states of these systems as results for benzene show ( bottom right panel of fig . 2 ) . lanczos results for the interacting hamiltonian are compared with b3lyp results for a charged benzene ideally restricted to a fixed geometric structure . actual energy differences are much higher than those predicted by the model . our results do also illustrate the lack of electron - hole symmetry that characterizes any realistic self - consistent field calculation as opposed to lieb s model . the ppp model with @xmath28 ev and values for the coulomb repulsion integrals from ref . , although greatly improves the fitting , still gives a symmetric curve by the well - known pairing between occupied and unoccupied molecular orbitals . a full fitting of charge and spin states will surely require including a larger number of parameters @xcite . we have proposed a new route to produce magnetic organic molecules that consists of hydrogenating pahs . in the case of alternant pahs the spin multiplicities of the molecules ground states agree with lieb s prediction , even though _ ab initio _ hamiltonians may significantly differ from hubbard s model . a probably related result is that _ ab initio _ energies of the spin states of these molecules can be very satisfactorily fitted by means of the simplest version of the hubbard model . it seems that the molecule topology is enough support for the main result . energies of charged molecules , instead , can not be described by the simple , most popular , interacting models , suggesting a critical examination of their use in , for instance , graphene . results for total spin densities in molecules having a magnetic ground state clearly show that the spin is localized in only one of the two sublattices . on the other hand , _ ab initio _ and model hamiltonian calculations for hydrogenated corannulene , place the frustrated bond at the same location . this produces an unbalance in the molecule which , using lieb s formula , gives s=2 for the ground state of the molecule , in agreement with the numerical results . it is also worth noting that in a recent study @xcite we have shown that dehydrogenation may also produce magnetic molecules . although dehydrogenation is a highly unlikely process , dehydrogenated pahs have been intensively investigated by astrophysicists @xcite who believe they to form part of interstellar matter . although the results presented here are encouraging , there is still a long way to go : finding procedures to synthesize these hydrogenated pahs and crystallize them into solids that may eventually show magnetic properties . the authors are grateful to j. feliu , m. yus and a. guijarro for useful suggestions and remarks . financial support by the spanish mcyt ( grants fis200402356 , mat2005 - 07369-c03 - 01 and nan2004 - 09183-c10 - 08 ) and the universidad de alicante is gratefully acknowledged . gc is thankful to the spanish mcyt for a ramn y cajal grant . miller , inorg . chem . * 39 * , 4392 ( 2000 ) . r. jain , k. kabir , j.b gilroy , k.a.r . mitchell , k .- c . wong , r.g . hicks , nature , * 445 * , 291 ( 2007 ) . a. rajca , in _ advances in physical organic chemistry 40 _ , edited by j. richard ( academic press , london , 2005 ) and references therein . a. rajca , j. wongsriratanakul , s. rajca , science * 294 * , 1503 ( 2001 ) . a. rajca , s. rajca , j. am . soc . * 118 * , 8121 ( 1996 ) . slipchenko , t.e . munsch , p.g . wenthold , a.i . krylov , angew . int . ed . * 43 * , 742 ( 2004 ) . krylov , j. phy . a * 109 * , 10638 ( 2005 ) . pez , a.r . scott , a.c . cooper , h. cheng , _ hydrogen storage by reversible hydrogenation of pi - conjugated substrates _ ( us patent no . 7101530 , issued on september 5 , 2006 ) . g. zhong , b. chan , l. radom , j. mol . : theochem * 811 * , 13 ( 2007 ) . e.h . lieb , phys . . lett . * 62 * , 1201 ( 1989 ) . a very restricted list of papers reporting on magnetic effects in fullerene and finite graphene systems follows : k. kusakabe and m. maruyama , phys . b * 67 * , 092406 ( 2003 ) ; y .- h . kim , j. choi , k. j. chang , and d. tomanek , phys . b , * 68 * , 125420 ( 2003 ) ; n. park , m. yoon , s. berber , j. ihm , e. osawa , and d. tomanek , phys . lett . * 91 * 237204 ( 2003 ) ; de - en jiang , b.g . sumpter , and s. dai , j. chem . phys . * 127 * , 124703 ( 2007 ) ; m. ezawa , phys . rev . b * 76 * , 245415 ( 2007 ) ; p. shemella , y. zhang , m. mailman , p. m. ajayan , and s. k. nayak , appl . phys . lett . * 91 * , 042101 ( 2007 ) ; j. fernndez - rossier and j.j . palacios , phys . * 99 * , 177204 ( 2007 ) ; o. hod , v. barone , and g.e . scuseria , phys . b * 77 * , 035411 ( 2008 ) ; k.n . kudin , acs nano * 2 * , 516 ( 2008 ) ; o. hod and g.e . scuseria , acs nano * 2 * , 2243 ( 2008 ) ; e. rudberg , p. salek , and y. luo , nano lett . * 7 * , 2211 ( 2007 ) . j. andzelm , m. klobukowski , e. radzio - andzelm , y. sakai , h. tatewaki in _ gaussian basis sets for molecular calculations _ , edited by s. huzinaga ( elsevier , amsterdam , 1984 ) . j. thom h. dunning , j. chem . * 90 * , 1007 ( 1989 ) . a. d. becke , j. chem . phys . * 98 * , 5648 ( 1993 ) . c. lee , w. yang and r. g. parr , phys . b * 37 * , 785 ( 1988 ) . r. colle and o. salvetti , theor . acta * 37 * , 329 ( 1975 ) . remember that unrestricted - hartree - fock calculations do not provide wavefunctions with a defined total spin ( only one component is a good quantum number ) . brooks , h.f . schaefer iii , j. chem . * 70 * , 5092 ( 1979 ) . m. w. schmidt , m. s. gordon , annu . phys . chem . * 49 * , 233 ( 1998 ) . b. roos , adv . phys . * 69 * , 399 ( 1997 ) . et al . _ , j. comput . * 14 * , 1347 ( 1993 ) . c. melero , r.p . herrera , a. guijarro , and m. yus , _ chem . j. _ * 13 * , 10096 ( 2007 ) . let us mention two recently published examples . antiferromagnetic correlations have been found both at the edges of graphene nanoribbons , see , y - w son , m.l . cohen and s.g . louie , nature * 444 * , 347 ( 2006 ) , and at the ( 110 ) surface of diamond , see , f. yndurain , phys . b * 75 * , 195443 ( 2007 ) . some of the papers listed in ref . include further examples . r. pariser and r.g . parr , j. chem . * 21 * , 466 ( 1953 ) . pople , trans . faraday soc . * 49 * , 1365 ( 1953 ) . j. hubbard , proc . london a - math and phys . sci . * 276 * , 238 ( 1963 ) . we use a straightforward lanczos transformation starting from a random ground state candidate to generate a small hamiltonian matrix that can be diagonalized to get a better approximation for the ground state . this process is iterated until convergence is reached . see g. chiappe , e. louis , j. galn , f. guinea and j. a. vergs , phys . b * 48 * , 16539 ( 1993 ) for details . k. ohno , theor . acta * 2 * , 219 ( 1964 ) . r. oszwaldowski , h. vzquez , p. pou , j. ortega , r. prez , and f. flores , j. phys . : condens . matter * 15 * , s2665 . g. chiappe , e. louis , e. sanfabin and , j.a . vergs , phys . b * 75 * , 195104 ( 2007 ) . j.a . vergs , e. san - fabin , l. pastor - abia , g. chiappe and , e. louis , phys . solidi ( to be published ) ( 2009 ) . v. le page , t.p . snow and and v.m . bierbaum , astrophys . journal * 584 * , 316 ( 2003 ) .	present routes to produce magnetic organic - based materials adopt a common strategy : the use of magnetic species ( atoms , polyradicals , etc . ) as building blocks . we explore an alternative approach which consists of selective hydrogenation of polycyclic aromatic hydrocarbons . self - consistent - field ( scf ) ( hartree fock and dft ) and multi - configurational ( cisd and mcscf ) calculations on coronene and corannulene , both hexa - hydrogenated , show that the formation of stable high spin species is possible . the spin of the ground states is discussed in terms of the hund rule and lieb s theorem for bipartite lattices ( alternant hydrocarbons in this case ) . this proposal opens a new door to magnetism in the organic world . = 1
_ markov chain monte carlo _ ( mcmc ) _ methods _ allow samples from virtually any target distribution @xmath0 , known up to a normalizing constant , to be generated . in particular , the celebrated _ metropolis hastings algorithm _ ( introduced in @xcite and @xcite ) simulates a markov chain evolving according to a reversible markov transition kernel by first generating , using some instrumental kernel , a candidate and then accepting or rejecting the same with a probability adjusted to satisfy the detailed balance condition @xcite . when choosing between several metropolis hastings algorithms , it is desirable to be able to compare the efficiencies , in terms of the asymptotic variance of sample path averages , of different @xmath0-reversible markov chains . despite the practical importance of this question , only a few results in this direction exist the literature . peskun @xcite defined a partial ordering for finite state space markov chains , where one transition kernel has a higher order than another if the former dominates the latter on the off - diagonal ( see definition [ defipeskunordering ] ) . this ordering was extended later by tierney @xcite to general state space markov chains and another even more general ordering , the covariance ordering , was proposed in @xcite . in general , it holds that if a homogeneous @xmath8-reversible markov transition kernel is greater than another according to one of these orderings , then the asymptotic variance of sample path averages for a markov chain evolving according to the former is smaller for all square integrable ( with respect to @xmath0 ) target functions . we provide an extension of this result to inhomogeneous markov chains that evolve alternatingly according to two different @xmath0-reversible markov transition kernels . to the best of our knowledge , this is the first work dealing with systematic comparison of asymptotic variances of inhomogeneous markov chains . the approach is linked with the operator theory for markov chains but does not make use of any spectral representation . after some preliminaries ( section [ secpreliminaries ] ) , our main result , theorem [ teomainresult ] , is stated in section [ secmain ] . in section [ secappl ] , we apply theorem [ teomainresult ] in the context of mcmc algorithms by comparing the efficiency , in terms of asymptotic variance , of some existing data - augmentation - type algorithms . moreover , we propose a novel pseudo - marginal algorithm ( in the sense of @xcite ) , referred to as the _ random refreshment _ algorithm , which on the contrary to the pseudo - marginal version of the _ monte carlo within metropolis _ ( mcwm ) algorithm turns out to be exact and more efficient than the pseudo - marginal version of the _ grouped independence metropolis hastings _ ( gimh ) algorithm . here , the analysis is again driven by theorem [ teomainresult ] . the proof of theorem [ teomainresult ] is given in section [ secproofmain ] and some technical lemmas are postponed to appendix [ app ] . finally , appendix [ secappb ] relates some existing mcmc algorithms to the framework considered in this paper . we denote by @xmath9 and @xmath10 the sets of nonnegative and positive integers , respectively . in the following , all random variables are assumed to be defined on a common probability space @xmath11 . let @xmath12 be a measurable space ; then we denote by @xmath13 and @xmath14 the spaces of positive measures and measurable functions on @xmath15 , respectively . the lebesgue integral of @xmath16 over @xmath17 with respect to the measure @xmath18 is , when well - defined , denoted by @xmath19 recall that a _ markov transition kernel _ @xmath1 on @xmath15 is a mapping @xmath20 $ ] such that : * for all @xmath21 , @xmath22 is a measurable function , * for all @xmath23 , @xmath24 is a probability measure . a kernel @xmath1 induces two integral operators , one acting on @xmath13 and the other on @xmath14 ; more specifically , for @xmath25 and @xmath26 , we define the measure @xmath27 and the measurable function @xmath28 moreover , the _ composition _ ( or _ product _ ) of two kernels @xmath1 and @xmath2 on @xmath15 is the kernel defined by @xmath29 we will from now on fix a distinguished probability measure @xmath0 on @xmath15 . given @xmath0 , we denote by @xmath30:= \ { f \in\mathcal{f}(\mathcal{x } ) \dvtx\pi f^2 < \infty\}$ ] the space of square integrable functions with respect to @xmath0 and furnish the same with the scalar product @xmath31 , g \in\ltwo[\pi ] \bigr)\ ] ] and the associated norm @xmath32 \bigr).\ ] ] here , we have expunged the measure @xmath0 from the notation for brevity . if @xmath1 is a markov kernel on @xmath15 admitting @xmath0 as an invariant distribution , then the mapping @xmath33 defines an operator on @xmath30 $ ] , and by jensen s inequality it holds that @xmath34 \dvtx \ltwonorm{f } \leq1 } \ltwonorm{pf } \leq1.\ ] ] recall that a kernel @xmath1 is _ @xmath0-reversible _ if and only if the detailed balance relation @xmath35 holds . if the markov kernel @xmath1 is @xmath0-reversible , then @xmath36 defines a self - adjoint operator on @xmath30 $ ] , that is , for all @xmath37 and @xmath38 belonging to @xmath30 $ ] , @xmath39 the following off - diagonal ordering of markov transition kernels on a common state space was , in the case of markov chains in a finite state space , proposed in @xcite . the ordering was extended later in @xcite to the case of markov chains in general state space . [ defipeskunordering ] let @xmath40 and @xmath41 be markov transition kernels on @xmath12 with invariant distribution @xmath0 . we say that _ @xmath41 dominates @xmath40 on the off - diagonal _ , denoted @xmath42 , if for all @xmath21 and @xmath0-a.s . all @xmath23 , @xmath43 the previous ordering allows the asymptotic efficiencies of different reversible kernels to be compared . more specifically , the following seminal result was established in @xcite , theorem 2.1.1 , for markov chains in discrete state space and extended later in @xcite , theorem 4 , to markov chains in general state space . [ teoefficiencyordering ] let @xmath40 and @xmath41 be two @xmath0-reversible kernels on @xmath44 . if , then for a.s . all @xmath45 $ ] , @xmath46 where we have defined , for a markov chain @xmath47[\mathbb{n}]$ ] with @xmath0-reversible transition kernel @xmath1 and initial distribution @xmath0 , @xmath48 note that according to @xcite , if @xmath49 is a @xmath8-reversible markov chain and @xmath45 $ ] , then @xmath50 is guaranteed to exist ( but may be infinite ) . nevertheless , the ordering in question does not allow markov kernels lacking probability mass on the diagonal , that is , kernels @xmath1 satisfying for all @xmath23 , to be compared . this is in particular the case for gibbs samplers in general state space . to overcome this limitation , one may consider instead the following covariance ordering based on lag - one autocovariances . [ deficovarordering ] let @xmath40 and @xmath41 be markov transition kernels on @xmath12 with invariant distribution @xmath0 . we say that _ @xmath41 dominates @xmath40 in the covariance ordering _ , denoted @xmath51 p_0 $ ] , if for all @xmath45 $ ] , @xmath52 the covariance ordering , which was introduced implicitly in @xcite , page 5 , and formalized in @xcite , is an extension of the off - diagonal ordering since according to @xcite , lemma 3 , @xmath53 implies @xmath51 p_0 $ ] . moreover , it turns out that for reversible kernels , @xmath51p_0 $ ] implies @xmath54 ( see the proof of @xcite , theorem 4 ) . all these results concern homogeneous markov chains , whereas many mcmc algorithms such as the gibbs or the metropolis - within - gibbs samplers use several kernels , for example , @xmath1 and @xmath55 in the case of two kernels @xcite . a natural idea would then be to apply theorem [ teoefficiencyordering ] to the homogeneous markov chain having the block kernel @xmath56 as transition kernel ; however , even when the kernels @xmath1 and @xmath57 are both @xmath0-reversible , the product @xmath56 of the same is usually not @xmath0-reversible , except in the particular case when @xmath1 and @xmath57 commute , that is , @xmath58 . thus , theorem [ teoefficiencyordering ] can not in general be applied directly in this case . in the following , let @xmath3 and @xmath59 , @xmath5 , be markov transition kernels on @xmath15 . define @xmath60 and @xmath61 as the markov chains evolving as follows : @xmath62{i } \stackrel{p_i } { \longrightarrow } \x[1]{i } \stackrel { q_i } { \longrightarrow } \x[2]{i } \stackrel{p_i } { \longrightarrow } \x[3]{i } \stackrel { q_i } { \longrightarrow } \cdots.\ ] ] this means that for all @xmath63 , @xmath5 and @xmath64 : * @xmath65{i } \in\mathsf{a } { \|}\mathcal{f}_{2 k}^{(i ) } ) = p_i(\x[2 k]{i } , \mathsf{a})$ ] , * @xmath66{i } \in\mathsf{a } { \|}\mathcal{f}_{2 k+1}^{(i ) } ) = q_i(\x[2 k+1]{i},\mathsf{a})$ ] , where @xmath67{i},\ldots , \x[n]{i})$ ] , @xmath68 . we impose the following assumption : @xmath69 \hspace{-100pt}\\[-8pt ] \hspace{-100pt}&&\mbox{(ii ) $ p_1 \pgeq[1 ] p_0 $ and $ q_1 \pgeq[1 ] q_0$.}\nonumber\end{aligned}\ ] ] as mentioned above , @xmath53 implies @xmath70p_0 $ ] ; thus , in practice , a sufficient condition for ( ii ) is that @xmath42 and @xmath71 . [ teomainresult ] assume that @xmath3 and @xmath59 , @xmath72 , satisfy and let @xmath73 , @xmath72 , be markov chains evolving as in ( [ eqeq1markov ] ) with initial distribution @xmath0 . then for all @xmath45 $ ] such that for @xmath72 , @xmath74{i } \bigr ) } { f \bigl(\x[k]{i } \bigr)}\bigr\|+\bigl\|\covardu { f \bigl ( \x[1]{i } \bigr ) } { f \bigl(\x[k+1]{i } \bigr)}\bigr\| \bigr ) < \infty,\ ] ] it holds that @xmath75 where @xmath76{i } \bigr ) } \qquad \bigl(i \in\{0 , 1\ } \bigr).\ ] ] at present , we have not been able to extend the arguments of our current proof of theorem [ teomainresult ] ( see section [ secproofmain ] ) to inhomogeneous markov chains evolving alternatingly according to _ more _ than two different kernels . on the other hand , we have not been able to find a counterexample rejecting the hypothesis that a similar result would hold true also in that case . we leave this as an open problem . [ remcounterexsummability ] condition ( [ eqassumpfuncthm ] ) is _ not _ a necessary condition for ( [ eqmainresult ] ) ; indeed , letting @xmath77 , @xmath78 , @xmath79 , where , as in @xcite , example 5 , @xmath80 , provides a straightforward counterexample . when verifying if a given @xmath37 satisfies the condition ( [ eqassumpfuncthm ] ) it may be convenient to consider the homogeneous markov chains @xmath81 or @xmath82 or even @xmath83 . typically , none of these chains are @xmath0-reversible . nevertheless , @xmath0-reversibility is not needed for checking conditions of type ( [ eqassumpfuncthm ] ) , which can be established using upper bounds on the _ @xmath84-norm _ between the distribution given by the @xmath85th iterate of a homogeneous kernel and its stationary distribution . this will be developed in the following section . for any measurable real - valued function @xmath37 on @xmath86 , define the _ @xmath84-norm _ of the _ function _ @xmath37 by @xmath87 moreover , let @xmath88 be a finite signed measure on @xmath89 . then by the jordan decomposition theorem there exists a unique pair of positive , finite and singular measures @xmath90 and @xmath91 on @xmath92 such that @xmath93 . the pair @xmath94 is referred to as the _ jordan decomposition _ of the signed measure @xmath88 . the finite measure @xmath95 is called the _ total variation _ of @xmath88 . let @xmath84 be a nonnegative function taking values in @xmath96 ; then the _ @xmath84-norm _ of the _ signed measure _ @xmath88 is defined by @xmath97 [ defvgeomerg ] a markov kernel @xmath1 on @xmath92 is _ @xmath84-geometrically ergodic _ if it admits a unique invariant distribution @xmath0 and there exists a measurable function @xmath98 satisfying @xmath99 and such that the following hold : there exist constants @xmath100 such that for all @xmath23 and all @xmath68 , @xmath101 there exist constants @xmath102 such that @xmath103 . @xcite , theorem 1.2 , provides sufficient conditions , in terms of drift towards a _ small set _ , for ( a ) in definition [ defvgeomerg ] to hold ; see also @xcite , fact 10 , for necessary and sufficient conditions under the assumption of aperiodicity and irreducibility . moreover , the coming developments require only the bound ( [ eqvgeom ] ) to hold @xmath0-a.s . we have now all necessary tools for giving sufficient conditions that imply the absolute summability assumption ( [ eqassumpfuncthm ] ) . let the chain @xmath49 evolve according to @xmath104 { } \stackrel{p } { \longrightarrow } \x [ 1 ] { } \stackrel{q } { \longrightarrow } \x[2 ] { } \stackrel{p } { \longrightarrow } \x[3 ] { } \stackrel{q } { \longrightarrow } \cdots\ ] ] with @xmath105 { } \sim\pi$ ] , for some markov kernels @xmath1 and @xmath2 . [ propaltcondition ] if the markov kernel @xmath56 is @xmath84-geometrically ergodic , then for all functions @xmath37 such that @xmath106 and @xmath107 , @xmath108 { } \bigr ) } { f \bigl(\x[k ] { } \bigr)}\bigr\|+\bigl\|\covardu{f \bigl(\x [ 1 ] { } \bigr ) } { f \bigl ( \x[k+1 ] { } \bigr)}\bigr\| \bigr ) < \infty,\ ] ] where @xmath109 { } ; k \in\mathbb{n}\}$ ] evolves as in ( [ eqeq1markovbis ] ) . the proof of proposition [ propaltcondition ] is found in appendix [ appproofpropaltcondition ] . before considering some applications of theorem [ teomainresult ] , we recall the following proposition , describing how to obtain a @xmath0-reversible markov chain using some instrumental kernel @xmath110 . although this result is fundamental in the metropolis hastings literature ( see , e.g. , @xcite and the references therein ) , it is restated here as it will be used in various situations in the sequel [ especially when there is no fixed reference measure dominating all the distributions @xmath111 . [ propderiveeradon ] let @xmath110 be a markov transition kernel on @xmath112 and @xmath8 a probability measure on @xmath15 . define the probability measures @xmath113 and @xmath114 . assume that the measures @xmath115 and @xmath116 are equivalent and such that for @xmath116-a.s . all @xmath117 , @xmath118 where @xmath119 denotes the radon nikodym derivative . then the markov kernel @xmath120 , where @xmath121 is @xmath0-reversible . a natural application of theorem [ teomainresult ] consists in using the result for comparing different data - augmentation - type algorithms . in the following , we wish to target a probability distribution @xmath122 defined on @xmath123 using a sequence @xmath124[\mathbb{n}]$ ] of valued random variables . to this aim , tanner and wong @xcite suggest writing @xmath125 as the marginal of some distribution @xmath0 defined on the product space in the sense that @xmath126 , where @xmath127 is some markov transition kernel on @xmath128 . in most cases , the marginal @xmath125 is of sole interest , while the component @xmath129 is introduced for convenience as a means of coping with analytic intractability of the marginal . ( it could also be the case that the marginal @xmath125 is too computationally expensive to evaluate . ) a first solution consists in letting @xmath124[\mathbb{n}]$ ] be the first - component process @xmath130[\mathbb{n } ] $ ] of the @xmath0-reversible markov chain @xmath131{1 } , \aux[k]{1 } ) ; k \in\mathbb{n}\}$ ] defined as follows . let @xmath132 and @xmath133 be instrumental markov transition kernels on @xmath134 and @xmath135 , respectively , and define a transition of the chain @xmath131{1 } , \aux[k]{1 } ) ; k \in\mathbb { n}\}$ ] by algorithm [ algalg1 ] . @xmath136{1 } , \aux[k]{1 } ) = ( y , u)$ ] : draw @xmath137 and call the outcome @xmath138 ( abbr . @xmath139 ) , draw @xmath140 , set @xmath141{1 } , \aux[k+1]{1 } \bigr ) \nonumber\\[10pt]\\[-22pt ] & & \qquad \gets \cases{\displaystyle ( \hat{y } , \hat{u } ) , & \quad with probability $ \displaystyle\alpha ( y , u , \hat{y } , \hat{u})$ \vspace{5pt}\cr & \qquad\quad $ \displaystyle:=1 \wedge\frac { \pi^{\ast}(\hat{y } ) r(\hat{y } , \hat{u})s(\hat{y } , \hat{u } ; y ) t(\hat{y } , \hat{u } , y ; u)}{\pi^{\ast}(y ) r(y , u ) s(y , u ; \hat{y } ) t(y , u , \hat{y } ; \hat{u})}$ , \vspace{5pt}\cr ( y , u ) , & \quad otherwise.}\nonumber\end{aligned}\ ] ] [ remgeneralradon ] in the expression ( [ eqacceptmetropolis ] ) of @xmath142 we assume implicitly that the families @xmath143 and @xmath144 of probability measures are dominated by a fixed nonnegative measure and we denote by @xmath145 and @xmath146 the corresponding transition kernel densities , respectively . in some cases ( see , e.g. , @xcite ) it may , however , happen ( typically when some dirac mass is involved ) that these kernels are not dominated by a nonnegative measure ; nevertheless , algorithm [ algalg1 ] as well as algorithm [ algalg2 ] defined below remain valid provided that the ratio in @xmath147 is replaced by the corresponding radon nikodym derivative @xmath148 where in this case , @xmath149 by applying proposition [ propderiveeradon ] , we deduce that the output @xmath131{1 } , \aux[k]{1});\break k\in\mathbb{n}\}$ ] is a @xmath8-reversible markov chain . as a consequence , the sequence @xmath150[\mathbb{n}]$ ] targets , although it is not itself a markov chain , the marginal distribution @xmath125 . note that the method requires the product @xmath151 to be known at least up to a multiplicative constant to guarantee the computability of the acceptance probability @xmath147 in ( [ eqacceptmetropolis ] ) . [ exgimh ] the grouped independence metropolis hastings ( gimh ) algorithm ( see @xcite ) is used in situations where @xmath125 is analytically intractable . in this algorithm , the quantity @xmath152 is in the acceptance probability replaced by an importance sampling estimate @xmath153 where @xmath154 is the density of some augmented target distribution @xmath155 defined on the product space @xmath156 , known up to a normalizing constant and allowing @xmath125 as marginal distribution , and @xmath157 are i.i.d . draws from the proposal @xmath158 . denoting by @xmath159 the density used for proposing new candidates @xmath138 , one obtains the acceptance probability ratio @xmath160 where @xmath161 and @xmath162 consequently , the gimh algorithm can be perfectly cast into the framework of the freeze algorithm , with the auxiliary variable @xmath163 playing the role of the dimensional monte carlo sample and @xmath164 . in the following , we use theorem [ teomainresult ] for comparing the performance of algorithm [ algalg1 ] to that of different modifications of the same obtained in the cases where : simulating @xmath127-transitions is feasible , simulating @xmath127-transitions is infeasible . @xmath165{2 } = y$ ] : draw @xmath166 , draw @xmath167 , draw @xmath168 , set @xmath169{2 } \gets \cases { \hat{y } , & \quad with probability $ \alpha(y , u , \hat{y } , \hat{u})$ [ defined in ( \ref{eqacceptmetropolis } ) ] , \vspace{2pt}\cr y , & \quad otherwise . } $ ] in this case , an alternative to algorithm [ algalg1 ] consists in letting @xmath170[\mathbb{n}]$ ] be the sequence @xmath171[\mathbb{n}]$ ] generated through algorithm [ algalg2 ] . note that algorithm [ algalg2 ] `` refreshes , '' in step ( i ) , systematically the second component of the markov chain , which advocates algorithm [ algalg2 ] to have better mixing properties than algorithm [ algalg1 ] . the main task of the present section is to establish rigorously this heuristics . the output @xmath172 of algorithm [ algalg2 ] is , on the contrary to @xmath173 , a markov chain . it is not a classical metropolis hastings markov chain due to the auxiliary variables @xmath174 and @xmath175 that appear explicitly in the acceptance probability . however , as established in the following proposition , whose proof is found in appendix [ appproofrevsystrefresh ] , the @xmath0-reversibility of @xmath176{1},\aux[k]{1 } ) ; k\in\mathbb{n}\}$ ] implies @xmath177-reversibility of @xmath172 . [ propinduces_rever ] the sequence @xmath172 generated in algorithm [ algalg2 ] is a @xmath125-reversible markov chain . in @xcite , the authors use the terminology _ randomized mcmc _ ( r - mcmc ) for a @xmath125-reversible metropolis hastings chain @xmath178 generated using a set of auxiliary variables@xmath179 with a particular expression of the acceptance probability . although only one of these auxiliary variables is sampled at each time step , one may actually cast this approach into the framework of algorithm [ algalg2 ] by creating artificially another auxiliary variable according to the deterministic kernel @xmath180 where @xmath37 is any continuously differentiable involution on @xmath181 . even though @xmath133 is not dominated , it is possible to verify ( [ eqcondradon ] ) using that @xmath37 is an involution . we prove in appendix [ apprmcmc ] that the r - mcmc algorithm is a special case of algorithm [ algalg2 ] with this particular choice of @xmath133 and with the general form of the acceptance probability described in remark [ remgeneralradon ] . the _ generalized multiple - try metropolis _ ( gmtm ) _ algorithm _ @xcite is an extension of the _ multiple - try metropolis hastings algorithm _ proposed in @xcite . given @xmath165{}=y$ ] , one draws @xmath85 i.i.d . possible moves @xmath182 according to @xmath183 . after this , a random index @xmath184 taking the value @xmath185 with probability proportional to @xmath186 is generated , whereupon a candidate is constructed as @xmath187 . the candidate is then accepted with some probability that is computed using @xmath85 additional random variables @xmath188 , where @xmath189 are i.i.d . draws from @xmath190 , and @xmath191 is set deterministically to @xmath192 ( see appendix [ appmtm ] for more details concerning the acceptance probability ) . in appendix [ appmtm ] , proposition [ lemmtm ] , it is shown that the gmtm algorithm is in fact a special case of algorithm [ algalg2 ] with @xmath193 and @xmath194 . when the function @xmath195 is known explicitly , one may obtain another @xmath196-reversible markov chain by means of the classical metropolis hastings ratio , that is , we use again algorithm [ algalg2 ] but replace the acceptance probability @xmath197 by @xmath198 the following proposition , which generalizes a similar result obtained in @xcite , section 2.3 , for the r - mcmc algorithm , shows , when combined with @xcite , theorem 4 , that the asymptotic variance of the classical metropolis hastings estimator is smaller than that of the estimator based on algorithm [ algalg2 ] . [ propmhalg2 ] the metropolis hastings kernel associated with the acceptance probability ( [ eqmhratio ] ) is larger , in the sense of definition [ defipeskunordering ] , than the transition kernel associated with algorithm [ algalg2 ] . set @xmath199 and note that @xmath116 is a probability measure . hence , as the mapping @xmath200 is concave , jensen s inequality implies that @xmath201 ( a similar technique was used in the proof of @xcite , lemma 1 ) . the previous computation shows that the off - diagonal transition density function of the metropolis hastings markov chain associated with the acceptance probability ( [ eqmhratio ] ) is larger than that of the chain in algorithm [ algalg2 ] . this completes the proof . however , in practice a closed - form expression of @xmath202 is rarely available , which prevents the classical metropolis hastings algorithm from being implemented . thus , if the transition density @xmath203 is known explicitly and can be sampled we have to choose between algorithms [ algalg1 ] and [ algalg2 ] for approximating @xmath177 . the classical tools ( such as the ordering in definition [ defipeskunordering ] ) for comparing and can not be applied here , since is not even a markov chain . nevertheless , theorem [ teomainresult ] allows these two algorithms to be compared theoretically by embedding and into inhomogeneous @xmath8-reversible markov chains . the construction , which will be carried through in full detail below , leads to the following result . [ teocompalg1alg2 ] let and be sequences of random variables generated by algorithms [ algalg1 ] and [ algalg2 ] , respectively , where @xmath204{1},\break \aux[0]{1 } ) \sim\pi$ ] and @xmath205{2 } \sim\pi^{\ast}$ ] . then for all @xmath206 $ ] satisfying @xmath207{i } \bigr ) } { h \bigl(\y[k]{i } \bigr ) } \bigr\| < \infty\qquad \bigl(i \in\{1,2 \ } \bigr)\ ] ] it holds that @xmath208{2 } \bigr ) } \leq\lim _ { n \to\infty } \frac{1}{n } \var{\sum_{k = 0}^{n - 1 } h \bigl(\y[k]{1 } \bigr)}.\ ] ] we preface the proof of theorem [ teocompalg1alg2 ] by the following lemma , which may serve as a basis for the comparison of _ homogeneous _ markov chains evolving according to @xmath209 ( or @xmath210 ) , @xmath5 , where @xmath3 and @xmath59 , @xmath5 , are kernels satisfying on some product space . [ lemhomogeneouscomp ] let @xmath3 and @xmath59 , @xmath5 , be kernels satisfying on @xmath15 , with @xmath211 and @xmath212 . in addition , assume that for all @xmath213 , @xmath214 then for all @xmath45 $ ] depending on only the first argument [ i.e. , @xmath215 for some @xmath216 and such that @xmath217 it holds that @xmath218 [ remcounterexlemma ] assumption ( [ eqassunitmass ] ) is essential in lemma [ lemhomogeneouscomp ] . indeed , let @xmath219 and @xmath220 , and define the kernels @xmath221 , @xmath222 for some @xmath223 , @xmath224 , and @xmath225 . then the kernels @xmath3 and @xmath59 , @xmath5 , satisfy , and consequently theorem [ teomainresult ] applies to the inhomogeneous chains evolving alternatingly according to the same . however , the similar result does not hold true for chains evolving according to the product kernels @xmath226 and @xmath227 , @xmath5 , as @xmath228 with @xmath37 being the identity mapping on @xmath17 . proof of lemma [ lemhomogeneouscomp ] define markov chains @xmath73 , @xmath5 , evolving as @xmath229{i } = \pmatrix{\y[k]{i } \vspace{3pt}\cr \aux[k]{i } } \stackrel{p_i } { \longrightarrow } \x[2k+1]{i}= \pmatrix{\cy[k]{i } \vspace{3pt}\cr \caux[k]{i } } \stackrel { q_i } { \longrightarrow } \x[2k+2]{i}= \pmatrix{\y[k+1]{i } \vspace{3pt}\cr \aux[k+1]{i } } \stackrel{p_i } { \longrightarrow } \cdots\ ] ] with @xmath105{i } \sim\pi$ ] . by construction , @xmath230{i } \bigr ) } { f \bigl(\x[k]{i } \bigr)}\bigr\|+\bigl\|\covardu { f \bigl ( \x[1]{i } \bigr ) } { f \bigl(\x[k+1]{i } \bigr)}\bigr\| \bigr ) \nonumber\\[-8pt]\\[-8pt ] & & \qquad = \pi f^2 - \pi^2 f + 4 \sum _ { k = 1}^\infty\bigl\|\covardu{h \bigl(\y[0]{i } \bigr ) } { h \bigl ( \y[k]{i } \bigr)}\bigr\| < \infty\qquad \bigl(i \in\{0 , 1\ } \bigr ) , \nonumber\end{aligned}\ ] ] where finiteness follows from the assumption ( [ eqassbddnessproduct ] ) . moreover , for all @xmath231 and @xmath232 , @xmath233{i } \bigr ) } = \var{\sum_{k = 0}^{n - 1 } h \bigl ( \cy[k]{i } \bigr ) } = \frac{1}{4 } \var{\sum_{k = 0}^{2n - 1 } f \bigl(\x[k]{i } \bigr)},\ ] ] which implies , by ( [ eqimpliedbddness ] ) , @xmath234{i } \bigr ) } \qquad \bigl(i \in\{0 , 1\ } \bigr).\ ] ] finally , by ( [ eqimpliedbddness ] ) we may now apply theorem [ teomainresult ] to the chains @xmath235{i } ; k \!\in\!\mathbb{n}\}$ ] , @xmath236 , which establishes immediately the statement of the lemma . proof of theorem [ teocompalg1alg2 ] we introduce the kernels : * @xmath237 , * @xmath238 , * @xmath239 being - defined implicitly as the transition kernel associated with the _ freeze algorithm _ ( algorithm [ algalg1 ] ) . it can be checked readily that the two sequences @xmath173 and @xmath172 generated by algorithms [ algalg1 ] and [ algalg2 ] , respectively , have indeed the same distributions as the marginal processes ( with respect to the first component ) of homogeneous chains evolving according to the products @xmath240 and @xmath241 , respectively . in addition , all kernels @xmath3 and @xmath59 , @xmath242 , are @xmath0-reversible , as : * @xmath41 is reversible with respect to any probability measure ( in particular , it is reversible ) , * @xmath243 is @xmath0-reversible as a gibbs - sampler sub - step transition kernel , * @xmath239 is @xmath0-reversible as a classical metropolis hastings transition kernel . since @xmath41 has no off - diagonal component , it holds that @xmath244 ; moreover , trivially , @xmath245 . thus , we may complete the proof by applying lemma [ lemhomogeneouscomp ] to the function @xmath215 , for which the condition ( [ eqassbddnessproduct ] ) is satisfied [ by ( [ eqcondh ] ) ] . _ pseudo - marginal algorithms _ ( see @xcite and @xcite ) are implemented using a markov kernel @xmath246 on @xmath247 and a family @xmath248 of real - valued nonnegative functions on @xmath249 such that @xmath250 for all @xmath251 . we denote by @xmath252 the transition density of the kernel @xmath253 with respect to some dominating measure . note that @xmath254 is a markov transition kernel as well . the problem at hand is to sample the target distribution @xmath255 under the assumption that : * for all @xmath256 , @xmath257 is known up to a normalizing constant , * for all @xmath251 , @xmath183 can be sampled from . the particular case where @xmath258 for all @xmath259 was discussed in the previous section , and we now turn to the case @xmath260 ( i.e. , sampling directly from @xmath127 is infeasible ) . the solution provided by pseudo - marginal algorithms consists in replacing , in algorithm [ algalg2 ] , the operation ( i ) by the sampling @xmath261 , and the computing the acceptance probability @xmath147 [ as defined in ( [ eqacceptmetropolis ] ) ] via the formula @xmath262 the output of this algorithm , which will be referred to as the _ noisy algorithm _ in the following , is typically not on the contrary to algorithm [ algalg2]@xmath125-reversible due to the replacement of @xmath127 by @xmath253 . this justifies the denomination . however , when @xmath263 is close to unity the noisy algorithm is close to algorithm [ algalg2 ] , which is , according to theorem [ teocompalg1alg2 ] , more efficient than algorithm [ algalg1 ] in terms of asymptotic variance . the monte carlo within metropolis algorithm ( mcwm ; see @xcite ) resembles closely the gimh algorithm ( see example [ exgimh ] ) , however , with the important difference that the importance sampling estimates @xmath264 [ given by ( [ eqgimhmcest ] ) ] are _ not _ stored and propagated through the algorithm along with the @xmath265-values . instead each estimate of the marginal density is recomputed using a `` fresh '' mc sample before the calculation of the acceptance probability . thus , the mcwm algorithm can be cast into the framework of the noisy algorithm with @xmath266 and with the auxiliary variables @xmath163 and @xmath267 playing the roles of @xmath268-dimensional monte carlo samples . considering this , we now propose a novel algorithm which will be referred to as the _ random refreshment _ _ algorithm _ and which is a hybrid between algorithm [ algalg2 ] and the noisy algorithm . this novel algorithm , which is described in algorithm [ algalgrefresh ] below , targets _ exactly _ @xmath125 and turns out to be more efficient than algorithm [ algalg1 ] . @xmath136{3 } , \aux[k]{3 } ) = ( y , u)$ ] : * draw @xmath269 , * set @xmath270 draw @xmath271 , draw @xmath272 , set @xmath273{3 } , \aux[k+1]{3 } ) \gets \cases { ( \hat{y } , \hat{u } ) , & \quad with probability $ \alpha(y , \check { u } , \hat{y } , \hat{u})$ , \vspace{3pt}\cr ( y , \check{u } ) , & \quad otherwise . } $ ] in step ( i ) in algorithm [ algalgrefresh ] , the auxiliary variable @xmath274 can be either `` refreshed , '' that is , replaced by a new candidate @xmath275 , or kept at the previous state @xmath276{3}$ ] according to an acceptance probability that turns out to be a standard metropolis hastings acceptance probability ( which will be seen in the proof of theorem [ teocompalg1alg4 ] below ) . interestingly , this allows the desired distribution @xmath0 as the target distribution of @xmath131{3 } , \aux[k]{3 } ) ; k \in\mathbb{n}\}$ ] . in comparison , the noisy algorithm described above differs only from algorithm [ algalgrefresh ] by step ( i ) , in that the new candidate is always accepted in the noisy algorithm . this `` systematic refreshment '' makes actually the noisy algorithm imprecise in the sense that @xmath0 is no longer the target distribution except when @xmath277 for all @xmath256 , in which case @xmath278 in ( [ eqdefrho ] ) becomes identically equal to unity and algorithm [ algalgrefresh ] translates into algorithm [ algalg2 ] . compared to algorithm [ algalg1 ] , step ( i ) allows the second component to be refreshed randomly according to the probability @xmath279 whereas this component remains unchanged in algorithm [ algalg1 ] . thus , in conformity with algorithm [ algalg2 ] , it is likely that algorithm [ algalgrefresh ] has better mixing properties than algorithm [ algalg1 ] . that this is indeed the case may be established by reapplying the embedding technique developed in the previous part . before formalizing this properly , we propose an example showing a typical situation where a random refreshment algorithm may be used . [ exrrabc ] in @xcite ( contributing to the discussion of @xcite ) , the authors propose a novel algorithm , _ rejuvenating gimh - abc _ @xcite , algorithm 1 , preventing the original _ gimh - abc _ @xcite , algorithm 2 ( termed _ mcmc - abc _ in the paper in question ) , from falling into possible trapping states . the gimh - abc is an instance of algorithm [ algalg1 ] targeting @xmath280 , where , in the abc context : @xmath281 is the desired posterior of a parameter @xmath282 given some observed data summary statistics @xmath283 , * @xmath183 is the likelihood of the data ( from which sampling is assumed to be feasible ) , * @xmath284 / \int\check{r}(y , \mathrm{d}u ' ) k[(s(u ' ) - { s_{\mathrm{obs } } } ) / h ] $ ] , where @xmath110 is a kernel integrating to unity , providing the classical abc discrepancy measure between the observed data summary statistics @xmath283 and that evaluated at the simulated data @xmath129 . rejuvenating gimh - abc comprises an intermediate step in which the simulated data @xmath129 , generated under the current parameter @xmath282 , are refreshed systematically . however , since sampling from @xmath285 is typically infeasible , the auxiliary variables are refreshed through @xmath253 in the spirit of algorithm [ algalg2 ] . therefore , in accordance with algorithm [ algalgrefresh ] , a _ @xmath0-reversible _ alternative to rejuvenating gimh - abc is obtained by , instead of refreshing systematically the data , performing refreshment with probability ( [ eqdefrho ] ) note that the fact that the constant in the denominator of @xmath286 is typically not computable does not prevent computation of ( [ eqdefrho ] ) , since this constant appears in @xmath286 as well as @xmath287 . this provides a _ random refreshment gimh - abc _ , which can be compared quantitatively , via the theorem [ teocompalg1alg4 ] below , to the gimh - abc while at the same time avoiding the possible gimh - abc trapping states mentioned in @xcite . [ teocompalg1alg4 ] let and be the sequences of random variables generated by algorithms [ algalg1 ] and [ algalgrefresh ] , respectively , where @xmath204{i},\break \aux[0]{i } ) \sim \pi $ ] , @xmath288 . then the following hold true : the output of algorithm [ algalgrefresh ] is @xmath0-reversible . for all @xmath289 $ ] satisfying @xmath290{i } \bigr ) } { h \bigl(\y[k]{i } \bigr ) } \bigr\| < \infty\qquad \bigl(i \in\{1,3 \ } \bigr)\ ] ] it holds that @xmath208{3 } \bigr ) } \leq\lim _ { n \to\infty } \frac{1}{n } \var{\sum_{k = 0}^{n - 1 } h \bigl(\y[k]{1 } \bigr)}.\ ] ] let the kernels @xmath41 and @xmath291 be defined as in the proof of theorem [ teocompalg1alg2 ] and introduce furthermore : * @xmath292 defined implicitly by the transition @xmath136{3 } , \aux [ k]{3 } ) \rightarrow(\y[k]{3 } , \check{u})$ ] according to step ( i ) in algorithm [ algalgrefresh ] ( note that the first component is held fixed throughout the transition ) , * @xmath293 . in conformity with the proof of theorem [ teocompalg1alg2 ] , it can be checked readily that the two sequences @xmath173 and @xmath294 generated by algorithms [ algalg1 ] and [ algalgrefresh ] , respectively , have indeed the same distributions as the marginal processes ( with respect to the first component ) of homogeneous chains evolving according to the products @xmath240 and @xmath295 , respectively . the @xmath0-reversibility of the kernels @xmath41 and @xmath296 was established in the proof of theorem [ teocompalg1alg2 ] . to verify @xmath0-reversibility of @xmath292 as well , note that @xmath292 is a metropolis hastings kernel associated with the target distribution @xmath0 , whose acceptance probability includes a radon nikodym derivative of the type given in proposition [ propderiveeradon ] ; it is therefore @xmath8-reversible . indeed , note that @xmath292 updates only the second component according to @xmath297 with the acceptance probability @xmath298 . assuming first that @xmath253 is dominated and denoting by @xmath252 its transition density , we have @xmath299 where @xmath300 in the density of the target @xmath0 . this shows that @xmath301 is indeed the acceptance probability of a metropolis hastings markov chain targeting @xmath0 , with proposal kernel @xmath302 ; the @xmath0-reversibility of @xmath292 follows . the proof can be adapted easily to the case where @xmath253 is not dominated . as a consequence , the product @xmath295 is also @xmath8-reversible , which establishes the statement ( i ) of the theorem . finally , since @xmath41 has zero mass on the off - diagonal , it holds that @xmath303 and , clearly , @xmath304 . the proof of ( ii ) is now concluded by applying lemma [ lemhomogeneouscomp ] along the lines of the proof of theorem [ teocompalg1alg2 ] . we preface the proof of theorem [ teomainresult ] with some preliminary lemmas . [ lemlem1 ] assume that @xmath305 are @xmath0-reversible markov transition kernels . then , for all @xmath306 \times\ltwo [ \pi]$ ] , @xmath307 as each @xmath308 is @xmath0-reversible , it holds that @xmath309 for all @xmath310 \times\ltwo[\pi]$ ] and @xmath311 . applying repeatedly this relation yields @xmath312 [ lemlem2 ] let @xmath313 and @xmath2 be markov transition kernels on @xmath12 such that @xmath314 and let @xmath315 be a markov chain evolving as @xmath316 { } \stackrel { p } { \longrightarrow } \x[1 ] { } \stackrel{q } { \longrightarrow } \x[2 ] { } \stackrel { p } { \longrightarrow } \x[3 ] { } \stackrel{q } { \longrightarrow } \cdots\ ] ] with initial distribution @xmath105{}\sim\pi$ ] . then , for all @xmath317 $ ] such that @xmath318 { } \bigr ) } { f \bigl(\x[k ] { } \bigr)}\bigr\|+\bigl\|\covardu { f \bigl(\x[1 ] { } \bigr ) } { f \bigl(\x[k+1 ] { } \bigr)}\bigr\| \bigr ) < \infty,\ ] ] the limit , as @xmath85 tends to infinity , of @xmath319{})}$ ] exists , and @xmath320 { } \bigr)}\nonumber \\ & & \qquad = \pi f^2 - \pi^2 f \\ & & \quad\qquad { } + \sum_{k = 1}^{\infty } \covardu{f \bigl(\x[0 ] { } \bigr ) } { f \bigl(\x[k ] { } \bigr ) } + \sum_{k = 1}^{\infty } \covardu{f \bigl(\x[1 ] { } \bigr ) } { f \bigl(\x[k+1 ] { } \bigr)}. \nonumber\end{aligned}\ ] ] as covariances are symmetric , @xmath321 { } \bigr)}=\pi f^{2 } - \pi^{2}f + 2 n^{-1 } \sum_{0 \leq i < j \leq n - 1}\covardu{f \bigl(\x[i ] { } \bigr ) } { f \bigl(\x[j ] { } \bigr)}.\ ] ] we now consider the limit , as @xmath85 tends to infinity , of the last term on the right - hand side . let @xmath322 and @xmath323 denote the two complementary subsets of @xmath324 consisting of the even and odd numbers , respectively . for all @xmath325 such that @xmath326 , we have @xmath327 { } \bigr ) } { f \bigl(\x[j ] { } \bigr ) } = \cases { \covardu{f \bigl(\x[0 ] { } \bigr ) } { f \bigl(\x[j - i ] { } \bigr ) } , & \quad if $ i \in\mathcal{e}$ , \vspace{3pt}\cr \covardu{f \bigl(\x[1 ] { } \bigr ) } { f \bigl(\x[j - i+1 ] { } \bigr ) } , & \quad if $ i \in \mathcal{o}$.}\ ] ] this implies that @xmath328 { } \bigr ) } { f \bigl(\x[j ] { } \bigr ) } \\ & & \qquad = \sum_{k = 1}^{n - 1 } n^{-1 } \biggl ( \biggl\lfloor\frac{n - 1 - k}{2 } \biggr\rfloor+ 1 \biggr ) \covardu { f \bigl(\x[0 ] { } \bigr ) } { f \bigl(\x[k ] { } \bigr)}\end{aligned}\ ] ] and @xmath329 { } \bigr ) } { f \bigl(\x[j ] { } \bigr ) } \\ & & \qquad = \sum_{k = 1}^{n - 2 } n^{-1 } \biggl ( \biggl\lfloor\frac{n - 2 - k}{2 } \biggr\rfloor+ 1 \biggr ) \covardu { f \bigl(\x[1 ] { } \bigr ) } { f \bigl(\x[k+1 ] { } \bigr)}.\end{aligned}\ ] ] under ( [ eqeq2asslem2 ] ) , the dominated convergence theorem applies , which provides that the limit , as @xmath85 goes to infinity , of @xmath330{})}$ ] exists and is equal to ( [ eqasvaraltexpression ] ) . [ lemlem4 ] let @xmath3 and @xmath59 , @xmath331 , be @xmath0-reversible markov kernels on @xmath92 such that @xmath332 and @xmath333 . for all @xmath334 and @xmath5 , denote by @xmath335{i}$ ] the markov kernel @xmath335{i}:=p_i \mathbh{1}_{\mathcal{e}}(n ) + q_i \mathbh{1}_{\mathcal{o}}(n)$ ] . in addition , let @xmath336 $ ] be such that for @xmath337 , @xmath338{i } \cdots\r[k-1]{i}f \bigr\rangle \bigr\vert < \infty.\ ] ] then for all @xmath339 , @xmath340{1 } \cdots\r[k-1]{1}f \bigr\rangle + \bigl\langle f , \r [ 1]{1 } \cdots\r[k]{1 } f \bigr\rangle \bigr ) \\ & & \qquad \leq\sum_{k=1}^{\infty } \lambda^k \bigl ( \bigl\langle f , \r[0]{0 } \cdots\r[k-1]{0}f \bigr\rangle + \bigl\langle f , \r [ 1]{0 } \cdots \r[k]{0}f \bigr\rangle \bigr).\end{aligned}\ ] ] for all @xmath334 and all @xmath341 , define @xmath342{\alpha}:=(1-\alpha)\r[n]{0}+\alpha\r[n]{1}$ ] . in addition , set , for @xmath339 , @xmath343{}(\alpha):= \kh [ \lambda]{\mathcal{e}}(\alpha)+\kh[\lambda]{\mathcal{o}}(\alpha ) $ ] , where @xmath344{\mathcal{e}}(\alpha ) & : = & \sum _ { k = 1}^{\infty } \lambda^k \bigl\langle f , \r[0 ] { \alpha } \cdots \r[k-1]{\alpha } f \bigr\rangle , \\ \kh[\lambda]{\mathcal{o}}(\alpha ) & : = & \sum_{k = 1}^{\infty } \lambda^k \bigl\langle f , \r[1]{\alpha } \cdots\r[k]{\alpha } f \bigr \rangle.\end{aligned}\ ] ] now , fix a distinguished @xmath339 ; we want show that for all @xmath345 $ ] , @xmath346{}}{\mathrm{d}\alpha } ( \alpha ) \leq0.\ ] ] thus , we start with differentiating @xmath343{\mathcal{e}}$ ] : @xmath347{\mathcal { e}}}{\mathrm{d}\alpha } ( \alpha ) = \frac{\mathrm{d}}{\mathrm{d } \alpha } \sum _ { k = 1}^{\infty } \lambda^k \bigl\langle f , \r[0 ] { \alpha}\cdots \r[k-1]{\alpha}f \bigr\rangle.\ ] ] to interchange @xmath348 and @xmath349 in the previous equation , we first note that @xmath350{\alpha } \cdots \r[k-1]{\alpha}f \bigr\rangle&= & \sum _ { \ell= 0}^{k - 1 } \frac{\partial}{\partial \alpha_\ell } \bigl\langle f , \r[0]{\alpha_0 } \cdots\r [ k - 1]{\alpha_{k - 1}}f \bigr \rangle \bigg\vert_{(\alpha_0,\ldots,\alpha_{k - 1})=(\alpha,\ldots,\alpha ) } \\ & = & \sum_{\ell= 0}^{k - 1 } \bigl \langle f , \r[0 { \nearrow}\ell-1]{\alpha } \bigl(\r[\ell]{1 } - \r[\ell ] { 0 } \bigr ) \r[\ell+ 1 { \nearrow}k - 1]{\alpha}f \bigr\rangle,\end{aligned}\ ] ] where @xmath351{\alpha}:=\r[s]{\alpha } \r[s+1]{\alpha } \cdots\r[t]{\alpha}$ ] for @xmath352 and @xmath351{\alpha } : = \operatorname{id}$ ] otherwise . by ( [ eqmajonormp ] ) , @xmath353{\alpha } \\|\leq1 $ ] , which implies that @xmath354}\|\frac{\mathrm{d } } { \mathrm{d } \alpha } \langle f , \r[0]{\alpha}\cdots\r[k-1]{\alpha}f \rangle \pi(f^2)$ ] . thus , as @xmath355 we may interchange , in ( [ eqderivh ] ) , @xmath356 and @xmath349 , yielding @xmath357{\mathcal{e}}}{\mathrm{d}\alpha } ( \alpha)=\sum_{k=1}^{\infty } \lambda^{k}\sum_{\ell=0}^{k-1 } \bigl \langle f , \r[0 { \nearrow}\ell-1]{\alpha } \bigl(\r[\ell ] { 1}-\r [ \ell]{0 } \bigr ) \r [ \ell+1 { \nearrow}k-1]{\alpha}f \bigr\rangle.\ ] ] similarly , it can be established that @xmath357{\mathcal{o}}}{\mathrm{d}\alpha } ( \alpha ) = \sum_{k=1}^{\infty } \lambda^{k}\sum_{\ell=1}^{k } \bigl \langle f , \r[1 { \nearrow}\ell-1]{\alpha } \bigl(\r[\ell ] { 1}-\r [ \ell]{0 } \bigr ) \r[\ell + 1 { \nearrow}k]{\alpha}f \bigr\rangle.\ ] ] we now apply lemma [ lemlem1 ] to the two previous sums . for this purpose , we will use the following notation : @xmath358{\alpha } : = \r[s]{\alpha } \r[s-1]{\alpha } \cdots\r[t]{\alpha}$ ] for @xmath359 and @xmath358{\alpha}:=\operatorname{id}$ ] otherwise . then @xmath360{}}{\mathrm { d}\alpha } ( \alpha ) & = & \sum_{k=1}^{\infty } \lambda^k \biggl\{\sum_{\ell=0}^{k-1 } \bigl\langle\r[\ell-1 { \searrow}0]{\alpha}f , \bigl(\r[\ell]{1}-\r[\ell]{0 } \bigr)\r [ \ell+1 { \nearrow}k-1]{\alpha}f \bigr\rangle \\[1pt ] & & \hspace{33pt}{}+\sum_{\ell=1}^{k } \bigl\langle\r[\ell-1 { \searrow}1]{\alpha}f , \bigl(\r[\ell]{1}-\r[\ell]{0 } \bigr)\r [ \ell+1 { \nearrow}k ] { \alpha}f \bigr\rangle \biggr\ } \\[1pt ] & = & \sum_{\ell=0}^{\infty}\sum _ { m=0}^{\infty}\lambda^{\ell+m+1 } \bigl\langle\r [ \ell-1 { \searrow}0]{\alpha}f , \bigl(\r[\ell]{1}-\r[\ell]{0 } \bigr)\r [ \ell+1 { \nearrow}\ell+m]{\alpha}f \bigr\rangle \\[1pt ] & & { } { } + \sum_{\ell=1}^{\infty}\sum _ { m=1}^{\infty}\lambda^{\ell+m-1 } \bigl\langle\r [ \ell-1 { \searrow}1]{\alpha}f , \bigl(\r[\ell]{1}-\r[\ell]{0 } \bigr)\r [ \ell+1 { \nearrow}\ell+m-1]{\alpha}f \bigr\rangle.\end{aligned}\ ] ] now , note that @xmath335{\alpha}=\r[n']{\alpha}$ ] for all @xmath361 and @xmath362{\alpha}=\r[m']{\alpha}$ ] for all @xmath363 ; hence , separating , in the two previous sums , odd and even indices @xmath364 provides @xmath360{}}{\mathrm { d}\alpha } ( \alpha ) & = & \sum_{\ell\in\mathcal{e}}\sum_{m=0}^{\infty } \lambda^{\ell+m+1 } \bigl\langle\r[1 { \nearrow}\ell]{\alpha}f , \bigl(\r[0]{1}- \r[0]{0 } \bigr)\r[1 { \nearrow}m]{\alpha}f \bigr\rangle \\[1pt ] & & { } + \sum_{\ell\in\mathcal{e}\setminus\{0\}}\sum_{m=1}^{\infty } \lambda^{\ell + m-1 } \bigl\langle\r[1 { \nearrow}\ell-1]{\alpha}f , \bigl(\r [ 0]{1}- \r[0]{0 } \bigr)\r[1 { \nearrow}m-1]{\alpha}f \bigr\rangle \\[1pt ] & & { } + \sum_{\ell\in\mathcal{o}}\sum_{m=0}^{\infty } \lambda^{\ell+m+1 } \bigl\langle\r[0 { \nearrow}\ell-1]{\alpha}f , \bigl ( \r[1]{1}- \r[1]{0 } \bigr)\r[0 { \nearrow}m-1]{\alpha}f \bigr\rangle \\[1pt ] & & { } + \sum_{\ell\in\mathcal{o}}\sum_{m=1}^{\infty } \lambda^{\ell+m-1 } \bigl\langle\r[0 { \nearrow}\ell-2]{\alpha}f , \bigl ( \r[1]{1}- \r[1]{0 } \bigr)\r[0 { \nearrow}m-2]{\alpha}f \bigr\rangle.\end{aligned}\ ] ] finally , by combining the even and the odd sums , @xmath360{}}{\mathrm { d}\alpha } ( \alpha ) & = & \biggl\langle\sum_{\ell=0}^{\infty } \lambda^{\ell}\r[1 { \nearrow}\ell]{\alpha}f , \bigl(\r[0]{1}-\r[0]{0 } \bigr ) \sum_{m=0}^{\infty } \lambda^m \r[1 { \nearrow}m]{\alpha}f \biggr\rangle \\ & & { } + \biggl\langle\sum_{\ell=0}^{\infty } \lambda^\ell\r[0 { \nearrow}\ell-1]{\alpha}f , \bigl(\r[1]{1}-\r[1]{0 } \bigr ) \sum_{m=0}^{\infty}\lambda^m\r [ 0{\nearrow}m-1]{\alpha}f \biggr\rangle.\end{aligned}\ ] ] since @xmath335{1}\pgeq[1 ] \r[n]{0}$ ] , the operator @xmath335{0}-\r [ n]{1}$ ] is nonnegative on @xmath30 $ ] ( by @xcite , lemma 3 ) , and for all @xmath336 $ ] it holds that @xmath365{1}-\r [ n]{0})f \rangle\leq0 $ ] . this shows ( [ eqnegder ] ) , which implies that the function @xmath366{}(\alpha)$ ] is nonincreasing on @xmath367 . the proof is complete . proof of theorem [ teomainresult ] according to lemma [ lemlem2 ] , for all functions @xmath368 $ ] and @xmath331 , @xmath369\\[-8pt ] & & { } + \sum_{k=1}^{\infty } \bigl ( \covardu{f \bigl ( \x[0]{i } \bigr ) } { f \bigl(\x[k]{i } \bigr ) } + \covardu{f \bigl(\x [ 1]{i } \bigr ) } { f \bigl ( \x[k+1]{i } \bigr ) } \bigr ) . \nonumber\end{aligned}\ ] ] for the kernels @xmath3 and @xmath59 , @xmath331 , in the statement of the theorem , let @xmath370 , @xmath331 , be defined as in lemma [ lemlem4 ] , which then implies that for all @xmath371 , @xmath372{1 } \bigr ) } { f \bigl(\x[k]{1 } \bigr ) } + \lambda^{k}\covardu{f \bigl(\x[1]{1 } \bigr ) } { f \bigl(\x[k+1]{1 } \bigr ) } \bigr ) \nonumber\\[-8pt]\\[-8pt ] & & \qquad \leq \sum_{k=1}^{\infty } \bigl ( \lambda^{k}\covardu{f \bigl(\x[0]{0 } \bigr ) } { f \bigl(\x[k]{0 } \bigr)}+ \lambda^{k}\covardu{f \bigl(\x[1]{0 } \bigr ) } { f \bigl(\x[k+1]{0 } \bigr ) } \bigr ) . \nonumber\end{aligned}\ ] ] we conclude the proof by letting @xmath373 tend to one on each side of the previous inequality . under ( [ eqassumpfuncthm ] ) , we may , by the dominated convergence theorem , interchange limits with summation , which establishes inequality ( [ eqth4proof ] ) also in the case @xmath374 . combining this with ( [ eqth4proof1 ] ) completes the proof . in this paper , we have extended successfully the theoretical framework proposed in @xcite and @xcite as a means of comparing the asymptotic variance of sample path averages for different markov chains and , consequently , the efficiency of different mcmc algorithms to the context of inhomogeneous markov chains evolving alternatingly according to two different markov transition kernels . it turned out that this configuration covers , although not apparently , several popular mcmc algorithms such as randomized mcmc @xcite , multiple - try metropolis @xcite and its generalization @xcite , and the pseudo - marginal algorithms @xcite . it should be remarked however that our results do not take possible additional computational cost into consideration , which may be of importance in practical applications . while these algorithms are inapproachable for the standard tools provided in @xcite and @xcite , our results allow , without heavy technical developments , rigorous theoretical justifications advocating the use of these algorithms . as illustrated by our novel _ random refreshment _ algorithm in the context of pseudo - marginal algorithms , the results of the present paper can also be used for designing new algorithms and improving , in terms of asymptotic variance , existing ones . first , set @xmath375 ; then by jensen s inequality , @xmath376 and since @xmath377 , @xmath378 now , without loss of generality we may assume that @xmath379 , @xmath380 , and @xmath381 . then applying ( [ eqineq - pn - v - half ] ) yields for all @xmath23 , @xmath382 hence , for all @xmath68 , @xmath383 { } \bigr ) } { f \bigl(\x[2n ] { } \bigr)}\bigr\| & = & \bigl\|\mathbb{e } \bigl(f(x_0 ) ( pq)^n f(x_0 ) \bigr)\bigr\| \\ & \leq & \bigl(2 c \rho^n \bigr)^{1/2 } \mathbb{e } \bigl(\bigl\|f(x_0)\bigr\|v^{1/2}(x_0 ) \bigr ) \leq \bigl(2 c \rho^{n } \bigr)^{1/2 } \pi v.\end{aligned}\ ] ] in the same way , for all @xmath384 , @xmath385 { } \bigr ) } { f \bigl(\x[2n+1 ] { } \bigr)}\bigr\|=\bigl\|\mathbb{e } \bigl(f(x_0 ) ( pq)^n pf(x_0 ) \bigr)\bigr\| \leq \bigl(2 c \rho^{n } \bigr)^{1/2 } \pi v.\ ] ] by applying successively the cauchy schwarz and jensen inequalities , we obtain @xmath386^{1/2 } \leq\pi v,\ ] ] where the last inequality follows from @xmath387 and @xmath388 . this implies that for all @xmath231 , @xmath389 { } \bigr ) } { f \bigl(\x[2n ] { } \bigr)}\bigr\| & = & \bigl\|\mathbb{e } \bigl(f(x_1)q(pq)^{n-1 } f(x_1 ) \bigr)\bigr\| \\ & \leq & \bigl(2 c \rho^{n-1 } \bigr)^{1/2 } \mathbb{e } \bigl(\bigl\|f(x_1)\bigr\|qv^{1/2}(x_1 ) \bigr ) \\ & \leq & \bigl(2 c \rho^{n-1 } \bigr)^{1/2 } \pi v.\end{aligned}\ ] ] in the same way , for all @xmath231 we have , using that @xmath390 , @xmath391 { } \bigr ) } { f \bigl(\x[2n+1 ] { } \bigr)}\bigr\| = \bigl\|\mathbb{e } \bigl(f(x_1)q(pq)^{n - 1 } pf(x_1 ) \bigr)\bigr\| \leq \bigl(2 c \rho^{n - 1 } \bigr)^{1/2 } \pi v.\ ] ] the statement of the proposition follows . let @xmath110 be the transition kernel of the markov chain @xmath392 , that is , for all @xmath393 , @xmath394 where @xmath395 . thus , establishing @xmath125-reversibility of @xmath110 amounts to verifying , for all @xmath37 and @xmath38 in @xmath396 , @xmath397 indeed , by @xmath0-reversibility of @xmath131{1},\aux[k]{1 } ) ; k\in \mathbb{n } \}$ ] it holds , for all @xmath398 and @xmath399 in @xmath400 , @xmath401 which establishes ( [ eqstargrev ] ) by letting @xmath402 and @xmath403 . this completes the proof . @xmath165{2 } = y$ ] : draw @xmath404 , draw @xmath405 , set @xmath406{2 } \gets\cases { \hat{y } , & \quad w.pr . $ \displaystyle \alpha^{(\mathrm{r})}(y , u , \hat{y})$ \vspace{5pt}\cr & \quad\qquad $ \displaystyle:=1\wedge\frac{\pi^{\ast}(\hat{y } ) \check{r}(\hat{y } , y ) \check{s}(\hat{y } , y ; f(u))}{\pi^{\ast}(y)\check{r}(y,\hat{y } ) \check{s}(y , \hat{y } ; u ) } \bigg\vert\frac{\partial f}{\partial u}(u ) \bigg\vert$ , \vspace{3pt}\cr y , & \quad otherwise.}\ ] ] as proposed initially by @xcite , the r - mcmc algorithm generates a markov chain @xmath172 with transitions given by algorithm [ algalgrmcmc ] below . denote by @xmath407 the jacobian determinant of a vector - valued transformation @xmath37 . in this algorithm , @xmath37 is any continuously differentiable involution on @xmath408 . in addition , @xmath253 and @xmath409 are instrumental kernels on @xmath410 and @xmath411 , respectively , having transition densities @xmath252 and @xmath412 with respect to some dominating measure and lebesgue measure on @xmath413 , respectively . [ apprmcmc ] the r - mcmc algorithm is a special case of algorithm [ algalg2 ] . since @xmath414 and @xmath163 , obtained in steps ( i ) and ( ii ) of algorithm [ algalgrmcmc ] , are not drawn in the same order as in algorithm [ algalg2 ] , we first derive the expression of the corresponding kernels @xmath127 and @xmath132 , that is , @xmath415 where @xmath416 is lebesgue measure on @xmath417 . also note that @xmath418 moreover , introduce another auxiliary variable @xmath267 taking values in @xmath181 and being drawn according to @xmath419 . note that the kernel @xmath133 is not dominated by a common nonnegative measure regardless the value of @xmath129 ; still , following remark [ remgeneralradon ] , the r - mcmc algorithm may be covered by algorithm [ algalg2 ] , provided that the ratio in the acceptance probability @xmath420 corresponds to the radon nikodym derivative in proposition [ propderiveeradon ] for @xmath421 and @xmath422 the proof is completed by applying lemma [ lemrmcmclem2 ] below . [ lemrmcmclem2 ] the acceptance probability @xmath423 in ( [ eqacceptrmcmc ] ) is equal to @xmath424 where @xmath425 , @xmath426 , and @xmath427 denotes the radon nikodym derivative between the measures @xmath428 and @xmath429 defined by @xmath430 write @xmath431 , where @xmath432 to show ( [ eqapp2 ] ) , we will prove that for all bounded measurable functions @xmath433 on @xmath434 it holds that @xmath435 & = & \int g(x,\hat{x } ) \nu^{(\mathrm{r } ) } ( \mathrm{d}x \times\mathrm{d}\hat{x } ) \\ & = & \int g(x , \hat{x } ) \gamma ^{(\mathrm{r } ) } ( y , u , \hat{y } ) \mu^{(\mathrm{r})}(\mathrm{d}x \times\mathrm{d } \hat{x})\end{aligned}\ ] ] [ where @xmath436 and @xmath437 . now , using the change of variables @xmath438 , which is equivalent to @xmath439 ( since @xmath37 is an involution ) and using the relation ( [ eqtechnicos ] ) we obtain @xmath440 \\ & & \qquad = \int g^{(\mathrm{r } ) } \bigl(y , f(\hat{u } ) , \hat{y } , \hat{u } \bigr ) \pi^{\ast}(\mathrm{d}\hat{y } ) r(\hat{y } , \hat{u } ) s(\hat{y } , \hat{u } ; \mathrm{d}y ) \lambda _ d(\mathrm{d } \hat{u } ) \\ & & \qquad = \int g^{(\mathrm{r } ) } \bigl(y , u , \hat{y } , f(u ) \bigr ) \\ & & \hspace{42pt } { } \times \pi^{\ast } ( \mathrm{d } \hat{y } ) r \bigl(\hat{y } , f(u ) \bigr ) s \bigl(\hat{y } , f(u ) ; \mathrm{d}y \bigr ) \bigl\| ( \partial f/\partial u ) ( u)\bigr\| \lambda_d ( \mathrm{d}u ) \\ & & \qquad = \int g^{(\mathrm{r } ) } \bigl(y , u , \hat{y } , f(u ) \bigr ) \frac{\pi ^{\ast } ( \hat{y } ) \check { r}(\hat{y } , y ) \check{s}(\hat{y } , y ; f(u ) ) } { \pi^{\ast}(y ) \check{r}(y , \hat{y})\check{s}(y , \hat{y } ; u ) } \bigg\vert\frac { \partial f}{\partial u}(u ) \bigg\vert \\ & & \hspace{9pt}\quad\qquad { } \times\pi^{\ast}(\mathrm{d}y ) \check{r}(y,\mathrm { d } \hat{y } ) \check{s } ( y , \hat{y } ; \mathrm{d}u ) \\ & & \qquad = \int g^{(\mathrm{r})}(x , \hat{x } ) \gamma^{(\mathrm{r})}(y , u , \hat{y } ) \mu ^{(\mathrm{r})}(\mathrm{d}x \times\mathrm{d}\hat{x}),\end{aligned}\ ] ] which completes the proof . the gmtm algorithm proposed in @xcite generates a markov chain @xmath172 with transitions given by algorithm [ algalgmtm ] below . @xmath165{2 } = y$ ] : draw @xmath441 , let @xmath184 take the value @xmath442 w.pr . @xmath443 , let @xmath444 , draw @xmath445 , let @xmath446 , let @xmath447{2 } \gets\cases { \hat{y } , & \quad with probability $ \displaystyle \alpha^{(\mathrm{m})}(y , v , \hat{y } , \hat{v})$ \vspace{5pt}\cr & \quad\qquad $ \displaystyle : = 1 \wedge\frac{\pi^{\ast}(\hat{y } ) \check{r}(\hat{y } , y ) \omega(\hat{y } , y ) \sum_{k = 1}^n \omega(y , v_k)}{\pi^{\ast } ( y ) \check{r}(y , \hat{y } ) \omega(y , \hat{y } ) \sum_{k = 1}^n \omega(\hat{y},\hat{v}_k)}$ , \vspace{3pt}\cr y , & \quad otherwise.}\ ] ] in algorithm [ algalgmtm ] , the auxiliary variables @xmath448 are defined on @xmath249 and for all @xmath251 and @xmath449 , @xmath450 are sample weights . moreover , @xmath253 is an instrumental kernel defined on @xmath451 having the transition density @xmath252 with respect to some dominating measure on @xmath123 . [ lemmtm ] the gmtm algorithm is a special case of algorithm [ algalg2 ] . denoting by @xmath182 the random variables generated in step ( i ) in algorithm [ algalgmtm ] , the proposed candidate @xmath414 is obtained as @xmath452 , where @xmath184 is generated in step ( ii ) . let @xmath453 , where @xmath454 to obtain the joint distribution of @xmath455 conditionally on @xmath456{2}$ ] , write for any bounded measurable function @xmath433 on @xmath457 , @xmath458{2 } = y \bigr ] \\ & & \qquad = \sum_{j = 1}^n \mathbb{e } \bigl[g ( v_j , v_{-j } ) \mathbh{1}_{j = j } { \|}\y[k]{2 } = y \bigr ] \\ & & \qquad = { \int\cdots\int}\check{r}(y , \mathrm{d}\hat{y } ) \prod_{k = 1}^{n - 1 } \check{r}(y , \mathrm{d}u_k ) \frac { n \omega(y , \hat{y})}{\sum_{\ell = 1 } ^{n - 1 } \omega(y , u_\ell ) + \omega(y , \hat{y})}g(\hat{y } , u ) \\ & & \qquad = { \int\cdots\int}r(y , \mathrm{d}u ) s(y , u ; \mathrm{d}\hat{y } ) g(\hat { y } , u),\end{aligned}\ ] ] where we introduced the kernels @xmath459 now , set @xmath460 where the @xmath461 s are sampled in step ( iv ) . the distribution of @xmath267 conditionally on @xmath136{2 } , u , { \hat{y } } ) = ( y , u , \hat{y})$ ] is given by @xmath462 if @xmath253 is dominated by a nonnegative measure , then ( [ eqmtmr ] ) , ( [ eqmtms ] ) and ( [ eqmtmt ] ) show that the kernels @xmath127 , @xmath132 and @xmath133 are dominated as well . denoting by @xmath203 , @xmath145 and @xmath146 the corresponding transition densities , it can be checked readily that @xmath463 so that @xmath464 defined in ( [ eqacceptmtm ] ) corresponds to the acceptance probability @xmath147 defined in ( [ eqacceptmetropolis ] ) with these particular choices of @xmath203 , @xmath145 and @xmath146 . consequently , the gmtm algorithm is a special case of algorithm [ algalg2 ] . note that in the previous proof , we have chosen the auxiliary variable @xmath163 as the vector of rejected candidates after step ( ii ) . another natural idea would consist in choosing @xmath465 , where the @xmath466s are obtained in step ( i ) ; however , since @xmath414 belongs to this set of candidates , the model would then not be dominated , which would make the proof more intricate . we thank the anonymous referees for insightful comments that improved significantly the presentation of the paper . a special thanks goes to the referee who provided the two counterexamples in remarks [ remcounterexsummability ] and [ remcounterexlemma ] , as well as the possible application of our methodology to the abc context in example [ exrrabc ] .	in this paper , we study the asymptotic variance of sample path averages for inhomogeneous markov chains that evolve alternatingly according to two different @xmath0-reversible markov transition kernels@xmath1 and @xmath2 . more specifically , our main result allows us to compare directly the asymptotic variances of two inhomogeneous markov chains associated with different kernels @xmath3 and @xmath4 , @xmath5 , as soon as the kernels of each pair @xmath6 and @xmath7 can be ordered in the sense of lag - one autocovariance . as an important application , we use this result for comparing different data - augmentation - type metropolis hastings algorithms . in particular , we compare some pseudo - marginal and propose a novel exact algorithm , referred to as the _ random refreshment _ algorithm , which is more efficient , in terms of asymptotic variance , than the grouped independence metropolis hastings algorithm and has a computational complexity that does not exceed that of the monte carlo within metropolis algorithm . ,
mesons in which the gluons are in an excited state have been discussed for some time . there are two main pictures that have evolved for treating these excited states . the first is the constituent gluon approach where the quarks and a gluon move in an mit bag @xcite or a potential model @xcite . the second picture envisions the quarks to be connected by a string or flux tube @xcite which has quantized transverse vibrations . in this case the flux tube can be thought of as a coherent gluonic state . in all of these models the resulting meson is analogous to the diatomic molecule where the gluonic degrees of freedom are the `` electronic state '' that can assume many levels of excitation . each excited state yields an interaction energy that acts as an adiabatic potential in which the quarks or `` ions '' move . the ground state of the glue corresponds to standard meson states and the excited glue to `` hybrid meson '' states . recently the excited glue states with fixed end points have been investigated in detail by lattice simulation @xcite . these calculations are done with an improved action in the quenched approximation for a variety of gluonic operators , and on several anisotropic lattices . it is our purpose here to point out that the systematics of the gluon states are extremely simple from the vibrating relativistic flux tube point of view . to a remarkable extent the gluon states group themselves into a few highly degenerate states at large quark separations , reflecting the well known degeneracy of the quantized two - dimensional harmonic oscillator . we further show that , given the ground state potential , the hybrid adiabatic potentials are uniquely predicted and agree well with the lattice results . our calculation is fully relativistic and does not introduce arbitrary procedures as required by previous work @xcite . it is a common misconception that a free relativistic string _ must _ be formulated in twenty six dimensions in order to be consistently quantized . in fact , quantized theories of a single non - interacting relativistic string can be defined consistently in any spacetime dimension smaller than twenty six using the standard string theoretic methods . long ago , brower and goddard and thorn @xcite showed that free bosonic string theories in spacetime dimensions @xmath0 are free of ghost ( negative norm ) states as long as the first excited state is not tachyonic . subsequently rohrlich @xcite found an oscillator quantization of the non - interacting relativistic string that is manifestly free of ghosts in any dimension , while polyakov @xcite quantized the string as a sum over random surfaces that is consistent in dimensions twenty - six or smaller . it is only in the context of dual models and their superstring offspring that the theory becomes consistent in a single ( critical ) dimension . this is because unphysical states that may be consistently eliminated from a free string can not be consistently eliminated from an interacting string theory @xcite . the nambu - goto action , with fixed end boundary conditions may be quantized consistently in @xmath1 using the gupta - bleuler method in the temporal gauge , @xmath2 . the energy of a string of tension @xmath3 and distance @xmath4 between the fixed ends , @xmath5 follows from the zero mode of the virasoro constraint @xcite . the index @xmath6 labels the mode level which is occupied by @xmath7 phonons of positive helicity and @xmath8 phonons of negative helicity . the constant @xmath9 is an arbitrary normal ordering constant subject only to the constraint , @xmath10 , of the no - ghost theorem . in the temporal gauge , lorentz invariance does not impose any restriction on spacetime dimension @xmath11 @xcite as it does in light - cone gauge and only requires that @xmath9 be chosen such that the system has a rest frame . this formally infinite constant is often calculated @xcite by summing the casimir zero - point energies using zeta function regularization , yielding the value @xmath12 . the standard brst quantization method also yields the values @xmath13 and @xmath14 . the gupta - bleuler method we use here yields a consistent quantum theory of a single string for any value of @xmath15 . it is well known that different methods of quantizing theories with constraints , such as the nambu - goto string , need not be equivalent and may differ in their energy spectra as well as in their dynamical degrees of freedom . the excited glue states are completely specified by the separation between the ends and the occupation number of each of the modes , @xmath16 states with @xmath17 are denoted @xmath18 . the second label is the cp value of the glue which appears as either a subscript @xmath19 or @xmath20 depending on whether cp is even or odd respectively . the @xmath21 states ( @xmath22 ) are labeled additionally by a superscript @xmath23 denoting their parity under reflection through a plane in which the axis lies . the cp value of the flux tube is determined by @xcite @xmath24 in the vibrating flux tube model the @xmath25 and @xmath26 states are always degenerate as well . for @xmath27 and @xmath28 the glue states are uniquely @xmath29 and @xmath30 respectively . for @xmath31 the flux tube can be excited in the @xmath32 , @xmath33 , and @xmath34 states . for an arbitrary @xmath35 , states with @xmath36 can be excited . this degeneracy is a firm prediction of the flux tube picture and is expected to hold for large @xmath37 separations . in this section we compare the predictions of eq . ( [ energies ] ) for flux tube vibrations with the results of a lattice simulation of quenched qcd @xcite . the lattice energies are given relative to the ground state ( @xmath29 ) energy at a quark separation of @xmath38 , where @xmath39 is a hadronic scale distance determined @xcite from the @xmath29 data at large @xmath4 . the ground state ( @xmath29 ) potential from eq . ( [ energies ] ) with @xmath27 , together with a short distance @xmath37 color singlet potential @xmath40 and an additive constant , gives @xmath41 the existence of a potential for all values of @xmath4 requires @xmath9 non - negative . if @xmath42 , the `` cornell '' potential is recovered , @xmath43 the best fit of eq . ( [ groundpotential ] ) to the lattice data occurs in fact where @xmath42 . assuming this , we obtain for the best representation of @xmath29 @xmath44 once the above values have been fixed by the ground state lattice data , all of the excited states are predicted uniquely , with @xmath45 the gluonic excitation energies obtained by lattice simulation are shown in figs . 13 in molecular notation . in fig . 1 the @xmath46 states @xmath47 are plotted . the lowest @xmath48 ( @xmath49 ) and @xmath50 ( @xmath51 ) states are displayed in figs . 2 and 3 respectively . for @xmath52 , the @xmath53 and @xmath54 states are degenerate . in the figures we also show the @xmath55 , @xmath56 , and @xmath57 predictions superimposed upon the @xmath21 , @xmath58 , and @xmath59 lattice data @xcite . we observe general agreement at large @xmath4 . the @xmath58 and @xmath59 states follow the vibrating string prediction even down to small @xmath37 separations . the @xmath26 states do not closely resemble either the @xmath31 or @xmath57 curves . one possible explanation is that they have admixtures of higher @xmath60 excitations . because the flux tube may be viewed as a coherent gluonic state , one might expect that for small quark separation it should continue smoothly to the perturbative configuration of one gluon and a @xmath37 color octet . we test this assumption by modifying the prediction of eq . ( [ highern ] ) by adding to the excited potentials a color octet short distance repulsion , @xmath61 where @xmath62 was determined ( [ rnaught ] ) by the short distance color singlet attraction of the @xmath29 ground state . our results ( dashed curves ) indicate that this short range behavior seems to improve most of the predictions , if only marginally . relativistic vibrating string models have been widely discussed in the literature . some authors maintain that a tachyonic ground state ( @xmath63 ) is required in fewer than 26 dimensions @xcite . others maintain that there are quantization ambiguities that allow , at least for a non - interacting string , consistent quantization in four dimensions @xcite . we advocate the later interpretation and show that the vibrating ( phonon ) excitations closely correspond to a lattice simulation of qcd . the success of this simple flux tube picture strongly supports the string vibration modes as being the correct low energy degrees of freedom for the gluonic excitations in hybrid mesons . the original work of isgur and paton @xcite has the same large @xmath4 limit as our result but is intrinsically non - relativistic . since waves propagate with the speed of light on a flux tube , a relativistic model is mandated . furthermore , the isgur - paton model introduces an arbitrary radial functional dependence to avoid unphysical behavior at @xmath64 . in this function a free parameter can be tuned so that the potential has zero slope at @xmath64 for the @xmath65 state . for larger @xmath60 the intercept rises as required , but now the slope becomes progressively more negative . this effect is demonstrated numerically in @xcite . our model predicts a unique potential for the excited states once the parameters are determined by the ground state . we thank c.j . morningstar for providing us the data of reference @xcite . this work was supported in part by the us department of energy under contracts nos . de - ac02 - 76ch03000 and de - fg02 - 95er40896 . 99 t. barnes , caltech ph.d . thesis , 1977 ; + p. hasenfratz , r.r . horgan , j. kuti , and j.m . richard , _ phys . lett . _ * 95b * ( 1981 ) 299 ; + k.j . juge , j. kuti , and c.j . morningstar , `` bag picture of the excited qcd vacuum with static @xmath37 source , '' hep - lat/9709132 .	recent lattice results for the energy of gluonic excitations as a function of quark separation are shown to correspond to transverse relativistic flux tube vibration modes . for large quark separations all states appear to degenerate into a few categories which are predicted uniquely , given the ground state . epsf 6.5 in 8.6 in 0.0 in + 0.5 cm theodore j. allen _ and _ m. g. olsson + _ department of physics , university of wisconsin , + 1150 university avenue , madison , wi 53706 _ + sinia veseli + 0.1 cm _ fermi national accelerator laboratory , + p.o . box 500 , batavia , il 60510 _ 0.7 cm
the energetic gamma ray experiment telescope ( egret ) aboard the _ compton gamma ray observatory _ is sensitive in the energy range 30 mev to 30 gev ( thompson 1993 ) . it has detected @xmath750 agn ( montigny 1995 , thompson 1995 , mattox 1996a ) in the blazar class ( by which we mean the ensemble of bl lac objects , high polarization quasars ( hpq ) , and optical violently variable ( ovv ) quasars ) . the absence of pair - production absorption in the egret spectra and the fact that only sources which show parsec scale radio jet structure have been identified as egret sources indicate that the hard @xmath8-rays are emitted in a relativistic jet directed toward us . most models feature inverse compton scattering as the emission mechanism , but there is not a consensus as to the origin of the low energy photons which are scattered . it has been suggested that they might originate within the jet as synchrotron emission ( maraschi ghisellini & celotti 1992 ; bloom & marscher 1993 ) . this is designated as the synchrotron self - compton ( ssc ) process . another possibility is that the low energy photons come from outside of the jet . this is designated as the external compton scattering ( ecs ) process . dermer , schlickeiser , & mastichiadis ( 1992 ) suggested that they come directly from an accretion disk around a blackhole at the base of the jet . it was subsequently proposed that the dominant source of the low energy photons for scattering could be re - processing of disk emission by broad emission line clouds ( sikora , begelman , & rees 1994 ; blandford & levinson 1995 , levinson & blandford 1995 ; levinson 1996 ) . ghisellini & madau ( 1996 ) suggest that the dominant source of low energy photons for scattering is broad - line - region re - processing of jet synchrotron emission . hartman ( 1996 ) find that the multiwavelength spectra of 3c 279 can be adequately fit with either a ssc model or an ecs model in both the high and low states . the correlation of multiwavelength variability promises a means to distinguish the ssc and the ecs models . however , this is difficult because the sensitivity of egret is insufficient to resolve variation on time scales shorter than @xmath71 week when blazars are faint , and intense @xmath8-ray flares are infrequent . because of this , we proposed that a `` quick look analysis '' of egret data be done to detect a flare in progress . this lead to our observation of pks1622@xmath0297 . pks1622@xmath0297 has not received much attention previously ( being located in the galactic center region , l=348.82 , b=13.32 ) . it is not cataloged by hewitt & burbidge ( 1987 , 1989 ) . no optical polarization measurement , nor search for rapid optical variability has been previously reported . however , the radio properties indicate that it belongs to the blazar class . a 5 ghz flux density of 1.92jy and a spectral index of @xmath9 ( @xmath10 ) were reported by khr ( 1981 ) . steppe ( 1993 ) report 90ghz flux densities at three epochs of 1.5 , 1.8 , and 2.0 jy ; and one 230ghz observation at a flux density of 1.0 jy . preston ( 1985 ) report a vlbi correlated flux density at 2.29ghz of 0.29jy , 13% of the total . impey & tapia ( 1990 ) report a 5 ghz radio polarization of 4.6% . it was optically identified by torres & wroblewski ( 1984 ) at 21 mag , and by saikia ( 1987 ) at 20.5 mag . a red shift of @xmath11 is reported in the pks catalog ( wright & otrupcek 1990 ) . pks1622@xmath0297 was detected by rosat during the sky survey at a flux of [email protected] erg @xmath2s@xmath3 in the 0.1 - 2.4 kev energy band ( voges 1996 ) . pks 1622@xmath0297 has been deeply exposed previously by egret . a likelihood analysis ( mattox 1996b ) of the sum of egret exposure for the first half of the mission ( 4/22/91 10/04/94 , a total exposure of 1.49 @xmath13 s ) yields a 95% confidence upper limit of @xmath14@xmath2s@xmath3 ( e @xmath4 100 mev ) . it was much brighter during our cycle 4 observation . the exposure is given in table 1 for each cycle 4 viewing period ( vp ) . . gro exposure for pks 1622@xmath0297 during cycle 4 . egret exposure is for the energy selection e @xmath4 100 mev . [ cols="^,^,^,^,^,^ " , ] the position determined by likelihood analysis ( mattox 1996b ) with the egret data ( e@xmath4100 mev , vp 421.0 423.5 ) is j2000 ra = 246.49 , [email protected] . the region of position uncertainty is nearly circular with a radius of 15 at 95% confidence . the significance of the detection is 25@xmath15 . the @xmath8-ray position estimate is consistent with pks1622@xmath0297 , differing by 6 . we use the method of mattox ( 1996a ) to assess the reliability of this identification . this method uses the number density of potentially confusing sources which are as flat as pks1622@xmath0297 and as bright at 5 ghz , the fraction of @xmath71 jy sources detected by egret , and considers where pks1622@xmath0297 is located in the egret position error ellipse . because flat - spectrum sources with a flux density of at least 1.9 jy are rare ( 1 per 500 square degrees ) , the identification is good . assuming a prior probability of 5.4% that pks1622@xmath0297 is a source ( this is the fraction of blazars of this radio flux which egret detects , mattox 1996a ) , the formal confidence of a correct identification is 99.6% . we show below that the @xmath8-ray source exhibits dramatic variability . the only type of identified egret source which shows this type of variability is the blazar type of agn . because pks1622@xmath0297 is the only bright radio source with blazar properties near the egret position , the identification is even more secure than the formal confidence given above . the observations shown in table 1 have been analyzed to obtain an egret light curve for the event energy selection e @xmath4 100 mev . the exposure was binned according to the quality of the egret sensitivity and the strength of the emission to obtain as much time resolution as possible given the statistical limitations imposed by the very sparse egret data . the egret sensitivity to pks 1622@xmath0297 was limited because it was observed at a substantial off - axis angle ( except during vp 423.5 ) and the response of egret at the time was about half of what it originally was ( sreekumar 1996 indicate a response in vp 423 of 42% of the original response for e @xmath4 100 mev ) due primarily to degradation of the spark - chamber gas . maps of counts and exposure were constructed for adjoining time intervals . a likelihood analysis ( mattox 1996b ) of these maps was done to obtain flux estimates . the substantial flux of pks 1622@xmath0253 ( egret team publication , in preparation ) was simultaneously fit in order to include it in a background model . we have also looked carefully at nearby source 2eg j1631 - 2845 during our observation . it is not significantly detected in the sum of vp 421.0 thru 423.5 . the 95% confidence flux upper limit is @xmath16@xmath2s@xmath3 ( e @xmath4 100 mev ) . a search for a short timescale flare of 2eg j1631 - 2845 found only the expected indication of variability due to nearby pks 1622@xmath0297 . the result for pks 1622@xmath0297 is shown in figure 1 . the flux is observed to vary dramatically during this observation . a major flare occurred at mjd 49894 ( mjd @xmath17 jd @xmath0 2400000.5 ) . the maximum flux was @xmath18@xmath2s@xmath3 ) for the 0.14 day bin centered on mjd 49892.77 . this small bin was chosen after the analysis of cumulative counts described below . it is expected that this maximum flux is somewhat enhanced by selection . the cumulative counts analysis described below indicates that the apparent decrease in flux between this bin and the subsequent bin has a 1.5% probability of being statistical . a more conservative estimate of the maximum flux of @xmath1@xmath2s@xmath3 ( e @xmath4 100 mev ) results from an analysis of the combination of these two bins and the following bin ( mjd 49892.7 49893.7 ) . the determination of the minimum time scale of variability of the high - energy @xmath8-ray flux of blazars is not straight forward because of the sparse egret statistics . the average flux of pks 1622@xmath0297 in cycle 4 divided by the average egret sensitivity is a count rate of @xmath71/hour . however , the flare at mjd 49894 to a flux @xmath75 times this average potentially provides the statistics to detect variability on a short time scale . therefore , the data has been examined in detail . the available information is summarized in the time history of cumulative counts shown in figure 2 . this figure indicates that the onset of the mjd 49894 flare was rapid . there were 5 counts detected in the orbit at mjd 49892.72 and 7 counts in the next orbit . in all eight preceding orbits , only 3 counts were detected . unfortunately , the last of these eight orbits had only a small amount of exposure due to limited availability of tdrss satellites for telemetry downlink . this increases the difficulty of determining the time and time scale of the flare onset ( intensifying the problem caused by sparse statistics ) . we will begin by finding a time scale for which the significance of a flux increase is beyond dispute . the significance of variability is assessed by a comparison of the observed cumulative counts and those expected for an invariant flux . the standard kolmogorov - smirnov ( ks ) test uses the maximum of the absolute value of the difference between observed and expected cumulative counts . kuiper s variant of this test ( press 1992 ) uses the sum of the maximum positive difference and the absolute value of the maximum negative difference . for figure 2 , the maximum positive and negative differences are 18.0 and 31.5 counts respectively . confirming what is obvious by glancing at figure 1 , the significance of variation using kuiper s variant of the ks test is for the interval mjd 49889.5 49897.5 . now that variability has been established , we may confidently seek variability in shorter intervals . we apply the standard ks test to the ten orbit interval described above , mjd 49892.192 thru 49892.812 . the maximum deviation of the observed counts from those expected for invariant flux is 8.5 counts at mjd 49892.68 ( where a cumulative 11.5 of 15 counts is expected for invariant flux ) . the corresponding significance of variability is 6 . the high significance reported above for flux variation in the 8 day interval allows us to obtain this result with not more than 10 trial intervals . now that we have established that a significant flux increase occurred at mjd 49892.7 , it is possible to examine the extent to which shorter time scale can be deduced . this can be done without substantial trials to dilute the significance of the result because we examine a very limited number of sub - intervals within the interval for which a significant flux increase has already been established . we apply the ks test to the five orbits ending with the two bright orbits ( mjd 49892.517 49892.812 ) . the expected time history of the invariant flux for this interval is shown with a short , thin line in figure 2 . the maximum deviation of the observed counts is 5.7 counts at mjd 49892.68 ( where a cumulative 6.7 of 13 counts is expected for invariant flux , and 1 is observed ) . the standard ks test indicates a flux increase with a significance of 9 . this implies a flux increase in less than 7.1 hours with 99% confidence . it is conventional to report the observed time for a doubling of the flux . however , it is not possible to measure a doubling time directly in this instance because of limited statistics . a likelihood analysis for the 3 day interval before the flare ( 9889.7 9892.7 ) yields a flux of @xmath19@xmath2s@xmath3 ( e @xmath4 100 mev ) . this flux uncertainty is the 1@xmath15 error . the conservative peak flux obtained above is @xmath20 . an integration of the probability distribution function for the factor by which the flux increases obtained from the ratio of these two gaussian distributions ( papoulis 1991 ) , @xmath21 indicates with 95% confidence that the flux increase is greater than a factor of 3.6 . assuming an exponential rise , the corresponding upper limit on the doubling time is 3.8 hours . a rapid flux decrease is also noted . a ks test shows variability ( with 99.996% confidence ) for the interval mjd 49886.8 49888.6 . two brief intervals of high flux are apparent at mjd 49886.85 and 49887.61 . the second is apparent with a ks test at 99.91% confidence for the interval mjd 49887.593 49887.996 . the expected time history of the invariant flux for the latter interval is also shown with a short , thin line in figure 2 . this interval indicates a flux decrease in less than 9.7 hours from a flux of @xmath22 ( during mjd 49887.468 49887.624 ) to @xmath23@xmath2s@xmath3 , e @xmath4 100 mev ( during mjd 49887.660 49887.946 ) . it is possible that the time scale of the flux change is not resolved with egret , and is much faster . if so , the observations of the proposed glast satellite ( michelson 1996 ) will be of great interest with 10 times the effective area of egret . the possibility of a flux change in less than 1 hour is very interesting . this would severely constrain the size ( or the doppler factor ) of the emission region . the too pointing in vp 423.5 lead to an osse detection ( kurfess 1995 ) . osse observes in the 50 kev to 10 mev energy range . its small field of view limits its observations to one object at a time so that coordinated observations of specific egret targets must be planned in advance or arranged via a too . subsequent to the too pointing , osse was able to continue monitoring pks 1622@xmath0297 during vp 424 with the reduced sensitivity provided by two of the four detectors at an off - axis angle of 2.3 . we have obtained an osse / egret spectrum for vp 423.5 which is strictly simultaneous . the counts matrix and the response matrix for the standard egret spectral analysis ( nolan 1993 ) were converted to the xspec format . the xspec program was used to simultaneously analyze the egret and osse results . the result is shown in figure 3 . the osse data alone are well represented ( reduced @xmath24 ) by a power law with a photon spectral index of @xmath25 . the best power - law fit to egret data alone has a photon spectral index of @xmath26 . with a reduced @xmath27 , the egret fit is adequate but not compelling . it appears to break gradually to a steeper spectrum with increasing energy . similar convexity is apparent in the egret spectra of several other blazars ( see montigny 1995 ) , especially 1633 + 382 ( mattox 1993 ) . although these osse and egret spectra have the same index , the normalization of the egret spectrum is a factor of 5 larger than that for the osse spectrum at 10 mev . when the osse and egret data are fit together with a single power law , the result shown with the solid line in fig . 3 is obtained . the photon spectral index is @xmath28 . the fit is not excellent , but acceptable with a reduced @xmath29 . for comparison , the reduced @xmath30 for the total osse and egret data for the two separate power - law fits described above is 0.86 . the f test indicates that the separate fits do not offer a significant improvement ( with a 14% chance probability ) . it is clear that the break at a few mev to a harder x - ray spectrum which has been reported for 4 of the 5 blazars previously detected by both egret and osse ( mcnaron - brown 1996 , observations were simultaneous for one object , only contemporaneous for the others ) is not apparent for pks 1622@xmath0297 . we examined the egret data for evidence of spectral evolution during the flare . the result is shown in the scatter plot of fig . we have done a likelihood analysis of the flux for two energy intervals : the 100 300 mev and e @xmath4 300 mev . we analyzed time intervals that were as short as possible to provide sensitivity to spectral evolution on the short timescales seen for flux changes . however , each interval had to be long enough to provide sufficient statistics . thus , we used shorter time intervals when the flux was large . the intervals were primarily formed by combining the time bins shown in figure 1 . however , the b and c intervals correspond to a single interval in figure 1 . this interval was split after an examination of a scatter plot of event energy verses time indicated that the emission appeared much harder during the first part of the interval . interval b corresponds to the interval of high flux apparent in figure 2 at mjd 49886.85 . the exact time intervals can be obtained from the caption of figure 4 . assuming an invariant spectrum , the ratio of these fluxes is expected to be consistent with that found for an analysis of the total exposure . under this assumption , it is appropriate to assume the flux ratio observed in the total exposure and fit a single intensity parameter for each time interval . a @xmath30 statistic is then obtained by summing the deviation of both energy ranges for each time interval . the result for all 12 time intervals ( @xmath31 for 12 degrees of freedom ) indicates spectral variation with only 77% confidence . we have also done the analysis with 3 time bins ( 49876.2 49892.7 , 49892.7 49895.6 , 49895.6 49908.6 ) and see colors consistent with an invariant spectrum . it is interesting that all three intervals which are at a peak in fig.1 ( b , e , & j ) are harder than the average . this is consistent with the report of mukherjee ( 1996 ) of a marginally significant indication that the egret spectrum of pks 0528 + 134 was harder during an interval of high emission than it was during lower emission level intervals . the hardest interval is b for which @xmath32 ( for 1 degree of freedom ) . this corresponds to a 2.4@xmath15 deviation from an invariant spectrum . the most energetic pks1622@xmath0297 event ( [email protected] gev ) occurred during this interval at mjd 49886.897 . we note that the spectrum appears to evolve from hard to soft ( clockwise in figure 4 ) for all 3 peaks . further observation is required to confirm this indication of spectral evolution of the emission of blazars . the large flux of pks 1622@xmath0297 was apparent to the egret team during a routine `` quicklook '' analysis of the vp 421 data . the news of the detection was conveyed to us on 6/14/95 and triggered our too program . fortuitously , this followed the first week of a 5 week observation of the galactic center region which included pks 1622@xmath0297 , and this source was opposite the sun . the detection was announced in an iau circular ( mattox 1995 ) . we also notified directly a group of multiwavelength observers who had been previously enlisted . unfortunately , the nearly anti - solar position of pks1622@xmath0297 at the time prevented x - ray observation by either rosat or asca , and the _ rossi _ xte satellite had not yet been launched . detailed multiwavelength results including multiwavelength spectra will be published latter ( mattox in preparation ) . we will also report subsequently on possible batse and comptel detections and a possible correlation between the egret and osse flux in vp 423.5 . extensively optical monitoring of pks 1622@xmath0297 commenced when we learned of the egret detection . we found that pks 1622@xmath0297 was 3 magnitudes brighter than its quiescent state , and that the brightness was observed to vary by as much as 150% in less than 24 hours . thus , pks1622@xmath0297 displays optical idv ( intra - day variability ) as do all other blazars detected by egret which have been observed frequently enough for idv to be detected ( wagner 1996 ) . several radio observations in the mm band occurred which showed a flux density significantly higher than previously observed . it was also detected by iue ( bonnell 1995 ) during observations on july 1 , 5 , 6 , 8 , and found to be variable . in collaboration with alan marscher , a vlba observation of pks 1622@xmath0297 was made on 7/25/95 . subsequent observations will be used to search for a new radio component corresponding to the flare . the maximum egret flux observed from pks1622@xmath0297 @xmath33@xmath2s@xmath3 , e @xmath4 100 mev ) was a factor of 2 larger than that of the vela pulsar which is normally the brightest egret source . it is a factor of 4 larger than the flux of 3c 279 at the peak of the 1991 flare ( kniffen 1993 ) . a peak energy flux of 1.6 ergs @xmath2s@xmath3 for the observed energy range 30 mev to 10 gev is obtained for pks1622@xmath0297 assuming a power law spectrum with photon index of @xmath01.9 . the corresponding isotropic luminosity is 2.9 erg / s assuming a friedmann universe with @xmath34 and @xmath35 . for this luminosity to be less than the standard eddington limit , the central black hole mass must exceed 2 m@xmath36 . however the standard eddington limit does not pertain to the luminosity because the compton scattering cross section in the klein - nishina regime is much smaller than the thomson cross section ( dermer & gehrels 1995 , pohl 1995 ) . using the expression of dermer & gehrels ( 1995 ) for the cross section in the klein - nishina regime for an egret flux of @xmath37@xmath2s@xmath3 ( e @xmath4 30 mev ) , the lower limit for the mass of the pks1622@xmath0297 black hole is 8 m@xmath36 . the corresponding limit on the schwarzschild radius is @xmath38 cm . we apply the `` elliot - shapiro argument '' ( elliot & shapiro 1974 ; dermer & gehrels 1995 ; pohl ; dondi & ghisellini 1995 , we note that they err in assuming the thomson cross section for compton scattering in the klein - nishina regime ) to establish that the emission must occur in a relativistic jet by showing that a contradiction follows otherwise . under the assumption that the emission is unbeamed , the observed upper limit on the doubling time of @xmath39 hours implies that the extent of the emission region is less than @xmath40 cm . the fact that this upper limit is less than lower limit for @xmath41 suggests that the emission occurs in a medium which is moving relativistically with a lateral extent less than @xmath42(@xmath43 ) cm , and an extent along the jet of less than @xmath44@xmath45 cm , where @xmath46 is the relativistic doppler factor , @xmath47 is the angle between the jet axis and the line of sight , and @xmath48 , where @xmath49 is the bulk velocity in units of the speed of light . this argument assumes that the accretion of optically thin material is the immediate source of power for the s. this may not be the case . accretion energy could be stored in the rotation of a black hole , or in magnetic fields to power the flare ; or the flare could result from the accretion of a star . also , for a kerr blackhole , emission could occur as close as @xmath50 . another argument for emission in a relativistic jet can be made from the lack of @xmath51 absorption in the spectrum . if the emission does not take place in a relativistic jet , the emission would be absorbed by /x - ray pair production . mattox ( 1993 ) derived the expected optical depth under the assumption that the x - rays are produced co - spatially with the s. their expression was in error due a misunderstanding of the definition of luminosity distance . the corrected expression ( assuming a friedmann universe with @xmath34 ) is @xmath52 where @xmath53 is the time scale of variation in units of @xmath54 seconds , @xmath55 is the observed x - ray flux at a kev , @xmath56 is the x - ray energy spectral index , @xmath57 , @xmath58 is the energy , and @xmath59 . the luminosity distance error of mattox ( 1993 ) also affected their expression for the lower limit of the doppler factor . the corrected expression is @xmath60^ { -(4 + 2\alpha)^{-1 } } \ \ \ \ \ ( 2)\ ] ] assuming a spectral index ( @xmath61 ) typical of blazars , the rosat sky survey flux for pks1622@xmath0297 is @xmath62 . the corresponding optical depth from equation ( 1 ) is @xmath63 no indication of such absorption is apparent in the spectrum of figure 3 . the lower limit for the doppler factor from equation ( 2 ) ( for @xmath64 , and @xmath65 for @xmath66 gev ) is @xmath67 . we note that the x - ray observation was not simultaneous . if it were a factor of 3 lower during the flare , the lower limit on @xmath68 would decrease to 6.6 . we report the brightest , most luminous blazar ever detected . it shows the most rapid flux change yet seen for any blazar , with a flux doubling time of less than 3.6 hours . this is the first observation of idv for a blazar . we show that the elliot - shapiro argument and the lack of @xmath51 absorption of s indicate that the emission occurred in a relativistic jet . our results are not inconsistent with the prediction of ghisellini & madau ( 1996 ) of an invariant spectrum throughout the flare . if their model of compton scattering of synchrotron photons reprocessed in the broad - line region pertains , the lightcurve shown in fig . 1 might indicate the radial profile near the jet of the density of the broad - line region . romanova and lovelace ( 1996 ) have also developed a model for blazar flares which is consistent with our data . in their model , flares occur when electrons are accelerated by a shock in the jet . our results do not demonstrate the soft to hard spectral evolution expected for a emitting jet component emerging from a photosphere ( blandford & levinson 1995 , levinson & blandford 1995 ; levinson 1996 ) . however , a comparison of a detailed calculation based on their model to these data must be done before quantitative limits can be placed on their model . although the statistics of this exposure are not sufficient to establish this beyond doubt , the high - energy spectrum appears to evolve from hard to soft as a flare progresses , the opposite of the prediction of this model . we are grateful to the egret team for the timely processing of telemetry and data to obtain the `` quicklook '' detection of pks 1622@xmath0297 , and to anonymous referees for useful comments . j. mattox acknowledges support from nasa grants nag5 - 2833 and nagw-4761 .	we report the observation by the _ compton gamma ray observatory _ of a spectacular flare of radio source pks1622@xmath0297 . a peak flux of @xmath1@xmath2s@xmath3 ( e @xmath4 100 mev ) was observed . the corresponding isotropic luminosity is 2.9 erg / s . we find that pks 1622@xmath0297 exhibits intra - day variability . a flux increase by a factor of at least 3.6 was observed to occur in less than 7.1 hours ( with 99% confidence ) . assuming an exponential rise , the corresponding doubling time is less than 3.8 hours . a significant flux decrease by a factor of 2 in 9.7 hours was also observed . without beaming , the rapid flux change and large isotropic luminosity are inconsistent with the elliot - shapiro condition ( assuming that gas accretion is the immediate source of power for the s ) . this inconsistency suggests that the emission is beamed . a minimum doppler factor of 8.1 is implied by the observed lack of pair - production opacity ( assuming x - rays are emitted co - spatially with the s ) . simultaneous observation by egret and osse finds a spectrum adequately fit by a power law with photon index of @xmath01.9 . although the significance is not sufficient to establish this beyond doubt , the high - energy spectrum appears to evolve from hard to soft as a flare progresses . # 1@xmath5 # 1@xmath6 to be published in ap j , february 20 , 1997 issue , vol . 476 quasars : general quasars : individual : pks 1622@xmath0297 gamma rays : observations
the ac stark shift of the atomic ground state level and the phase shift of the transmitted field are both governed by the response of the two - level atomic system to the laser field , and in particular , the mean optical coherence induced in the ions . we denote the lower and upper levels @xmath85 and @xmath86 , and we assume that we only excite the ions very weakly , i.e. , the ground state population @xmath87 to first order in the rabi frequency @xmath88 , where @xmath89 is the dipole matrix element and @xmath90 denotes the driving field amplitude . this leads to the equation of motion for the optical coherence @xmath91 where the decoherence rate @xmath40 may include both radiative ( @xmath92 ) and non - radiative contributions , and where we have defined @xmath93 as the detuning , measured in radians per second . the steady state solution reads , @xmath94 , and the resulting macroscopic polarization of the doped cantilever ( with an ion density @xmath55 ) is then given by @xmath95 , leading to the optical phase shift and energy @xmath96 analyzed in the article . the purpose of this appendix is to analyze the case where @xmath7 performs oscillatory motion , @xmath97 , and the strain induced coupling leads to a temporally modulated detuning , @xmath98 . due to the long lifetime of the ion coherence and excited state , we can not merely assume that the coherence @xmath99 attains the stationary solution @xmath100 , evaluated at the time dependent value of the detuning . we can assume , however , that the ions reach a periodic steady state , and for weak modulation , we make the ansatz , @xmath101 applying this ansatz in , and isolating terms @xmath102 leads to the relations , @xmath103 , @xmath104 where @xmath105 . inserting these results in the equation for the non - oscilating terms in the periodic solutions to , we obtain , @xmath106 , which holds to first order in the small modulation amplitude of the detuning @xmath107 . if @xmath109 , and if @xmath110 , the expression simplifies , @xmath111 \simeq \frac{-\omega } { 2\delta_r(t)},\ ] ] and the coherence , indeed , follows the stationary solution adiabatically in this coupling regime . when @xmath112 is comparable to or larger than @xmath113 , we see a more complicated dependence of the real ( dispersive ) and imaginary ( absorptive ) part of the coherence , and also a weaker dependence on the position in the high frequency limit , where the coherence is unable to follow the time dependent steady state . i. yeo , p .- de assis , a. gloppe , e. dupont - ferrier , p. verlot , n. s. malik , e. dupuy , j. claudon , j .- m . g ' erard , a. auffves , g. nogues , s. seidelin , j. poizat , o. arcizet , and m. richard , nat . nanotechnol . 9 , 106 ( 2014 ) .	by spectrally hole burning an inhomogeneously broadened ensemble of ions while applying a controlled perturbation , one can obtain spectral holes that are functionalized for maximum sensitivity to different perturbations . we propose to use such hole burnt structures for the dispersive optical interaction with rare - earth ion dopants whose frequencies are sensitive to crystal strain due to the bending motion of a crystal cantilever . a quantitative analysis shows that good optical sensitivity to the bending motion is obtained if a magnetic field gradient is applied across the crystal during hole burning , and that the resulting opto - mechanical coupling strength is sufficient for observing quantum features such as zero point vibrations . the ability to control and probe physical systems at the quantum mechanical level has been demonstrated in a variety of examples , ranging from single photons and atoms , over currents and voltages in electronic circuits to the motion of mechanical devices . many of these systems are promising candidates for sensitive and high precision measurements and for transmission , processing and storage of quantum information , but one single system is typically not adequate for all of these functions . this has spurred the interest in so - called hybrid system @xcite which combine physical components which are separately optimized for different tasks . outside the technical challenge of handling physically very different systems in a single laboratory experiment , the mismatch of their physical properties ( resonance excitation frequencies , spatial overlap , and coherence time scales ) presents a main obstacle against the efficient transfer of quantum states between them . for instance , single atoms have microscopic dipole moments and interact only weakly with physical observables of mesoscopic quantum systems which occupy orders of magnitude larger spatial volumes . one successful remedy to this weak coupling is to use ensembles of many atomic particles with correspondingly increased coherent coupling strength . this is , indeed , the rationale behind the use of atomic ensembles for optical interfaces and memories and of spin ensembles in conjunction with superconducting circuits @xcite . in this article , we address the application of rare - earth ion doped crystals for hybrid quantum technologies . such crystals have found rich applications in quantum communication protocols where their strong inhomogeneous broadening is favorable for speed and bandwidth . one particular hybrid technology which holds promise for an efficient coupling between radically different degrees of freedom relies on opto - mechanical interactions @xcite . major achievements such as the ability to prepare a mechanical oscillator in the quantum ground state @xcite have spurred ambitious goals to prepare non - classical states of motion and use such systems in precision measurements and quantum information applications . while mechanical oscillators can be coupled via light beams to other systems , such as atomic ensembles @xcite we propose a simpler set - up in which the bending motion of an inorganic crystal is coupled to the optical transitions in an ensemble of rare - earth ion dopants in the same crystal . the crystal strain generated by the harmonic motion of a bulk mechanical oscillation provides an intrinsically stable coupling to the dopant optical properties as , contrary to other coupling mechanisms to externally imposed fields , it circumvents instabilities arising from drifts in position of field source and oscillator . weak strain coupling to single emitters has been previously observed @xcite , but using a rare - earth ions ensemble , we benefit from their strong collective coupling as well as their narrow homogeneous linewidths and long coherence times , allowing unambiguous observations of quantum features . we suggest to use spectral hole burning to prepare a transmission window with no optical absorption and obtain the opto - mechanical coupling through the off - resonant , dispersive interaction with ions with transition frequencies outside the hole . as a concrete application , we will focus on @xmath0 ions in a @xmath1 host matrix as they exhibit record - narrow optical transitions among solid - state emitters . despite its long coherence time @xcite , the 580 nm optical transition @xmath2 @xmath3 in @xmath0 ions is sensitive to crystal strain @xcite , making it an attractive candidate system for strain - coupled optomechanics . we investigate a crystal in the shape of a micro - meter scale cantilever whose bending produces a local strain - induced perturbation to the ions resonant frequencies . schematics of a cantilever anchored in the foreground of the drawing , and with the far end ( a ) in its equilibrium position , and ( b ) bent upwards . the colors on the cantilever indicate different values of the strain : green corresponds to a non - strained material , blue and red to compressive and tensile strain , respectively . the white dashed lines indicate planes of equal strain , neglecting variations along @xmath4 . the bottom graphs illustrate how a single emission line is shifted due to the strain ( arbitrary units).,width=302 ] during bending , there will be a strain gradient across the cantilever , ranging from tensile to compressive strain ( see fig . [ peda_fig ] ) , and ions with identical frequencies in the unbent crystal ( a ) experience different frequency shifts in a bent cantilever ( b ) . in the latter case , ions located in three different layers ( subject to identical strain within each layer ) experience different strain and the emission lines shift accordingly : if the oscillator is bent upwards , emitters on the top face experience a compression , whereas emitters on the bottom face are subject to a tensile strain . as the cantilever vibrates , the collective line shape periodically broadens and narrows , and the motion should be readily detected by an optical probe . however , if static local perturbations of the emitters lead to additional inhomogeneous broadening with a very wide absorption line shape , the strain induced frequency shifts can not be probed dispersively . to circumvent this , we propose to prepare the ion ensemble by burning of narrow spectral holes with the cantilever at rest , and subsequently use their optical response to the cantilever motion . spectral holes are formed by resonant optical excitation and decay of the ions into other long - lived ( dark ) states , typically other hyperfine state within the electronic ground state manifold . due to the narrow homogenous linewidths , one can scan the excitation laser over a finite frequency interval and burn a spectral hole which must be narrower than the ground state hyperfine splitting to avoid repopulating the hole by subsequent excitation of the dark ions . a weak optical field with a frequency at the center of the hole burnt interval will subsequently interact dispersively with the ions with frequencies outside the spectral hole , and we will in the following propose a method to make this dispersive coupling sensitive to the cantilever motion . as indicated by the example in fig . [ peda_fig ] , bending of the crystal shifts the transition frequency by an amount proportional to the strain experienced by the individual ions . this implies that , to lowest order , the weak probe field is not sensitive to the bending , unless we can ensure an asymmetric response of the ion frequency distribution . such an asymmetry can be prepared by a particular hole burning protocol as discussed in the following sections . for our analysis we assume a spatially homogeneous ion distribution , and we also consider the frequency distribution due to the inhomogeneous broadening to be constant within a wide detuning range . we assume that the bending of the beam shifts the resonance frequency of all ions by a quantity proportional to the strain at their position in the crystal . the bending of the cantilever resonator with thickness @xmath5 and length @xmath6 is represented by a single collective coordinate @xmath7 , measured as the vertical displacement of the cantilever tip shown in fig . [ peda_fig ] . the ions are all located near the anchoring of the cantilever with spatial coordinates @xmath8 , where @xmath9 is in the same direction as @xmath7 and runs from @xmath10 to @xmath11 , and @xmath4 is along the beam axis , and takes values from @xmath12 to @xmath6 . from the euler - bernouilli theory of beams , we know that the local strain of the matrix around ion @xmath13 is proportional to both @xmath7 , @xmath9 and @xmath14 . we optically address only atoms near the base of the cantilever , and we therefore neglect the @xmath4 dependence , so the strain induced frequency shift reads @xmath15 , with @xmath16 a constant depending on the beam geometry and physical properties of the doped material . to enhance the position sensitivity of the cantilever , we prepare our system by applying conventional spectral hole burning while the unbent crystalline cantilever is subject to a magnetic gradient . we assume a convenient choice of light polarization and magnetic field directions with reference to the crystalline axes so that all ions are shifted by the linear zeeman effect such that the overall transition frequency of the @xmath17 ion at position @xmath8 can be written as @xmath18 where @xmath19 is the chosen central frequency for our hole burnt structure , @xmath20 is the intrinsic inhomogeneous frequency shift of the @xmath17 ions ( without the effect of cantilever motion or magnetic field ) , @xmath21 is the externally applied magnetic gradient and @xmath22 the linear zeeman effect sensitivity . we finally also assume that @xmath20 and @xmath9 are uncorrelated . our functionalized hole burning procedure consists of two steps : * 1 ) * apply a static magnetic gradient @xmath23 and scan the hole burning laser across the spectral range @xmath24 $ ] within the inhomogeneous linewidth . the value of @xmath25 can be chosen arbitrarily , but must fulfill the condition @xmath26 . this step is represented in fig . 2a . * 2 ) * apply the reverse static magnetic gradient @xmath27 and scan the hole burning laser across the interval @xmath28 $ ] . this step is represented in fig . 2b . as shown in fig . 2 , after the magnetic gradient has been switched off , we obtain an inhomogeneous absorption profile of the ions with a transmission hole around the central frequency @xmath19 , and with a width that depends on the location @xmath9 in the cantilever but is on average close to @xmath29 . we are then ready to probe and interact with the system by transmission of a weak laser beam with frequency @xmath19 , at the center of the hole . in fig . 2c , we show how the frequency shift @xmath15 due to bending of the cantilever transforms the shape of the transmission window . in particular , we note that our hole burning procedure ensures a non - vanishing phase shift of the probe beam due to changes in @xmath7 , because contributions from ions with positive @xmath9 are not canceled by the ions with negative @xmath9 as they are further detuned . at a finite temperature , corrections to the hole shape arise because the thermal brownian motion of the cantilever amplitude during hole burning exposes the whole inhomogeneous profile to strain distortions like the one shown in fig . 2c . these distortions lead to excitation and decay into dark states of those ions that are brought into resonance with the hole burning laser during their random brownian motion . we estimate the consequences of this mechanism by assuming that ions with frequencies within @xmath30 of the spectral hole in fig . 2a are also transferred to dark states , where @xmath31 is of same order of magnitude as the root - mean - square thermal excursions of @xmath7 . fig . 2d shows how the spectral hole is modified by the thermal brownian motion under hole burning and effectively excludes ions at vertical position @xmath9 that are detuned with respect to @xmath19 by an amount @xmath32 $ ] , where @xmath33 and @xmath34 for positive ( negative ) values of @xmath9 . to describe the interaction between the probe laser field and the hole burnt ensemble , we use a semi - classical approach based on the maxwell - bloch equations , which , on the one hand , accounts for the state of the individual atoms by solution of the quantum optical master equation and , on the other hand , describe the damping and dispersion of the maxwell electric field amplitude due to the interaction with the mean atomic dipole density . due to the spectral hole , the laser is far off - resonant from all atomic transitions , and we are in the dispersive and perturbative regime of linear excitation of the atoms . the cantilever is a mechanical resonator of eigenfrequency @xmath35 , for which we assume a small modulation amplitude . under this assumption , and imposing the extra condition that @xmath36 for adiabatic following ( see appendix ) , the phase shift of the probe field with wavelength @xmath37 and cross section @xmath38 interacting with a two - level system with detuning @xmath39 and radiative decay rate @xmath40 is given by @xmath41 per atom . the phase change @xmath42 of the classical ( coherent state ) field amplitude is equivalent to a phase change of the same value occurring on each single photon due to the presence of the ion . the combined state of the ion and the stream of photons thus accumulates a quantum phase @xmath43 that changes at a rate @xmath44 , where @xmath45 ( @xmath46 ) is the rate of photons passing through @xmath38 with intensity @xmath47 . the ion is left in its ground state and the photons leave with unchanged frequency , and the time evolving phase due to the light - matter interaction is equivalent to the light induced ac stark shift of the ion due to the laserfield , @xmath48 where @xmath49 . the expression for @xmath50 has , indeed , the familiar form @xmath51 , valid for large detunings . we note that we applied a simple two - level description and to evaluate the accurate value of the optical phase shift and atomic energy shift , one must specify the actual transition , the light polarization , and also the index of refraction of the host material . since the ion transition frequencies are modified by the crystal strain , their ac stark shifts become functions of @xmath7 as well as a of the optical intensity , i.e. , we obtain the desired optomechanical coupling . the total energy shift of all ions - due to the dispersive light matter interaction , is the sum of eq.([phaseshift ] ) over all ions . schematics of the result of the hole burning procedure on the ground state ion distribution as a function of the ion position and `` intrinsic '' inhomogeneous frequency detuning @xmath39 . the lines denote the border between the zones with large @xmath52 values where the ions are left in the ground state , and the central zones where they transferred to the dark state during burning . in the first step of the hole burning procedure ( a ) , the hole is burned in the @xmath24 $ ] interval , indicated with dashed lines , while the magnetic gradient @xmath23 is on . the solid lines show the hole after the magnetic gradient is turned off , realizing the `` right '' edge of the final hole . the second step ( b ) realizes the `` left '' edge by hole burning the interval @xmath28 $ ] while the magnetic gradient @xmath27 is on . the solid lines show the final hole after the 2 steps are completed and the magnetic gradient is turned off . panel ( c ) shows the asymmetric effect of bending after burning both edges ( in black , the unbent cantilever , and in red , the cantilever bent by a positive amount @xmath7 ) . panel ( d ) shows the effect on the hole due to thermal brownian motion during the burning procedure : dashed lines correspond to burning at zero temperature and solid lines to burning with finite brownian motion of the cantilever . , width=302 ] to evaluate the dependence of the total interaction energy on the collective cantilever coordinate @xmath7 , we integrate the contributions over the thickness @xmath5 of the cantilever and the corresponding distribution of @xmath53 from all ions outside the spectral hole : @xmath54.\end{gathered}\ ] ] here , @xmath55 is the density of emitters per unit of frequency and unit vertical distance , and the interaction limits which correspond to the edges of the spectral holes are expressed in terms of the functions defined previously in the text . by applying a first order taylor expansion of @xmath56 in @xmath57 , the integrals can be readily evaluated and we obtain @xmath58.\end{gathered}\ ] ] note that for temperatures below 100 mk during hole burning , the brownian motion is negligible , @xmath59 , and the expression for @xmath60 simplifies , and a third order taylor expansion in @xmath61 yields @xmath62 the interaction strength given by this expression is explicitly proportional to @xmath7 and @xmath47 as desired for the opto - mechanical coupling . note that it is also proportional to the bias magnetic gradient @xmath23 , applied during hole burning . the parameter @xmath23 thus serves as an ( adjustable ) gain coefficient for the optomechanical coupling . to obtain a maximum coupling strength , it is desirable to match the bias magnetic gradient to the width of the spectral hole @xmath63 . the coupling of the lightfield with the emitters causes an ac stark shift that depends on the bending of the cantilever and hence results in an displacement @xmath64 of the resonator equilibrium position given by @xmath65 where @xmath66 is the angular frequency of oscillation of the resonator . conversely , the coupling between the light field and the emitters produces a phase shift on the laser . the same proportionality applies between the optical phasehift and interaction energy for the whole system as for a single ion , and we can thus directly write the phase shift of the optical beam as @xmath67 where the dependence on the intensity @xmath47 cancels due to the explicit proportionality between @xmath60 and @xmath47 in the dispersive coupling regime . this phase - shift can be detected interferometrically , which provides a readout of the position of the resonator . in particular , for the light intensity that leads to the displacement expressed in eq . [ staticdisplacement ] , the corresponding phase shift of the carrier optical frequency is @xmath68 . the vibratory movements , on the other hand , can be detected on the transmitted laser field by observing spectral sidebands at the mechanical oscillator frequency @xmath69 from the carrier . thermal excitation of the cantilever motion yields a total integrated power of sideband phase fluctuations equal to @xmath70 , which converges to the value @xmath71 at zero temperature . to detect such sidebands requires their amplitude to exceed the shot noise limit for the given optical power and integration time ( measurement bandwidth ) . the resolution of the phase fluctuation is thus ultimately limited by photon absorption and transfer of ions to other long lived states , which gradually `` overburns '' the spectral hole and destroys it . the transition line in eu : yso , however , has a very small natural linewidth @xmath40 and therefore , even for high optical power , overburning will not occur before a typical time @xmath72 , which sets the limit for the integration time . increasing optical power within the integration time @xmath73 can lead to arbitrarily high resolution , although of course practical consideration ( heat dissipation , etc ... ) limits the maximum power to a few mw . further improvement of the resolution can be obtained by repeating the experiment ( erasing , re - imprinting and probing the spectral hole ) and averaging the successive results . as an example we consider a single - clamped cantilever with the dimensions @xmath74 interacting with a laser beam traversing the cantilever near its fixed end for maximum strain as illustrated in fig . [ holes ] . we consider a cantilever which consists of @xmath1 ( young modulus of 135 gp ) with an effective mass @xmath75 kg , and of which the first excited mechanical mode vibrates at @xmath35= 890 khz . the cantilever contains a 0.1 % doping of @xmath0 ions , with a @xmath2 @xmath3 transition centered at 580 nm and with @xmath76 hz ( and @xmath77 typically @xmath78 khz ) at t=3 k. we choose a power of 1 mw and a hole width of @xmath29=6 mhz to fulfill the condition for large detuned adiabatic following , see appendix . to calculate the proportionality constant @xmath16 we use the values from @xcite -211.4 hz / pa for crystal site 1 , which is the most sensitive of the two non - equivalent sites . the sensitivity to magnetic field is @xmath79 khz / g @xcite . in order to maximize the coupling while fulfilling the condition @xmath80 we choose @xmath81 t / m , which seems possible to achieve in a cryogenic environment over short distances ( mm or less ) using small superconducting coils . in this configuration , the static displacement of the tip of the resonator due to the light field amounts to @xmath64=0.4 pm and the corresponding phase shift of the laser ( the carrier ) equals 0.2 mrad . this shift is easily observable as , for the 1 mw laser power , the shot noise limited phase resolution is 0.45 @xmath82rad within the allowed detection time , before hole - overburning becomes non - negligible ( approximately 16 ms for the 122 hz linewidth ) . each of the spectral sidebands due to brownian motion at 3 k contains an integrated phase of 0.11 mrad , while the phase resolution for an integration time equal to the inverse of the thermal linewidth ( 25 @xmath82s ) is 14 @xmath82rad , hence the sidebands are readily observed . moreover , by increasing the integration time , it is possible to observe the detailed shape of the brownian motion induced sidebands . the sidebands due to zero point fluctuations contain a total integrated phase of 0.4 @xmath82rad , a value close to the phase resolution corresponding to the maximum integration time before hole - overburning . the shot noise limited resolution is therefore sufficient to observe the effect of the quantization of the mechanical resonator . note that the resolution can be further increased by repeating the full hole imprinting and measurement sequence several times , or use optical repumpers to preserve the spectral hole . the effect of the zero point fluctuations can be observed by measuring the total integrated phase fluctuations in the sidebands and/or their amplitude and linewidth at a given temperature . at the shot - noise limited highest resolution achievable , the few percent deviation from the purely brownian - motion induced sidebands due to zero - point fluctuations can be observed even at 3 k temperature . several measurements at various temperatures may also be used to estimate the various necessary parameters ( exact temperature of the resonator , q factor , ... ) within a few percent accuracy . furthermore , we emphasize that even though , for technical simplicity , we evaluated the different parameters at 3k , a dilution fridge with temperature around 0.1k or lower can also be used , reducing the thermal fluctuations in order to make quantum features even more easily discernible . a last potentially perturbing effect arises due to fluctuations of the laser power . as the static displacement corresponding to 1mw of laser power is approximately 0.4 pm , this laser power must be stable to within @xmath83 10@xmath84 to ensure a perturbation much smaller than the zero point fluctuations ( @xmath83 1 fm ) , a power stability requirement well within reach of standard stabilization techniques . in summary , we have proposed a hybrid scheme which provides an efficient optomechanical strain based coupling between a mechanical resonator and an ensemble of narrow linewidth ions . the coupling mechanism via a functionalized spectrally burnt hole does not suffer from inhomogeneous broadening of the emitters , nor from the inhomogeneity of the strain across the resonator during displacement . it is even possible to further increase the coupling by exploiting not just one but an ensemble of functionalized spectral holes within the inhomogeneous linewidth , which can be obtained by using a comb of optical frequencies . the narrow linewidth and dispersive nature of the coupling allows use of a relatively large optical power without destruction of the functionalized hole and observation of the cantilever motion at the quantum level . all authors thank stefan krll and philippe goldner for discussions , and ss thanks daniel estve for help and discussions . km acknowledges support from the villum foundation , ylc from the ville de paris emergence program and ss from la r ' egion rhne - alpes ( cmira explora - pro ) .
nonlinear charge transport in weakly coupled , undoped , photo - excited , type - i semiconductor superlattices ( sls ) is well described by spatially discrete drift - diffusion equations.@xcite as in the much better known case of doped sls , nonlinear phenomena include formation and dynamics of electric - field domains , self - sustained oscillations of the current through voltage - biased sls , chaos , etc . experimentally , the formation of static electric - field domains in undoped , photo - excited sls was already reported many years ago.@xcite the first experimental observation of dynamical aspects of domain formation in undoped , photo - excited sls such as self - sustained oscillations of the photo - current were reported by kwok _ _ et al.__@xcite due to the excitation condition , the oscillations were damped . subsequently , undamped self - sustained oscillations of the photo - current in undoped sls were observed for a type - ii gaas / alas@xcite and for a direct - gap gaas / alas sl.@xcite tomlinson _ _ et al.__@xcite reported the detection of undamped photo - current oscillations in an undoped gaas / al@xmath0ga@xmath1as sl , where the transport is governed by resonant tunnelling between @xmath2 states . the evolution from a static state at low carrier densities to an oscillating state at higher carrier densities was demonstrated in an undoped , photo - excited sl by increasing the photo - excitation intensity.@xcite an investigation of the bifurcation diagrams for undoped , photo - excited sls showed the existence of a transition between periodic and chaotic oscillations.@xcite for a detailed review of the nonlinear static and dynamical properties of doped and undoped superlattices , see ref . . previous theoretical studies of undoped photo - excited sls , including studies of bifurcation and phase diagrams,@xcite are based on a discrete drift model having a constant recombination coefficient.@xcite so far , there have been no reports on considering field - dependent recombination or the fact that the time scale of the electron - hole dynamics depends strongly on the optical excitation intensity . however , the consequences of including these effects for the dynamics of electric - field domains can be striking . in this work , we incorporate into the previously studied discrete model the dependence of the electron - hole recombination on the electric field and on the photo - excitation intensity using a straightforward model that takes into account the overlap integral between the electron and hole wave functions . the field - dependent recombination coefficient decreases with increasing electric field , which has far reaching consequences . at high photo - excitation intensities , it is possible to find only one stable electric - field domain , not two as in the case of a constant recombination coefficient.@xcite in this case , self - sustained oscillations of the current ( ssoc ) may appear under dc voltage bias . the field profile during ssoc can exhibit nucleation of dipole waves inside the sample , the splitting of one wave into two , and the motion of the resulting waves in opposite directions . these dipole waves resemble the pulses in excitable reaction - diffusion systems such as the fitzhugh - nagumo model for nerve conduction @xcite and are quite different from field profiles for a constant recombination coefficient.@xcite in an excitable dynamical system , an appropriate finite disturbance of the unique stable fixed point may cause a large excursion in phase space before returning to the stable state . when diffusion is added , the resulting reaction - diffusion system may support wave fronts , pulses , and wave trains . it is also possible to find sl configurations at high photo - excitation intensities for which there exist no stable electric - field domains . in these cases , there are ssoc , whose corresponding field profiles are wave trains , comprising a periodic succession of dipole waves . these cases are similar to wave trains in oscillatory media such as those appearing in the fitzhugh - nagumo model in the presence of a sufficiently large external current.@xcite the equations governing nonlinear charge transport in weakly coupled , undoped , photo - excited , type - i sls are @xmath3 in these equations , the tunneling current densities between the quantum wells ( qws ) as well as between the sl and the contact regions are @xmath4 the voltage bias condition is @xmath5 here @xmath6 , @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 and @xmath12 denote the electron charge , the average permittivity , the conductivity of the injecting contact , the doping density of the collecting contact , the average electric field , as well as the two - dimensional electron and hole densities of the @xmath13th period of the sl , respectively . equation ( [ 1 ] ) corresponds to the averaged poisson equation . equation ( [ 2 ] ) denotes ampre s law : the total current density @xmath14 equals the sum of the displacement current density and @xmath15 , the electron tunneling current density across the @xmath13th barrier that separates the quantum wells @xmath13 and @xmath16 . charge continuity is obtained by differentiating eq . ( [ 1 ] ) with respect to time and using eq . ( [ 2 ] ) in the result . tunneling of holes is neglected so that only photogeneration and recombination of holes with electrons enter into eq . ( [ 3 ] ) . for high temperatures , i.e. @xmath17/m^*$ ] , where @xmath18 denotes boltzmann s constant , @xmath19 the temperature , @xmath20 the effective electron mass , and @xmath21 $ ] the order of magnitude of the electron density , the tunneling current is given by eq . ( [ 4 ] ) , in which @xmath22 and @xmath23 are functions of the electric field given in ref . and @xmath24 is the length of one sl period ( @xmath25 and @xmath26 denote the individual widths of the quantum well and barrier , respectively ) . even for lower temperatures , the qualitative behavior of the solutions of the discrete drift - diffusion model is similar to one of the more general tunneling current models described in ref . . for @xmath27 and a constant recombination coefficient @xmath28 , eqs . ( [ 1])([4 ] ) describe the well known discrete drift model introduced in ref . . the indices @xmath29 and @xmath30 represent the sl injecting and collecting contacts , and eq . ( [ 2 ] ) holds for them with the phenomenological currents given by eq . ( [ 5 ] ) . the total current density follows from the voltage bias condition in eq . ( [ 6 ] ) @xmath31 photogeneration and recombination are given by @xmath32 and @xmath33 respectively . here @xmath34 , @xmath35 , @xmath36 , @xmath37 , and @xmath38 denote the photo - excitation intensity , the frequency of the exciting photon , the refractive index , the intrinsic carrier density , and the speed of light , respectively . @xmath39 and @xmath40 correspond to the two- ( 2d ) and three - dimensional ( 3d ) absorption coefficients . the 2d absorption coefficient is proportional to the square of the modulus of the electron - hole overlap integral for a constant electric field @xmath41 ( cf . @xmath42 where @xmath43 denotes the energy of the bandgap at the @xmath2 point and @xmath44 as well as @xmath45 solve the stationary schrdinger equation inside one sl period , @xmath46 $ ] , for the electrons and holes , respectively . in this equation , the electric field @xmath41 is considered to be constant and @xmath47 , where the penetration length @xmath48 solves the cubic equation @xmath49}= 1\ ; , \nonumber\ ] ] and therefore @xmath48 depends self - consistently on the eigenvalue @xmath50 . for a fixed value of @xmath34 , the recombination coefficient decreases with increasing electric - field strength @xmath41 , as depicted in fig . [ fig1 ] . vs electric field @xmath41 for the four configurations listed in table [ t1 ] . ] to study the model described by eqs . ( [ 1])([7 ] ) , it is convenient to render them dimensionless . for a fixed value of the photo - excitation intensity @xmath34 and a constant electric field @xmath51 , the stationary solution of eq . ( [ 3 ] ) is @xmath52 . we use the maximum values @xmath21=[p]=\sqrt{\gamma(i)/r(0,i)}$ ] to define typical values for @xmath11 and @xmath12 . the field @xmath53 at the maximum of the drift velocity is a typical value of the field when there are ssoc . therefore , we adopt it as the field unit @xmath54=f_{m}$ ] . similarly , @xmath55=v_{m}$ ] and , therefore , @xmath56=e[n]v_{m}/l$ ] and @xmath57=lv_{m}$ ] . there are two possible time scales , the first one being @xmath58 $ ] , which balances maxwell s displacement current with the current density in eq . ( [ 2 ] ) , and the second one @xmath59/\gamma(i)=1/\sqrt{\gamma(i ) r(0,i)}$ ] , which balances both sides of eq . ( [ 3 ] ) . it is reasonable to choose the time unit as the longer of the two times @xmath60 and @xmath61 . we have chosen four representative gaas / al@xmath62ga@xmath63as sl configurations with 10 nm wells and 4 nm barriers . the configurations are i : @xmath64 and @xmath65 kw/@xmath66 ; ii : @xmath64 and @xmath67 kw/@xmath66 ; iii : @xmath68 and @xmath69 kw/@xmath66 ; and iv : @xmath70 and @xmath71 kw/@xmath66 . we assume a circular cross section with a diameter of 160 @xmath72 m , a photo - excitation intensity of 60 mw , a wavelength of 413 nm , and four different beam diameters yielding the previously listed values of the laser intensity . for most of the cases listed in table [ t1 ] , @xmath73 , and therefore we choose @xmath74=t_{n}$ ] . all scaled variables are listed in table [ t1 ] . .units used to achieve a set of equations with dimensionless variables . [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] vs electric field @xmath41 . for a fixed value of the total current density @xmath75 , there may be one or three zeros of @xmath76-@xmath75 depending on the al content @xmath77 . ] it is interesting to depict the phase plane corresponding to spatially uniform solutions of eqs . ( [ 10])([12 ] ) with @xmath78 , @xmath51 @xmath79 generically and for a fixed value of @xmath75 , the nullclines @xmath80 and @xmath81 intersect in one or three fixed points , depending on the al content @xmath77 in the barriers . at these fixed points , @xmath82 the function @xmath76 is depicted in fig . [ fig2 ] for the sl configurations listed in table [ t1 ] . the calculations for arbitrary values of aluminum content @xmath77 show that for @xmath83 , there are three fixed points of the system in eq . ( [ 17 ] ) , one on each of the three branches of @xmath84 , and the ratio of @xmath85 to the average current @xmath86 is sufficiently large . as we shall see later , some nonlinear phenomena occurring in these sls are quite similar to the ones observed in doped sls : static electric - field domains with domain walls joining the stable branches of @xmath84 , ssoc due to the recycling of pulses formed by two moving domain walls having a high - field region between them , etc . for @xmath87 , @xmath76 is either increasing for positive @xmath41 ( for @xmath88 ) or the ratio of @xmath85 to the average current @xmath89 is small ( for @xmath90 ) . for @xmath88 , there is a unique fixed point at @xmath91 , which , for an appropriate value of @xmath75 , may be located on any of the three branches of @xmath84 . if the fixed point is located on one of the two stable branches of @xmath84 for which @xmath92 has a positive slope , the dynamical system of eq . ( [ 17 ] ) is excitable , whereas it is oscillatory if the fixed point is located on the second branch of @xmath84 with @xmath93 . in an excitable dynamical system , an appropriate finite disturbance of the unique stable fixed point may cause a large excursion in phase space before returning to the stable state . when diffusion is added , the resulting excitable reaction - diffusion system may support a variety of wave fronts , pulses and wave trains.@xcite a dynamical system having an unstable fixed point and a stable limit cycle around it is oscillatory . again in the presence of diffusion , oscillatory systems may support different spatio - temporal patterns.@xcite an undoped photo - excited sl with excitable or oscillatory dynamics exhibits quite unusual phenomena . under dc current bias , it is possible to have pulses moving to the right or to the left and periodic wave trains . under dc voltage bias , these pulses and wave trains may give rise to a variety of sscos . similar phenomena are observed in the case @xmath90 for which @xmath76 has a shallow valley for a narrow interval of current densities . the behavior of a dc voltage - biased sl is quite different depending on its al content and photo - excitation intensity . for an al content smaller than 45% , @xmath76-@xmath75 in eq . ( [ j - f ] ) has a single zero for any value of @xmath75 , and the only stable states of the sl are stationary ones unless the photo - excitation intensity is sufficiently large ( cf . next section ) . let us assume that the al content is larger than 45% so that @xmath76-@xmath75 in eq . ( [ j - f ] ) may have three zeros for an appropriate range of @xmath75 values . in this case , the undoped sl behaves similarly to an @xmath94-doped sl.@xcite the most interesting limit is that of small photo - excitation intensity , i. e. , @xmath95 . and @xmath96 , respectively . the other parameter values correspond to configuration iv in table [ t1 ] and are @xmath97 , @xmath98 , @xmath99 , @xmath100 , and al content @xmath101.,title="fig : " ] and @xmath96 , respectively . the other parameter values correspond to configuration iv in table [ t1 ] and are @xmath97 , @xmath98 , @xmath99 , @xmath100 , and al content @xmath101.,title="fig : " ] for the time scale @xmath102 , eqs . ( [ 11 ] ) and ( [ 12 ] ) can be rewritten as @xmath103 in the limit of large photo - excitation intensities , @xmath104 , eq . ( [ 50 ] ) indicates that the @xmath105 do not depend on @xmath106 . in this case , eq . ( [ 49 ] ) corresponds to a sl doped with a density @xmath105 , and we may expect phenomena similar to the ones observed in an @xmath94-doped sl . in the opposite limit of small photo - excitation intensities , @xmath107 , @xmath108 is a slow scale . the @xmath12 are functions of @xmath109 determined by solving eq . ( [ 50 ] ) with a zero left hand side @xmath110 the resulting equation for @xmath111 is then obtained by inserting eq . ( [ 51 ] ) into eq . ( [ 49 ] ) @xmath112d(f_{i})\;. \label{52}\end{aligned}\ ] ] this equation is similar to the one describing the electric field in a doped sl , but now there are drift and diffusion terms which are nonlinear in the differences @xmath109 . under dc voltage bias , there are ssoc mediated by pulses of the electric field . the current density varies on a slow time scale , whereas the electric - field profile consists of a varying number of wave fronts joining the stable constant solutions of eq . ( [ 52 ] ) at the instantaneous value of the current.@xcite depending on the conductivity of the injecting contact @xmath8 , there are different types of ssoc due to the periodic generation of dipoles or monopoles at the injecting contact . if the curve @xmath113 intersects the bulk current - field characteristic curve @xmath76 before its maximum ( cf . [ fig2 ] ) , sscos due to recycling and motion of charge monopole waves ( moving charge accumulation layers ) appear as shown in fig . [ fig3 ] . if @xmath113 intersects @xmath76 after its maximum ( cf . [ fig2 ] ) , sscos due to recycling and motion of dipole waves are obtained , as depicted in fig . we have indicated in figs . [ fig3 ] and [ fig4 ] whether there are one or three uniform and time - independent ( constant ) solutions of eq . ( [ 17 ] ) , solving @xmath114 and @xmath81 for the instantaneous value of the current density @xmath115 during the ssoc . and @xmath116 , respectively . the other parameter values are the same as the ones used for fig . [ fig3 ] corresponding to configuration iv in table [ t1 ] . , title="fig : " ] and @xmath116 , respectively . the other parameter values are the same as the ones used for fig . [ fig3 ] corresponding to configuration iv in table [ t1 ] . , title="fig : " ] if @xmath87 , the phase plane in eq . ( [ 17 ] ) may have only one fixed point located on any branch of the nullcline @xmath84 . for small photo - excitation intensities , the only stable state is a stationary one . however , for sufficiently large photo - excitation intensities , an infinite , dc current - biased sl may exhibit pulses moving downstream or upstream and also wave trains moving downstream.@xcite the counterpart of these stable solutions for a dc voltage - biased sl is very interesting and different from anything observed in an @xmath94-doped sl . our simulations correspond to sls with different al contents , whose current - field characteristics @xmath76 and injecting contact curve @xmath117 are shown in fig . [ fig2 ] . in all cases , sscos appear for an average bias roughly in the region of negative differential resistance ( ndr ) , where @xmath118 ( e.g. , @xmath8=0.5625 in fig . [ fig2 ] ) . and @xmath64 ( cf . [ fig2 ] ) . these oscillations correspond to having a finite wave train . the other parameter values correspond to configuration ii in table [ t1 ] and are @xmath97 , @xmath119 , @xmath120 , @xmath121 , and @xmath122.,title="fig : " ] and @xmath64 ( cf . [ fig2 ] ) . these oscillations correspond to having a finite wave train . the other parameter values correspond to configuration ii in table [ t1 ] and are @xmath97 , @xmath119 , @xmath120 , @xmath121 , and @xmath122.,title="fig : " ] and the other parameters correspond to configuration i in table [ t1 ] . , title="fig : " ] and the other parameters correspond to configuration i in table [ t1 ] . , title="fig : " ] when the conductivity of the injecting contact is such that @xmath113 intersects @xmath76 near the maximum thereof , it is possible to have sscos that are quite different from the ones appearing in @xmath94-doped sls . for an injecting contact conductivity @xmath123 ( cf . fig . [ fig2 ] ) , pulses ( charge dipoles ) may be triggered at the injecting contact , move toward the receiving contact , and cause sscos as shown in fig . for all the instantaneous values of @xmath14 during these sscos , there is only one constant solution of eq . ( [ 17 ] ) : most of the times this solution is on the second ndr branch of @xmath76 . only when @xmath14 is near its maximum value , the constant solution is on the third branch of @xmath76 . for a constant current density , the constant solution of eq . ( [ 17 ] ) is on the ndr branch , which implies that the system is oscillatory and that periodic wave trains are possible . the realization of wave trains for a long dc - biased sl are displayed in fig . [ fig6 ] : at any time during ssoc , there are only two fully developed pulses present in this sl . these pulses experience variations in their shape and velocity when they are generated or arrive at the contacts , but longer sls allow for realizations of wave trains , in which more pulses exist simultaneously inside the sl . [ fig2 ] ) . the other parameter values correspond to configuration iii in table [ t1 ] and are @xmath124 , @xmath125 , @xmath126 , @xmath127 , and @xmath128.,title="fig : " ] ( cf . [ fig2 ] ) . the other parameter values correspond to configuration iii in table [ t1 ] and are @xmath124 , @xmath125 , @xmath126 , @xmath127 , and @xmath128.,title="fig : " ] in the previous example , pulses always move downstream , from left to right . for slightly larger conductivity of the injecting contact ( @xmath129 in fig . [ fig2 ] ) , fig . [ fig7 ] shows that two pulses are formed inside the sl and move with opposite velocities toward the contacts . these pulses are similar to the ones constructed above for the case of dc current bias ( with positive or negative velocity ) , except that the current changes slowly with time during the self - oscillation and the pulses accommodate their form to the instantaneous value of the current . note that the pulses are triggered inside the sl , not at the injecting contact . a pulse moving with positive speed has a long trailing region , in which the field increases as we move away from the pulse . however , there is a depletion layer near the injecting contact , in which the field decreases as we move away from the injecting contact . in a long , but finite sl , a local maximum is formed inside the sl , when the decreasing field near the contact meets the increasing field in the trailing region of the exiting pulse with positive speed . the current increases as the exiting pulse is absorbed by the receiving contact , until it surpasses a critical value . in this case , the local maximum of the field profile inside the sl is split , and two new pulses are created . the pulse closer to the injecting contact moves toward it with negative speed whereas the other pulse moves toward the receiving contact with positive speed . the upward moving pulse reaches the injecting contact and is absorbed there before the downward moving pulse arrives at the other contact . in this case , the field profile close to the injecting contact is quasi - stationary , and a local maximum of the field is formed when we match this region with the trailing region of the downward moving pulse . after the critical value of the current is reached , another pulse pair is nucleated , and the same process is periodically repeated . it is interesting to evaluate in some detail the process of nucleation and disappearance of pulses during these sscos . in the current vs time diagram in fig . [ fig7 ] , we distinguish different regions depending on the number of constant solutions of eq . ( [ 17 ] ) that exist for the corresponding instantaneous value of @xmath14 . in region a , there is only one constant solution on the third branch of @xmath76 . in region b , there is one constant solution on the third branch and two on the second branch of @xmath76 . in region c , there are three constant solutions , one on each branch of @xmath76 , whereas two of these solutions are on the second branch and one of the first branch of @xmath76 , if @xmath14 is in region d. there is only one constant solution located on the first branch of @xmath76 , if @xmath14 is in region e. for a constant voltage bias , a pulse moving upstream may be generated only if @xmath14 surpasses a critical value ( 1.007454 ) , which is located in region c and marked by an arrow . once generated , the upstream moving pulses persist for any instantaneous value of the current density . these ssoc are apparently weakly chaotic : we have calculated the corresponding lyapunov exponents and found that there is a single positive exponent with a rather small value of @xmath130@xmath131@xmath132 . [ fig2 ] ) . the other parameter values correspond to configuration i in table [ t1 ] and are @xmath133 , @xmath134 , @xmath135 , @xmath136 , and @xmath137.,title="fig : " ] ( cf . [ fig2 ] ) . the other parameter values correspond to configuration i in table [ t1 ] and are @xmath133 , @xmath134 , @xmath135 , @xmath136 , and @xmath137.,title="fig : " ] we have also observed sscos mediated by dipole waves that nucleate alternatively at two different qws of the sl as shown in fig . [ fig8 ] . during these sscos , there is only one constant solution of eq . ( [ 17 ] ) for any instantaneous value of @xmath14 . this solution is either on the first or the second branch of @xmath76 . at about @xmath138 , where @xmath14 reaches its global maximum , two dipole waves are nucleated at two different qws . they become fully developed pulses and move with positive speed toward the receiving contact . when the first one arrives there , the current increases so that the corresponding constant solution of eq . ( [ 17 ] ) is on the second branch of @xmath76 . in this case , a small dipole wave is nucleated at the local field maximum , where the tail of the first dipole meets the depletion layer near the injecting contact . the small dipole wave never grows into a fully developed field pulse , and it continues advancing , until the old large dipole wave disappears at the receiving contact . at this time ( about 4296 ) , a large current spike appears , and a new dipole wave is formed closer to the injecting contact than the small dipole . @xmath14 decreases abruptly , while the small dipole disappears , and the newly created dipole reaches a large size and moves toward the receiving contact . shortly afterward , a new current spike marks the creation of another small dipole . this small dipole travels toward the receiving contact , and it grows only when the only existing large pulse reaches the receiving contact and disappears . a small dipole formed closer to the injecting contact does not grow , until the large pulse reaches the receiving contact and disappears without triggering a new dipole wave . in this case , the corresponding pulse is close to the receiving contact . when it reaches the contact and disappears into it , two new dipoles are simultaneously triggered and become fully developed . a scenario similar to the one previously described follows , marked again by a large current spike . the situation is not repeated exactly : there are small differences in the qws at which pulses are nucleated , differences in the size and lifetimes of the small dipoles , etc . these oscillations also seem to be weakly chaotic , in which a single lyapunov exponent is small and positive . if the injecting contact conductivity is smaller so that the contact current @xmath113 intersects @xmath76 on the ndr branch thereof , there appear standard sscos due to repeated dipole pulse nucleation at the injecting contact and motion toward the receiving contact . we have calculated the recombination coefficient as a function of the applied electric field for undoped , photo - excited , weakly coupled gaas / al@xmath62ga@xmath63as superlattices . depending on the al content @xmath77 , the superlattice may have only one static domain for small @xmath77 or two stable differentiated static domains for @xmath139 . in the latter case , a dc voltage biased sl under weak photoexcitation may exhibit self - sustained oscillations of the current due to repeated nucleation of a charge monopole or dipole waves at the injecting contact and their motion toward the collector . for high photo - excitation intensities , the walls separating electric field domains are mostly pinned , and self - sustained oscillations of the current occur only in narrow voltage intervals . for small @xmath77 among other unusual phenomena , there may appear weakly chaotic ssoc due to dipole dynamics in dc - voltage - biased sls for high photo - excitation intensities , during which nucleated dipole waves can split in two oppositely moving dipoles . these dipoles are pulses of the electric field with shapes and behavior similar to pulses in excitable media , where a sufficiently large disturbance about the unique stable domain may induce them . for other parameter values , the unique static domain is unstable , and the underlying dynamics is oscillatory so that wave trains formed by succession of pulses give rise to self - sustained oscillations of the current . the work was financially supported in part by the spanish ministry of science and innovation under grant fis2008 - 04921-c02 - 01 .	a model for charge transport in undoped , photo - excited semiconductor superlattices , which includes the dependence of the electron - hole recombination on the electric field and on the photo - excitation intensity through the field - dependent recombination coefficient , is proposed and analyzed . under dc voltage bias and high photo - excitation intensities , there appear self - sustained oscillations of the current due to a repeated homogeneous nucleation of a number of charge dipole waves inside the superlattice . in contrast to the case of a constant recombination coefficient , nucleated dipole waves can split for a field - dependent recombination coefficient in two oppositely moving dipoles . the key for understanding these unusual properties is that these superlattices have a unique static electric - field domain . at the same time , their dynamical behavior is akin to the one of an extended excitable system : an appropriate finite disturbance of the unique stable fixed point may cause a large excursion in phase space before returning to the stable state and trigger pulses and wave trains . the voltage bias constraint causes new waves to be nucleated when old ones reach the contact .
an open book is a connected orientable surface @xmath1 with boundary , together with a diffeomorphism @xmath2 fixing the boundary pointwise . positive hopf plumbing is an operation that adds a @xmath3-handle @xmath4 to @xmath1 and composes @xmath5 with a right - handed dehn twist along an embedded circle that runs once through @xmath4 . in case an open book @xmath6 represents the @xmath7-sphere , the boundary of @xmath1 is a fibred link with monodromy @xmath5 . positive hopf plumbing was originally defined in this classical setting by stallings @xcite , where it corresponds to gluing a positive hopf band on top of a fibre surface . in this paper , we consider positive @xmath0-plumbing , an @xmath0-times iterated version of positive hopf plumbing , as sketched in figure 1 , for @xmath8 . successive @xmath3-handles . in order to get an embedded picture , one would have to add a positive full twist to every @xmath3-handle . ] the special case @xmath9 is called positive trefoil plumbing , since it amounts to plumbing the fibre surface of a positive trefoil knot on top of a surface . we say that an open book admits a positive @xmath0-plumbing summand , if it is obtained from a suitable open book by one positive @xmath0-plumbing operation . open books that arise from the trivial open book @xmath10 by iterated positive hopf plumbing are known to be right - veering @xcite . our first result is a plumbing criterion for such open books . an open book @xmath6 with right - veering monodromy @xmath5 admits a positive @xmath0-plumbing summand , if and only if there exists a relatively embedded arc @xmath11 , such that @xmath12 are homologically independent and pairwise disjoint , except at their common boundary points , after a suitable relative isotopy . this is a generalisation of the hopf deplumbing criterion proved by etnyre and li @xcite ( theorem 3.3 ) . in terms of the homological arc complex defined in their paper , the last condition means that the arcs @xmath12 form an ( @xmath13)-simplex . a virtually equivalent criterion is given by buck et al . @xcite ( corollary 1 ) . the case @xmath9 has a remarkable consequence concerning positive braids . the fibre surface of positive braid knots is obtained from the trivial open book by iterated positive trefoil plumbing . at first sight , this seems to be a consequence of corollary 3 in giroux - goodman @xcite . however , the discussion there only implies that positive braid knots are connected to the trivial knot by a sequence of positive trefoil plumbing and deplumbing operations . we conclude the introduction with an application of theorem 2 , the precise meaning of which we will explain in section 5 . the fibre surface of a positive braid knot of genus @xmath14 can be untwisted by @xmath14 ribbon twists . in the last section , we present a criterion by hironaka @xcite obstructing the existence of @xmath0-plumbing summands . as an application , we detect the largest @xmath0-plumbing summand for fibre surfaces of all 3-stranded torus links . let @xmath15 be an open book with a positive @xmath0-plumbing summand . the monodromy @xmath5 can be written as a composition @xmath16 where @xmath17 are right - handed dehn twists along the core curves of @xmath0 hopf bands , in the order of plumbing , and the support of @xmath18 is disjoint from the @xmath0 attached @xmath3-handles . let @xmath11 be an essential relative arc in the outermost @xmath3-handle . the iterated images @xmath19 are easy to determine : they are indeed homologically independent and pairwise disjoint , as shown in figure 2 , for @xmath8 . this proves one implication of theorem 1 , without any further assumption on @xmath5 . for the reverse implication , we assume that @xmath5 is right - veering . loosely speaking , this means that every relatively embedded interval @xmath20 is mapped to the right by @xmath5 . technically speaking , choose @xmath21 to have minimal intersection number with @xmath22 . then the tangent vectors of @xmath22 and @xmath21 are required to form a positive basis of @xmath23 , at their common boundary points ( compare figures 2 and 3 ) . right - veering diffeomorphisms were introduced by honda et al . @xcite in the context of contact geometry . one of their main results is that all closed tight contact @xmath7-manifolds are supported by right - veering open books . let @xmath11 be a relatively embedded interval , such that @xmath24 , @xmath25 are homologically independent and have pairwise disjoint interiors . setting @xmath26 we obtain @xmath0 homologically independent embedded circles @xmath27 . after a suitable isotopy , these circles form a chain , i.e. the only intersection points arise from pairs of consecutive circles . here the right - veering property is absolutely essential . this is illustrated in figure 3 for @xmath28 , where the bold and thin lines represent @xmath29 and @xmath30 , respectively ( the vertical line being @xmath31 ) . let @xmath32 be right - handed dehn twists along the curves @xmath33 . then the composition @xmath34 leaves the arcs @xmath35 invariant . in particular , @xmath18 is defined on the surface @xmath1 cut open along @xmath35 , which is again connected , due to the homological assumption . we conclude that @xmath15 contains a positive @xmath0-plumbing summand . a braid is positive , if it contains positive crossings only . standard generators @xmath36 are used to denote positive crossings . braid words are read from top to bottom , in accordance with the second author s convention @xcite . we use brick diagrams as an alternative notation for positive braids , as well as their fibre surfaces , see figure 4 . the diagram on the right shows the fibre surface of the braid @xmath37 , an embedded orientable surface whose boundary is the closure of that braid and which naturally retracts on the brick diagram . it is instructive to verify that the isotopy type of the fibre surface is invariant under the braid relations @xmath38 . throughout this paper , we will only consider reduced braids , i.e. braids with at least two crossings between all strands . in other words , every column of the brick diagram contains at least one rectangle . every rectangle defines a simple closed curve whose neighbourhood in the fibre surface is a positive hopf band . the left- and topmost rectangle defines a hopf band which can be deplumbed , since it lies on top of the surface ( compare figure 4 ) . on the level of braids , this amounts to removing one crossing on the top left . therefore , fibre surfaces of reduced positive braids admit a positive hopf plumbing structure . this well - known fact is due to stallings @xcite . as explained in detail by dehornoy @xcite , the monodromy can be written as a product of right - handed dehn twists , one for each rectangle , in the order of plumbing , i.e. from the bottom right to the top left , filling up columns from the right to the left . we will make use of this in the proof of theorem 2 . the proof of theorem 2 relies on the following well - known fact , which can be found at various places in the literature , e.g. in franks - williams @xcite . every positive braid whose closure is a non - trivial link contains the square of a generator , @xmath39 , up to braid relations . here braid relations are understood in a generalised sense , including conjugation and markov moves . let @xmath40 be a positive braid representing a non - trivial link . by the braid relation , we may remove all instances of @xmath41 in @xmath40 . if there is only one single crossing of type @xmath42 left , we remove it by a markov move . otherwise @xmath40 contains a section of the form @xmath43 with either no @xmath44 or at least two @xmath44 s in @xmath45 . in the first case we are done ; in the second case @xmath40 contains a section of the form @xmath46 with no @xmath42 in @xmath45 and either no @xmath47 or at least two @xmath47 s in @xmath45 . this argument eventually leads to a proof of the statement . the fibre surface associated with the braid @xmath48 is a positive hopf band . more generally , let @xmath40 be a positive braid that contains a square @xmath39 , then the corresponding fibre surface contains a hopf band which can actually be deplumbed , since it lies on top of the surface . this provides an alternative proof that fibre surfaces of positive braids admit a positive hopf plumbing structure . the monodromy of positive braid links is right - veering , since it can be written as a product of right - handed dehn twists @xcite . therefore , we may apply the criterion of theorem 1 to the open books associated with positive braids . instead of looking at iterates of an arc @xmath11 under the monodromy @xmath5 , we will consider a simple closed curve @xmath49 of the form @xmath50 and verify that @xmath51 does not intersect @xmath31 . let @xmath40 be a positive braid representing a non - trivial knot . by lemma 1 , we may assume that @xmath40 contains a square @xmath39 . after a suitable conjugation , we may further assume that @xmath40 starts with @xmath52 for some numbers @xmath53 . moreover , @xmath40 contains at least one more generator @xmath54 , since it represents a knot . let @xmath11 be an essential relative arc in the uppermost band and let @xmath55 be the simple closed curve running once through the hopf band associated with @xmath39 . in order to determine the image @xmath51 , we decompose the monodromy as @xmath56 where @xmath57 is the number of columns of the brick diagram , and each @xmath58 is a product of right - handed dehn twists in the @xmath59-th column , from the bottom to the top . the last @xmath60 factors @xmath61 do not affect the curve @xmath49 . indeed , a dehn twist ` along a rectangle ' affects another rectangle , if and only if their curves intersect . this happens precisely when the two rectangles are arranged as in the braids @xmath62 , @xmath63 , or @xmath64 . the effect of the monodromy on a single rectangle is described in detail in @xcite . the map @xmath65 shifts the curve @xmath49 down , as shown on the left of figure 5 . the map @xmath66 adds a certain union of rectangles to its left . all subsequent factors @xmath58 ( @xmath67 ) add one layer of rectangles to the curve @xmath68 , as long as their union forms a staircase . figure 5 illustrates this process for the braid @xmath69 the curves @xmath70 , @xmath71 and @xmath72 are shown from left to right . in that example , the last factor @xmath73 does not have any effect on the curve @xmath72 . we observe that @xmath51 does not intersect the arc @xmath31 . now theorem 1 allows us to apply trefoil deplumbing , which amounts to cutting @xmath1 along the arcs @xmath31 and @xmath74 , i.e. cutting the upper two bands associated with @xmath39 . on the level of braids , this simply means removing @xmath39 . it is instructive to verify that a neighbourhood of the theta graph @xmath75 is indeed isotopic to the fibre surface of a positive trefoil knot ( see again figure 5 ) . we conclude by induction . a ribbon twist is an operation on seifert surfaces that inserts a full twist into a ribbon , as shown in figure 6 . two seifert surfaces are twist equivalent if they are related by a finite number of ribbon twists . generic seifert surfaces are unlikely to be twist equivalent , since ribbon twists preserve the homotopy type of the surface complement in @xmath76 . here we will show that fibre surfaces of positive braid knots can be untwisted in a natural way . a seifert surface of genus @xmath14 is trivial if it admits @xmath14 disjoint non - isotopic compression discs on either side .. ] in particular , the complement of a trivial seifert surface is homeomorphic to a handlebody of genus @xmath77 . a simple example for @xmath78 is depicted in figure 6 , on the right . @xmath79 @xmath80 the sequence of diagrams in that figure shows that the fibre surface of the positive trefoil knot can be untwisted in one move . the last isotopy is best seen by dragging one end of the hopf band once around the untwisted annulus . in fact , the same sequence of diagrams , combined with the isotopy of figure 7 , demonstrates that positive trefoil plumbing can be expressed as a connected sum with a compressible torus , followed by a single ribbon twist . as a consequence , a seifert surface @xmath81 that is obtained by @xmath14-times iterated positive trefoil plumbing can be untwisted by @xmath14 ribbon twists . moreover , the number @xmath14 is minimal , since the unknotting number of the boundary knot @xmath82 is not smaller than @xmath14 ( see rudolph @xcite ) . this proves corollary 1 . @xmath80 we conclude this section with a basic question : which fibre surfaces of genus @xmath14 in @xmath76 can be untwisted by @xmath14 ribbon twists ? the purpose of this section is to detect large @xmath0-plumbing summands in fibre surfaces of torus links . using a result of hironaka on the behaviour of the alexander polynomial under iterated trefoil plumbing @xcite , we will determine the largest @xmath0-plumbing summand for fibre surfaces of all 3-stranded torus knots . we define a torus link of type @xmath83 as the closure of the braid @xmath84 the fibre surface of the torus knot @xmath85 is shown in figure 8 . as explained in section 3 , the left- and topmost rectangle defines a hopf band which can be deplumbed . let @xmath49 be the core curve of that hopf band . as in the proof of theorem 2 , we decompose the monodromy as @xmath86 , where @xmath58 denotes the product of dehn twists in the @xmath59-th column , from the bottom to the top . we claim that for all @xmath87 , the curve @xmath88 runs once around the uppermost rectangle in the @xmath89-th column . indeed , let us suppose this is true for some @xmath90 , and let us write @xmath91 ( @xmath92 ) for the curve defined by the uppermost rectangle in the @xmath93-th column ( in particular , @xmath94 ) . the first factor of the monodromy @xmath5 that affects the curve @xmath95 is @xmath96 . in fact , @xmath97 is a curve running once around the union of @xmath95 and @xmath98 . the next factor @xmath99 then removes @xmath95 from that curve , thus @xmath100 the subsequent factors @xmath101 do not affect the curve @xmath98 . figure 8 illustrates this process for the torus knot @xmath85 and @xmath102 . for a more detailed analysis of the monodromy of lorenz links , including torus links , we refer the reader to section 2 of @xcite . keeping in mind that @xmath55 , where @xmath11 is an essential relative arc in the uppermost band , we deduce that the family of arcs @xmath103 is pairwise disjoint . as a consequence , the fibre surface @xmath104 of the torus link @xmath83 contains a @xmath105-plumbing summand . in the special case @xmath106 , we obtain the entire surface as a @xmath105-plumbing summand , as expected . in general , the fibre surface obtained by deplumbing a @xmath105-summand from @xmath104 is not the fibre surface of a torus link ( and not even the fibre surface of a positive braid link , for that matter ) . in case a fibre surface @xmath1 admits an @xmath0-plumbing summand , we naturally obtain an embedded subsurface of type @xmath107 in @xmath1 . with this in mind , we will look for a large subsurface of type @xmath107 in @xmath108 . applying the braid relation three times , we obtain @xmath109 in particular , the fibre surface @xmath110 contains @xmath111 as a subsurface ( compare figure 9 ) . roughly speaking , @xmath108 contains a surface of type @xmath107 which accounts for three quarters of the genus . however , this bound can not be achieved by a plumbing summand . 1 . the fibre surface @xmath112 of the torus knot @xmath113 admits an @xmath0-plumbing summand , if and only if @xmath114 . the fibre surface @xmath115 of the torus knot @xmath116 admits an @xmath0-plumbing summand , if and only if @xmath117 . by the above discussion , we are basically left to prove that @xmath118 does not admit an @xmath0-plumbing summand for @xmath119 resp . this is a direct consequence of hironaka s criterion , which involves the alexander polynomial @xmath121 of links . let @xmath122 be obtained by plumbing a @xmath107-summand on the fibre surface @xmath1 of a link @xmath123 with @xmath124 components , and let @xmath125 be the boundary link of @xmath122 . then there exists a polynomial @xmath126 $ ] of degree @xmath127 , such that @xmath128 in order to appreciate this criterion , let us first analyse what it says about 2-stranded torus knots . the alexander polynomial of torus knots is calculated in most textbooks about knot theory , e.g. in rolfsen @xcite : @xmath131 the special case @xmath132 boils down to @xmath133 . comparing this with hironaka s criterion , we obtain @xmath134 and @xmath135 , in accordance with the fact that @xmath136 is itself a plumbing summand , and the remaining surface is a disc . \i ) @xmath137 . an elementary calculation shows @xmath138 multiplying this with @xmath139 yields @xmath140 comparing this again with hironaka s criterion , we obtain @xmath141 thus @xmath142 and @xmath143 . for example , @xmath144 @xmath145 and @xmath146 . we conclude that @xmath147 is the largest possible plumbing summand of the fibre surface @xmath112 . in our convention , @xmath112 admits at most a positive @xmath148-plumbing summand . moreover , this bound is attained , by the discussion preceding proposition 1 . \ii ) @xmath149 . a similar calculation yields @xmath150 @xmath141 @xmath142 and @xmath151 . this leaves room for a @xmath152-plumbing summand , rather than @xmath153 . in order to detect this summand , we work with the 3-stranded diagram of @xmath116 . as usual , let @xmath55 be the curve running once around the top left rectangle . a careful inspection of figure 10 reveals that the third power of the monodromy @xmath154 shifts the curve @xmath49 down by three rectangles . repeating this process @xmath59 times , we arrive at the bottom left rectangle . the latter is mapped to the bottom right rectangle under @xmath5 . as a consequence , we obtain a family of pairwise disjoint arcs @xmath155 , in turn a @xmath156-plumbing summand . the genus of the fibre surface @xmath104 of a torus knot @xmath83 is @xmath160 . applying this to the knots of proposition 2 , we see that the largest plumbing summand of @xmath158 fills up more than half of the genus ( the precise ratio is @xmath161 and is again given by hironaka s criterion ) . as in the above proofs , it is all about tracing the image of the top left rectangle @xmath55 under iterates of the monodromy @xmath5 . the @xmath157-th power of the monodromy shifts @xmath49 down by @xmath157 rectangles . this process can be repeated @xmath162 times , without ever intersecting the initial curve @xmath31 . another @xmath163 iterations map the last curve @xmath164 to a rectangle in the right column . the whole process is shown in figure 11 , for the torus knot @xmath165 . summing up , the first @xmath166 iterates of the arc @xmath31 are pairwise disjoint . this gives rise to a plumbing summand of type @xmath159 , as required .	we give a criterion for an open book to contain an @xmath0-times iterated hopf plumbing summand . as an application , we show that fibre surfaces of positive braid knots admit a trefoil plumbing structure .
the most spectacular phenomenon associated with granular matter is certainly the phase segregation that one observes when a granular mixture is shaked , poured or rotated @xcite . respectively , stripe @xcite , axial @xcite and radial @xcite segregation patterns have been reported . recent experiments @xcite and theoretical models @xcite have been extensively performed in the particular case of stripe segregation . however , the least studied type of segregation is the case of the axial segregation in the so - called oyama s drum @xcite . the experiment consists in rotating a tube which is half filled with a sand mixture . the mixture is composed by 2 granular species differing in size and having different angles of repose . due to the rotation , avalanches occur in the tube . bands of the different granular species appear after a few minutes . a sketch and a picture of the axial segregation are illustrated in figure 1 . it has been reported @xcite that the bands merge and only three stripes remain after very long times . one should also remark that magnetic resonance imaging ( mri ) experiments have shown that a radial segregation is also present in the center of the tube @xcite . small grains being located in the center of the tube . 0.3 cm on the theoretical side , it has been proposed @xcite that the axial segregation is similar to a spinodal decomposition . a simple and elegant model for diffusion of the different sand species along the tube axis has also been proposed by savage @xcite . this model considers the difference @xmath0 of dynamical angles for both species as a relevant parameter . if @xmath1 is the density of one species along the tube axis @xmath2 , the different slopes involve a segregation flux @xmath3 for that species . the positive coefficient @xmath4 is a function of the slope difference @xmath0 and the rotation speed @xmath5 of the tube . in addition , a fickian diffusion @xmath6 is taken into account . the diffusion coefficient @xmath7 is a function of the rotation speed @xmath5 . summing both fluxes and taking into account the continuity equation for @xmath8 , one @xcite obtains the classical diffusion equation @xmath9 with an effective diffusion constant @xmath10 which could be negative for defined @xmath0 and @xmath5 values . a negative coefficient explains the formation of bands because any inhomogeneity in @xmath11 is then amplified towards full segregation . the savage s model is a first step in the understanding of the phenomenon . the most elaborated model remains the one proposed by zik and coworkers @xcite . in the zik s model , @xmath4 scales as @xmath12 and the occurence of bands is attributed to a spinodal decomposition instead of a simple diffusion . more recently , molecular dynamics simulations have also been performed @xcite in order to study the effect of friction . axial segregation can be simulated with thousand of virtual grains . we have performed new experiments in the oyama s drum . we have produced axial segregation patterns and studied the phenomenon during several hours / days . in some experiments , we have obtained spectacular patterns that we call _ instabilities_. the present paper deals with such unusual patterns . they are discussed in the light of previous theoretical arguments and earlier experiments . mechanisms for instabilities appearance are proposed . our experimental setup is the following . a glass tube ( diameter @xmath13 mm , length @xmath14 mm ) is half filled with a binary sand mixture . the tube is rotated at a constant angular frequency @xmath5 which can be controlled . the rotation speed can range from 20 rotations per minute ( rpm ) to 40 rpm . the tube in rotation is perfectly horizontal . two different sands have been selected for our study . for each type of sand , the granulometry has been selected by sifting . white grains have a mean diameter around @xmath15 mm . black grains have a mean diameter around @xmath16 mm . the angle of repose for dark grains is @xmath17 . the angle of repose for the white grains is @xmath18 . repose angles of different species have been measured by averaging the slope of a stable pile of the considered species along its surface . a ccd camera captures the pattern along the tube axis every minute . a computer collects the successive pictures . each picture of the glass tube is rescaled into a line of pixels . this line represents the average density of the sand species along the @xmath2 axis . the successive lines of pixels are numerically glued together in order to form a space - time diagram of the evolving segregation pattern . figure 2 presents a typical spatio - temporal graph built with 3840 images of the pattern , for a total period of 64 hours . in the picture , time goes down . the segregation pattern emerges within a few minutes . after a long time ( several hours ) , the bands begin to merge . the merging of adjacent bands is rare . however , such an event is a rapid process ( typically less than one minute ) since the end part of a vanishing band is always abrupt . see three vanishing bands in the top of figure 2 . the number @xmath19 of bands seems to decrease exponentially with time . figure 3 emphasizes such a decay for the particular data of figure 2 . we have fitted those data with the stretched exponential @xmath20 with @xmath21 being the initial number of bands and @xmath22 being a characteristic time for the decay . the data for others space - time graphs obtained with our setup are also well fitted with this empirical law . a systematic study of that particular dynamics is under way and will be published elsewhere @xcite . before looking at the details , we can make 3 general observations : \(1 ) the segregation begins first at the extremities of the tube and then nucleates randomly in the tube . the spreading pattern formation is emphasized in figure 4 . the presence of tube ends ( walls ) affects locally the friction and induces the formation of bands at the extremities . we have noticed that at low rotation speeds ( @xmath23 rpm ) , bands appear only at the extremities while the middle part of the tube is still mixed , even for very long times . \(2 ) the topography of the granular surface is deeply affected by the segregation pattern . a close look at the surface reveals different slopes for the different sand species ( see the top picture of figure 1 ) . those slopes correspond to the dynamical angles of the species , which are @xmath24 larger than the repose angles . this is the main ingredient of both savage @xcite and zik @xcite models . \(3 ) for very long times , the pattern tends to a complete segregation of both granular species . only two bands are observed . this complete segregation is clearly observed at the end of figure 2 . one should remark that frette and stavans @xcite did not always obtain a complete segregation pattern even after having performed very long experiences ( but at a lower speed than in our cases ) . in the great majority of our experiments , the successive merging of bands towards a full segregation of both granular species have been observed . however , looking at the details of figure 2 , one observes a turbulent - like pattern at the interface between white and dark bands . this particular event occured after 18 hours of rotation . the turbulent pattern is emphasized in figure 5 . the duration of that event lasted about 5 hours . the instability of white / dark interfaces began when a new dark band nucleated in the middle part of a white band . this nucleation process involved the development of secondary bands in the neighborhood . extremely slow undulations of the borders are also seen in the bottom of figure 2 , a long time after the ` turbulent ' event emphasized in figure 5 . the occurrence of such behavior is still mysterious . we did not discover the precise conditions on which they appear and disappear . however , it should be remarked that the occurence of such patterns is not an artefact . indeed , they appeared in different experiments under various conditions , i.e. at various rotation speeds . moreover , initial mixtures were premixed as homogeneously as possible . figures 6 and 7 present additional patterns that we have also obtained . in those examples , the instability does not affect the total number of bands along the tube axis . in figure 5 , one observes undulations of the borders of white bands . in fact , thin dark bands nucleate periodically in the center of white bands before being expelled . in figure 6 , the situation is slightly different : two small bands are seen to detach from a white band and to travel from that band towards the next one . some instabilities are provoked by the nucleation of thin dark bands ( see figures 5 and 6 ) . others appear at the border of apparently stable white bands ( see figure 7 ) . we will discuss the occurence of that phenomena in the next section . it should be noted that an earlier experimental work @xcite put into evidence some travelling waves in the tube . this behavior was present for higher filling fraction of the tube ( @xmath25% ) than in our experiments . such travelling waves were associated with the early stages of the segregation ( about 20 minutes ) , when inhomogeneities of the initial mixture provoke the nucleation and the motion of bands . herein , the instabilities are seen to occur after very long times and last several hours . the nature of the instabilities could be quite different from those reported in @xcite . as mentionned at the beginning of this paper , mri experiments @xcite have put into evidence that radial segregation is always present . small grains ( the dark species herein ) tend to segregate in the bulk . this implies that in our experiments , the white bands may have a dark core . we assist thus at the competition between axial and radial segregation in that regions . any fluctuation of the radial pattern may provoke the nucleation of a thin dark band . thereafter , this thin band travels slowly towards one border to be expelled . this coupling between surface and volume compositions could explain the instabilities . surface and bulk processes are characterized by different time scales . axial segregation has a larger characteristic time scale than radial segregation . this rate difference could also explain why our instabilities last several hours : the phenomenon comes from bulk fluctuations . mri experiments could corroborate those mechanisms . consider the fraction of dark species @xmath2 observed along the tube , measured by image analysis . figure 8 presents the time evolution of the ratio @xmath26 in the long experiment of figure 5 . the ratio @xmath27 emphasizes the surface ` equilibrium ' between both species along the tube . during the whole process , the ratio fluctuates around a linear trend close to @xmath28 . however , during a period of instability , we observe giant fluctuations of @xmath27 up to @xmath29 . this means that the compositions of both bulk and surface deviate strongly from equilibrium composition . bulk segregation should be taken into account when axial patterns are discussed . oyama s drum experiment is an example of 3-dimensional segregation . this has been already suggested by hill et al @xcite . moreover , we have reported hereabove additional effects such as long lasting instabilities which arise from bulk processes . the dynamics of those patterns are extremely slow with respect to the formation / vanishing of the bands . we stress that those instabilities originate from the competition between _ axial _ and _ radial _ segregations . as a consequence , the savage model , which considers only surface effects should be extended to take into account the bulk distribution of granular species . the physical conditions , such as rotation speed and granulometry , which imply the formation of instabilities , are still unclear . this point needs a systematic experimental study . this represents a huge amount of work since each experiment lasts several hours . hc is financially supported by the fria ( brussels , belgium ) . this work is also suppported through the arc contract n@xmath3002/07 - 293 . the authors would also like to thank m. ausloos for fruitful discussions .	we have performed experiments of axial segregation in the oyama s drum . we have tested binary granular mixtures during very long times . the segregation patterns have been captured by a ccd camera and spatio - temporal graphs are created . we report the occurence of instabilities which can last several hours ! we stress that those instabilities originate from the competition between axial and radial segregations . we put into evidence the occurence of giant fluctuations in the fraction of grain species along the surface during the unstable periods . granular flow : mixing , segregation and stratification , porous materials ; granular materials 45.70.mg , 81.05.rm
during the last 70 years , significant progress has been achieved in the study of the physics in the interior of the sun and stars @xcite . this advance has been possible due to the great development of several key fields of experimental and theoretical physics , such as statistical physics , magneto - hydrodynamics , particle physics , and nuclear physics , among others . astronomers are now able to describe with high - precision the physics that takes place inside stars like the sun , as well as many other classes of stars , with masses different than that of the sun , and in quite different stages of stellar evolution . in general , the physics of stars is well understood since the first moments of their formation , up to the most advanced stages of stellar evolution , including the formation of highly compact objects , like white dwarfs , neutron stars and black holes . the progress in the understanding of the physical principles operating inside stars was accompanied and challenged by important developments in the observational fields of astrophysics , such as astrometry , photometry and high - resolution spectroscopy , as well as the new fields of helioseismology and asteroseismology , which create powerful tools to probe the interior of stars . the recent past has shown that the prosperity of modern stellar physics was possible due to the powerful partnership built between theoretical physics and astrophysics . in this new phase of stellar physics , the large amount of data made available by several observational projects permits to use the sun and stars as a tool to challenge our knowledge about fundamental physics,@xcite and in doing so it opens new branches of research , such as gravitation tests and probes of the existence of new particles . among other applications , to validate the new gravitational theories proposed as an alternative to general relativity @xcite , and to probe the existence of dark matter ( dm ) inside stars to investigate the dm problem @xcite . currently , the interest of studying the interaction of dm with stars is twofold : on one hand to identify which type of particles dm is made of , and , on the other hand , to understand the physical mechanisms by which dm contributes to the formation of stars . in the former , stars are used as a complementary tool to test dm candidates , being in that way an alternative method to test the candidates proposed by modern theories of particle physics , or alternatively , to test candidates detected by experiments of direct or indirect dm searches . in the latter , the aim is to explore how dm contributes for the structure formation in the universe , comprising galaxies and the first generation of stars , not only by locally changing the gravitational field where stars are formed , but also to explore how the interaction of dm with baryons changes the evolution of stars . our universe is constituted by 5% of _ baryonic matter _ , a type of matter in which we have become great experts during the last two centuries , by developing several branches of physics ; 27% is constituted by _ dark matter _ , which plays a major role in the formation of structure in the universe , but its fundamental characteristics are yet poorly understood ; and another 68% of the total energy density of the universe is usually refereed to as _ dark energy _ , which physical origin is even more uncertain @xcite . although the basic properties of dm are not known , namely , which type of particles is dm made of , there is strong evidence of its existence , both from astrophysical and cosmological observations , as well as from numerical simulations @xcite . among other direct evidence of the existence of a gravitational field caused by the presence of dm , we make reference to the velocity of galaxies in clusters , the rotation curves of galaxies , the cosmic microwave background anisotropies , the velocity dispersions of dwarf spheroidal galaxies and the inference of the dm by gravitational lensing . all these observational and theoretical results suggest that most of the formation of structure in our universe can only be explained by the presence of a gravitational field caused by the presence of a new type of particles that must be non - baryonic and cold @xcite , such as the particles belonging to the group of the wimps ( for weakly interacting massive particles ) . therefore , it is no surprise that with such an amount of observational evidence for the existence of dm , a large effort is being devoted to theoretical work and experimentation in several branches of astrophysics , cosmology and particle physics , with the intent of discovering such fundamental particles . if dm particles exists , then in the near future we should expect to detect such particles in the large hadron collider at cern or in other direct detection experiments . another possibility is the confirmation of the existence of dm by the indirect detection of dm by - products , like the production of high - energy neutrinos or gamma rays caused by the annihilation of dm pairs . in the last 30 years , several classes of particles have been proposed as dm candidates , among others , axions , wimps , asymmetric dm particles , and other more exotic types of matter . at present , two groups of particles merit special attention , because they sum up most of the critical properties necessary to be the ideal dm particle . first , the well known weakly interacting massive particles ( wimps ) , which interact gravitationally with other particles and have weak interaction with baryons . wimps are among the most popular dm candidates . such class of particles occur in several extensions of the standard model of particle physics , like super - symmetric ( susy ) models.@xcite in such models , the lightest susy particle , the neutralino , a stable particle with a self - annihilation cross section of the order of the weak - scale interaction , is the most suitable candidate for dm . the second type of candidates are known as asymmetric dm particles,@xcite which like wimps have interactions with baryons at the weak - scale , even if they do not self - annihilate inside compact objects.@xcite unlike wimps , these particles carry a conserved charge analogous to the baryon number asymmetry . as a consequence , dm becomes asymmetric , _ i.e. _ , there is an unbalanced amount of particles and antiparticles , introducing an asymmetric parameter in the dm sector identical to baryon - anti - baryon asymmetry parameter , the so - called baryonic asymmetry . furthermore , these particles are expected to have a mass of the order of a few gev @xcite . in recent years , several underground experiments have been built to search for direct signatures of the interactions between dm particles and a baryons @xcite . most of such dm searches have not detected any dm signal . this is the case of experiments like xenon10/100,@xcite picasso,@xcite simple,@xcite and lux.@xcite however , in disagreement with such results are the positive detections of dama / libra,@xcite cogent,@xcite cresst - ii @xcite and cdms ii @xcite experiments , which all found evidence of events that can be credited to dm particles with similar properties . the former two experiments report evidence of an annual modulation in the differential event rate , which is explained as a consequence of the motion of the earth around the sun , which in turn moves through a cloud of dm particles @xcite . nevertheless , these results remain utterly controversial and further experimental work must be done in order to converge on a plausible answer about these detections . a possible solution to accommodate all experimental results comes from a new theoretical interpretation of the interaction of dm particles with baryons in the detectors . the simpler interpretation , used so far to analyse the data , is to assign the annual modulation to a collision of a dm particle with nucleons inside the detector . in such scenario , the dm particle is estimated to have a mass of the order of a few gev ( likely between @xmath0 and @xmath1 gev ) , and a dm - nucleon scattering cross - section of the order of @xmath2 for spin - independent ( si ) interactions or @xmath3 for spin - dependent ( sd ) interactions on protons . dama and cogent experiments show very similar positive detections , yet the dama experiment is favourable to a larger scattering cross - section than the cogent experiment . nevertheless , this problem is resolved by several theoretical solutions that have been proposed to overcome the inconsistency between the different experimental results @xcite . under such interpretations the data obtained by the different experiments can be reconciled . among the various possible theoretical explanations , one of the more appealing suggestions is the possibility that the dm particle couples unequally to the protons and neutrons of the collision nuclei because of isospin violation . usually , the dm particle is considered to couple equally with protons and neutrons . accordingly , the si scattering cross - section of heavy elements scales with @xmath4 , where @xmath5 is the atomic number of the nucleus . if the dm particles couple differently to protons and neutrons , that leads to a quite distinct interpretation of direct dm searches , leading to the reconciliation of almost all the experiments @xcite . this nuclear mechanism is known as isospin coupling violation @xcite . in such cases , the proton scattering cross section for the si interaction increases by @xmath6 to @xmath7 relatively to the usual interpretation of the experimental results , leading to an effective si scattering cross - section with values between @xmath8 and @xmath9.@xcite . other models propose momentum and velocity - dependent interactions to explain the data.@xcite we point out here that these types of dm particles can modify the internal properties of stars like the sun by conducting energy very efficiently,@xcite leading to a quite different flux of solar neutrinos @xcite and helioseismology data @xcite . dark matter impacts the evolution of a star by means of two mechanisms : through the change of the transport of energy @xcite and by the creation of an additional source of energy.@xcite the former mechanism can even be important for stars in our galactic neighbourhood , like the sun @xcite and other main - sequence stars.@xcite the latter mechanism is more pronounced in environments with a very high dm density , several million times the dm density of the solar neighbourhood . this type of scenario occurs in stellar populations located in the centre of galaxies , including the milk way,@xcite or during the formation of the first generation of stars in the primordial universe.@xcite on the other hand , the gravitational influence of the dm mass accumulated in the interior of stars similar to the sun is several tens of orders of magnitude smaller than the star s total mass , so it can be neglected . most dm particles cross the stellar interior without undergoing any type of interaction , yet a few of them scatter off nuclei losing part of their kinetic energy . in some cases , the loss of energy will result in the particle being captured by the star , _ i.e. _ , the particle can no more escape the star s gravitational field . as expected , a star with a large mass has a large gravitational field , and consequently captures a larger amount of dm during its evolution . the capture rate of dm particles by a star is inversely proportional to the mass and the dispersion velocity of the dm particle , but proportional to the local dm density of the halo and the scattering cross - section off baryons @xcite . two leading parameters define the scattering of the dm particles with the nuclei of the stellar isotopes : the sd scattering cross - section that is only relevant for hydrogen ; and the si scattering cross - section that defines the interaction of the dm particles with the heavy nuclei . the values of the scattering cross sections used in the calculations with stellar codes are usually the maximum allowed by the null results of direct detection experiments , or those in agreement with the values suggested by the interpretation of positive experimental results . however , it is worth noticing that , if the value of the si scattering cross - section is larger than a hundreth of the sd scattering cross - section , then the capture of dm particles is dominated by collisions with heavy nuclei , rather than by collisions with hydrogen @xcite . the scenario that is usually considered corresponds to a dm halo with a density in the solar neighbourhood of @xmath10 , constituted by particles with a mass of a few gev and with a maxwellian velocity distribution with a dispersion of @xmath11 . the interactions of dm particles with stars have been implemented in a stellar code @xcite that explicitly follows the capture rate of the dm particles by the different chemical elements present inside stars , some of them changing in isotopic abundance during the star s evolution . capture rates of dm were first calculated by ref . in the case of the sun , by ref . for generic massive bodies , and by ref . for main sequence stars . presently , the capture rate is computed numerically from the integral expression of ref . implemented as indicated in ref . . a detailed discussion about the dependence of the capture on the properties of dm particles , as well as the impact that the uncertainties in the dm and stellar parameters have on the capture rate by the sun and other stars is presented in ref the accumulation of dm inside a star leads to the formation of a very small dm core , with a radius of the order of the few percent of the stellar radius . the impact on the stellar properties comes from the additional energy transport mechanism provided by dm conduction , or by the dm self - annihilation . in both cases , the evolution of the star is only affected when the changes produced by the dm particles compete with the classical transport or production of energy mechanisms . mostly , only low - mass stars can be influenced , and the impact of dm becomes insignificant for more massive stars . the efficiency of the energy transport provided by dm particles depends on the average distance travelled by particles between consecutive collisions , _ i.e. _ , the mean free path of the dm particle . if the mean free path is short compared with the dm scale height , then the stellar plasma and the dm are in local thermal equilibrium . alternatively , if the mean free path is large , the successive collisions are widely separated , so that the energy transfer proceeds within the knudsen regime @xcite . the energy generation rate due to pair annihilation of dm particles is more significant in dm halos of high density . in such cases , the dm in the stellar core provides an extra source of energy . it follows that every pair of captured dm particles annihilates , being converted into additional energy for the star . without loss of generality , all the products of dm annihilation , except neutrinos , are assumed to interact with the stellar plasma , so these new particles have a very short mean free paths in the star s core and rapidly reach the thermal equilibrium . the process can be very efficient : for each annihilation pair most of the energy is converted in thermal energy and only a small fraction is lost in the form of high - energy neutrinos that escape the star s gravitational field . recent simulations of the self - annihilation of two neutralinos ( which are majorana particles in most susy models ) have shown that the energy loss in a star as the sun could be as low as 10% of the total energy produced by the annihilation of the dm pair @xcite . [ cols="^,^ " , ] in other scenarios , the dm annihilation provides stars with an extra source of energy that can dramatically change their evolution path.@xcite however , environmental dm densities millions of times greater than the local dm density are needed for these important modifications to occur . these huge dm densities are only expected in very particular locations in the local universe , such as the galactic center or the dwarf spheroidal galaxies , and in the early stages of the universe . the study of the impacts of dm in stars in environments with high dm densities is subjected to important difficulties from the observational side and larger uncertainties from the theoretical side . the most important ones are those on the stellar mass and velocity , as well as on the density of dm around the star , which strongly depends on the model . on the other hand , in these cases the potential effects of dm on low - mass stars can be dramatic : the creation of an unexpected convective core that may be detected with asteroseismology ( see figure [ fig - seism_isoc ] left),@xcite the evolution through a different path in the hertzsprung - russell diagram at a lower evolutionary speed,@xcite and the changes in the global properties of a whole cluster of stars ( see figure [ fig - seism_isoc ] right).@xcite similarly , the first generation of stars may have also been strongly influenced by the presence of dm @xcite . in the last fifteen years , the use of the sun and stars as cosmological tools to probe and test the dm particle candidates has become a regular procedure in dm research . this is testified by the arrival of several groups working in such a new research field @xcite , as well as by a significant increase on the number of publications in this subject . the increase of our knowledge of the physics of the solar interior made possible by the solar neutrinos as well as by helioseismolgy will allow accurate tests of the existence of different dm particle candidates , and also the extension of such studies to other fields of fundamental physics , such as alternative theories of gravitation . presently , with the large amount of data made available by observational asteroseismology , mostly by the spacial missions kepler@xcite and corot , the pulsation spectrum has been identified and measured in more than ten thousand stars . some of these stars are identical to the sun , but most of them have quite different masses and they are in distinct phases of stellar evolution . with this large and diverse set of stars it should be possible to further constrain the properties of the dm particles . the approach highlighted here provides a complementary contribution to the multidisciplinary effort of dm research . only a collaborative work across the several fields involved will allow the scientific community to produce fruitful results , and in doing so to overcome one of the most exciting and difficult problems of modern astrophysics , particle physics and cosmology , which is the discovery of the constituent particle(s ) of dm . j.c . acknowledges the support from the alexander von humboldt foundation . i.l . acknowledges the support from the fundao para a cincia e tecnologia .	during the last century , with the development of modern physics in such diverse fields as thermodynamics , statistical physics , and nuclear and particle physics , the basic principles of the evolution of stars have been successfully well understood . nowadays , a precise diagnostic of the stellar interiors is possible with the new fields of helioseismology and astroseismology . even the measurement of solar neutrino fluxes , once a problem in particle physics , is now a powerful probe of the core of the sun . these tools have allowed the use of stars to test new physics , in particular the properties of the hypothetical particles that constitute the dark matter of the universe . here we present recent results obtained using this approach .
wireless sensor and actor networks ( wsans ) are composed of sensor nodes and actors that are coordinated via wireless communications to perform distributed sensing and acting tasks . in wsans , sensor nodes collect information about the physical world , while actors use the collected information to take decisions and perform appropriate actions upon the environment . the sensor nodes are usually small devices with limited energy resources , computation capabilities and short wireless communication range . in contrast , the actors are equipped with better processing capability , stronger transmission powers and longer battery life . the number of actors in wsan is significantly lower than the number of sensor nodes @xcite . the wsans technology has enabled new surveillance applications , where sensor nodes detect targets of interest over a large area . the information collected by sensor nodes allows mobile actors to achieve surveillance goals such as target tracking and capture . several examples of the wsan - based surveillance applications can be found in the related literature , including land mine destruction @xcite , chasing of intruders @xcite , and forest fires extinguishing @xcite . the surveillance applications of wsans require real - time data delivery to provide effective actions . a fast response of actors to sensor inputs is necessary . moreover , the collected information must be up to date at the time of acting . on the other hand , the sensor readings have to be transmitted to the mobile actors through multi - hop communication links , which results in transmission delays , failures and random arrival times of packets . the energy consumption , transmission delay , and probability of transmission failure can be reduced by decreasing the amount of transmitted data @xcite . thus , minimization of data transmission is an important research issue for the development of the wsan - based surveillance applications @xcite . it should be noted that other methods can be used in parallel to alleviate the above issues , e.g. , optimisation of digital circuits design for network nodes @xcite . this paper introduces an approach to reduce the data transmission in wsan by means of suppression methods that were originally intended for wireless sensor networks ( wsns ) . the basic idea behind data suppression methods is to send data to actors only when sensor readings are different from what both the sensor nodes and the actors expect . in the suppression schemes , a sensor node reports only those data readings that represent a deviation from the expected behaviour . thus , the actor is able to recognize relevant events in the monitored environment and take appropriate actions . the data suppression methods available in the literature were designed for monitoring applications of wsns . in such applications , a sink node needs to collect information describing a given set of parameters with a defined precision or recognize predetermined events . these state - of - the - art suppression methods are based on an assumption that a large subset of sensor readings does not need to be reported to the sink as these readings can be inferred from the other transferred data @xcite . in order to infer suppressed data , the sink uses a predictive model of the monitored phenomena . the same model is used by sensor nodes to decide if particular data readings have to be transmitted . a sensor node suppresses transmission of a data reading only when it can be inferred within a given error bound . temporal suppression techniques exploit correlations between current and historical data readings of a single sensor node . the simplest scheme uses a nave model , which assumes that current sensor reading is the same as the last reported reading @xcite . when using this method , a sensor node transmits its current reading to sink only if difference between the current reading and previously reported reading is above a predetermined threshold . parameters monitored by wsns usually exhibit correlations in both time and space @xcite . thus , several more sophisticated spatiotemporal suppression methods were proposed that combine the basic temporal suppression with detection of spatially correlated data from nearby nodes @xcite . according to the spatiotemporal approach , sensor nodes are clustered based on spatial correlations . sensor readings within each cluster are collected at a designated node ( cluster head ) , which then uses a spatiotemporal model to decide if the readings have to be transmitted to sink . in previous work of the first author @xcite a decision - aware data suppression approach was proposed , which eliminates transfers of sensor readings that are not useful for making control decisions . this approach was motivated by an observation that for various control tasks large amounts of sensor readings often do not have to be transferred to the sink node as control decisions made with and without these data are the same . the decision - aware suppression was used for optimizing transmission of target coordinates from sensor nodes to a mobile sink which has to track and catch a moving target . according to that approach only selected data are transmitted that can be potentially useful for reducing the time in which the target will be reached by the sink . according to the authors knowledge , there is a lack of data suppression methods in the literature dedicated for the surveillance applications of wsans . in this paper the available data suppression methods are adapted to meet the requirements of the wsans . effectiveness of these methods is evaluated by using a model of wsan , where mobile actors have to capture randomly distributed targets in the shortest possible time . the paper is organized as follows . details of the wsan model are discussed in section 2 . section 3 introduces algorithms that are used by actors to navigate toward targets as well as algorithms of sensor - actor communication that are based on the data suppression concept . results of simulation experiments are presented in section 4 . finally , conclusions are given in section 5 . in this study a model of wsan is considered , which includes 16 actors and 40000 sensor nodes . the monitored area is modelled as a grid of 200 x 200 square segments . discrete coordinates @xmath0 are used to describe positions of segments , sensor nodes , actors , and targets ( @xmath1 , @xmath2 ) . the sensor nodes are placed in centres of the segments . each sensor node detects presence of a target in a single segment . communication range of a sensor node covers the segment where this node is located as well as the eight neighbouring segments . radius of the actor s communication range equals 37 segments . in most cases , the sensor nodes have to use multi - hop transmission for reporting their readings to actors . due to the long communication range , each actor can transmit data to a large number of nodes ( up to 4293 ) directly in one hop . the task of sensor nodes is to detect stationary targets in the monitored area and report their positions to actors . on the basis of the received information , each actor selects the nearest target and moves toward it . this process is executed in discrete time steps . maximum speed of actor equals two segments per time step . at each time step three new targets are created at random positions . a target is eliminated if an actor reaches the segment in which the target was detected . the targets may correspond to fires , intruders , landmines , enemy units , etc . default ( initial ) positions of actors were determined to ensure that the communication ranges of the 16 actors cover the entire monitored area ( fig . [ fig : fig1bp ] ) . an actor , which has not received information from sensor nodes about current target locations moves toward its default position . such situation occurs when there is no target within the actor s range or the information about detected target is suppressed by sensor node . for the above wsan model , data communication cost is evaluated by using two metrics : number of data transfers ( packets sent from sensor nodes to actors ) , and total hop count . the hop count is calculated assuming that the shortest path is used for each data transfer . performance of the targets elimination by actors is assessed on the basis of average time to capture , i.e. , the time from the moment when a target is created to the moment when it is captured by an actor and eliminated . during target chasing , the mobile actors decide their movement directions based on the navigation algorithm , which is presented in tab . [ tab : tab1bp ] . each actor holds a target map @xmath3 to collect the information delivered by particular sensor nodes . an element of target map @xmath4 equals 1 if the @xmath5-th actor has received information that there is target detected in segment @xmath0 . in opposite situation , the target map element equals 0 . the following symbols are used in the pseudo - code of the navigation algorithm : @xmath6 denotes a segment in which the @xmath5-th actor is currently located , @xmath7 is the nearest target according to the information collected by @xmath5-th actor in its target map , @xmath8 is the selected destination segment , and @xmath9 denotes the default actor position . it should be remembered that discrete coordinates @xmath0 are used to identify the segments . current position of actor , the destination segment as well as the targets map are broadcasted by the actor to all sensor nodes in its communication range . an actor moves toward its default position unless a target is registered in its target map . each actor takes a decision regarding the segment @xmath10 into which it will move during the next time step . the actor s decision is taken by solving the following optimization problem : @xmath11 where @xmath12 denotes the euclidean distance between segments and @xmath13 ( segments per time step ) is the maximum speed of actor . .pseudo - code of navigation algorithm executed by @xmath5-th actor [ cols= " < , < " , ] [ tab : tab3bp ] it should be noted that for @xmath14 the sts algorithm corresponds to ts . similarly , das corresponds to ts for @xmath15 as the radius of actor s communication range equals 37 . the shortest time to capture was achieved by using the das-1 algorithm with @xmath16 segments . the lowest hop counts were obtained for the sts algorithms with high @xmath17 values . figures [ fig : fig4bp ] and [ fig : fig5bp ] present detailed results for selected settings that allow the compared algorithms to achieve the maximum performance , i.e. , minimum average time to capture . the error bars show the range between minimum and maximum of the metrics obtained from the 20 simulation runs . average values are depicted as columns . ) , and das-1 ( @xmath18),width=415 ] ) , and das-2 ( @xmath19),width=415 ] according to the presented results , it can be concluded that the spatiotemporal and decision aware suppression methods reduce the number of data transfers and hop counts in comparison with the temporal suppression . moreover , these approaches decrease the average time in which actors eliminate the targets . the effect of decreased time to capture is especially visible for the algorithms that are based on the decision aware suppression . the reason underlying these results arises from the fact that when using the sts and das algorithms the sensor nodes do not report the detected targets if it is not necessary for effective navigation of actors . the information about target is transmitted from a sensor node to an actor when the distance between them is shorter . therefore , the probability that a new target will appear closer to the selected actor before it reaches the previously reported target is diminished and there is smaller chance that the assignment of targets to actors will be non - optimal . reduction of data transmission is an important issue for the development of wsan - based surveillance applications that require real - time data delivery , energy conservation , and effective utilization of the bandwidth - limited wireless communication medium . in this paper an approach is introduced to reduce the data transmission in wsan by means of suppression methods that were originally intended for wireless sensor networks . communication algorithms based on temporal , spatiotemporal , and decision aware data suppression methods are proposed for a wsan system in which mobile actors have to capture distributed targets in the shortest possible time . effectiveness of the proposed data communication algorithms was verified in computational experiments by using a wsan model . the experimental results show that the spatiotemporal and decision aware suppression methods reduce the number of data transfers and hop counts in comparison with the temporal suppression , which ensures that the actors receive complete information about targets detected in their communication ranges . further research will be conducted to test the proposed approach in more complex network scenarios . moreover , an interesting topic for future works is to investigate the impact of transmission failures on performance of the presented algorithms . vedantham , r. , zhuang , z. , sivakumar , r. : mutual exclusion in wireless sensor and actor networks . in : 3rd annual ieee communications society on sensor and ad hoc communications and networks secon06 , 1 , pp . 346355 ( 2006 ) alippi , c. , anastasi , g. , di francesco , m. , roveri , m. : an adaptive sampling algorithm for effective energy management in wireless sensor networks with energy - hungry sensors . ieee transactions on instrumentation and measurement , vol . 59 , no . 2 , pp . 335344 ( 2010 ) evans , w. c. , bahr , a. , martinoli , a. : distributed spatiotemporal suppression for environmental data collection in real - world sensor networks . in : ieee international conference on distributed computing in sensor systems dcoss , pp . 7079 , ( 2013 ) silberstein , a. , gelfand , a. , munagala , k. , puggioni , g. , yang , j. : making sense of suppressions and failures in sensor data : a bayesian approach . in : proceedings of the 33rd international conference on very large data bases , 842853 ( 2007 ) paczek , b. : communication - aware algorithms for target tracking in wireless sensor networks . in : kwiecien , a. , gaj , p. , and stera , p. ( eds . ) ccis , vol . springer , heidelberg ( 2014 ) paczek b. , bernas m. : optimizing data collection for object tracking in wireless sensor networks . in : kwiecien , a. , gaj , p. , and stera , p. ( eds . ) ccis , vol . 485 - 494 . springer , heidelberg ( 2013 )	this paper introduces algorithms for surveillance applications of wireless sensor and actor networks ( wsans ) that reduce communication cost by suppressing unnecessary data transfers . the objective of the considered wsan system is to capture and eliminate distributed targets in the shortest possible time . computational experiments were performed to evaluate effectiveness of the proposed algorithms . the experimental results show that a considerable reduction of the communication costs together with a performance improvement of the wsan system can be obtained by using the communication algorithms that are based on spatiotemporal and decision aware suppression methods . sensor and actor networks , data suppression , target tracking , surveillance applications
the intra - cluster medium ( icm ) in clusters of galaxies is known to possess magnetic fields whose origin and properties are not yet well known . the presence of cluster magnetic fields is demonstrated by a ) the existence of diffuse cluster - wide radio emission ( radio halo ) as revealed in some clusters ( e.g. coma , see giovannini et al . 1993 , and references therein ) , b ) the detection of inverse compton hard x - ray emission ( bagchi et al . 1998 , fusco - femiano et al . 1998 ) , c ) the study of variations of the faraday rotation of background sources shining through different lines of sight across the clusters , d ) the analysis of faraday rotation gradients in extended sources embedded within the cluster . kim et al . ( 1991 ) analyzed the rotation measure ( rm ) of radio sources in a sample of abell clusters and found that @xmath0 g level fields are widespread in the icm , regardless whether they do or do not have a strong radio halo . stronger magnetic fields , from about 5 up to the extreme value of 30 @xmath0 g ( as in 3c 295 , perley & taylor 1991 ; and hydra a , taylor & perley 1993 ) have been found in `` cooling flow '' clusters where extremely large faraday rotations have been revealed , suggesting that the generation of very strong icm magnetic fields may be connected with the cooling flow process ( soker & sarazin 1990 , godon et al . 1998 ) . in the coma cluster , a magnetic field of about 6@xmath1 @xmath0 g was found by feretti et al . ( 1995 ) from the analysis of the rotation measure in the cluster radio galaxy ngc 4869 . this large value is comparable to the magnetic field strength observed in cooling flow clusters , and it is considerably larger than the `` equipartition '' field derived by the radio data of the diffuse radio halo coma c , permeating the coma cluster center . the icm magnetic field can be tangled on scales much smaller than the typical galaxy size . crusius - wtzel et al . ( 1990 ) , studying the depolarization in 5 strong double sources , found tangling on smaller scales ( 1 - 4 kpc ) . this is confirmed by the results of feretti et al . ( 1995 ) on the coma cluster . the knowledge of the properties of the large - scale magnetic fields in clusters is important to study the cluster formation and evolution , and has significant implications for primordial star formation ( pudritz & silk 1989 ) . it has been suggested that strong fields can originate either by large scale dynamo amplification ( ruzmaikin 1989 ) or by turbulence following a cluster merger ( tribble 1993 ) . these magnetic fields can be traced by studying the rotation measures of radio sources located within or behind the cluster . [ cols= " < , < , < , < , < , < " , ] caption . column 1 : source name ; column 2 : maximum absolute value of rm ; column 3 : average value of rm ; column 4 : rm dispersion ; column 5 : average depolarization defined as m@xmath2/m@xmath3 ; column 6 : distance from the cluster center in units of core radii . in the case of a tangled magnetic field , with cells of uniform size , same strength , and random orientation , the observed rm along any given line of sight will be generated by a random walk process , and the distribution of rm results in a gaussian with zero mean , and the dispersion related to the number of cells along the line of sight . the source will also depolarize at long wavelength , if the external faraday screen is not fully resolved by the observing beam . the good @xmath4 fits to the polarization angle favour the interpretation that external faraday rotation is the dominant mechanism in the present sources . in table [ polar ] we summarize the results for the present radio galaxies . the most striking result is the trend of rm dispersion and depolarization with distance from the cluster center . the innermost source , 0053 - 015 , has the largest @xmath5 , the highest absolute values of rm , and the highest depolarization at long wavelengths . the source 0053 - 016 , located just beyond 1 core radius from the cluster center , still shows high values of rm , but lower than in 0053 - 015 , and also the depolarization is lower . finally , the peripheral source 3c 29 shows little rm and little depolarization . this result points to the interpretation that the external faraday screen is the same for all 3 sources , i.e. it is the intergalactic medium in a 119 , which plays different roles according to how much magneto - ionized medium is crossed by the polarized emission . this is consistent with the two nat s being really located in the inner cluster region , and not simply projected onto it . as suggested by tribble ( 1991 ) , unresolved external rm fluctuations produce a fall - off of the polarization degree with @xmath6 . a consistent picture is thus that the structure of the intergalactic magnetic field is tangled on small scales , this accounting for the observed depolarization . from the polarization degree of 0053 - 015 and 0053 - 016 ( see sect . 3 ) , there is evidence that the 3.5 observing beam does not fully resolve the screen . thus , we can argue that the scale of tangling of the magnetic field is @xmath74 kpc . moreover , field reversals must take place . the indirect detection of the magnetic field associated with the intergalactic medium of a 119 is an important result , since so far a significant intergalactic magnetic field has been only found at the center of clusters with strong cooling flows ( ge & owen 1993 , taylor et al . 1994 ) . moreover , direct evidence of a cluster magnetic field is provided in the few clusters containing a radio halo ( see e.g. feretti & giovannini 1996 ) . the magnetic field present in a 119 is spread on a size larger at least than one cluster core radius . the existence of a magnetic field component in the intergalactic medium therefore seems to be a common feature in clusters of galaxies . the determination of the strength of the magnetic field in a 119 depends on the model for the x - ray gas distribution and on several assumptions , including the rm structure function , and the field structure . in the simplest case that a constant magnetic field fills the whole cluster , using the central gas density and a core radius as the path - length , eq . ( 2 ) requires @xmath8 = 1.2 @xmath0 g . we have found , however , that the magnetic field is likely to be tangled on scales of the order of a few kpc . in this case , the intensity of magnetic field is larger by the square root of the number of cells crossed by the line of sight . a more realistic approach is to assume a magnetic field in randomly oriented cells of uniform size and strength , and a gas density distribution given by the hydrostatic isothermal model . the gas density @xmath9 has the functional form @xmath10 where the parameters have the same meaning as in eq . ( 1 ) . in this case , the rm dispersion at different projected distance from the cluster center was evaluated by felten ( 1996 ) by solving the integral of eq . ( 2 ) : @xmath11 where @xmath12 is the gamma function , @xmath13 is the core radius in kpc , and @xmath14 is the size of each cell in kpc ; the central gas density @xmath15 is in @xmath16 and @xmath8 is in @xmath0 g . the constant @xmath17 depends on the integration path over the gas density distribution : @xmath17 = 624 , if the source lies completely beyond the cluster , and @xmath17 = 441 if it is as distant from the observer as the cluster center . for the cluster a 119 , using @xmath17 = 441 and the gas parameters given in table [ clusd ] , the previous formula becomes @xmath18 the values of @xmath5 obtained for the 2 nat s of a 119 fit fairly well in the model described above and are consistent with a magnetic field strength in the range 9.6 - 13.6 @xmath0 g , for a cell size between 2 and 4 kpc . the rm of the peripheral source 3c 29 is also consistent with the previous model . however , the magnetic field probably can not be constant over 4 core radii ( @xmath19 1.5 mpc ) , otherwise the magnetic pressure would exceed the thermal pressure in the outer parts of the cluster . so it is reasonable to believe that the very low rm of 3c 29 derives not only from the low density of the foreground medium , but also from a weaker magnetic field in this region . the model used to derive the magnetic field strength is very simplified : first , the two radio galaxies can be at different locations along the line of sight ; second , the cluster shows possible substructures both in optical and in x - rays , and therefore the model used for the gas density distribution may be inaccurate ; third , the magnetic field structure could be more complicated than assumed . we can not exclude the possibility that the cell size of the magnetic field in the region around the radio galaxies would be smaller than that of the the cluster - wide magnetic field , due to the interaction between the radio sources themselves and the intergalactic medium . with cells of 10 - 20 kpc , the magnetic field derived from eq . ( 5 ) is 4.5 - 6.1 @xmath0 g . even allowing for the uncertainties related to the previous computations , the observational evidence favours the existence of a strong magnetic field in the intergalactic medium of a 119 , over a scale larger than the cluster core radius . a plausible range for the magnetic field strength is about 5 - 10 @xmath0 g . three radio galaxies belonging to the cluster abell 119 , located at different projected distance from the cluster center , are studied in total intensity and polarization . the two radio galaxies closer to the cluster center show similar nat structure , while the most peripheral radio galaxy is a classical fri source . the spectra of the nat radio galaxies progressively steepen with the distance , as expected under the effect of synchrotron energy losses . the electron lifetimes inferred from the spectral curvature lead to drift velocities of @xmath194000 - 4500 km s@xmath20 . still higher values are obtained if the magnetic field is different from the equipartion value . such large values of the velocities can derive from projection effects , or from the existence of reacceleration processes within the tails , or bulk motions along the tails and/or in the integalactic medium . we also note that non - uniformities in the magnetic field can affect the interpretation of the spectral curvature . large faraday rotations are found in the nat s . the magnitude and scale of the rm is consistent with a magnetic field embedded in the hot x - ray emitting cluster medium . to account for the observed faraday rotation , the cluster magnetic field must be present up to a distance of more than 1 cluster core radius . the strength of the field is between 5 and 10 @xmath0 g , depending on the scale of the magnetic field tangling . bagchi j. , pislar v. , lima neto g.b . , 1998 , mnras 296 , l23 bliton m. , rizza e. , burns j.o . , owen f.n . , ledlow m.j . , 1998 , mnras 301 , 609 cirimele g. , nesci r. , trevese d. , 1997 , apj 475 , 11 colless m. , burstein b. , wegner g. , et al . , 1993 , mnras 262 , 475 crusius - wtzel a.r . , biermann p.l . , lerche i. , schlickeiser r. , 1990 , apj 360 , 417 . de ruiter h.r . , parma p. , fanti c. , fanti r. , 1990 , a&a 227 , 351 eilek j.a . , arendt p.n . , 1996 , apj 457 , 105 fasano g. , falomo r. , scarpa r. , 1996 , mnras 282 , 40 felten j.e . , 1996 , in : _ clusters , lensing , and the future of the universe _ , asp conference series , vol . v.trimble and a.reisenegger , p. 271 feretti l. , giovannini g. : 1996 , in : _ extragalactic radio sources _ , iau symp . 175 , eds . r. ekers , c. fanti & l. padrielli , kluwer academic publisher , p. 333 feretti l. , dallacasa d. , giovannini g. , tagliani a. , 1995 , a&a 302 , 680 feretti l. , giovannini g. , klein u. , et al . , 1998 , a&a 331 , 475 fusco - femiano r. , dal fiume d. , feretti l. , et al . , 1998 , proceedings of 32nd cospar scientific assembly , nagoya , japan , in press , astro - ph/9808012 ge j.p . , owen f.n . , 1993 , aj 105 , 778 giovannini g. , feretti l. , venturi t. , kim k .- , kronberg p.p . , 1993 , apj 406 , 399 godon p. , soker n. , white r. e. iii , 1998 , apj 116 , 37 jaffe w.j . , perola g.c . , 1973 , a&a 26 , 423 kardashev n.s . , 1962 , sov . astron . aj 6 , 317 kim k.t . , tribble p.c . , kronberg p.p , 1991 , apj 379 , 80 . loken c. , roettiger k. , burns j.o . , norman m. , 1995 , apj 445 , 80 mack k .- h . , feretti l. , giovannini g. , klein u , 1993 , a&a 280 , 63 markevitch m. , forman w.r . , sarazin c.l . , vikhlinin a. , 1998 , apj 503 , 77 morganti r. , killeen n.e.b . , tadhunter c.n . , 1993 , mnras 263 , 1023 murgia m. , fanti r. , 1996 , internal technical report ira 228/96 napier j.p . , 1989 , in _ synthesis imaging in radio astronomy _ , eds . perley , f.r . schwab , a.h . bridle , astron . soc . of the pacific conf . , vol . 6 , p. 39 odea c.p . , owen f.n . , 1985 , aj 90 , 927 pacholczyk a.g . , 1970 , radio astrophysics , freeman , san francisco peres c.b . , fabian a.c . , edge a.c . , allen s.w . , johnstone r.m . , 1998 , mnras 298 , 416 perley r.a . , taylor g.b . , 1991 , aj 101 , 1623 postman m. , lauer t.r . , 1995 , apj 440 , 28 pudritz r.e . , silk j. , 1989 , apj 342 , 650 ruzmaikin a. , sokoloff d. , shukurov a. , 1989 , mnras 241 , 1 saglia r.p . , burstein d. , baggley g. , et al . , 1997 , mnras 292 , 499 simard - normandin m. , kronberg p.p . , button s. , 1981 , apjs 45,97 soker n. , sarazin c.l . , 1990 , apj 348 , 73 smith e.p . , heckman t.m . , 1989 , apjs 69 , 365 taylor g.b . , perley r.a . , 1993 , apj 416 , 554 taylor g.b . , barton e.j . , ge j.p . , 1994 , aj 107 , 1942 tribble p.c . , 1991 , mnras 250 , 726 tribble p.c . , 1993 , mnras 263 , 31 van der laan h. , perola g.c . , 1969 , a&a 3 , 468 wardle j.f.c . , kronberg p.p . , 1974 , apj 249 , 255 way m.j . , quintana h. , infante l. , 1998 , submitted to aj , astroph/9709036 wiita p.j . , gopal - krishna , 1990 , apj 353 , 476 zirbel e.l . , baum s.a . , 1998 , apjs 114 , 177	we present new , multiwavelength very large array observations of the 3 radio galaxies in the cluster abell 119 : 0053 - 015 , 0053 - 016 and 3c 29 . the first two radio galaxies , which lie close to the cluster center , show a narrow - angle - tailed structure , with many twists in the tails . the third radio source is located at the cluster periphery , and shows an undistorted fri morphology . all three radio sources are strongly polarized at the highest frequencies , and all three show both more depolarization and higher faraday rotation measures with increasing proximity to the cluster center . we interpret this polarization behaviour as induced by the magneto - ionized intracluster medium , whose magnetic field is estimated to be in the range 5 - 10 @xmath0 g . 27a 40l 0 1
the observation of a relation between the masses of supermassive black holes at the centers of galaxies and the stellar velocity dispersion ( @xmath2 relation ; * ? ? ? * ; * ? ? ? * ) suggests a direct link between black holes and their host galaxies . energy and momentum driven out from the central regions push gas and dust away , halting star formation and stopping agn fueling (; see for a review ) . the action of agn feedback could be achieved through the powerful radio jets in the kinetic mode . in the radiative mode , accretion disks drive powerful winds that could contribute significantly to the energy budget of the bh - galaxy system . observing the properties of such a wind is of great importance , particularly in x - rays where most of the radiation from the expelled material is produced . although warm absorber winds are common in the x - ray spectra of agn , with outflow velocities of @xmath3 and column densities of @xmath4 @xmath5 , they are weak , providing only @xmath6 of the agn bolometric luminosity . the more powerful winds seen in several objects with outflow velocities of @xmath7 and column densities of @xmath8 @xmath5 could carry power that is a few percent of the bolometric luminosity . these ultra - fast outflows ( ufo ) seem to be present in at least @xmath9 of observed agn in x - rays . however , this number could be an overestimate when alternative modeling and more conservative statistical analyses are considered @xcite . establishing how common these outflows are , their physical and geometrical properties is therefore crucial to understanding their contribution to the energy and momentum budget of black holes and their hosts . in this letter , we present analysis of the _ nustar_@xcite observation of the quasar pg 1211 + 143 . _ nustar_band ( @xmath10 kev ) with the unprecedented sensitivity at hard ( @xmath11 kev ) x - rays , fixes the continuum and thus allows a meaningful search for blue - shifted absorption below 10 kev . pg 1211 + 143(@xmath12 ) is the archetypical case for the ultra - fast outflows in active galaxies . the first observation with _ xmm - newton _ in 2001 showed evidence for highly blue - shifted absorption lines that are reminiscent of mildly relativistic disk winds ( @xmath7 ; * ? ? ? the same dataset was analyzed by @xcite who find a best fit outflow velocity of 3000 instead of the high 24,000 . letg observation showed two redshifted ( instead of blueshifted ! ) absorption lines at 4.56 and 5.33 kev in the source frame @xcite , which , when identified as the h - like k@xmath13 lines corresponds to inflowing velocities of @xmath14 . later _ xmm - newton _ observations in 2004 and 2007 showed weaker lines but seem to be consistent with the original observations @xcite , or possibly with no absorption lines at all _ nustar_observed pg 1211 + 143 in four exposures between february and july 2014 ( the exact dates are : 18 february 2014 , 08 and 09 april 2014 and 07 july 2014 ) . the four observations had net exposures 111 , 48 , 64 and 74 ks , totaling to nearly 300 ks . the data were reduced using heasoft v6.16 with the latest calibration ( version 20141020 ) . we used the scripts ` nupipeline ` and ` nuproducts ` to extract the spectral products . source and background spectra were extracted from regions on ( with a radius of 2 arcmin ) and off source respectively and grouped so that there are least 100 source counts per bin . the spectra were analyzed using xspec v12.8.2 . spectral analysis was performed on individual and combined spectra as discussed in section [ sec : spec_analysis ] . spectra from the two focal point modules a and b ( fpma and fpmb ) and from different epochs were combined using ` addspec ` tool in heasoft . the response files were combined using ` addrmf ` with the proper weighting . the resulting 330 kev light curves from the four exposures are shown in fig . [ fig : lc_spec]-left . the average 310 flux ( from a power - law fit ) is @xmath15 ergs @xmath5 s@xmath16 which is about the same as the first _ xmm - newton _ observation of 2001 , which had a 310 kev flux of @xmath17 ergs @xmath5 s@xmath16 . the source showed some flux variability between observations . no strong spectral changes are seen apart from a normalization change in the main power - law continuum ( see section [ sec : spec_analysis ] ) . one of the goals of the _ nustar_observation was to search for absorption lines from high velocity outflows . the spectrum from the new datasets is shown in fig . [ fig : lc_spec]-right . it has a clear iron k emission line and an excess above 10 kev that is most likely due to the compton reflection hump . to be systematic in the search , we consider several baseline models , including simple fits to the 3 - 10 kev band so we can directly compare with the baseline model used in . in the following discussions , we search for absorption ( and emission ) features by adding a narrow improvements . ] gaussian line and doing a systematic scan of the residuals for each baseline model . we use a grid of energies between 3.5 and 9.5 kev in 100 ev steps , and normalization values between @xmath18 and @xmath19 ( to account for both emission and absorption ) . we use a monte carlo method to obtain significance estimates . as pointed out in @xcite , a simple _ f_-test done by comparing the improvement in @xmath20 after adding a gaussian line is _ not _ appropriate for obtaining meaningful statistical confidence intervals . the baseline model is not known a priori , and neither is the reference ( null ) distribution of the @xmath21-statistic . this inapplicability of the _ f_-test is not related to the quality of data , but rather it is fundamental to any case of an added spectral component . using monte carlo methods provides a way of _ estimating _ the unknown reference distribution and then using it to assess the significance of deviations from the null model . the following steps are used to obtain a reference distribution that takes into account both the uncertainty in the baseline model parameters . these are similar to those in and but additionally take into account the baseline model uncertainties : + a ) start with a ` baseline ` model with its best fit @xmath22 and covariance matrix that encodes the uncertainty of best value parameters . + b ) draw @xmath23 random sets of parameters using the covariance matrix ( @xmath24 in our case ) . this can be achieved for instance by running a markov chain monte carlo ( mcmc ) parameter search and taking parameters directly from the resulting chains ( after they have converged ) . + c ) for each parameter set , fake spectra using the observed response and background files . this produces @xmath23 simulated spectra that are drawn from the baseline model taking into account its uncertainty . + d ) add a narrow gaussian line and scan in energy and normalization and record the maximum improvement in @xmath25 . this gives @xmath23 values of @xmath25 that are used to construct a reference distribution to which observed @xmath25 values are compared . the significance of an observed @xmath26 is @xmath27 , where @xmath28 is the number of simulated @xmath25 values greater than @xmath26 . several baseline models are used to assess their effect on estimating absorption features : 1 . ` powerlaw + zgauss ` . this consists of a power - law plus a broad gaussian fitted to the 310 kev spectrum similar to those used in and @xcite . powerlaw+reflionx ` @xcite . powerlaw+xillver ` @xcite . relxill ` @xcite . for models in 2,3 and 4 , the whole _ nustar_band where the source is significantly detected above background is used . in all models , galactic absorption is fixed at @xmath29 @xmath5 . for the relativistic reflection models , we assume a power - law emissivity and we fit for its index ( @xmath30 ) along with the black hole spin , photon index , ionization parameter , iron abundance , inclination , high energy cutoff and reflection fraction . the details of the model parameters are shown in table [ tab : params ] . for each baseline model , we repeat steps a - d . the results are shown in fig . [ fig : pg1211_sig ] . .fit parameters for the baseline models discussed in section [ sec : line_search ] . [ cols= " < , < " , ] it is clear from fig . [ fig : pg1211_sig ] that there is _ no _ significant absorption feature that is persistent with all baseline models . the phenomenological description of the first model gives an excellent description to the data in the 3 10 kev band with no significant residuals . in the ` reflionx ` fit , an absorption feature at 7.1 kev is apparent along with the emission feature at 5.8 kev . the same feature at 5.8 kev persists for the ` xillver ` fit . both the ` reflionx ` and ` xillver ` models account for the fe k@xmath13 emission at 6.4 kev . the residuals at @xmath31 indicate that additional emission is required red - ward of the line , suggesting line broadening . if broadening is included in the fit ( ` relxill , relxill+xillver ` ) , the feature disappears . the absorption in the ` reflionx ` fit at 7.1 kev disappears too when relativistic blurring is included . the last model ( ` relxill+xillver ` ) is included for completeness . it models reflection from both the inner ( with relativistic broadening ) and outer regions . although , the additional component improves the @xmath20 slightly compared to ` relxill ` alone , its requirement is not very significant . the conclusion from fig . [ fig : pg1211_sig ] is that absorption features depend on the chosen continuum , and when the continuum is pinned down with _ nustar_coverage at high energies and the inclusion of relativistic reflection , no significant absorption is seen . the analysis summarized in fig . [ fig : pg1211_sig ] was done using a combination of all the spectra ( fpma and fpmb from four observations ) to obtain the best signal . we tested for any possible flux dependency by grouping observations 1 and 4 together and 2 and 3 together ( see light curve in fig . [ fig : lc_spec]-left ) , and the results do not change ; no additional absorption lines are required by the data . the first _ xmm - newton _ observation in 2001 showed an absorption line at @xmath32 kev in the observer s frame plus possible additional lines at lower energies @xcite . we reanalyzed this _ xmm - newton _ dataset using the same method discussed in section [ sec : line_search ] and find the line at @xmath33 kev to be significant at the @xmath34 level with an equivalent width of @xmath35 ev . the flux levels of the first _ xmm - newton _ observation and the _ nustar_observation discussed here are comparable ; therefore , we also assess whether an absorption line similar to that seen in _ xmm - newton _ would have been detectable in the current _ nustar_data . we fitted the _ xmm - newton _ pn spectrum with a model consisting of a power - law , relativistic reflection and an absorption line . we then used this model to fake _ nustar_data , and followed the procedure of section [ sec : line_search ] to search for absorption line . the line is detected significantly in this simulated data showing that an absorption line would have been easily ( @xmath36 significance ) detected if it was present in the data . adding a gaussian line to our best fit model ( ` relxill ` ) with the energy and the width of absorption line seen in the _ xmm - newton _ spectrum provides a fit improvement of @xmath37 for one degree of freedom . the equivalent width of this line is @xmath38 ev . subsequent to the first observation , _ xmm - newton _ observed pg 1211 + 143several times , and when we analyze these observations self - consistently , none of them show any significant absorption features when using a baseline model consisting of a power - law and a relativistic line . the best fit model to the _ nustar_data , shown in fig . [ fig : pg1211_spec ] , consists of an absorbed power - law and relativistic reflection , both modeled with ` relxill ` . only data below 10 kev are shown to highlight the differences among models . as was clear from fig . [ fig : pg1211_sig ] , the data show an excess at @xmath39 kev when fitted with simple reflection models . the addition of relativistic blurring ( ` relxill ` ) accounts for the excess . the data , however , do not allow very strong constraints on the spin of the black hole . the parameters of the best fit model are shown in table [ tab : params ] . if we fit for the inner radius of emission instead of the black hole spin , we obtain the marginal probability contours shown in the bottom panel of fig . [ fig : pg1211_spec ] . this figure is obtained by running a monte carlo markov chain on the best fit ` relxill ` to obtain posterior probability distributions of the fit parameters and then marginalize over the parameters that are not of interest . the inner radius of emission and emissivity index are degenerate in the fit , but it is clear that although the extra broadening of the fe k@xmath13 line is required , it is not very large . the inner radius of emission ranges from @xmath40 to 40 gravitational radii ( @xmath41 ) depending on the disk emissivity . the solution with a steep emissivity and large radii might be unphysical , and although this data do not allow us to firmly locate the x - ray emission , the detection of reverberation time delays in this objects in the _ xmm - newton _ data @xcite would favor the solution with a small radius for the emission . plotted in the 310 kev only where the models differ . the x - axis represents the rest - frame energy of the source . _ bottom : _ marginal probability density contours in the ` relxill ` model for the inner radius of emission and the emissivity index . the contours are for the 68 , 90 and 99@xmath42 confidence levels . ] we have analyzed a deep 300 ks observation of the quasar pg 1211 + 143 , obtained with _ nustar_in four time intervals spanning six months . when fit with a simple power - law model , the total spectrum shows clear signatures of disk reflection ( fig . [ fig : lc_spec]-right ) . fits with the newest , most sophisticated reflection models currently available provide significantly improved fits . the residuals to fits with simple reflection show an excess around @xmath39 kev indicating that relativistic blurring is required , clearly signaling that x - ray emission probes the very innermost portion of the accretion flow in pg 1211 + 143 . the data do not permit a strong constraint on the spin of the black hole , but when reverberation delays seen previously in _ xmm - newton _ data are considered , a solution with a small inner radius ( @xmath43 , corresponding to a black hole spin of @xmath44 ) is supported by the data . we thoroughly examined the _ nustar_data both in segments and in total , and the spectra show no compelling evidence of a steady or variable ultra - fast outflow seen previously in this source . the limits are particularly tight once disk reflection has been modeled . _ nustar_clearly has the sensitivity and spectral resolution required to detect extraordinary outflows , when they are present . observations of pds 456 , for instance , reveal a likely p - cygni profile in that quasar , signaling a massive and powerful outflow from that black hole into its host galaxy ( nardini et al . 2014 , submitted ) . _ nustar_spectra from observations of ngc 1365 were able to disentangle variable absorption , outflows , and disk reflection in that source , leading to a robust measure of the black hole spin @xcite . at the opposite end of the mass scale , _ nustar_observations of the black hole candidate 4u 1630@xmath45472 revealed strong absorption that may be a ufo - like outflow , in addition to disk reflection @xcite . the true extent of these ultra - fast outflows and their persistence is unclear . the non - detection of ultra - fast outflows in the prototypical object pg 1211 + 143 in the current observation ( and also previous observations subsequent to the first _ xmm - newton _ detection ) raises several questions on how common these might be in this and other sources . systematic searches suggest that @xmath46 of objects observed with _ xmm - newton _ and _ suzaku _ show them . this number could be much lower if a careful statistical search is performed that takes into account physical baseline models along with their observational uncertainties . this comes in the form of including the best fit uncertainties in simulating spectra during the monte carlo tests and also allowing for that uncertainty when fitting for narrow features at arbitrary energies in the spectrum . the true extent of ufo s could be affected by the significance of any single detection , which we have discussed , and also by their variability . a non - detection could simply be due to the fact they are transient in nature as suggested by the the simulations @xcite . the case of pg 1211 + 143shows that the ufo is seen in one out of eleven observations ( _ xmm - newton _ , _ suzaku _ , _ chandra _ and _ nustar _ ) between 2001 and 2014 . with the caveat that we still have small numbers , if one assumes the observations are done randomly as far as the agn is concerned , it appears that a variability argument would also suggest that the contribution of ufo s is an order of magnitude lower than previously inferred . this work made use of data from the _ nustar _ mission , a project led by the california institute of technology , managed by the jet propulsion laboratory , and funded by the national aeronautics and space administration .	we report on four epochs of observations of the quasar pg 1211 + 143using _ nustar_. the net exposure time is 300 ks . prior work on this source found suggestive evidence of an `` ultra - fast outflow '' ( or , ufo ) in the fe k band , with a velocity of approximately @xmath0 . the putative flow would carry away a high mass flux and kinetic power , with broad implications for feedback and black hole - galaxy co - evolution . _ nustar_detects pg 1211 + 143out to 30 kev , meaning that the continuum is well - defined both through and above the fe k band . a characteristic relativistic disk reflection spectrum is clearly revealed , via a broad fe k emission line and compton back - scattering curvature . the data offer only weak constraints on the spin of the black hole . a careful search for ufo s show no significant absorption feature above @xmath1 confidence . the limits are particularly tight when relativistic reflection is included . we discuss the statistics and the implications of these results in terms of connections between accretion onto quasars , seyferts , and stellar - mass black holes , and feedback into their host environments . .
in complex dynamics , the mandelbrot set is a source of inspiration for much current research . this set , @xmath9 is a complicated and interesting subset of the moduli space of degree two polynomials . in the past two decades , much research has been done on dynamical systems in a nonarchimedean setting . see , for example , @xcite . for a survey of the subject , see @xcite or @xcite . if one examines the mandelbrot set over a @xmath4-adic field , one finds the object to be much less inspiring . for any prime @xmath4 , the @xmath4-adic mandelbrot set for quadratic polynomials as defined above , replacing @xmath10 with @xmath11 , is simply the unit disk . but when we consider an analogous set for polynomials of higher degree , the @xmath4-adic mandelbrot set for @xmath7 can have a complicated and interesting structure . let @xmath12 denote the parameter space of monic polynomials @xmath2 of degree @xmath1 defined over @xmath11 with @xmath13 . note that every degree @xmath1 polynomial can be put in this form via conjugation by an affine linear transformation . we call a map @xmath2 _ post - critically bounded _ ( pcb ) if all of its critical points have bounded orbit under iteration of @xmath2 . let @xmath14 denote the subset of @xmath12 that is pcb . we define the following quantity , which measures the critical radius of the @xmath4-adic mandelbrot set in @xmath12 : @xmath15 one may wonder why we define @xmath16 using the @xmath4-adic valuations of critical points of polynomials in @xmath12 rather than using the @xmath4-adic valuations of their coefficients . when @xmath17 , as we will see in theorem [ pgtd ] , the two notions are equivalent . in other situations , however , parameterizing by the critical points rather than by the coefficients is more natural . for example , when @xmath1 is a power of @xmath4 , it is easier to describe @xmath14 in terms of the critical points rather than the coefficients , as we will see in proposition [ primepower ] . this is because there is one uniform bound on the absolute value of critical points for polynomials in @xmath14 , but the bounds for the coefficients @xmath18 vary depending on the @xmath4-adic valuation of @xmath19 . knowing @xmath16 can be useful in searching for all post - critically finite polynomials over a given number field , as is done for cubic polynomials over @xmath20 in @xcite . for small primes , in particular @xmath7 , the set @xmath14 may be complicated and have a fractal - like boundary . we use @xmath16 as a way to measure its complexity . just as the critical values for quadratic polynomials in the classical mandelbrot set over @xmath10 are contained in a disk of radius 2 ( * ? ? ? * theorem 9.10.1 ) , the critical points for polynomials in @xmath14 are contained in a disk of radius @xmath21 . for @xmath17 or @xmath22 , it is known that @xmath23 , but for lack of a suitable reference we will provide an elementary proof . the following is the main result of this paper , which gives the exact value of @xmath16 for certain values of @xmath7 . [ mainthm ] for @xmath24 we have @xmath25 further , for @xmath26 we have @xmath27 . it may also be interesting to pursue such questions in berkovich space . for some work related to critical behavior for polynomials in berkovich space , see @xcite . in section 2 , we describe the notation and tools used throughout this paper . section 3 consists of some lemmas that are frequently employed in the proofs that follow . in section 4 , we discuss the known results in this realm and provide elementary proofs for when @xmath17 or @xmath22 . we prove our main result in section 5 . finally , we conclude the paper in section 6 with a study of a one - parameter family of cubic polynomials over @xmath28 to illustrate the fact that @xmath14 can indeed be quite complicated . throughout this paper , we fix a prime number @xmath4 and we let @xmath29 be a degree @xmath1 polynomial in @xmath30 $ ] . we suppress the @xmath4 from our notation for absolute values and valuations . we denote the critical points of @xmath2 by @xmath31 , not necessarily distinct , labeled so that @xmath32 we denote the closed disk centered at @xmath33 of radius @xmath34 in @xmath11 by @xmath35 the _ filled julia set of @xmath2 _ is the set @xmath36 we let @xmath37 be the smallest number such that @xmath38 equivalently , as shown in @xcite , we can define @xmath39 as follows : @xmath40 we also set @xmath41 we will often use the fact that @xmath42 where @xmath43 denotes the @xmath44 symmetric function of the critical points . whenever we count critical points , roots , or periodic points for @xmath2 , we do so with multiplicity . the newton polygon is a useful object in @xmath4-adic analysis that we will use frequently . consider a polynomial @xmath45 the newton polygon for @xmath46 is the lower convex hull of the set of points @xmath47 . if any @xmath48 , that point is omitted . ( one can think of that point as being at infinity . ) this object encodes information about the roots of @xmath46 . in particular , it tells us that @xmath46 has @xmath49 roots of absolute value @xmath50 if the newton polygon for @xmath46 has a segment of horizontal length @xmath49 and slope @xmath51 . for proofs of these facts , see @xcite . one consequence of these facts is that for polynomials , or more generally , for power series over @xmath11 , a disk in @xmath11 is mapped everywhere @xmath52-to-@xmath53 ( counting with multiplicity ) onto its image , which is also a disk . the following proposition , whose proof can be found in ( * ? * corollary 3.11 ) , will prove useful . [ diskbehavior ] let @xmath54 $ ] be a degree @xmath1 polynomial and let @xmath55 be a disk in @xmath11 . then @xmath56 , where @xmath57 moreover , @xmath58 is everywhere @xmath51-to-@xmath53 for some positive integer @xmath51 , counting with multiplicity . [ risr ] let @xmath59 , and let @xmath60 and @xmath39 be as defined by and . if @xmath61 and @xmath62 , then @xmath63 . first note that if @xmath2 is post - critically bounded , then @xmath64 is necessary . recall that @xmath65 since @xmath66 the only way that @xmath39 could be strictly greater than @xmath60 is if @xmath67 is maximal in the formula for @xmath39 , with @xmath68 in this case , we see that @xmath69 could be as large as @xmath70 . but if this is true , then @xmath71 is dominated by a single term , namely @xmath72 with @xmath73 contradicting the fact that @xmath2 is pcb . thus @xmath63 . [ pto1 ] let @xmath74 $ ] be a degree @xmath1 polynomial , let @xmath75 be a disk in @xmath11 , and let @xmath51 be an integer with @xmath76 . if @xmath2 maps @xmath75 @xmath51-to-1 onto its image , then @xmath75 contains exactly @xmath77 critical points of @xmath2 , counted with multiplicity . without loss of generality , replace @xmath2 with a conjugate so @xmath78 . let @xmath79 then , counting with multiplicity , @xmath80 has @xmath51 roots in the unit disk , which implies that @xmath51 is the largest positive integer such that @xmath81 now consider the newton polygon for @xmath82 . since @xmath51 is the largest integer such that @xmath83 is minimal and @xmath84 , we see that @xmath51 is also the largest integer such that @xmath85 is minimal among all @xmath86 . therefore , the newton polygon for @xmath82 has exactly @xmath77 non - positive slopes , which implies that there are @xmath77 critical points , counted with multiplicity , in @xmath87 . [ pgtd ] let @xmath17 . then @xmath88 is pcb if and only if @xmath89 for all critical points @xmath90 of @xmath2 . this is a known result , but since it does not appear in the literature , we present an elementary proof here . first , suppose @xmath89 for all @xmath19 . then @xmath91 for all @xmath19 , and so @xmath92 . therefore , @xmath2 is pcb . now let @xmath2 be pcb and suppose , for contradiction , that @xmath93 . let @xmath51 be maximal such that @xmath94 . ( in other words , there are exactly @xmath51 critical points with absolute value @xmath95 . ) first , we show that there are exactly @xmath51 roots @xmath96 of @xmath2 such that @xmath97 . since @xmath17 , the newton polygons for @xmath2 and @xmath82 are the same , up to horizontal translation . thus , the rightmost segment of the newton polygon for @xmath2 has the same slope and horizontal length as the rightmost segment of the newton polygon for @xmath82 , and therefore , @xmath2 has exactly @xmath51 roots @xmath98 such that @xmath97 . next , we use lemma [ pto1 ] to reach a contradiction . consider @xmath99 . this is a union of up to @xmath1 smaller disks @xmath100 , where the @xmath98 are the roots of @xmath2 . note that , since @xmath2 is pcb , each critical point must lie in one of these disks . by proposition [ diskbehavior ] , we know that each @xmath101 . so , each of the @xmath51 large critical points @xmath102 must lie in the following set : @xmath103 @xmath104 is a disjoint union of @xmath105 disks . relabel the subscripts so that we can express @xmath104 as follows : @xmath106 let @xmath100 map @xmath107-to-1 onto @xmath108 . then , since @xmath104 contains exactly @xmath51 preimages of 0 , counted with multiplicity , we have @xmath109 let @xmath110 be the number of critical points in @xmath100 . then , lemma [ pto1 ] tells us that @xmath111 , and so the number of critical points , counted with multiplicity , in @xmath104 is @xmath112 this is a contradiction . thus , if @xmath2 is pcb , all the critical points lie in the unit disk . in particular , theorem [ pgtd ] implies that @xmath27 for @xmath113 . the same is also true if @xmath22 for some positive integer @xmath114 . this result follows immediately upon comparing the newton polygons for @xmath2 and @xmath82 . a proof of this result can be found in @xcite , but we present a proof below that is simple and tailored to the normal form used in this paper . [ primepower ] let @xmath22 for some positive integer @xmath114 , and let @xmath88 . then @xmath2 is pcb if and only if @xmath89 for all critical points @xmath90 of @xmath2 . first , suppose all the critical points for @xmath2 lie in the unit disk . then all the coefficients of @xmath2 are @xmath4-integral , and so @xmath92 . therefore , @xmath2 is pcb . now , suppose @xmath2 is pcb . by comparing the newton polygons for @xmath2 and @xmath82 , one sees that the slope of the rightmost segment of the newton polygon for @xmath82 is greater than the slope of the rightmost segment of the newton polygon for @xmath2 . in other words , the largest critical point of @xmath2 is strictly larger than the largest root . if this critical point @xmath115 were outside the unit disk , then @xmath116 , and @xmath2 would not be pcb . therefore , @xmath2 is pcb if and only if all critical points lie in the unit disk . so far , we have seen a few situations in which the @xmath4-adic mandelbrot set can be very easily described : it is simply a product of unit disks . for primes smaller than @xmath1 , this is often not the case . in section 6 , we investigate a one - parameter family of cubic polynomials over @xmath6 to illustrate that @xmath14 can be quite intricate . while we can not hope to describe @xmath14 exactly for @xmath7 , we can get a sense of the size of @xmath14 by calculating @xmath16 . we now give a lower bound for @xmath16 when @xmath7 . [ lowerbound ] suppose that @xmath7 and that @xmath1 is not a power of @xmath4 . let @xmath114 be the largest integer such that @xmath118 and let @xmath119 be the largest integer such that @xmath120 . write @xmath121 , where @xmath122 and @xmath123 . then , @xmath124 let @xmath125 satisfy the following equation : @xmath126 then , the lower bound given in this proposition is realized by the following map : @xmath127 this map has either two or three critical points : @xmath128 , @xmath129 , and possibly 0 . ( zero is a critical point if @xmath130 . ) we choose @xmath128 above so that @xmath131 and @xmath13 . thus , @xmath2 is post - critically finite , and therefore pcb , with @xmath132 since @xmath119 is necessarily less than or equal to @xmath114 , this gives a positive lower bound for @xmath16 in most situations in which @xmath7 . it does not give a positive lower bound when @xmath133 , where @xmath134 , because in this situation @xmath135 and thus @xmath136 . the next proposition is the beginning of an exploration of that situation . [ twop ] let @xmath88 and suppose @xmath137 . then @xmath2 is pcb if and only if @xmath89 for all critical points @xmath90 of @xmath2 . proposition [ primepower ] proves this statement if @xmath138 , so we proceed assuming @xmath139 . again , one direction is straightforward . if all the critical points are in the unit disk , then all the coefficients of @xmath2 are @xmath4-integral , and @xmath2 is pcb . now let @xmath2 be pcb and suppose for contradiction that @xmath2 has a critical point outside the unit disk , with @xmath140 . by comparing the rightmost segments of the newton polygons for @xmath2 and @xmath82 , since the rightmost vertex for @xmath82 is one unit above the rightmost vertex for @xmath2 , we get that in most cases the largest critical point for @xmath2 is larger than its largest root . since we can write @xmath141 , where the @xmath98 are the roots of @xmath2 counted with multiplicity , in this situation @xmath142 for all @xmath19 , and therefore @xmath143 . more generally , @xmath144 , and thus @xmath2 can not be pcb . the only situation in which this does not happen is if the rightmost segment of the newton polygon for @xmath2 has horizontal length equal to @xmath4 , in which case it is possible for the largest root of @xmath2 to have the same absolute value as the largest critical point . in this situation , there are exactly @xmath4 roots @xmath98 with @xmath97 , and there are at least @xmath4 critical points @xmath90 with @xmath145 . suppose there are exactly @xmath114 such critical points , counted with multiplicity , where @xmath146 . now we use lemma [ pto1 ] to show that this is only possible if they are all contained in a disk centered at a root @xmath147 that maps @xmath4-to-1 onto @xmath148 . recall that if @xmath2 is pcb , then each critical point lies in a disk @xmath100 mapping via @xmath2 onto @xmath149 , where @xmath150 . proposition [ diskbehavior ] implies that @xmath151 for all @xmath19 . since @xmath152 , it is necessary that each of the @xmath114 large critical points lie in one of these disks centered at a root @xmath98 with @xmath153 . since there are at least @xmath4 such critical points and only @xmath4 such roots , lemma [ pto1 ] implies that this is only possible if there is one disk @xmath154 mapping via @xmath2 onto @xmath149 containing all @xmath4 such roots and all @xmath114 of the largest critical points . writing @xmath155 for @xmath156 , we calculate @xmath71 : @xmath157 we will reach a contradiction if the coefficient of @xmath158 is a @xmath4-adic unit , because this would imply that @xmath159 , and thus that @xmath2 is not pcb . modulo @xmath4 , the coefficient of @xmath158 is congruent to @xmath160 this is because @xmath161 and @xmath162 . thus , we reach the desired conclusion , that @xmath2 is pcb if and only if all the critical points lie in the unit disk . in particular , proposition [ twop ] implies that @xmath163 . a similar but more elaborate argument shows that @xmath164 as well , but the techniques used do not generalize to arbitrary @xmath165 . now we turn our attention to the case where @xmath166 to prove the remainder of theorem [ mainthm ] . suppose that @xmath166 . note that proposition [ lowerbound ] shows that @xmath167 . it remains to show that @xmath168 is also an upper bound for @xmath16 . suppose there is a polynomial @xmath59 with a critical point @xmath169 such that @xmath93 . let @xmath170 , where @xmath171 . lemma [ risr ] implies that the critical orbits for @xmath2 are all contained in @xmath172 . let @xmath51 denote the number of critical points with absolute value @xmath95 ( with multiplicity ) , i.e. , @xmath173 we break the proof into two cases . the first case we consider is @xmath174 . we will refer to @xmath175 as the _ large critical points_. each large critical point must lie in one of the disks in the following set , where @xmath176 and @xmath101 : @xmath177 by lemma [ pto1 ] , we must have more than @xmath51 roots @xmath98 such that @xmath97 . since the newton polygons for @xmath2 and @xmath82 can only differ at one place ( namely , at the @xmath4th place ) , this is only possible if there are exactly @xmath114 roots of @xmath2 ( and at most @xmath178 critical points ) with absolute value @xmath95 . this implies that @xmath179 . let @xmath180 be the largest critical point such that @xmath181 and let @xmath182 . since @xmath183 , we must have @xmath184 , which implies that @xmath185 . looking at @xmath186 , the sole largest term is @xmath187 , which implies that @xmath188 if @xmath2 is pcb , then @xmath189 , which gives the inequality @xmath190 , and the desired bound follows . now suppose the number of large critical points is @xmath191 . then , by analysis of the newton polygons for @xmath2 and @xmath82 , either @xmath2 has a root @xmath147 with @xmath192 , or @xmath2 has exactly @xmath51 roots of absolute value @xmath95 . the first possibility does not occur , because if @xmath193 , then @xmath147 must be in the basin of infinity , by lemma [ risr ] . this is a contradiction , since @xmath147 is preperiodic , as 0 is a fixed point for @xmath2 . so , the largest root @xmath147 of @xmath2 satisfies @xmath194 and the number of large critical points is equal to the number of roots of absolute value @xmath95 . by lemma [ pto1 ] , the only way for @xmath2 to be pcb is if there is a disk @xmath195 mapping @xmath4-to-1 onto @xmath149 containing at least @xmath4 of the large critical points , where @xmath196 by proposition [ diskbehavior ] . we will again divide into two cases . first , suppose @xmath197 for all critical points @xmath198 . let @xmath155 , where @xmath199 . then we have @xmath200 we will use the fact that @xmath201 to simplify our expression for @xmath71 to the following : @xmath202 it remains to check that the coefficient of @xmath203 is a @xmath4-adic unit and to determine the largest possible absolute value for @xmath204 . first , we look at the coefficient of @xmath203 in ( [ fc1 ] ) . this coefficient can be rewritten as follows : @xmath205 since the full alternating sum from 0 to @xmath1 of binomial coefficients is always zero , we see that the coefficient of @xmath203 is either @xmath53 or @xmath206 , depending on whether @xmath1 is even or odd . either way , it is a @xmath4-adic unit , and so the first term in @xmath71 has absolute value @xmath207 . since we must have @xmath208 in order for @xmath2 to be pcb , it is necessary that @xmath209 as well . the only term that can possibly be that large is the one corresponding to @xmath72 . let @xmath210 denote the @xmath211 symmetric function on the @xmath212 . then , the portion of @xmath72 contributing to @xmath204 is : @xmath213 note that since @xmath214 is a multiple of @xmath4 for all @xmath215 , the last term is the only one that can possibly realize the absolute value @xmath207 . looking at @xmath216 , we see that @xmath217 since @xmath218 , we have @xmath219 , which implies that @xmath220 . this is strictly smaller than @xmath168 for @xmath221 , and we obtain the desired result . we will now treat the @xmath222 case separately . suppose @xmath223 and all @xmath4 critical points are in a disk centered at @xmath169 of radius @xmath224 . the above argument shows that , if @xmath2 is pcb , then we must have @xmath225 . we will improve our upper bound for @xmath34 to prove the result in this case . we know that @xmath195 maps @xmath4-to-1 onto @xmath149 . writing @xmath226 so that it is centered at @xmath169 , we have @xmath227 this allows us to see that , by proposition [ diskbehavior ] , @xmath228 in particular , @xmath229 , which implies that @xmath34 satisfies the inequality @xmath230 to calculate @xmath231 , we note that @xmath232 plugging in @xmath233 , we calculate @xmath234 this implies that @xmath235 . plugging this into ( [ sbound ] ) , we get that @xmath236 . finally , since @xmath237 , we see that @xmath238 , as desired . now we return to the general situation to treat the final case , in which there are at least @xmath4 large critical points and there exist @xmath198 such that @xmath239 . without loss of generality , let @xmath240 , where @xmath169 is in the disk @xmath195 which contains at least @xmath4 critical points . there must be a fixed point @xmath128 for @xmath2 such that @xmath241 . we see this by examining the equation @xmath242 here the @xmath243 are the fixed points for @xmath2 . since the left hand side of this equation must have absolute value at most @xmath95 , the same must be true of the right hand side . since 0 is a fixed point , we can let @xmath244 . then , since @xmath245 , we are left with @xmath246 this implies that there is some @xmath243 satisfying @xmath247 . call this fixed point @xmath128 . next , conjugate @xmath2 by the affine linear transformation @xmath248 . the new map @xmath249 is of the desired form ( monic with @xmath13 ) and is pcb because @xmath2 is pcb . note that @xmath250 has at least @xmath4 , but no more than @xmath251 , critical points in @xmath252 , where @xmath253 . this implies that the number of large critical points for @xmath250 is strictly less than @xmath4 . we have already dealt with this case , and so we know that all the critical points @xmath254 for @xmath250 satisfy @xmath255 . so , we can conclude that @xmath256 for all critical points @xmath90 of @xmath2 . if any @xmath90 satisfies @xmath257 , we can conclude that @xmath258 as well , and we reach the desired conclusion . however , if all critical points have the same absolute value as @xmath128 , the result does not yet follow . suppose that @xmath145 for all @xmath19 and that there exists @xmath90 such that @xmath259 . we have just shown that @xmath260 , which implies that @xmath261 for all @xmath262 . therefore , we can proceed assuming that @xmath263 . since not all critical points are in @xmath195 , there is another disk in the set @xmath264 containing @xmath265 critical points and ( by lemma [ pto1 ] ) @xmath266 roots . thus , since all the large critical points and roots must be contained in @xmath104 in accordance with lemma [ pto1 ] , the number of roots of absolute value @xmath95 outside @xmath267 must exceed the number of critical points outside @xmath195 by at least one , and therefore , the number of critical points inside @xmath267 must be greater than @xmath4 , as there are exactly @xmath4 roots in @xmath267 . let @xmath268 and write all critical points as before , with @xmath269 . here , @xmath270 for @xmath271 , and @xmath272 for @xmath273 . looking at equation , we examine the size of @xmath204 . once again , in order to have the necessary cancellation with the leading term , it must be true that @xmath209 , which can only be achieved if @xmath274 . but , @xmath275 this gives the following inequality : @xmath276 this reduces ( after a bit of algebra ) to : @xmath277 which is the desired result . we have alluded to the fact that @xmath14 can be an interesting set when @xmath7 . the following example of a one - parameter family of cubic polynomials reveals that the boundary of this mandelbrot set can be complicated and fractal - like . consider the following one - parameter family of cubic polynomials , where the parameter @xmath278 : @xmath279 of the two critical points , one ( zero ) is a fixed point and the other is @xmath280 . note that @xmath281 corresponds to a post - critically finite map ( i.e. , all critical points are preperiodic ) , with the free critical point 1 mapping to the fixed point @xmath282 . [ bdrybehavior ] consider the one - parameter family of cubic polynomials in defined in . for the sequence of parameters @xmath283 converging to @xmath281 , the corresponding polynomials @xmath284 for @xmath285 are not pcb . there is another sequence , @xmath286 , also converging to @xmath281 , for which the corresponding polynomials for @xmath287 are all pcb . proposition [ bdrybehavior ] shows that @xmath281 is on the boundary of the @xmath4-adic mandelbrot set for this family of polynomials , in that it is arbitrarily close in the parameter space to parameters corresponding to both pcb and non - pcb maps . for @xmath17 , such examples do not exist , as @xmath288 is simply the unit polydisk in @xmath289 , which has empty boundary . first , consider the sequence @xmath291 , where @xmath292 . as @xmath114 approaches infinity , @xmath293 approaches @xmath53 in @xmath28 . we will now show that for @xmath285 , the orbit of the critical point @xmath293 under @xmath284 is unbounded . in fact , we can say something stronger . if @xmath294 for some @xmath295 with @xmath285 , then @xmath296 is not pcb . let @xmath297 and @xmath285 such that @xmath298 . we begin by calculating the first few iterates of @xmath280 under @xmath296 : @xmath299 @xmath300 @xmath301 from here , each iterate moves further away from @xmath282 , so that for @xmath156 we have @xmath302 thus , @xmath303 and we can write @xmath304 , where @xmath305 . a quick calculation then shows that @xmath306 . note that in this case , @xmath307 . so , since @xmath308 , the orbit of @xmath280 is unbounded . now we turn our attention to the other sequence . let @xmath286 . we will show that the orbit of @xmath309 under iteration of @xmath310 is bounded . for ease of notation , let @xmath311 for some @xmath287 . once again , we begin by calculating the first few iterates : @xmath312 @xmath313 @xmath314 in general , for @xmath315 , we have @xmath316 more specifically , @xmath317 . this shows that we can write @xmath318 for some odd integer @xmath319 . calculating one more iterate , we see that @xmath320 this puts @xmath280 in the basin of attraction of @xmath321 , because @xmath296 maps @xmath322 onto itself . this example shows that this @xmath327-adic mandelbrot set , i.e. , the set @xmath328 , has a complicated boundary . while this set is difficult to visualize over @xmath28 , we can begin to draw this set if we restrict to @xmath6 . let @xmath329 . for @xmath330 , note that @xmath331 . we now calculate @xmath332 : @xmath333 therefore , the orbit of @xmath280 is unbounded for all @xmath280 outside the unit disk . next note that @xmath296 maps @xmath87 into itself for @xmath334 . so @xmath335 for @xmath334 . if we are interested in the boundary of @xmath336 , we therefore only have to consider @xmath280 for which @xmath290 . we can represent a neighborhood in @xmath6 as a binary tree , as every disk in @xmath6 is comprised of two disjoint disks . for example , the disk @xmath337 , which will be the root of our tree , is comprised of @xmath338 and @xmath339 . each disk in turn branches into two smaller disks . traversing down the tree , one `` zooms in '' on a point in @xmath6 . see figure [ fig : fig1 ] for a depiction of the first few levels of this tree . we color a node black if the entire disk is in @xmath336 , we color a node white if the entire disk is outside @xmath336 , and we color a node gray if it contains some points in @xmath336 and some points outside it . the number that labels each node denotes the center of the disk the node represents . as one moves down the left side of the tree , one zooms in on the post - critically finite boundary point @xmath281 . this tree is symmetrical , because @xmath340 , and so @xmath296 is pcb if and only if @xmath341 is pcb . in figure [ fig : fig2 ] , we depict the disk @xmath342 and the tree that emanates from it to give a sense of the complexity of @xmath336 . note that as one zooms in on @xmath281 , a self - similar pattern emerges , as illustrated in figure [ fig : fig3 ] and as shown in proposition [ bdrybehavior ] . this is reminiscent of the classical mandelbrot set over @xmath10 and its fractal - like boundary . beginning at @xmath326 for any @xmath221 , we see in figure [ fig : fig3 ] that the pattern repeats every time we move two levels down the tree toward @xmath53 . the disk @xmath343 corresponds to non - pcb maps , while the disks @xmath344 and @xmath345 correspond to pcb maps . the parameter values in the disk @xmath346 , labeled in figure [ fig : fig3 ] with a question mark , exhibit quite complicated behavior . some data collected with sage shows that for many values of @xmath114 one often has to move twenty or more levels down the tree before one finds a disk that is entirely pcb or entirely non - pcb . furthermore , the locations of these disks vary with @xmath114 , and there is no apparent pattern . this is where the boundary of @xmath336 seems to be most intricate . the author would like to thank her advisor , joseph silverman , for his guidance and advice , robert benedetto for an email exchange that prompted this project , the referee for his or her valuable feedback , and everyone involved in icerm s semester program in complex and arithmetic dynamics for providing the stimulating research environment in which much of this work was completed .	let @xmath0 $ ] be a degree @xmath1 polynomial . we say @xmath2 is _ post - critically bounded _ , or pcb , if all of its critical points have bounded orbit under iteration of @xmath2 . it is known that if @xmath3 and @xmath2 is pcb , then all critical points of @xmath2 have @xmath4-adic absolute value less than or equal to 1 . we give a similar result for @xmath5 . we also explore a one - parameter family of cubic polynomials over @xmath6 to illustrate that the @xmath4-adic mandelbrot set can be quite complicated when @xmath7 , in contrast with the simple and well - understood @xmath8 case . & pdflatex
there are several reasons to believe that a population of intergalactic globular clusters ( igcs ) should exist outside of galaxies : \(1 ) the jeans mass at recombination was @xmath0 solar masses , and hence globular cluster sized objects could have formed wherever the local density of matter was high enough . \(2 ) many galaxies may have met their demise over a hubble time as a result of collisions and tidal disruption . globular clusters are likely to survive the disruption of their parent galaxy , resulting in the gradual accumulation of a population of igcs . intergalactic stars , planetary nebulae , supernovae and hii regions have already been found ; it would be surprising if there were no igcs . \(3 ) the existence of igcs might explain high specific frequencies , bimodal globular cluster metallicity distributions and other current puzzles in the study of globular cluster systems . jordn et al . ( 2003 ) reported a tentative detection of igcs in the center of the rich galaxy cluster a1185 ( @xmath1 ) based on @xmath2-band images obtained with wfpc2 on the hubble space telescope . we ( ct , jordn , marzke , west ) recently obtained very deep , multicolored ( @xmath3 and @xmath2 ) images of the same a1185 field using hst with the new acs . the goals of these new observations are to 1 ) detect the peak of the assumed universal gaussian - like globular cluster luminosity function ( which should occur at @xmath4 at a1185 s distance ) and thereby confirm that these candidate igcs are bona fide globular clusters and 2 ) use color information to infer their metallicities . preliminary analysis indicates that we are reaching sufficiently faint magnitudes to reliably detect the luminosity function turnover . the number and colors ( metallicities ) of igcs will provide constraints on the number and types of galaxies that have been destroyed or stripped over a hubble time . using the keck telescope , we ( ferguson , gregg , tanvir , von hippel , west ) recently measured the redshift of a candidate igc in the nearby virgo galaxy cluster that was found serendipitously on an hst image obtained for another project . preliminary data reductions show that this object , which is slightly resolved in the hst image and appears to be a distant globular cluster , has a recessional velocity of @xmath5 km / s , and hence is most likely in the virgo cluster . its apparent magnitude , @xmath6 , is consistent with it being a bright globular cluster . using telescopes on mauna kea we have since obtained optical and nir colors of this object , as well as a medium - resolution spectrum that should yield its velocity dispersion . these data are presently being analyzed .	we confirm and extend our previous detection of a population of intergalactic globular clusters in abell 1185 , and report the first discovery of an intergalactic globular cluster in the nearby virgo cluster of galaxies . the numbers , colors and luminosities of these objects can place constraints on their origin , which in turn may yield new insights to the evolution of galaxies in dense environments . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
on february 1974 , at the same aaas meeting in san francisco where the discovery of grbs by the vela satellites was publicly announced @xcite , the possible relation of grbs with the `` moment of gravitational collapse '' leading to a bh formation was advanced ( see * ? ? ? @xcite considered , for definiteness , the vacuum polarization process occurring in an overcritical kerr - newman bh ( knbh ) . evidence was given for : a ) the formation of a vast amount of @xmath7-baryon plasma ; b ) the energetics of grbs of the order of @xmath8 erg , where @xmath9 is the bh mass , implying their cosmological origin ; c ) the ultra - high energy cosmic rays with energy up to @xmath10 ev originating from such an extreme electrodynamical process . soon after , the role of an @xmath7 plasma for the origin of grbs was also considered by @xcite . it took almost thirty years to clarify some of the analogies and differences between these two processes of @xmath7-pair creation leading , respectively , to the alternative concepts of fireball " and fireshell " @xcite . already in @xmath11 , well before the establishment of the grb cosmological nature and energetics , @xcite gave support to the cosmological interpretation of grbs and indicated in merging ns binaries their possible origin . they also pointed out the relevance of such ns ns mergers for the occurrence of r - process , as well as for the emission of gravitational radiation , indicating the uncertainty in the determination of their rate of occurrence . following the launch of the compton satellite and the observations by the batse detector @xcite , a phenomenological classification based on the prompt @xmath12 duration was advanced : grbs were classified into long grbs for @xmath13 s , and short grbs for @xmath14 s @xcite . shortly after @xcite indicated the possible cosmological origin of short grbs originating in binary ns mergers . they also introduced the clear indication of the role of @xmath15 annihilation leading to the formation of an @xmath7 plasma . this paper was followed by a large number of theoretical works including the gravitational wave emission in newtonian , post - newtonian , and general relativistic treatments ( see , e.g. , * ? ? ? * ) , as well as the @xmath15 annihilation leading to an @xmath7 plasma ( see , e.g. , @xcite and @xcite and references therein ) . soon after the paper by @xcite , @xcite also supported the cosmological origin of grbs and introduced the concept of bh - accretion - disks , produced by the collapse of a very massive star . such a system was indicated by its author as a _ collapsar _ and was assumed to be the origin of ultrarelativistic jets expected to occur by the same author in long grbs . for a recent review see . after the determination of the cosmological nature of grbs @xcite and the confirmation of their outstanding energy ( @xmath16 erg ) , we returned to our grb scenario @xcite . in a period of four years , from 1997 to 2001 , we developed a fully relativistic grb theoretical model examining , as well , the dynamics of the @xmath7 plasma originating the grb emission ( the fireshell model , see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * and section [ sec : fireshell ] ) . the fireshell model applies to both short and long grbs . the origin of short grbs from ns ns ( or ns - bh ) binaries as `` in - states '' has been confirmed by strong observational and theoretical evidences . in this article we address specifically some of the latest results within the fireshell model @xcite on the possible presence or absence of a bh formation in ns ns mergers , the consequent classification of short bursts into s - grfs , when no bh is formed ( see section [ sec : descr_s - grfs ] ) , and s - grbs , when a bh is formed ( see section [ sec : descr_sgrbs ] ) , and the computation of their occurrence rate ( see section [ sec : rates ] ) . the application of the fireshell model to the case of long grbs followed a longer path for reaching a proper understanding of the overall phenomenon . the first application of our model to a long grb was implemented on grb 991216 @xcite . in these papers a clear difference between the thermal component observed at the transparency of the @xmath7 plasma , the proper grb ( p - grb ) emission @xcite , and the non - thermal remaining part , later called prompt emission @xcite , was evidenced . this fully relativistic approach was not readily accepted by the grb community , also in view of its objective technical complexity and novelties in the theoretical physics scenario . some authors attempted to describe the grb phenomenon by simplified newtonian approaches , e.g . , those based on the concept of _ magnetars _ . as the detailed observations of the x - ray afterglow by the _ swift_-xrt were obtained , as well as the high energy emission by the fermi - lat @xcite , our model has correspondingly evolved pointing out the precise common power - law behavior of the rest - frame @xmath17@xmath18 kev x - ray luminosity light curves @xcite , as well as the nesting properties @xcite . as pointed out in the present article , the concept of long grbs has evolved into xrfs and bdhne , depending on the possible presence or absence of a bh in their formation process ( see also * ? ? ? it is appropriate to recall that the quest for having progenitors for the collapsar hypothesized by @xcite led to an interesting direction of research dealing with a binary system composed of two very massive stars of @xmath19 m@xmath20 each . the large masses involved in these systems were introduced in order to form a bh at the end of their evolution . similarly , the large amount of angular momentum of the system would guarantee the formation of an accretion disks needed in the collapsar model @xcite . up to six different scenarios were there envisaged leading to a _ collapsar _ , as well as a few leading , alternatively , to a variety of binary compact systems . the need for choosing low - metallicity massive stars followed from the expectation of the formation large bhs in their evolution @xcite . the elimination of h from metal - rich massive stars would follow naturally , but the formation of a bh was not expected in their final stage of evolution @xcite . the spatial and temporal coincidence of a long grb explosion with an optical sn , first observed in the association between grb 980425 and sn 1998bw , created a profound conceptual turmoil in the grb community . woosley and collaborators postulated the birth of a sn out of a collapsar . in our approach , grbs were supposed to originate from the bh formation , while sne were expected to lead only to nss ( see , e.g. , * ? ? ? * and references therin ) . we consequently introduced a new paradigm to explain the coincidence of these two qualitative and quantitative different astrophysical events in space and time : the birth of a sn and the occurrence of a grb . the induced gravitational collapse ( igc ) paradigm was then introduced . this approach differs from alternative descriptions of the grb - sn coincidences occurring , e.g. , in the _ magnetars _ and the _ collapsar _ models , where the two events are coming from a single progenitor star , and takes full advantage of the recent observations of sne ibc in interacting binary systems . in a first formulation we considered a finely tuned process : the grb triggering the explosion of a binary companion star very close to the onset of the sn @xcite this scenario soon led to the alternative igc paradigm in which a co@xmath0 undergoes a sn explosion in presence of a ns companion in a tight binary system . this is also by itself an unlikely event which in order to occur needs a fine tuning of the initial conditions of the binary system . this scenario was shown to be consistent with population synthesis analysis @xcite . the sn explosion induces a hypercritical accretion of its ejecta onto the companion ns , leading to the formation of a more massive ns ( mns ) , when the ns critical mass @xmath21 is not reached , or to the formation of a bh with the associated grb emission in the opposite case ( see , e.g. , * ? ? ? * ; * ? ? ? the igc scenario was first tested and verified in grb 090618 . it soon became clear that the occurrence of a grb is far from being a single event , but it is part of an authentic laboratory composed of a variety of astrophysical relativistic phenomena preceeding and following the prompt grb emission . we adopted as progenitor of our co@xmath0ns binary system the massive binaries independently considered in @xcite and @xcite . in our case , the late evolution of such massive binary systems do not lead to a collapsar , nor to an hypernova , but to a much richer and vast number of possibilities , made possible by our igc paradigm . consistently with the considerations by @xcite , indicating that xrfs , x - ray rich bursts , and all long grbs are part of a same population which we show to originate in the hypercritical accretion process of the sn ejecta onto a binary companion ns . in agreement with the considerations by @xcite for a sub - classification of long bursts into low - luminous and high - luminous grbs , we have divided the long bursts into two different scenarios depending on the distance between the co@xmath0 and the ns binary companion @xcite . correspondingly two different sub - classes of long bursts , both originating in the hypercritical accretion process of the igc scenario , have been shown to exist @xcite : the xrfs , which clearly include low - luminous grbs , such as grb 060218 @xcite , when no bh is formed ( see section [ sec_3_prot_rat ] ) , and the bdhne , such as grb 130427a @xcite , when a bh is formed ( see section [ sec : descr_bdhne ] ) . their occurrence rates have been computed ( see section [ sec : rates ] ) . instead of proposing just a new classification , we also give the description of its underlying physical origin : the hypercritical accretion process of the sn ejecta onto a binary companion ns , with the full associated theoretical treatments at the basis of the igc paradigm . to the above four sub - classes of long and short bursts , we have recently added a new hybrid sub - class of ultrashort grbs ( u - grbs ) which , as recently pointed out in @xcite , can originate during the further evolution of the bdhne out - states . indeed , nearly 100@xmath22 of the ns - bh binaries which are the outcome of bdhne remain bound . their orbital velocities are high and even large kicks are unlikely to unbind these systems . they represent a new family of ns - bh binaries unaccounted for in current standard population synthesis analyses ( see , e.g. , * ? ? ? * and section [ sec : rate5 ] ) . the above considerations based on the igc paradigm and the ns - ns paradigm as progenitors encompass and classify into sub - classes most of the known astrophysical systems related to grbs . we finally recall the existence of a class of long grbs occurring in a low density circumburst medium ( cbm ) with density @xmath23 @xmath24 , with hybrid short / long burst properties in their @xmath25-ray light curves : 1 ) an initial spike - like harder emission and 2 ) a prolonged softer emission observed up to a hundred seconds . these bursts do not have an associated sn , even though in the case of a low value of the cosmological redshift its detection would not be precluded . the prototype of such systems is grb 060614 @xcite . the progenitor for this class of long bursts has been identified in a binary system composed of a ns and a white dwarf ( wd ) @xcite . their merger leads to a mns with additional orbiting material , but not to an authentic grb . we refer to these systems , historically addressed as disguised short grb , as gamma - ray flashes ( grfs ) , see , e.g. , grb 060614 , @xcite and grb 071227 , @xcite ) . in the following we adopt the term _ burst _ only for those systems leading to the bh formation , namely to s - grbs , u - grbs and bdhne . we refer to the term _ flash _ , instead , only for those systems not leading to the bh formation , namely to s - grfs , grfs and xrfs ( see fig . [ fig : summary ] ) . the main topic addressed in the present article is to estimate the rates of occurrence of the xrfs , bdhne , s - grfs , s - grbs , u - grbs , and grfs and to give a generel description of these grb sub - classes . in section [ sec : fireshell ] we present a short summary on the fireshell model . in section [ sec:1052erg ] we discuss the @xmath26 erg lower limit in binary systems leading to bh formation . after describing the observational properties of the above sub - classes , their interpretation within the igc paradigm , the ns - ns merger scenario and the fireshell scenario , we present some prototypes ( see sections [ sec_3_prot_rat ] , [ sec : descr_bdhne ] , [ sec : descr_s - grfs ] , [ sec : descr_sgrbs ] , [ sec : rate5 ] , and [ sec : ds ] , respectively ) . we then proceed in section [ sec : rates ] to estimate their observed occurrence rates and to compare and contrast our results with those outlined in the literature ( see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? we then draw some general conclusions in section [ sec : conclusions ] . a standard flat @xmath27cdm cosmological model with @xmath28 , @xmath29 , and @xmath30 km s@xmath31 mpc@xmath31 is adopted throughout the paper . a summary of acronyms used throughout the paper is shown in table [ acronyms ] . .alphabetic ordered list of the acronyms used in this work . [ cols="<,^",options="header " , ] the prototype of grfs is grb 060614 and has been analyzed by @xcite . from the analysis of its p - grb emission a baryon load @xmath32 and a lorentz factor at the transparency condition @xmath33 have been found . from the fit of the light curve of the prompt emission it has been infereed that this burst occurred in a cbm with density @xmath34@xmath35 @xmath24 , which is typical of galactic halos where ns ns and ns wd mergers occur . analogous results were obtained for the grf 071227 : @xmath36 and @xmath37@xmath38 @xmath24 @xcite . further analyses on other grf examples will be presented elsewhere . the observed grb occurrence rate is defined by the convolution of both ( likely redshift - dependent ) luminosity function , which describes the fraction of bursts with isotropic equivalent luminosities in the interval @xmath39 and @xmath40 , and cosmic grb occurrence rate , which gives the number of sources at different redshifts . the definition of both of these functions is still an open issue and depends on _ a priori _ assumptions and some investigations have been carried out in the literature ( see , e.g. , , for long bursts , @xcite , for short bursts , and @xcite , for both long and short bursts ) . to complicate the matter , also the instrumental sensitivity threshold , the field of view @xmath41 , and the operational time @xmath42 of the various detectors @xmath43 observing grbs introduce additional uncertainties to the problem . in the following we ignore the possible redshift - evolution of the luminosity function . thus , if @xmath44 events are detected by various detectors in a finite logarithmic luminosity bin from @xmath39 to @xmath45 , the total local event rate density of bursts between observed minimum and maximum luminosities , @xmath46 and @xmath47 respectively , is defined as ( see * ? ? ? * ) @xmath48 where @xmath49 and the comoving volume is given by @xmath50^{1/2}}\ , \ ] ] where @xmath51 is the luminosity distance . the dimensionless function @xmath52 describes the grb cosmic redshift - dependent event rate density . in the following we assume no redshift dependency , therefore we set @xmath53 . the maximum redshift @xmath54 in eq . ( [ eq : rate2 ] ) defines the maximum volume inside which an event with luminosity @xmath55 can be detected . this redshift can be computed from the @xmath56 s - bolometric peak luminosity @xmath55 , @xmath57-corrected from the observed detector energy band into the burst cosmological rest - frame energy band @xmath56@xmath58 kev @xcite , and the corresponding @xmath56 s - threshold peak flux @xmath59 , which is the limiting peak flux to allow the burst detection ( see * ? ? ? * for details ) . therefore , @xmath60 can be defined via @xmath61 where we duly account for the @xmath57-correction . within the assumptions that the grb luminosity function does not evolve with redshift and that @xmath53 , we investigate the evolution with the redshift of the grb rates by separating the bursts into several redshift bins . as suggested in @xcite , this can be done in each redshift interval @xmath62 by changing the integration limits of eq . ( [ eq : rate2 ] ) into @xmath63 and min@xmath64 $ ] , where @xmath65 is the maximum redshift for the j@xmath66 redshift bin . finally , from eq . ( [ eq : rate ] ) we derive an event rate @xmath67 in each redshift bin around @xmath68 . in the following we adopt the following fields of view and operational times for various detectors : _ beppo_-sax , @xmath69 sr , @xmath70 y ; batse , @xmath71 sr , @xmath72 y , hete-2 , @xmath73 sr , @xmath74 y ; _ swift_-bat , @xmath75 sr , @xmath76 y ; _ fermi_-gbm , @xmath77 sr , @xmath78 y. we assume no beaming correction in computing the rates of the grb sub - classes . the local rate of s - grfs , obtained from the sample of sources listed in table [ tab : s - grfs ] , is @xmath79 gpc@xmath80 y@xmath31 and it is in agreement with the estimates obtained from the whole short burst population detected by the _ swift_-bat detector ( and , therefore , including also s - grbs and grfs ) and reported in the literature ( @xmath56@xmath18 gpc@xmath80 y@xmath31 , see , e.g. , * ? ? ? * and references therein ) . in particular our local rates with @xmath53 agrees with recent more precise estimates : a ) @xmath81 gpc@xmath80 y@xmath31 for @xmath82 erg / s and for @xmath52 described by a power law merger delay model @xcite ; b ) @xmath83 , @xmath84 , and @xmath85 gpc@xmath80 y@xmath31 for @xmath86 erg / s and @xmath52 described as gaussian , log - normal and power law merger delay models , respectively @xcite . the evolution of the s - grf rate in various redshift bins is shown in fig . [ fig : ratesfam ] ( c ) . this rate decreases as a power law from the local value in the interval @xmath87 to a value of @xmath88gpc@xmath80 y@xmath31 in the interval @xmath89 . also in the case of s - grfs the increasing sampled comoving universe volume and the threshold of the detectors play a fundamental role in the observed decrease of their rate at larger distances . previously we have identified and described four s - grbs in @xcite , e.g. , grb 081024b , grb 090227b , grb 090510 , and grb 140619b . here we present two additional new members of this class : grb 060801 ( at @xmath90 and with @xmath91 , in this work ) and grb 140402a ( at @xmath92 and with @xmath93 , ruffini et al . in preparation ) . from these six s - grbs detected by the _ fermi _ and the _ swift _ satellites , we obtain via eqs . ( [ eq : rate])([eq : rate2 ] ) a local rate @xmath94 gpc@xmath80 y@xmath31 . with only six sources , we could not build the evolution with the redshift of such systems . in @xcite , we have estimated an updated observed rate for the xrfs at @xmath95 based on the method outlined in @xcite and @xcite . in this work , we consider the complete list of xrfs shown in table [ tab : xrfs ] and the method outlined in @xcite . from eq . ( [ eq : rate])([eq : rate2 ] ) the local rate of xrfs is @xmath96gpc@xmath80 y@xmath31 , where the attached errors are determined from the @xmath97 confidence level of the poisson statistic @xcite . within the extent of our different classification criteria and different choices for @xmath52 , our estimate is in agreement with those reported for low - luminous long grbs in @xcite and @xcite , and in particular with the value of @xmath98gpc@xmath80 y@xmath31 , obtained by @xcite with the same method . in the igc scenario the xrf out - states are ns ns binary systems . for a reasonable sets of binary initial conditions , populations synthesis simulations performed by @xcite provide a ns ns formation rate @xmath99@xmath100 gpc@xmath80 y@xmath31 . ns formation rate accounts for other possible channels in the population synthesis models , in addition to the one we considered from the xrfs . it is interesting , nevertheless , that our predicted rate is consistent with that obtained by @xcite obtained . for the same above reason , our rate of xfrs can be also compared with the ns ns merger rate proposed by @xcite . in this historical paper , the ns ns merger rate is derived from the strong assumption that each merger ejects always the same amount of material r - process classified and the heavy r - process material . @xcite obtain a rough estimate of @xmath101@xmath102 gpc@xmath80 y@xmath31 , which is marginally consistent to the upper value of the local xrf rate . the evolution of the xrf rate in various redshift bins is shown in fig . [ fig : ratesfam ] ( a ) . it decreases from a value of @xmath103gpc@xmath80 y@xmath31 in the interval @xmath104 to a value of @xmath105gpc@xmath80 y@xmath31 in the interval @xmath106 . this effect is mainly due to the intrinsic low luminosities of the bulk of the xrf population ( @xmath107@xmath108 erg / s , see , e.g. , @xcite ) and to the threshold of the detectors : at increasing sampled universe comoving volumes , these low luminous xrfs become undetectable , therefore at higher redshifts the total xrf rate decreases . we proceed now in estimating the rate of bdhne from the total sample of @xmath109 sources ( see table [ tab : bdhne ] ) . from eq . ( [ eq : rate])([eq : rate2 ] ) the local rate of bdhne is @xmath110 gpc@xmath80 y@xmath31 . our estimate is in agreement with two recent estimates obtained from long bursts with @xmath111 erg / s and by assuming @xmath112 : a ) the value of @xmath113 gpc@xmath80 y@xmath31 obtained by @xcite , even though limited to the _ swift _ long bursts and including long some bursts with @xmath114 erg , obtained from a grb inferred cosmic rate independent on the star formation rate ; b ) the value of @xmath115 gpc@xmath80 y@xmath31 obtained by @xcite with the same method and including the star formation rate dependence . in the igc scenario the bdhne out - states are ns bh binary systems . following again the work by @xcite , populations synthesis simulations ( which accounts also for alternative scenarios to that of the igc model ) provide a ns bh formation rate of @xmath116@xmath117 gpc@xmath80 y@xmath31 . also in this case , even though a straightforward comparison is not possible , the bdhne rate is consistent with the ns bh formation rate obtained by @xcite . the evolution of the bdhn rate in various redshift bins is shown in fig . [ fig : ratesfam ] ( b ) . it slightly decreases from the local value in the interval @xmath87 to a value of @xmath118gpc@xmath80 y@xmath31 in the interval @xmath119 . as stated for the case of xrfs , this effect occurs because for increasing sampled universe comoving volumes , only the most luminous bdhne are detectable , even though in a less marked way than the case of the xrfs . as pointed out in section [ sec : rate5 ] , nearly 100@xmath22 of the bdhne leads to bound ns - bh binaries , which are the progenitor systems of u - grbs @xcite . if we include the possibility of other channels of formation for these ns - bh binaries , we can safely assume the bdhn local rate as a lower limit for these u - grbs , e.g. , @xmath110 gpc@xmath80 y@xmath31 . from this consideration , it appears that the u - grbs have the second higher rate among the short bursts after the s - grfs . we proceed now in estimating the rate of grfs from the total sample of @xmath18 sources ( see table [ tab : dsgrbs ] ) . from eq . ( [ eq : rate])([eq : rate2 ] ) we obtain a local rate @xmath120 gpc@xmath80 y@xmath31 , and represent the first estimate for these kind for bursts originating from ns wd mergers . due to the limited number of sources in our sample , we limited the study of the grf rate evolution in two redshift bins , as shown in fig . [ fig : ratesfam ] ( d ) . the rate starts from a value consistent with the above local rate , in the redshift interval @xmath121 , to a value of @xmath122gpc@xmath80 y@xmath31 , in the interval @xmath123 . also for grfs , the cutoff in the rate at higher redshift occurs because for increasing sampled universe comoving volumes only the most luminous sources are detectable . however this effect , as in the case of s - grfs , is more pronounced due to their intrinsically weaker luminosities , when compared to those of s - grbs . ( a ) ( b ) ( c ) ( d ) the remarkable progress reached in understanding the grb phenomenon has been made possible by the outstanding spectral and temporal information acquired from x - rays , @xmath25-rays and high energy observatories , as well as from optical and radio data obtained by telescopes all over the planet . at the same time this result has been fostered by a novel deeper theoretical understanding in the physics and astrophysics of wd ( see , e.g. , * ? ? ? * ) , ns ( see , e.g. , * ? ? ? * ; * ? ? ? * ) and bh ( see , e.g. , * ? ? ? consequently the understanding of the grb phenomenon has evolved from an elementary paradigm based on a single jetted emission process as postulated in the fireball model ( see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * and reference therein ) to an authentic astrophysical laboratory involving many - body interactions between different astrophysical systems encountering previously unexplored regimes and observational evidence . in the introduction we review the increasing number of grb observations which have led likewise to the theoretical progress in the understanding of the grb phenomena . while the role of ns ns ( or ns - bh ) binaries as `` in - states '' of short grbs has been widely accepted and confirmed by strong observational and theoretical evidence , the identification of the progenitor systems for long grbs followed a more difficult path . initially , theoretical models based on a single progenitor were proposed : a _ collapsar _ @xcite , or a _ magnetar _ ( see , e .g . then , the role of binary progenitor systems composed of two very massive stars for long grbs was recognized by @xcite , where several different scenarios were there envisaged leading to a _ collapsar _ @xcite , as well as a few leading , alternatively , to a variety of binary compact systems . these considerations were addressed by our group in a set of papers assuming that the birth of a sn and the occurrence of a grb were qualitatively and quantitatively different astrophysical events in space and time . this led to the necessity of introducing the igc paradigm . in the igc paradigm the long grb - sn coincidence originates from co@xmath0ns binary progenitors system . this approach differs from alternative descriptions , e.g. , the _ magnetars _ and the _ collapsar _ models , where the two events are coming from a single progenitor star . in section [ sec : fireshell ] we review the fireshell model for grbs ( see , e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) and its general description which can be applied to any source of an optically thick baryon - loaded @xmath7 plasma , i.e. , in the quantum - elctrodynamical process expected in the formation of a bh ( see , e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , as well as in the case of a pair plasma created via @xmath124 mechanism in a ns ns merger @xcite , or in the hyper - accretion disks around bhs @xcite . in section [ sec:1052erg ] , we discuss the role of the @xmath26 erg energy critical value introduced to discriminate between binary systems leading to the formation of a mns ( xrfs , s - grfs and grfs ) , with energy lower than the above critical value , and those leading to the formation of a bh ( bdhne , s - grbs and u - grbs ) , with energy larger than the above critical value . the value of @xmath26 erg is derived by considering the hypercritical accretion process onto a ns leading to an energy release in form of neutrinos and photons , given by the gain of gravitational potential energy of the matter accreted in the ns . this includes the change of binding energy of the ns while accreting both matter and angular momentum @xcite . a typical ns mass of @xmath125 m@xmath20 has been assumed , as observed in galactic ns binaries . a ns critical mass in the range of @xmath126 m@xmath20 up to @xmath127 depending on the equations of state and angular momentum ( see * ? ? ? * ; * ? ? ? * ; * ? ? ? * for details ) has been assumed . in section [ sec_3_prot_rat ] , we describe the properties of xrfs ( see fig . [ fig : ffff ] ) . in these systems the distance between the co@xmath0 and the ns companion is @xmath128 cm . the hypercritical accretion process is not sufficient to push the ns beyond its @xmath129 and an mns is formed ( see , e.g. , * ? ? ? * ; * ? ? ? * ) . in table [ tab : xrfs ] we list the xrfs considered in this work , as well as the spectral , temporal and luminosity analysis of selected prototype , e.g. , grb 060218 . the complete theoretical simulation of this prototype is presented in @xcite . in section [ sec : descr_bdhne ] we consider the bdhne , for which the binary separation between the co@xmath0 and the ns binary companion is @xmath130 cm and the hypercritical accretion process trigger the gravitational collapse of the ns into a bh ( see , e.g. , * ? ? * ; * ? ? ? we show here an updated list of bdhne ( see table [ tab : bdhne ] ) , as well as a diagram summarizing some of the key properties and prototypes ( see fig . [ fig : ffff ] ) , analyzed within the igc paradigm and the fireshell model ( see , e.g. , grb 090618 and grb 130427a ) . in section [ sec : descr_s - grfs ] , we outline the properties of s - grfs listed in table [ tab : s - grfs ] and shown in fig . [ fig : ffff ] . these systems coincide with the short bursts considered in . they originate in ns - ns mergers leading to the formation of a mns and possibly a binary companion , in order to fulfill the conservation of energy and momentum @xcite . in section [ sec : descr_sgrbs ] we present s - grbs originating in ns - ns mergers leading to the formation of a bh ( see fig . [ fig : ffff ] ) . we give , in table [ tab : sgrbs ] , their updated list . we then describe their prototypes , analyzed within the fireshell model ( see , e.g. , grb 090227b and grb 140619b ) , and outline the key role of the p - grb identification for their description , as well as the analysis of the gev emission . in section [ sec : rate5 ] , motivated by the results obtained by @xcite , where it was shown that nearly 100@xmath22 of the ns - bh binaries , namely the out - states of the bdhne , remain bound , we add the description of this not yet observed , but theoretically predicted sub - class of u - grbs , unaccounted for in current standard population synthesis analyses . in section [ sec : ds ] , we review the properties of the grfs listed in table [ tab : dsgrbs ] and shown in fig . [ fig : ffff ] . we recall and describe the results obtained from the sources analyzed within the fireshell model ( see , e.g. , grb 060614 , @xcite and grb 071227 , @xcite ) . the most important result of the present article is the estimate the rates of occurrence of the xrfs , bdhne , s - grfs , s - grbs , u - grbs , and grfs sub - classes . in section [ sec : rates ] , we introduce the procedure outlined in @xcite for estimating the local rates and their evolution with the redshift of the above sub - classes of long and short bursts , assuming no beaming ( note : the recent observation of the absence of gev emission associated to a bdhn may limit this assumption ) . by ignoring possible redshift - evolution of the grb sub - classes luminosity functions and assuming that the grb cosmic event rate density is redshift - independent ( e.g. , @xmath53 ) , the above method duly takes into account observational constraints , i.e. , the detector solid angle coverage of the sky @xmath131 and sensitivities which in turn define a maximum volume of observation depending on the intrinsic luminosity of the sources ( see section [ sec : rates ] and @xcite for details ) . we obtain : * an s - grf local rate of @xmath79 gpc@xmath80 y@xmath31 ( see section [ sec : sgrfs3 ] ) ; * an s - grb local rate of @xmath132 gpc@xmath80 y@xmath31 ( see section [ sec : sgrbs3 ] ) ; * an xrf local rate of @xmath96gpc@xmath80 y@xmath31 ( see section [ sec : xrfs3 ] ) ; * a bdhn local rate of @xmath110 gpc@xmath80 y@xmath31 ( see section [ sec : bdhne3 ] ; for the above reason this rate coincides with that of the u - grbs , see section [ sec : ugrbsrate ] ) ; * a grf local rate of @xmath120 gpc@xmath80 y@xmath31 ( see section [ sec : dsgrb3 ] ) . the local rates of s - grfs , xrfs , and bdhne , are in agreement , within the extent of the different classification criteria , with those reported in the literature . the local rates of s - grbs and grfs are , instead , new ones following from the classification proposed in this work . the evolution with the redshift of the rates of xrfs , bdhne , s - grfs , and grfs is shown in fig . [ fig : ratesfam ] . it is certainly of interest to compare and contrast these results obtained from the direct observations of the sources in our new classification with the results computed from population synthesis models . any possible disagreement will give the opportunity to identify possible missing links in the evolutionary phases within population synthesis analysis . we are now in a position to apply the above rates of s - grfs , s - grbs and u - grbs to assess the detectability and the expected number of gravitational wave detections by ligo from ns - ns and ns - bh binaries ( ruffini et al . , in preparation ) . we are also ready to apply the above bdhn rate to give an estimate of the contribution of grbs to cosmic rays ( ruffini et al . , in preparation ) . before concluding , in support of the classification proposed in this article , we recall that the luminosity light curves of the gev emission is uniquely observed in both bdhne and s - grbs . in both cases it follow a precise power - law behavior with time @xmath133 ( see @xcite , @xcite and fig . [ gevtotal ] ) . an outstanding conclusion of this paper is that in both bdhne and s - grbs , where the presence of the bh is predicted , the turn - on of this gev emission occurs after the p - grb emission and at the beginning of the prompt emission phase ( see figs . [ rad_ind_tot ] ( d ) and [ shortxo ] ( d ) ) . this commonality , in such different systems , as well as their energy requirements ( see tab . [ tab : lat ] and fig . [ gevtotal ] ) are naturally explained if we assume , as indicated in @xcite , that this gev emission originates by accretion processes in the newly - born bh . we have pointed out in @xcite how the total energy of the gev emission can be expressed in term of the gravitational binding energy of matter accretion into kerr bhs ( see ruffini & wheeler 1969 , in problem 2 of @xmath134 104 in @xcite ) . this energetics requirement could not be fulfilled in the case of accretion onto a ns , in view of the much smaller value of the gravitational binding energy when compared to the case of a rotating bh ( see e.g. * ? ? ? * ) . on the general issue of the origin of the jetted gev emission , and not just of its energetics , we refer to the last paragraph of the conclusions of the paper by @xcite . we have added a table . [ tab : lat ] with the values of the gev emission for both the case of s - grbs and bdhne . these energy releases up to @xmath16 erg can be explained by the occurrence of accretion onto a rotating bh with mass in the range of @xmath135@xmath18 m@xmath20 . it is also clear from fig . [ gevtotal ] that s - grbs and bdhne have gev emission sharing a common luminosity pattern and originating , in both cases , from a newly - born kerr bh @xcite . this picture includes also the first scenario of an igc considered in @xcite where an exploding @xmath136 is in a close binary system with an already formed bh companion . in view of the hypercritical accretion process of the sn ejecta onto an already formed bh , these systems have @xmath137 erg and @xmath138 mev . their out - states are a binary composed of a more massive bh and a @xmath139ns . such systems , which we refer to as bh - sne , are expected to be the late evolutionary stages of x - ray binaries such as cyg x-1 or cyg x-3 ( see , e.g. , * ? ? ? * ) . lcccc source & z & @xmath140 & @xmath141 & @xmath142 + & & ( mev ) & ( @xmath26 erg ) & ( @xmath26 erg ) + + 081024b & @xmath143 & @xmath144 & @xmath145 & @xmath146 + 090510 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + 140402a & @xmath151 & @xmath152 & @xmath153 & @xmath154 + 140619b & @xmath155 & @xmath156 & @xmath157 & @xmath158 + + 080916c & @xmath159 & @xmath160 & @xmath161 & @xmath162 + 090902b & @xmath163 & @xmath164 & @xmath165 & @xmath166 + 110731a & @xmath167 & @xmath168 & @xmath169 & @xmath170 + 130427a & @xmath171 & @xmath172 & @xmath173 & @xmath174 + in conclusion we have computed the occurence rate of short and long bursts following a new classification and obtaining figures in good agreement with the ones derived from population synthesis models . essential to the classification have been the following new considerations : * the binary nature of the progenitors and their separation ; * the essential role of the hypercritical accretion process onto a ns member of a close binary system . the possible reaching of @xmath21 by the accretion process and the formation of a bh ; * the activity of the newly - born bhs originating the energetic prominent gev emission , which can be explained in terms of the gravitational energy release by accreting matter onto a kerr bh . this classification is now open to a verification by the addition of new grbs sources and offer new possibilities of theoretical and observational activities including : * the reaching of new observational constraints on the value of the ns critical mass @xmath21 and the minimum mass of a bh , which play a fundamental role in defining the separatrix among the different classes of our classification . * having elucidated the role of the activities of the newly - born bh in explaining the energetics of the gev emission , in order to identify its microphysical process , the study of fundamental issues of general relativistic quantum electrodynamical processes appears to be open to further lines of inquiry ( see , e.g. , * ? ? ? * and references therein ) ; * it is conceivable that the sizable enlargement of the database of grbs and of their spectral and luminosity time varability may open the possibilty of further enlarging the above classification . we thank the editor and the referee for their comments which helped to improve the presentation and the contextualization of our results this work made use of data supplied by the uk swift science data center at the university of leicester . j.a.r . acknowledges the support by the international cooperation program capes - icranet financed by capes - brazilian federal agency for support and evaluation of graduate education within the ministry of education of brazil . m.k . and y. a. are supported by the erasmus mundus joint doctorate program grant n. 20131471 and 2014 - 0707 , respectively , from eacea of the european commission . m.m . acknowledges the partial support of the project n 3101/gf4 ipc-11 , and the target program f.0679 of the ministry of education and science of the republic of kazakhstan . in fig . [ fig : fluence ] we compare and contrast the sources in tab . [ tab : lat ] , all exhibiting a gev emission , with 2 s - grfs and 1 xrf which , as theoretically expected within the fireshell model , do not exhibit any gev emission . all these sources were in the optimal position ( @xmath175 from the lat boresight ) for the detection of the gev emission . in the left panel we plot the values of @xmath141 and of the isotropic energy in the _ fermi_-lat energy band or the corresponding upper limits if not observed . these upper limits were obtained by using the unbinned likelihood analysis which was performed assuming an integration time of @xmath176 s after the flash trigger , a radius of the source region of @xmath177 and a zenith angle cut of @xmath178 . this plot observationally supports the theoretical expectation , made in @xcite and quoted in section [ sec : descr_s - grfs1b ] above , that s - grfs have , if any , gev fluxes necessarily @xmath179@xmath180 times weaker than those of s - grbs , although their @xmath141 is only a factor @xmath181 smaller . motivated by a request of the referee , we also plotted in the right panel , the values of the fluence observed by _ fermi_-gbm and by _ fermi_-lat or the corresponding upper limits if not observed ( computed as above ) . from both plots it is clear that the upper limits to the gev emission of s - grfs and xrfs are much lower than what one may expect from the extrapolation to lower energies of the one observed in bdhne and s - grbs . this is a further clear observational support to the absence , theoretically implied by the fireshell model , of any gev emission associated to s - grfs and xrfs ( see sections [ sec : xrfs1b ] and [ sec : descr_s - grfs1b ] above ) . , h. , & ruffini , r. , eds . 1975 , astrophysics and space science library , vol . 48 , neutron stars , black holes and binary x - ray sources ; proceedings of the annual meeting , san francisco , calif . , february 28 , 1974 , j. 2003 , in american institute of physics conference series , vol . 662 , gamma - ray burst and afterglow astronomy 2001 : a workshop celebrating the first year of the hete mission , ed . g. r. ricker & r. k. vanderspek , 229236 , r. , bernardini , m. g. , bianco , c. l. , et al . 2005 , in american institute of physics conference series , vol . 782 , xith brazilian school of cosmology and gravitation , ed . m. novello & s. e. perez bergliaffa , 42 , i. b. , klebesadel , r. w. , & evans , w. d. 1975 , in annals of the new york academy of sciences , vol . 262 , seventh texas symposium on relativistic astrophysics , ed . p. g. bergman , e. j. fenyves , & l. motz , 145158	there is mounting evidence for the binary nature of the progenitors of gamma - ray bursts ( grbs ) . for a long grb , the induced gravitational collapse ( igc ) paradigm proposes as progenitor , or in - state " , a tight binary system composed of a carbon - oxygen core ( co@xmath0 ) undergoing a supernova ( sn ) explosion which triggers hypercritical accretion onto a neutron star ( ns ) companion . for a short grb , a ns - ns merger is traditionally adopted as the progenitor . we divide long and short grbs into two sub - classes , depending on whether or not a black hole ( bh ) is formed in the merger or in the hypercritical accretion process exceeding the ns critical mass . for long bursts , when no bh is formed we have the sub - class of x - ray flashes ( xrfs ) , with isotropic energy @xmath1 erg and rest - frame spectral peak energy @xmath2 kev . when a bh is formed we have the sub - class of binary - driven hypernovae ( bdhne ) , with @xmath3 erg and @xmath4 kev . in analogy , short bursts are similarly divided into two sub - classes . when no bh is formed , short gamma - ray flashes ( s - grfs ) occur , with @xmath1 erg and @xmath5 mev . when a bh is formed , the authentic short grbs ( s - grbs ) occur , with @xmath3 erg and @xmath6 mev . we give examples and observational signatures of these four sub - classes and their rate of occurrence . from their respective rates it is possible that `` in - states '' of s - grfs and s - grbs originate from the `` out - states '' of xrfs . we indicate two additional progenitor systems : white dwarf - ns and bh - ns . these systems have hybrid features between long and short bursts . in the case of s - grbs and bdhne evidence is given of the coincidence of the onset of the high energy gev emission with the birth of a kerr bh .
a _ graph _ @xmath0 consists of a _ _ vertex set _ _ @xmath1 and an _ _ edge set _ _ @xmath2 , where each edge corresponds to a pair @xmath3 of vertices . a graph in which each edge has a _ weight _ associated with it is called a _ network_. a graph @xmath4 is a _ subgraph _ of @xmath0 if @xmath5 and @xmath6 . a graph @xmath7 is a _ homeomorph _ of @xmath8 if @xmath7 can be obtained by subdividing edges of @xmath8 with new vertices . we say that a graph @xmath9 is a _ square of a graph _ @xmath0 if @xmath10 . a two - terminal directed acyclic graph ( _ st - dag _ ) has only one source and only one sink . we consider a _ labeled graph _ which has labels attached to its edges . each path between the source and the sink ( a _ sequential path _ ) in an st - dag can be presented by a product of all edge labels of the path . we define the sum of edge label products corresponding to all possible sequential paths of an st - dag @xmath7 as the _ canonical expression _ of @xmath7 . an algebraic expression is called an _ st - dag expression _ ( a _ factoring of an st - dag _ in @xcite ) if it is algebraically equivalent to the canonical expression of an st - dag . an st - dag expression consists of literals ( edge labels ) , and the operators @xmath11 ( disjoint union ) and @xmath12 ( concatenation , also denoted by juxtaposition ) . an expression of an st - dag @xmath7 will be hereafter denoted by @xmath13 . we define the total number of literals in an algebraic expression as the _ complexity of the algebraic expression_. an equivalent expression with the minimum complexity is called an _ optimal representation of the algebraic expression_. a _ series - parallel _ _ graph _ is defined recursively so that a single edge is a series - parallel graph and a graph obtained by a parallel or a series composition of series - parallel graphs is series - parallel . as shown in @xcite and @xcite , a series - parallel graph expression has a representation in which each literal appears only once . this representation is an optimal representation of the series - parallel graph expression . for example , the canonical expression of the series - parallel graph presented in figure [ fig1 ] is @xmath14 . since it is a series - parallel graph , the expression can be reduced to @xmath15 , where each literal appears once . ( 5,3.5)(-5,0)(1,3)(3,0)4 ( 1,3)(1,0)3 ( 2.5,3.3)(0,0)@xmath16 ( 4,3)(5.5,5)(7,3 ) ( 7.085,3)(3,-2)0 ( 5.5,4.3)(0,0)@xmath17 ( 4,3)(5.5,1)(7,3 ) ( 7.085,3)(3,2)0 ( 5.5,2.3)(0,0)@xmath18 ( 7,3)(8.5,5)(10,3 ) ( 10.085,3)(3,-2)0 ( 8.5,4.3)(0,0)@xmath19 ( 7,3)(8.5,1)(10,3 ) ( 10.085,3)(3,2)0 ( 8.5,2.3)(0,0)@xmath20 ( 1,3)(4,-2)(7,3 ) ( 7.085,3)(4,3)0 ( 4,0.8)(0,0)@xmath21 a _ fibonacci graph _ @xcite has vertices @xmath22 and edges @xmath23@xmath24 . as shown in @xcite , an st - dag is series - parallel if and only if it does not contain a subgraph which is a homeomorph of the _ forbidden subgraph _ positioned between vertices @xmath25 and @xmath26 of the fibonacci graph illustrated in figure [ fig2 ] . thus , a fibonacci graph gives a generic example of non - series - parallel graphs . ( 5,2)(-0.9,-0.5)(0,0)(1.5,0)9 ( 0,-0.3)(0,0)1 ( 1.5,-0.3)(0,0)2 ( 3,-0.3)(0,0)3 ( 4.5,-0.3)(0,0)4 ( 7.5,-0.3)(0,0)n-3 ( 9,-0.3)(0,0)n-2 ( 10.5,-0.3)(0,0)n-1 ( 12,-0.3)(0,0)n ( 0,0)(1.5,0)8(1,0)1.5 ( 0.75,0.2)(0,0)@xmath27 ( 2.25,0.2)(0,0)@xmath28 ( 3.75,0.2)(0,0)@xmath29 ( 8.25,0.2)(0,0)@xmath30 ( 9.75,0.2)(0,0)@xmath31 ( 11.25,0.2)(0,0)@xmath32 ( 0,0)(1.5,2)(3,0 ) ( 1.5,0)(3,2)(4.5,0 ) ( 3,0)(4.5,2)(6,0 ) ( 4.5,0)(6,2)(7.5,0 ) ( 6,0)(7.5,2)(9,0 ) ( 7.5,0)(9,2)(10.5,0 ) ( 9,0)(10.5,2)(12,0 ) ( 3.085,0)(1.5,0)7(3,-2)0 ( 1.5,1.3)(0,0)@xmath33 ( 3,1.3)(0,0)@xmath34 ( 9,1.3)(0,0)@xmath35 ( 10.5,1.3)(0,0)@xmath36 ( 5.55,-0.2)(0.2,0)6 mutual relations between graphs and expressions are discussed in a number of works . specifically , @xcite , @xcite , and @xcite consider the correspondence between series - parallel graphs and read - once functions . a boolean function is defined as _ read - once _ if it may be computed by some formula in which no variable occurs more than once ( _ read - once formula _ ) . on the other hand , a series - parallel graph expression can be reduced to the representation in which each literal appears only once . hence , such a representation of a series - parallel graph expression can be considered as a read - once formula ( boolean operations are replaced by arithmetic ones ) . problems related to computations on graphs have applications in different areas . specifically , many network problems , which are either intractable or have complicated solutions in the general case are solvable for series - parallel graphs . for example , some efficient algorithms for flow problems on series - parallel networks are presented in @xcite , @xcite , @xcite , @xcite . papers @xcite , @xcite , @xcite , @xcite consider sequencing and scheduling in relation to precedence series - parallel constraints . linear algorithms for reliability problems on series - parallel networks are presented in @xcite , @xcite . an expression of a homeomorph of the forbidden subgraph belonging to any non - series - parallel st - dag has no representation in which each literal appears once . for example , consider the subgraph positioned between vertices @xmath25 and @xmath26 of the fibonacci graph shown in figure [ fig2 ] . possible optimal representations of its expression are @xmath37 or @xmath38 . for this reason , an expression of a non - series - parallel st - dag can not be represented as a read - once formula . however , for arbitrary functions , which are not read - once , generating the optimum factored form is np - complete @xcite . the problem of factoring boolean functions into shorter , more compact formulae is one of the basic operations in algorithmic logic synthesis . in logic synthesis , one standard measure of the complexity of a logic circuit is the number of literals . computation time also depends on the number of literals . some algorithms developed in order to obtain good factored forms are described in @xcite , @xcite . a symbolic approach to scheduling of a robotic line is considered in @xcite . the method uses the max - algebra tools and allows the shortest - path problem to be interpreted as the computation of the st - dag expression . the complexity of this problem is determined by the complexity of the st - dag expression . for a robotic line represented by a fibonacci graph , the proposed algorithm generates the processing sequence in polynomial time . a method for automated composition of algebraic expressions in complex business process modeling based on acyclic directed graph reductions is introduced in @xcite . the method transforms business step dependencies described by users into digraphs and finally generates algebraic expressions . if a graph is not series - parallel , the algorithm checks potential structural conflicts whose presence complicates certain aspects , such as execution control and system scalability . in this case , the expression generation may require exponential time . a lot of problems have polynomial algorithms on _ planar graphs _ , i.e. , the graphs that can be drawn in the plane without any edges crossing . for example , some polynomial - time algorithms for the exact computation of network reliability applied to planar graphs are surveyed in @xcite . in @xcite a polynomial algorithm for the connectivity augmentation problem is presented . this problem has applications in upgrading telecommunication networks to be invulnerable to link or node failures . there are also many practical applications with a graph structure in which crossing edges are a nuisance , including design problems for circuits , subways , utility lines @xcite . in our work we consider expressions with a minimum ( or , at least , a polynomial ) complexity as a key to generating efficient algorithms on distributed systems . in @xcite we presented an algorithm , which generates the expression of @xmath39 complexity for an @xmath40-vertex fibonacci graph . in this paper we investigate a non - series - parallel st - dag called a _ square rhomboid _ ( figure [ rhom_fig12 ] ) . this graph looks like a planar approximation of the square of a _ rhomboid _ , which is a series composition of _ rhomb _ graphs . a square rhomboid consists of the same vertices as the corresponding rhomboid . however , edges labeled by letters @xmath16 , @xmath17 , and @xmath18 ( see figure [ rhom_fig12 ] ) are absent in a rhomboid . geometrically , a square rhomboid can be considered to be a gluing of two fibonacci graphs , i.e. , it is the next harder one in a sequence of increasingly non - series - parallel graphs . ( 10,4)(-0.2,-1.7)(0,0)(3,0)7 ( 1.5,1.5)(3,0)6 ( 1.5,-1.5)(3,0)6 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 15,0.5)(0,0)n-1 ( 18,0.35)(0,0)n ( 1.5,1.8)(0,0)1 ( 4.5,1.8)(0,0)2 ( 13.5,1.8)(0,0)n-2 ( 16.5,1.8)(0,0)n-1 ( 1.5,-1.8)(0,0)1 ( 4.5,-1.8)(0,0)2 ( 13.5,-1.8)(0,0)n-2 ( 16.5,-1.8)(0,0)n-1 ( 0,0)(3,0)6(1,1)1.5 ( 0,0)(3,0)6(1,-1)1.5 ( 1.5,1.5)(3,0)6(1,-1)1.5 ( 1.5,-1.5)(3,0)6(1,1)1.5 ( 0,0)(3,0)6(1,0)3 ( 1.5,1.5)(3,0)5(1,0)3 ( 1.5,-1.5)(3,0)5(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 16.5,0.3)(0,0)@xmath41 ( 3,1.75)(0,0)@xmath42 ( 15,1.75)(0,0)@xmath43 ( 3,-1.75)(0,0)@xmath27 ( 15,-1.75)(0,0)@xmath31 ( 0.55,0.95)(0,0)@xmath44 ( 15.4,0.95)(0,0)@xmath45 ( 2.4,0.95)(0,0)@xmath46 ( 17.56,0.95)(0,0)@xmath47 ( 0.6,-0.95)(0,0)@xmath48 ( 15.4,-0.95)(0,0)@xmath49 ( 2.4,-0.95)(0,0)@xmath50 ( 17.56,-0.95)(0,0)@xmath51 ( 4,0.2)(0.2,0)6 ( 13,0.2)(0.2,0)6 ( 5.5,1.7)(0.2,0)6 ( 11.5,1.7)(0.2,0)6 ( 5.5,-1.7)(0.2,0)6 ( 11.5,-1.7)(0.2,0)6 the set of vertices of an @xmath52-vertex square rhomboid consists of @xmath53 _ middle _ ( _ basic _ ) , @xmath54 _ upper _ , and _ _ _ _ @xmath54 _ lower _ vertices . upper and lower vertices numbered @xmath55 will be denoted in formulae by @xmath56 and @xmath57 , respectively . the square rhomboid ( @xmath58 for brevity ) including @xmath40 basic vertices will be denoted by @xmath59 and will be called an @xmath58 of size @xmath40 . our intention in this paper is to simplify the expressions of square rhomboids and eventually find their optimal representations . with that end in view , we present an algorithm based on a _ decomposition method_. the method is based on revealing subgraphs in the initial graph . the resulting expression is produced by a special composition of subexpressions describing these subgraphs . ( 10,4)(-1.1,-2.5)(0,0)(3,0)7 ( 1.5,1.5)(3,0)6 ( 1.5,-1.5)(3,0)6 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 6,0.35)(0,0)3 ( 9,0.35)(0,0)4 ( 12,0.35)(0,0)5 ( 15,0.35)(0,0)6 ( 18,0.35)(0,0)7 ( 1.5,1.8)(0,0)1 ( 4.5,1.8)(0,0)2 ( 7.5,1.8)(0,0)3 ( 10.5,1.8)(0,0)4 ( 13.5,1.8)(0,0)5 ( 16.5,1.8)(0,0)6 ( 1.5,-1.8)(0,0)1 ( 4.5,-1.8)(0,0)2 ( 7.5,-1.8)(0,0)3 ( 10.5,-1.8)(0,0)4 ( 13.5,-1.8)(0,0)5 ( 16.5,-1.8)(0,0)6 ( 0,0)(3,0)6(1,1)1.5 ( 0,0)(3,0)6(1,-1)1.5 ( 1.5,1.5)(3,0)6(1,-1)1.5 ( 1.5,-1.5)(3,0)6(1,1)1.5 ( 0,0)(3,0)6(1,0)3 ( 1.5,1.5)(3,0)5(1,0)3 ( 1.5,-1.5)(3,0)5(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 4.5,0.3)(0,0)@xmath34 ( 7.5,0.3)(0,0)@xmath60 ( 10.5,0.3)(0,0)@xmath61 ( 13.5,0.3)(0,0)@xmath62 ( 16.5,0.3)(0,0)@xmath63 ( 3,1.75)(0,0)@xmath42 ( 6,1.75)(0,0)@xmath64 ( 8.7,1.75)(0,0)@xmath65 ( 12,1.75)(0,0)@xmath66 ( 15,1.75)(0,0)@xmath67 ( 3,-1.75)(0,0)@xmath27 ( 6,-1.75)(0,0)@xmath28 ( 8.7,-1.75)(0,0)@xmath29 ( 12,-1.75)(0,0)@xmath68 ( 15,-1.75)(0,0)@xmath69 ( 0.55,0.95)(0,0)@xmath44 ( 3.55,0.95)(0,0)@xmath70 ( 6.55,0.95)(0,0)@xmath71 ( 9.55,0.95)(0,0)@xmath72 ( 12.55,0.95)(0,0)@xmath73 ( 15.5,0.95)(0,0)@xmath74 ( 2.4,0.95)(0,0)@xmath46 ( 5.4,0.95)(0,0)@xmath75 ( 8.4,0.95)(0,0)@xmath76 ( 11.4,0.95)(0,0)@xmath77 ( 14.45,0.95)(0,0)@xmath78 ( 17.45,0.95)(0,0)@xmath79 ( 0.6,-0.95)(0,0)@xmath48 ( 3.6,-0.95)(0,0)@xmath80 ( 6.6,-0.95)(0,0)@xmath81 ( 9.6,-0.95)(0,0)@xmath82 ( 12.6,-0.95)(0,0)@xmath83 ( 15.6,-0.95)(0,0)@xmath84 ( 2.4,-0.95)(0,0)@xmath50 ( 5.4,-0.95)(0,0)@xmath85 ( 8.4,-0.95)(0,0)@xmath86 ( 11.4,-0.95)(0,0)@xmath87 ( 14.45,-0.95)(0,0)@xmath88 ( 17.45,-0.95)(0,0)@xmath89 ( 9,2.5)(9,0.55 ) ( 9,0.1)(9,-2.5 ) for a non - trivial @xmath58 subgraph with a source @xmath90 and a sink @xmath91 we choose any _ decomposition vertex _ @xmath92 ( @xmath93 ) located in the basic group of a subgraph . we conditionally split each @xmath58 through its decomposition vertex ( see the example in figure [ rhom_fig3 ] ) . two kinds of subgraphs are revealed in the graph in the course of decomposition . the first of them is an @xmath58 with a fewer number of vertices than the initial @xmath58 . the second one is an @xmath58 supplemented by two additional edges at one of four sides . possible varieties of this st - dag ( we call it a _ _ single - leaf square rhomboid _ _ and denote by @xmath94 ) which are subgraphs revealed from an @xmath58 in figure [ rhom_fig3 ] , are illustrated in figure [ rhom_fig4](a , b , c , d ) . let @xmath95 ( an @xmath94 of size @xmath40 ) denote an @xmath94 including @xmath40 basic vertices . we denote by @xmath96 a subexpression related to an @xmath58 subgraph with a source @xmath90 and a sink @xmath91 . we denote by @xmath97 , @xmath98 , @xmath99 , @xmath100 subexpressions related to @xmath94 subgraphs of figure [ rhom_fig4](a , b , c , d , respectively ) with a source @xmath90 and a sink @xmath101 , a source @xmath102 and a sink @xmath91 , a source @xmath90 and a sink @xmath103 , and a source @xmath104 and a sink @xmath91 . ( 10,3)(-1.3,-1)(-1,0)(0,0)(a ) ( 0,0)(3,0)3 ( 1.5,1.5)(3,0)3 ( 1.5,-1.5)(3,0)2 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 6,0.35)(0,0)3 ( 1.5,1.8)(0,0)1 ( 4.3,1.8)(0,0)2 ( 7.5,1.8)(0,0)3 ( 1.5,-1.8)(0,0)1 ( 4.3,-1.8)(0,0)2 ( 0,0)(3,0)3(1,1)1.5 ( 0,0)(3,0)2(1,-1)1.5 ( 1.5,1.5)(3,0)2(1,-1)1.5 ( 1.5,-1.5)(3,0)2(1,1)1.5 ( 0,0)(3,0)2(1,0)3 ( 1.5,1.5)(3,0)2(1,0)3 ( 1.5,-1.5)(3,0)1(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 4.2,0.3)(0,0)@xmath34 ( 3,1.75)(0,0)@xmath42 ( 6,1.75)(0,0)@xmath64 ( 3,-1.75)(0,0)@xmath27 ( 0.55,0.95)(0,0)@xmath44 ( 3.55,0.95)(0,0)@xmath70 ( 6.55,0.95)(0,0)@xmath71 ( 2.4,0.95)(0,0)@xmath46 ( 5.4,0.95)(0,0)@xmath75 ( 0.6,-0.95)(0,0)@xmath48 ( 3.6,-0.95)(0,0)@xmath80 ( 2.4,-0.95)(0,0)@xmath50 ( 5.4,-0.95)(0,0)@xmath85 ( 10,4.5)(-1.3,-0.85)(-1,0)(0,0)(c ) ( 0,0)(3,0)3 ( 1.5,1.5)(3,0)2 ( 1.5,-1.5)(3,0)3 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 6,0.35)(0,0)3 ( 1.5,1.8)(0,0)1 ( 4.5,1.8)(0,0)2 ( 1.5,-1.8)(0,0)1 ( 4.5,-1.8)(0,0)2 ( 7.5,-1.8)(0,0)3 ( 0,0)(3,0)2(1,1)1.5 ( 0,0)(3,0)3(1,-1)1.5 ( 1.5,1.5)(3,0)2(1,-1)1.5 ( 1.5,-1.5)(3,0)2(1,1)1.5 ( 0,0)(3,0)2(1,0)3 ( 1.5,1.5)(3,0)1(1,0)3 ( 1.5,-1.5)(3,0)2(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 4.5,0.3)(0,0)@xmath34 ( 3,1.75)(0,0)@xmath42 ( 3,-1.75)(0,0)@xmath27 ( 6,-1.75)(0,0)@xmath28 ( 0.55,0.95)(0,0)@xmath44 ( 3.55,0.95)(0,0)@xmath70 ( 2.4,0.95)(0,0)@xmath46 ( 5.4,0.95)(0,0)@xmath75 ( 0.6,-0.95)(0,0)@xmath48 ( 3.6,-0.95)(0,0)@xmath80 ( 6.6,-0.95)(0,0)@xmath81 ( 2.4,-0.95)(0,0)@xmath50 ( 5.4,-0.95)(0,0)@xmath85 ( 10,0)(-1.2,-1.5)(10.5,0)(0,0)(d ) ( 12,0)(3,0)3 ( 13.5,1.5)(3,0)2 ( 10.5,-1.5)(3,0)3 ( 12,0.35)(0,0)5 ( 15,0.35)(0,0)6 ( 18,0.35)(0,0)7 ( 13.5,1.8)(0,0)5 ( 16.5,1.8)(0,0)6 ( 10.3,-1.8)(0,0)4 ( 13.5,-1.8)(0,0)5 ( 16.5,-1.8)(0,0)6 ( 12,0)(3,0)2(1,1)1.5 ( 12,0)(3,0)2(1,-1)1.5 ( 13.5,1.5)(3,0)2(1,-1)1.5 ( 10.5,-1.5)(3,0)3(1,1)1.5 ( 12,0)(3,0)2(1,0)3 ( 13.5,1.5)(3,0)1(1,0)3 ( 10.5,-1.5)(3,0)2(1,0)3 ( 13.5,0.3)(0,0)@xmath62 ( 16.5,0.3)(0,0)@xmath63 ( 15,1.75)(0,0)@xmath67 ( 12,-1.75)(0,0)@xmath68 ( 15,-1.75)(0,0)@xmath69 ( 12.55,0.95)(0,0)@xmath73 ( 15.5,0.95)(0,0)@xmath74 ( 14.45,0.95)(0,0)@xmath78 ( 17.45,0.95)(0,0)@xmath79 ( 12.6,-0.95)(0,0)@xmath83 ( 15.6,-0.95)(0,0)@xmath84 ( 11.45,-0.95)(0,0)@xmath87 ( 14.45,-0.95)(0,0)@xmath88 ( 17.45,-0.95)(0,0)@xmath89 ( 10,0)(-1.2,-6.85)(10.5,0)(0,0)(b ) ( 12,0)(3,0)3 ( 10.5,1.5)(3,0)3 ( 13.5,-1.5)(3,0)2 ( 12,0.35)(0,0)5 ( 15,0.35)(0,0)6 ( 18,0.35)(0,0)7 ( 10.3,1.8)(0,0)4 ( 13.5,1.8)(0,0)5 ( 16.5,1.8)(0,0)6 ( 13.5,-1.8)(0,0)5 ( 16.5,-1.8)(0,0)6 ( 12,0)(3,0)2(1,1)1.5 ( 12,0)(3,0)2(1,-1)1.5 ( 10.5,1.5)(3,0)3(1,-1)1.5 ( 13.5,-1.5)(3,0)2(1,1)1.5 ( 12,0)(3,0)2(1,0)3 ( 10.5,1.5)(3,0)2(1,0)3 ( 13.5,-1.5)(3,0)1(1,0)3 ( 13.5,0.3)(0,0)@xmath62 ( 16.5,0.3)(0,0)@xmath63 ( 12,1.75)(0,0)@xmath66 ( 15,1.75)(0,0)@xmath67 ( 15,-1.75)(0,0)@xmath69 ( 12.55,0.95)(0,0)@xmath73 ( 15.5,0.95)(0,0)@xmath74 ( 11.4,0.95)(0,0)@xmath77 ( 14.45,0.95)(0,0)@xmath78 ( 17.45,0.95)(0,0)@xmath79 ( 12.6,-0.95)(0,0)@xmath83 ( 15.6,-0.95)(0,0)@xmath84 ( 14.45,-0.95)(0,0)@xmath88 ( 17.45,-0.95)(0,0)@xmath89 any path from vertex @xmath25 to vertex @xmath105 in fig . [ rhom_fig3 ] passes through decomposition vertex @xmath26 or through edge @xmath65 or through edge @xmath29 . therefore , @xmath96 is generated by the following recursive procedure ( _ decomposition procedure _ ) : 1 . [ mr1]@xmath106 @xmath107 2 . [ mr2]@xmath106 @xmath108 3 . [ mr3]@xmath106 @xmath109 4 . [ mr4]@xmath110 + @xmath111 the expression related to a one - vertex @xmath58 is defined formally as @xmath25 ( line [ mr1 ] ) . line [ mr2 ] describes the expression of @xmath112 which is a trivial subgraph . the procedure @xmath113 in line [ mr3 ] chooses an arbitrary number @xmath92 on the interval @xmath114 so that @xmath93 . this number is a number of a decomposition vertex in a basic group of the @xmath58 subgraph positioned between vertices @xmath90 and @xmath91 . a current subgraph is decomposed into six new subgraphs in line [ mr4 ] . subgraphs described by subexpressions @xmath115 and @xmath116 include all paths from vertex @xmath90 to vertex @xmath91 passing through vertex @xmath92 . subgraphs described by subexpressions @xmath117 and @xmath118 include all paths from vertex @xmath90 to vertex @xmath91 passing via edge @xmath119 . subgraphs described by subexpressions @xmath120 and @xmath121 include all paths from vertex @xmath90 to vertex @xmath91 passing via edge @xmath122 . @xmath123 is the expression of the initial @xmath58 of size @xmath40 . hence , the decomposition procedure is initially invoked by substituting @xmath25 and @xmath40 instead of @xmath90 and @xmath91 , respectively . we now consider the algorithm that generates the algebraic expression for a square rhomboid using 1-vdm as its base . the algorithm is based on decomposition of the initial graph and all kinds of subgraphs to new subgraphs . we conjecture that the representation with the minimum complexity of @xmath124 derived by 1-vdm is generated when the decomposition vertex is a middle vertex in the basic group of the @xmath58 . that is , the number @xmath92 of the decomposition vertex for a current @xmath58 subgraph which is positioned between vertices @xmath90 and @xmath91 is chosen as @xmath125 ( @xmath126 or @xmath127 ) . in other words , @xmath92 is equal to @xmath125 for odd @xmath128 and @xmath92 equals @xmath129 or @xmath130 for even @xmath128 . an @xmath94 subgraph is decomposed through a decomposition vertex selected in its basic group into six new subgraphs in the same way as an @xmath58 ( see the examples in figure [ rhom_fig13 ] ) . the decomposition vertex is chosen so that the location of the split is in the middle of the subgraph . ( 10,3.3)(-0.5,-1.9)(0,0)(3,0)3 ( 1.5,1.5)(3,0)3 ( 1.5,-1.5)(3,0)2 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 6,0.35)(0,0)3 ( 1.5,1.8)(0,0)1 ( 4.5,1.8)(0,0)2 ( 7.5,1.8)(0,0)3 ( 1.5,-1.8)(0,0)1 ( 4.5,-1.8)(0,0)2 ( 0,0)(3,0)3(1,1)1.5 ( 0,0)(3,0)2(1,-1)1.5 ( 1.5,1.5)(3,0)2(1,-1)1.5 ( 1.5,-1.5)(3,0)2(1,1)1.5 ( 0,0)(3,0)2(1,0)3 ( 1.5,1.5)(3,0)2(1,0)3 ( 1.5,-1.5)(3,0)1(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 4.5,0.3)(0,0)@xmath34 ( 2.7,1.75)(0,0)@xmath42 ( 6,1.75)(0,0)@xmath64 ( 2.7,-1.75)(0,0)@xmath27 ( 0.55,0.95)(0,0)@xmath44 ( 3.55,0.95)(0,0)@xmath70 ( 6.55,0.95)(0,0)@xmath71 ( 2.4,0.95)(0,0)@xmath46 ( 5.4,0.95)(0,0)@xmath75 ( 0.6,-0.95)(0,0)@xmath48 ( 3.6,-0.95)(0,0)@xmath80 ( 2.4,-0.95)(0,0)@xmath50 ( 5.4,-0.95)(0,0)@xmath85 ( 3,2.5)(3,0.55 ) ( 3,0.1)(3,-2.5 ) ( 10,0)(-8.7,-2.5)(0,0)(3,0)4 ( 1.5,1.5)(3,0)4 ( 1.5,-1.5)(3,0)3 ( 0,0.35)(0,0)1 ( 3.02,0.35)(0,0)2 ( 6,0.35)(0,0)3 ( 9,0.35)(0,0)4 ( 1.5,1.8)(0,0)1 ( 4.5,1.8)(0,0)2 ( 7.5,1.8)(0,0)3 ( 10.5,1.8)(0,0)4 ( 1.5,-1.8)(0,0)1 ( 4.5,-1.8)(0,0)2 ( 7.5,-1.8)(0,0)3 ( 0,0)(3,0)4(1,1)1.5 ( 0,0)(3,0)3(1,-1)1.5 ( 1.5,1.5)(3,0)3(1,-1)1.5 ( 1.5,-1.5)(3,0)3(1,1)1.5 ( 0,0)(3,0)3(1,0)3 ( 1.5,1.5)(3,0)3(1,0)3 ( 1.5,-1.5)(3,0)2(1,0)3 ( 1.5,0.3)(0,0)@xmath33 ( 4.5,0.3)(0,0)@xmath34 ( 7.5,0.3)(0,0)@xmath60 ( 3,1.75)(0,0)@xmath42 ( 5.7,1.75)(0,0)@xmath64 ( 9,1.75)(0,0)@xmath65 ( 3,-1.75)(0,0)@xmath27 ( 5.7,-1.75)(0,0)@xmath28 ( 0.55,0.95)(0,0)@xmath44 ( 3.55,0.95)(0,0)@xmath70 ( 6.55,0.95)(0,0)@xmath71 ( 9.55,0.95)(0,0)@xmath72 ( 2.4,0.95)(0,0)@xmath46 ( 5.4,0.95)(0,0)@xmath75 ( 8.4,0.95)(0,0)@xmath76 ( 0.6,-0.95)(0,0)@xmath48 ( 3.6,-0.95)(0,0)@xmath80 ( 6.6,-0.95)(0,0)@xmath81 ( 2.4,-0.95)(0,0)@xmath50 ( 5.4,-0.95)(0,0)@xmath85 ( 8.4,-0.95)(0,0)@xmath86 ( 6,2.5)(6,0.55 ) ( 6,0.1)(6,-2.5 ) three kinds of subgraphs are revealed in an @xmath94 in the course of decomposition . the first and the second of them are an @xmath58 and an @xmath94 , respectively . the third one is an @xmath58 supplemented by two additional pairs of edges ( one pair is on the left and another one is on the right ) . possible varieties of this st - dag ( we call it a _ _ dipterous square rhomboid _ _ and denote it by @xmath131 ) are illustrated in figure [ rhom_fig6](a , b , c , d ) . we define subgraphs illustrated in figure [ rhom_fig6](a , c ) as _ trapezoidal _ @xmath131 _ graphs _ and subgraphs illustrated in figure [ rhom_fig6](b , d ) as _ parallelogram _ @xmath131 _ graphs_. let @xmath132 ( an @xmath131 of size @xmath40 ) denote an @xmath131 including @xmath40 basic vertices . we denote by @xmath133 , @xmath134 , @xmath135 , @xmath136 subexpressions related to subgraphs of figure [ rhom_fig6](a , b , c , d , respectively ) with a source @xmath102 and a sink @xmath101 , a source @xmath137 and a sink @xmath101 , a source @xmath137 and a sink @xmath103 , and a source @xmath102 and a sink @xmath138 . ( 10,2)(5,0.4)(5.3,0)(0,0)(a ) ( 6,0)(3,0)2 ( 4.5,1.5)(3,0)3 ( 7.5,-1.5)(3,0)1 ( 6,0.35)(0,0)6 ( 9,0.35)(0,0)7 ( 4.5,1.8)(0,0)5 ( 7.5,1.8)(0,0)6 ( 10.5,1.8)(0,0)7 ( 7.5,-1.8)(0,0)6 ( 6,0)(3,0)2(1,1)1.5 ( 6,0)(3,0)1(1,-1)1.5 ( 4.5,1.5)(3,0)2(1,-1)1.5 ( 7.5,-1.5)(3,0)1(1,1)1.5 ( 6,0)(3,0)1(1,0)3 ( 4.5,1.5)(3,0)2(1,0)3 ( 7.5,0.3)(0,0)@xmath63 ( 6,1.75)(0,0)@xmath67 ( 9,1.75)(0,0)@xmath139 ( 6.55,0.95)(0,0)@xmath74 ( 9.55,0.95)(0,0)@xmath140 ( 5.4,0.95)(0,0)@xmath78 ( 8.4,0.95)(0,0)@xmath79 ( 6.6,-0.95)(0,0)@xmath84 ( 8.4,-0.95)(0,0)@xmath89 ( 10,1)(0,-0.6)(5.3,0)(0,0)(b ) ( 6,0)(3,0)2 ( 7.5,1.5)(3,0)2 ( 4.5,-1.5)(3,0)2 ( 6,0.35)(0,0)6 ( 9,0.35)(0,0)7 ( 7.5,1.8)(0,0)6 ( 10.5,1.8)(0,0)7 ( 4.5,-1.8)(0,0)5 ( 7.5,-1.8)(0,0)6 ( 6,0)(3,0)2(1,1)1.5 ( 6,0)(3,0)1(1,-1)1.5 ( 7.5,1.5)(3,0)1(1,-1)1.5 ( 4.5,-1.5)(3,0)2(1,1)1.5 ( 6,0)(3,0)1(1,0)3 ( 7.5,1.5)(3,0)1(1,0)3 ( 4.5,-1.5)(3,0)1(1,0)3 ( 7.5,0.3)(0,0)@xmath63 ( 9,1.75)(0,0)@xmath139 ( 6,-1.75)(0,0)@xmath69 ( 6.55,0.95)(0,0)@xmath74 ( 9.55,0.95)(0,0)@xmath140 ( 8.4,0.95)(0,0)@xmath79 ( 6.6,-0.95)(0,0)@xmath84 ( 5.4,-0.95)(0,0)@xmath88 ( 8.4,-0.95)(0,0)@xmath89 ( 10,0)(-10,-1.2)(5.3,0)(0,0)(d ) ( 6,0)(3,0)2 ( 4.5,1.5)(3,0)2 ( 7.5,-1.5)(3,0)2 ( 6,0.35)(0,0)6 ( 9,0.35)(0,0)7 ( 4.5,1.8)(0,0)5 ( 7.5,1.8)(0,0)6 ( 7.5,-1.8)(0,0)6 ( 10.5,-1.8)(0,0)7 ( 6,0)(3,0)1(1,1)1.5 ( 6,0)(3,0)2(1,-1)1.5 ( 4.5,1.5)(3,0)2(1,-1)1.5 ( 7.5,-1.5)(3,0)1(1,1)1.5 ( 6,0)(3,0)1(1,0)3 ( 4.5,1.5)(3,0)1(1,0)3 ( 7.5,-1.5)(3,0)1(1,0)3 ( 7.5,0.3)(0,0)@xmath63 ( 6,1.75)(0,0)@xmath67 ( 9,-1.75)(0,0)@xmath141 ( 6.55,0.95)(0,0)@xmath74 ( 5.4,0.95)(0,0)@xmath78 ( 8.4,0.95)(0,0)@xmath79 ( 6.6,-0.95)(0,0)@xmath84 ( 9.6,-0.95)(0,0)@xmath142 ( 8.4,-0.95)(0,0)@xmath89 ( 10,0)(-5,-1.9)(5.3,0)(0,0)(c ) ( 6,0)(3,0)2 ( 7.5,1.5)(3,0)1 ( 4.5,-1.5)(3,0)3 ( 6,0.35)(0,0)6 ( 9,0.35)(0,0)7 ( 7.5,1.8)(0,0)6 ( 4.5,-1.8)(0,0)5 ( 7.5,-1.8)(0,0)6 ( 10.5,-1.8)(0,0)7 ( 6,0)(3,0)1(1,1)1.5 ( 6,0)(3,0)2(1,-1)1.5 ( 7.5,1.5)(3,0)1(1,-1)1.5 ( 4.5,-1.5)(3,0)2(1,1)1.5 ( 6,0)(3,0)1(1,0)3 ( 4.5,-1.5)(3,0)2(1,0)3 ( 7.5,0.3)(0,0)@xmath63 ( 6,-1.75)(0,0)@xmath69 ( 9,-1.75)(0,0)@xmath141 ( 6.55,0.95)(0,0)@xmath74 ( 8.4,0.95)(0,0)@xmath79 ( 6.6,-0.95)(0,0)@xmath84 ( 9.6,-0.95)(0,0)@xmath142 ( 5.4,-0.95)(0,0)@xmath88 ( 8.4,-0.95)(0,0)@xmath89 an @xmath131 subgraph is decomposed into six new subgraphs in the same way as an @xmath58 and an @xmath94 ( see the examples in figure [ rhom_fig14 ] ) . the number @xmath92 of the decomposition vertex in the basic group for a current @xmath131 subgraph which is positioned between vertices @xmath90 and @xmath91 , is chosen as @xmath130 ( @xmath143 or @xmath144 ) . ( 10,3)(-4,-1.5)(9,0)(3,0)3 ( 10.5,1.5)(3,0)3 ( 7.5,-1.5)(3,0)3 ( 9,0.35)(0,0)5 ( 12,0.35)(0,0)6 ( 15,0.35)(0,0)7 ( 10.5,1.8)(0,0)5 ( 13.5,1.8)(0,0)6 ( 16.5,1.8)(0,0)7 ( 7.5,-1.8)(0,0)4 ( 10.5,-1.8)(0,0)5 ( 13.5,-1.8)(0,0)6 ( 9,0)(3,0)3(1,1)1.5 ( 9,0)(3,0)2(1,-1)1.5 ( 10.5,1.5)(3,0)2(1,-1)1.5 ( 7.5,-1.5)(3,0)3(1,1)1.5 ( 9,0)(3,0)2(1,0)3 ( 10.5,1.5)(3,0)2(1,0)3 ( 7.5,-1.5)(3,0)2(1,0)3 ( 10.5,0.3)(0,0)@xmath62 ( 13.5,0.3)(0,0)@xmath63 ( 11.7,1.75)(0,0)@xmath67 ( 15,1.75)(0,0)@xmath139 ( 9,-1.75)(0,0)@xmath68 ( 11.7,-1.75)(0,0)@xmath69 ( 9.55,0.95)(0,0)@xmath73 ( 12.55,0.95)(0,0)@xmath74 ( 15.5,0.95)(0,0)@xmath140 ( 11.4,0.95)(0,0)@xmath78 ( 14.45,0.95)(0,0)@xmath79 ( 9.6,-0.95)(0,0)@xmath83 ( 12.6,-0.95)(0,0)@xmath84 ( 8.4,-0.95)(0,0)@xmath87 ( 11.4,-0.95)(0,0)@xmath88 ( 14.45,-0.95)(0,0)@xmath89 ( 12,2.5)(12,0.55 ) ( 12,0.1)(12,-2.5 ) ( 10,0)(10.7,-2.2)(12,0)(3,0)4 ( 10.5,1.5)(3,0)5 ( 13.5,-1.5)(3,0)3 ( 12,0.35)(0,0)6 ( 15,0.35)(0,0)7 ( 18,0.35)(0,0)8 ( 21.02,0.35)(0,0)9 ( 10.5,1.8)(0,0)5 ( 13.5,1.8)(0,0)6 ( 16.5,1.8)(0,0)7 ( 19.5,1.8)(0,0)8 ( 22.5,1.8)(0,0)9 ( 13.5,-1.8)(0,0)6 ( 16.5,-1.8)(0,0)7 ( 19.5,-1.8)(0,0)8 ( 12,0)(3,0)4(1,1)1.5 ( 12,0)(3,0)3(1,-1)1.5 ( 10.5,1.5)(3,0)4(1,-1)1.5 ( 13.5,-1.5)(3,0)3(1,1)1.5 ( 12,0)(3,0)3(1,0)3 ( 10.5,1.5)(3,0)4(1,0)3 ( 13.5,-1.5)(3,0)2(1,0)3 ( 13.5,0.3)(0,0)@xmath63 ( 16.5,0.3)(0,0)@xmath145 ( 19.5,0.3)(0,0)@xmath146 ( 12,1.75)(0,0)@xmath67 ( 15,1.75)(0,0)@xmath139 ( 17.7,1.75)(0,0)@xmath147 ( 21,1.75)(0,0)@xmath148 ( 15,-1.75)(0,0)@xmath141 ( 17.7,-1.75)(0,0)@xmath149 ( 12.55,0.95)(0,0)@xmath74 ( 15.5,0.95)(0,0)@xmath140 ( 18.5,0.95)(0,0)@xmath150 ( 21.5,0.95)(0,0)@xmath151 ( 11.45,0.95)(0,0)@xmath78 ( 14.45,0.95)(0,0)@xmath79 ( 17.45,0.95)(0,0)@xmath152 ( 20.45,0.95)(0,0)@xmath153 ( 12.6,-0.95)(0,0)@xmath84 ( 15.6,-0.95)(0,0)@xmath142 ( 18.6,-0.95)(0,0)@xmath154 ( 14.45,-0.95)(0,0)@xmath89 ( 17.45,-0.95)(0,0)@xmath155 ( 20.45,-0.95)(0,0)@xmath156 ( 18,2.5)(18,0.55 ) ( 18,0.1)(18,-2.5 ) in the course of decomposition , two kinds of subgraphs are revealed in an @xmath131 . they are an @xmath94 and an @xmath157 . therefore , no new kinds of subgraphs are formed and subexpressions related to all subgraphs can be generated . hence , @xmath124 is computed by the following recursive relations : [ mfda_1 ] @xmath158 @xmath159 @xmath160 @xmath161 @xmath162 @xmath163 @xmath164 @xmath165 @xmath166 @xmath167 @xmath168 @xmath169 @xmath170 @xmath171 @xmath172 @xmath173 @xmath174 @xmath175 @xmath176 @xmath177 @xmath178 [ mfda_11]@xmath179 @xmath180 @xmath181 @xmath182 @xmath183 @xmath184 @xmath185 @xmath186 @xmath187 @xmath188 @xmath189 @xmath190 @xmath191 @xmath192 [ mfda_18]@xmath193 @xmath194 [ mfda_19]@xmath195 , @xmath196 @xmath197 @xmath198 @xmath199 , @xmath196 @xmath197 @xmath200 @xmath201 , @xmath196 @xmath202 @xmath203 @xmath204 , @xmath196 @xmath202 @xmath205 [ mfda_23]@xmath206 , @xmath207 @xmath202 @xmath208 @xmath209 , @xmath207 @xmath202 @xmath210 @xmath211 , @xmath207 @xmath202 @xmath212 [ mfda_26]@xmath213 , @xmath207 @xmath202 @xmath214 [ mfda_27]@xmath215 , @xmath216 @xmath197 . subgraphs of sizes @xmath25 and @xmath217 are trivial and their expressions are in lines ( 1 18 ) . specifically , expressions of subgraphs @xmath218 and @xmath219 are presented in the minimum factored form ( lines 11 18 ) . for example , the expression of the square rhomboid of size @xmath220 derived by 1-vda is @xmath221 it contains @xmath222 literals . the following lemma results from lines ( 23 26 ) of 1-vda . [ lem_mfda]complexities of expressions @xmath223 and @xmath224 derived by 1-vda are equal for @xmath225 . the following proposition results from lemma [ lem_mfda ] and from lines ( 1 27 ) of 1-vda . [ th_mfda]the total number of literals @xmath226 in the expression @xmath227 derived by 1-vda is defined recursively as follows : \1 ) @xmath2282 ) @xmath2293 ) @xmath2304 ) @xmath231 \5 ) @xmath232 6)@xmath2337 ) @xmath2348 ) @xmath235 \9 ) @xmath23610 ) @xmath237 11 ) @xmath23812 ) @xmath239 \13 ) @xmath24014 ) @xmath241 15 ) @xmath24216 ) @xmath243 \17 ) @xmath244 \18 ) @xmath245 \19 ) @xmath246 , where @xmath247 , @xmath248 , @xmath249 , and @xmath250 are the total numbers of literals in @xmath251 , @xmath252 , @xmath253 , and @xmath254 , respectively . hence , the expression @xmath227 derived by 1-vda consists of six subexpressions related to six revealed subgraphs of size @xmath255 @xmath256 and two additional literals . each of these subexpressions is constructed in its turn from six subexpressions related to six subgraphs of size @xmath257 @xmath258 and two additional literals , etc . thus , by the master theorem , @xmath259 , i.e. , 1-vda provides the representation of @xmath227 with a polynomial complexity . it is of interest to obtain exact formulae describing complexity of the expression @xmath227 derived by 1-vda . we attempt to do it for @xmath40 that is a power of two , i.e. , @xmath260 for some positive integer @xmath261 . statements ( 17 19 ) of proposition [ th_mfda ] are presented for @xmath262 @xmath263 as @xmath264{l}% t\left ( n\right ) = t\left ( \frac{n}{2}+1\right ) + t\left ( \frac{n}% { 2}\right ) + 2\widehat{t}\left ( \frac{n}{2}\right ) + 2\widehat{t}\left ( \frac{n}{2}-1\right ) + 2\\ \widehat{t}\left ( n\right ) = t\left ( \frac{n}{2}+1\right ) + 3\widehat { t}\left ( \frac{n}{2}\right ) + 2\widehat{\widehat{t}}\left ( \frac{n}% { 2}-1\right ) + 2\\ \widehat{\widehat{t}}\left ( n\right ) = \widehat{t}\left ( \frac{n}% { 2}+1\right ) + \widehat{t}\left ( \frac{n}{2}\right ) + 2\widehat{\widehat{t}% } \left ( \frac{n}{2}\right ) + 2\widehat{\widehat{t}}\left ( \frac{n}% { 2}-1\right ) + 2\text { , } % \end{array } \right . \label{rhomf3}%\ ] ] respectively . after a number of transformations , the following explicit formulae for simultaneous recurrences ( [ rhomf3 ] ) * * * * are obtained by the method for linear recurrence relations solving @xcite : @xmath265 another decomposition method ( called 2-vdm ) for generating algebraic expressions of square rhomboids is presented in @xcite and @xcite . the method consists in splitting a square rhomboid through two decomposition vertices one of which belongs to the upper group and another one to the lower group . the following algorithms based on this method are considered in @xcite and @xcite : full decomposition algorithm ( fda ) , combined decomposition algorithm ( cda ) and improved fda ( ifda ) . complexities of generated by these algorithms expressions are also @xmath266 . the values of @xmath226 for these algorithms and for 1-vda are presented in table [ tab1 ] ( @xmath267 ) . [ c]\|\|c\|\|c\|c\|c\|c\|\|@xmath40 & @xmath226 , fda & @xmath268 , cda & @xmath226 , ifda & @xmath269 , 1-vda + @xmath26 & @xmath270 & @xmath271 & @xmath271 & @xmath272 + @xmath273 & @xmath274 & @xmath275 & @xmath276 & @xmath277 + @xmath278 & @xmath279 & @xmath280 & @xmath281 & @xmath282 + @xmath105 & @xmath283 & @xmath284 & @xmath285 & @xmath286 + @xmath287 & @xmath288 & @xmath289 & @xmath290 & @xmath291 + @xmath292 & @xmath293 & @xmath294 & @xmath295 & @xmath296 + @xmath297 & @xmath298 & @xmath299 & @xmath300 & @xmath301 + one can see that 1-vda is more efficient than fda , cda , and ifda . the numerical advantage of 1-vda in contrast to other algorithms follows also from presented in @xcite and @xcite explicit formulae analogous to ( [ rhomf28]).the coefficient of the leading term of the formulae - @xmath302 is equal to @xmath303 for fda , @xmath304 for cda , @xmath305 for ifda , and @xmath306 for 1-vda . the decomposition method divides all subgraphs , which are revealed in the current recursive step into a constant number of subgraphs of proportionally decreasing sizes . the existence of a decomposition method for an st - dag @xmath7 is a sufficient condition for the existence of a representation of @xmath307 with a polynomial complexity . the complexity depends on the number of revealed subgraphs in each recursive step . a. jain and n. chandrasekharan , _ an efficient parallel algorithm for min - cost flow on directed series - parallel networks _ , in : proceedings of the 7th international parallel processing symposium , newport beach , california , usa , april 1993 , 188192 . m. korenblit and v. e. levit , _ on algebraic expressions of series - parallel and fibonacci graphs _ , in : discrete mathematics and theoretical computer science , proc . conf . , dmtcs 2003 , lncs * 2731 * , springer , 2003 , 215224 . m. korenblit and v. e. levit , _ an improved full decomposition algorithm for generating algebraic expressions of square rhomboids _ , in : proc . conf . on theoretical and mathematical foundations of computer science ( tmfcs-10 ) , 2010 , 31 - 38 . v. e. levit and m. korenblit , _ a symbolic approach to scheduling of robotic lines _ , in : intelligent scheduling of robots and flexible manufacturing systems , the center for technological education holon , israel , 1996 , 113125 . m. k. oikawa , j. e. ferreira , s. malkowski , and c. pu , _ towards algorithmic generation of business processes : from business step dependencies to process algebra expressions _ , in : business process management , proc . conf . , bpm 2009 , lncs * 5701 * , springer , 2009 , 8096 .	the paper investigates relationship between algebraic expressions and graphs . we consider a digraph called a square rhomboid that is an example of non - series - parallel graphs . our intention is to simplify the expressions of square rhomboids and eventually find their shortest representations . with that end in view , we describe the new algorithm for generating square rhomboid expressions , which improves on our previous algorithms .
the ability to controllably engineer spin splittings of electronic states is a key goal in the search for spintronic materials @xcite . a particularly successful strategy has been the lifting of spin degeneracy via spin - orbit coupling in the presence of structural inversion asymmetry . termed the rashba or bychov - rashba effect @xcite this manifests through a spin - momentum locking of the quasiparticles , stabilising a pair of fermi surfaces which are typically assumed to exhibit counter - rotating chiral spin textures @xcite . the ability to electrostatically control the strength of such spin splitting @xcite has led to prospects for all - electrical manipulation of electron spin precession @xcite , offering new prototypical schemes of semiconductor spintronics @xcite . the quest to practically realise such devices , and to operate them without cryogenic cooling , has motivated a major search for materials which can host stronger spin splittings than can typically be achieved in conventional semiconductors @xcite . a giant rashba - like spin splitting was recently discovered for bulk conduction and valence band states of the bismuth tellurohalide semiconductors @xcite . arising due to a combination of bulk inversion asymmetry ( fig . [ f : overview](b ) ) , strong atomic spin - orbit coupling , and a negative crystal field splitting of the valence bands @xcite , rashba parameters amongst the highest of any known material have been uncovered , together with a counter - rotating chiral fermi surface spin texture @xcite . exploiting element- and orbital - selective angle - resolved photoemission ( arpes ) , we show here that a complex interplay between atomic , orbital , and spin degrees of freedom significantly enriches this picture . we expect our findings to be broadly applicable across other strong spin - orbit rashba systems . * surface electronic structure of bitei . * ( a ) arpes measurements of the fermi surface and near-@xmath0 band dispersions measured ( @xmath1 ev , @xmath2-polarisation ) from the te - terminated surface of bitei . a lack of inversion symmetry of the bulk crystal structure ( b ) together with strong spin - orbit coupling mediates a large rashba - like spin splitting . additionally , the polar nature of the te - terminated surface induces a strong downward band bending , causing a ladder of rashba - split subband ( sb ) states to emerge in the near - surface quantum well . ( c ) these are clearly resolved in measurements of the dispersion along @xmath3 . the conventional spin texture associated with such rashba splitting is shown schematically by coloured arrows , with the spin expected to lie predominantly in the surface plane . ] first , we summarise the generic electronic structure of the te - terminated surface of bitei ( fig . [ f : overview ] ) . previous measurements have shown this termination to support a near - surface electron accumulation @xcite . we clearly observe two 2d subbands ( sb1 and sb2 in fig . [ f : overview](c ) , see also supplementary fig . s1 ) formed within the resulting quantum well . each subband hosts two branches with a separation that grows approximately linearly with momentum away from their crossing at @xmath4 . this is a hallmark of rashba - like spin splitting . we extract a large rashba energy @xmath5 mev ( @xmath6 mev ) and momentum offset of the band bottom @xmath7 @xmath8 ( @xmath9 @xmath8 ) for the first ( second ) subband , respectively , supporting previous studies which established this material as a model host of giant spin splittings @xcite . while typically treated in a single - band picture , the electronic wavefunction for each branch of a rashba - split state can more generally be written as @xmath10 , where , following the notation in ref . @xcite , @xmath11 is the atomic index , @xmath12 and @xmath13 are the orbital and spin indices , respectively , @xmath14 are atomic wavefunctions , and @xmath15 are complex coefficients . neglecting spin - orbit coupling , our calculations predict a conduction band in bitei predominantly derived from bi @xmath16 orbitals ( see supplementary fig . including such effects , however , not only permits it to become strongly spin - split via rashba - like interactions , but also promotes significant orbital mixing . in general , therefore , multiple @xmath15 can be expected to become non - negligible . for a complete description of the rashba - split states , it is therefore essential to consider the interplay of the underlying atomic , orbital , and spin components . to disentangle these , we combine two powerful features of arpes : characteristic selection rules for photoemission using linearly - polarised light , allowing us to directly probe the orbital wavefunction @xcite , and resonant photoemission to provide elemental sensitivity @xcite . such resonant enhancements are evident in fig . [ f : disp](a , b ) . they cause the spectral weight of the conduction band states to become strongly peaked at photon energies around 26 and 28 ev , close in energy to the binding energy of the bi @xmath17 core levels , with functional forms that are well described by fano lineshapes . this points to a significant bi - derived atomic character of the lowest conduction band states , consistent with theoretical calculations @xcite . we exploit this , selectively probing `` on - resonance '' to unveil the bi - projected spectral function . the resulting measurements of the dispersion reveal pronounced momentum - dependent spectral weight variations ( fig . [ f : disp](c ) ) . we focus on the first subband ( sb1 ) , which is most clearly visible across our measurements . when probed using @xmath2-polarised light , we find stronger spectral weight for the inner branch of this subband , while the outer branch is significantly more pronounced when measured using @xmath18-polarisation . selection rules @xcite dictate that , of the @xmath2-orbitals , the former measurement should be sensitive to @xmath16- and @xmath19-derived orbital character , while the transition matrix element is only non - vanishing for photoemission from @xmath20 orbitals in the latter case ( our measurement geometry is shown in fig . [ f : cec](a ) ) . the asymmetric spectral weight distributions within and between these measurements immediately establishes that the two spin - split branches of the dispersion host a markedly different orbital makeup . moreover , when measuring with a photon energy only 2 ev higher ( fig . [ f : disp](d ) ) , we find an almost complete reversal of these matrix element variations , with greater spectral weight for the inner branch of the lowest subband when measuring using @xmath18-polarised light . no longer on resonance , these measurements will not _ a priori _ be dominated by the bi - derived spectral weight , and we therefore conclude that the orbital textures of the rashba - split states are also strongly dependent on their underlying atomic character , as discussed in detail below . * @xmath20-projected spectral weight distribution . * ( a ) experimental geometry for our measurements , and ( b - d ) resulting constant binding energy surfaces measured using @xmath18-polarised light to probe the @xmath20 orbital character at 200 mev above , exactly at , and at 100 mev below the dirac point ( @xmath21 ) formed by the crossing of the two spin - split branches of the lowest subband . on - resonance measurements ( @xmath22 ev ) with the scattering plane aligned to ( b ) @xmath3 and ( c ) @xmath23 , and ( d ) off - resonance ( @xmath24 ev , scattering plane along @xmath23 ) measurements show pronounced angular variations in spectral weight , characteristic of strongly momentum - dependent orbital textures . ] fig . [ f : cec](b , c ) shows constant binding energy surfaces ( cess ) measured on - resonance with @xmath18-polarised light , to selectively probe the momentum - space distribution of the bi @xmath20 orbital character . the outer band of each ces exhibits strong spectral weight along our measurement direction ( at positive and negative @xmath25 ) , with a pronounced suppression at positive and negative @xmath26 , over the entire occupied bandwidth of the conduction bands . this remains qualitatively unchanged under rotation of the azimuthal orientation of the sample . such behaviour is indicative of a @xmath20-like orbital aligned along our measurement direction , irrespective of whether the @xmath3 or @xmath23 crystallographic direction is oriented to this @xcite . the in - plane bi - projected charge density must therefore be oriented such that its lobes point predominantly radially outwards from the outer ces . a similar conclusion can be drawn for the inner band of the ces at energies below the dirac point formed by the crossing of the two spin - split branches of the dispersion . as already evident 100 mev below the dirac point in fig . [ f : cec](b , c ) , and more clearly shown in the angular dependence of spectral weight extracted around the ces in fig . [ f : orb](a ) , the spectral weight is still peaked along the @xmath25 direction ( @xmath27 and @xmath28 in fig . [ f : orb](a ) ) , with a suppression of spectral weight along @xmath26 ( @xmath29 . indeed , while the band edge turning point in the dispersion leads to a van hove singularity in the density of states @xcite the inner band of the ces below the dirac point smoothly evolves into the outer band as it moves through this turning point , and thus hosts qualitatively the same orbital texture . moving up through the dirac point , however , the angular distribution of spectral weight around the inner ces gradually flattens and then inverts to become peaked at @xmath30 . this points to a tangential alignment of the in - plane bi - derived orbital character above the dirac point , with stronger @xmath20 character at positive and negative @xmath26 than positive and negative @xmath25 . this behaviour is well - reproduced by our first - principles calculations . as shown in fig . [ f : orb](c ) , the in - plane @xmath31 orbital polarisation around the calculated cess of the lower subband show a clear peak at @xmath30 above the dirac point , and 0 and @xmath28 below . to capture this generic behaviour , we extract a relative spectral weight angular distribution factor from our measurements : @xmath32 where @xmath33 is the spectral weight at angle @xmath34 and energy @xmath35 . this provides an experimental measure of the calculated `` orbital polarisation ratio '' used to describe topological states in ref . @xcite . here , this parameterises the relative strength of spectral weight , and thus @xmath20 orbital character , along the @xmath26 and @xmath25 directions of our measured cess , reflecting the strength ( @xmath36 ) and alignment of radial ( @xmath37 ) and tangential ( @xmath38 ) in - plane orbital textures . as shown in fig . [ f : orb](b ) , we find this is positive at energies above the dirac point for the on - resonance measurements , consistent with the dominantly tangential in - plane bi - projected charge density distribution of the inner band assigned above . with decreasing energy towards the dirac point , @xmath39 is suppressed to zero , indicating a loss of orbital polarisation , before growing again but with opposite sign below the dirac point . thus , we have experimentally observed a gradual crossover from a radial to a tangential alignment of the in - plane bi orbital character with the switch occurring exactly at the dirac point within our experimental error . this is strikingly similar to a switch in orbital texture that was recently reported at the dirac point of the spin - polarised surface states of topological insulators @xcite . our observation of such a crossover in a topologically - trivial compound establishes this as a general feature of strongly spin - orbit coupled systems , a point we return to below . intriguingly , extracting @xmath40 from our measurements performed off - resonance , we find an opposite trend ( fig . [ f : orb](b ) ) . above the dirac point , @xmath37 indicating a radial alignment of the in - plane orbital - projected charge density , as also clearly evident in the spectral weight distribution visible in fig . [ f : cec](d ) . this is again smoothly suppressed to zero approaching the dirac point before becoming positive , revealing a tangential orbital configuration , at energies below dirac point . we attribute this as a signature of the in - plane te orbital polarisation , which well matches that found in our calculations ( fig . [ f : orb](d ) ) . it can be seen in our measurements here at 30 ev photon energy as , with no bi resonant enhancement , the photoemission cross - section is higher for te @xmath41 than bi @xmath42 orbitals @xcite and , with te being located right at the surface , there is no depth - dependent attenuation due to the surface sensitivity of photoemission . our calculations ( supplementary fig . s2 ) reveal a strong in - plane te character of the inner branch of the dispersion , which can thus dominate the spectral weight of the inner branch of the ces in these off - resonance measurements . for the outer branch , however , there is little calculated in - plane te weight ( supplemental fig . s2 ) , and so the weak spectral features visible for the outer branch in fig . [ f : cec](d ) is still reflective of the bi in - plane orbital texture . together , these measurements and calculations reveal that spin - orbit coupling induces a complex atomic- and momentum - dependent hierarchy of orbitally - polarised components of the underlying electronic structure in bitei , summarised schematically in fig . [ f : cd ] . our calculations additionally reveal how each orbital component is , in turn , coupled to a disparate spin texture . we illustrate this for the bi - derived states in fig . [ f : cd](b - d ) ; the te - derived component is additionally shown in supplementary fig . the in - plane spin - texture @xmath43 projected onto bi @xmath16 orbitals yields a conventional counter - rotating chiral spin texture of neighbouring cess at energies above the dirac point , characteristic of classic rashba systems @xcite and indeed observed experimentally for bitei @xcite . in contrast , the spin texture is significantly more complex for the @xmath19 and @xmath20 orbital projection , with the in - plane spin component switching between tangential and radial around the ces . this results from a coupling of the spin to the characteristic orbital textures , and is similar to that found recently for topological surface states @xcite . for example , combined with our uncovered tangential ( radial ) in - plane bi orbital texture above ( below ) the dp , this leads to a net clockwise spin rotation for the in - plane orbital projection of _ both _ the inner and outer cess . this is in strong departure from the conventional picture of rashba - like spin polarisation . such in - plane orbital textures likely also underlie the unconventional spin topology predicted for certain @xmath44-derived states in bi / cu(111 ) @xcite and pb / cu(111 ) @xcite surface alloys . they are also broadly consistent with recent first - principles calculations that suggest the rashba parameter can be orbital - dependent in bismuth tellurohalides @xcite , although here we reveal how orbital mixing can lock the spin components on the nominally `` spin - split '' cess together , stabilising pronounced components of the underlying wavefunction that host markedly non - rashba - like spin textures in this system . similar considerations hold for the te - derived components , although with additional variations in the magnitude of the spin components projected onto out - of - plane vs. in - plane orbital components between the cess due to a greater out vs. in - plane orbital polarisation for te ( supplementary figs . s2 and s3 ) . with each orbital component locked to a different spin texture , the fundamental requirement from time - reversal symmetry of spin - degeneracy at the kramer s point ( the dirac point formed in this system at @xmath4 ) ensures equal contribution of in - plane @xmath19 and @xmath20 orbitals at this point . thus , the vanishing of orbital polarisation at the dirac point observed experimentally above can be simply viewed as an orbital analogue of kramer s spin degeneracy in time - reversal symmetric systems . away from the dirac point , our calculations additionally predict a strong canting of the spin out of the surface plane for the bi - derived orbitals . this grows in magnitude with increasing energy away from the dirac point , where the cess become increasingly hexagonally warped @xcite . we show evidence for this through circular dichroism measurements . we attribute such dichroism as a signature of unquenched orbital angular momentum @xcite , which our calculations reveal is large and locked approximately opposite to the spin angular momentum due to the strong spin - orbit coupling . circular dichroism is known to show a complex dependence on photon energy in this system @xcite as well as in other layered compounds such as bi@xmath45te@xmath46 ( ref . ) , which can naturally arise as a consequence of inter - layer photoelectron interference @xcite . to simplify such effects , we again perform our measurements on - resonance , selectively enhancing states of bi character ( fig . [ f : cd](f ) ) . we find that the outer ( hexagonally - warped ) band in such measurements develops a pronounced six - fold modulation , which has previously been shown to reflect out - of - plane spin - canting in topological insulators @xcite . here , this provides the first experimental evidence for deviations from simple in - plane chiral spin textures in bitei . as evident in fig . [ f : cd](b - d ) , @xmath47 is substantially larger when projected onto the in - plane orbital components than for the @xmath16 projection . the emergence of a large out - of - plane spin canting in this system is thus intricately tied to the development of the in - plane orbital texture away from the dirac point observed here . together , our measurements and calculations establish a powerful role of in - plane atomic orbitals shaping the spin structure of the model rashba compound bitei , revealing a complex interplay between atomic makeup , anisotropic orbital textures , and spin - momentum locking . while small quantitative variations may occur due to near - surface potential contributions to the rashba spin splitting and possible surface - induced orbital reconstructions here , our findings should be broadly applicable to the spin - split bulk electronic structure of bitei as well as being generic to other strong spin - orbit rashba systems , suggesting new routes to control spin splitting through the orbital sector . for example , exploiting structure - property relations to tune the competition of atomic spin - orbit coupling and crystal field splitting will allow control over the ratio of in - plane and out - of - plane orbital polarisation , thus modulating the degree of non - rashba - like spin components and out - of - plane spin canting . together , this promises new prospects for the targeted design of optimised spintronic materials . * angle - resolved photoemission : * arpes measurements were performed at the cassiopee beamline of soleil synchrotron ( france ) and i05 beamline of diamond light source ( uk ) . single - crystal samples of bitei , grown by chemical vapour transport , were cleaved _ in - situ _ at the measurement temperature of 10 k. measurements were performed using linear - horizontal , linear - vertical , and circularly - polarised synchrotron light at the photon energies described in the text , and reproduced on multiple samples . scienta r4000 hemispherical electron analysers were employed , with a vertical entrance slit and the light incident in the horizontal plane as shown in fig . [ f : cec](a ) . bitei is known to have domains of mixed te and i termination , supporting surface electron and hole accumulation layers , respectively @xcite . by monitoring the relative spectral weight of electron- and hole - like bands crossing the fermi level and from characteristic core - level shifts in x - ray photoemission spectra , we aligned the synchrotron light spot on te - terminated domains of the sample . from core - level spectra , we estimate an upper limit of 2% i - terminated regions within our probing region for the data shown here . * density - functional theory calculations : * density - functional theory ( dft ) was performed within the generalized gradient approximation ( gga ) . we used the dft code , openmx @xcite , based on the linear - combination - of - pseudo - atomic - orbitals ( lcpao ) method @xcite . spin - orbit interaction was included via the norm - conserving , fully relativistic @xmath48-dependent pseudopotential scheme in the non - collinear dft formalism @xcite . we model the te - terminated bitei surface electronic structure via a supercell calculation including a slab of 60 atomic layers and a vacuum region of @xmath49 thickness . to calculate the spin and orbital angular momentum for specific @xmath50-points , we used the lcpao coefficients of local atoms . we acknowledge m.s . bahramy for useful discussions . this work was supported by the engineering and physical sciences research council , uk ( grant nos . ep / i031014/1 , ep / m023427/1 and ep / g03673x/1 ) , the ministry of science and technology in taiwan ( project number most-102 - 2119-m-002 -004 ) , trf - sut grant rsa5680052 , the national science foundation ( grant nos . dmr-0847385 and dmr-1120296 ) , and the office of naval research ( grant no . n00014 - 12 - 1 - 0791 ) . pdck acknowledges support from the royal society through a university research fellowship . hiw acknowledges support from the nsf igert program ( dge-0903653 ) and the national science foundation graduate research fellowship under grant no . we acknowledge soleil for provision of synchrotron radiation facilities ( cassiopee beamlime ) and also thank diamond light source for access to beamline i05 ( proposal number si9427 ) that contributed to the results presented here . the authors declare that they have no competing interests . s1 : photon energy - dependent arpes + fig . s2 : calculated surface electronic structure of bitei + fig . s3 : coupled te spin - orbital texture + h. c. koo , j. h. kwon , j. eom , j. chang , s. h. han , m. johnson , control of spin precession in a spin - injected field effect transistor , _ science _ * 325 * , 1515 ( 2009 ) . y. a. bychkov , e. i. rashba , properties of a 2d electron gas with lifted spectral degeneracy , _ j.e.t.p . lett . _ * 39 * , 78 ( 1984 ) . s. lashell , b. a. mcdougall , e. jensen , spin splitting of an au(111 ) surface state band observed with angle resolved photoelectron spectroscopy , _ phys . lett . _ * 77 * , 3419 ( 1996 ) . j. nitta , t. akazaki , h. takayanagi , t. enoki , gate control of spin - orbit interaction in an inverted in@xmath51ga@xmath52as / in@xmath53al@xmath54as heterostructure , _ phys . lett . _ * 78 * , 1335 ( 1997 ) . d. grundler , large rashba splitting in inas quantum wells due to electron wave function penetration into the barrier layers , _ phys . _ * 84 * , 6074 ( 2000 ) . p. d. c. king , r. c. hatch , m. bianchi , r. ovsyannikov , c. lupulescu , g. landolt , b. slomski , j. h. dil , d. guan , j. l. mi , e. d. l. rienks , j. fink , a. lindblad , s. svensson , s. bao , g. balakrishnan , b. b. iversen , j. osterwalder , w. eberhardt , f. baumberger , ph . hofmann , large tunable rashba spin splitting of a two - dimensional electron gas in bi@xmath55se@xmath56 , _ phys . lett . _ * 107 * , 096802 ( 2011 ) . m. bahramy , p. d. c. king , a. de la torre , j. chang , m. shi , l. patthey , g. balakrishnan , ph . hofmann , r. arita , n. nagaosa , f. baumberger , emergent quantum confinement at topological insulator surfaces , _ nat . commun . _ * 3 * , 1159 ( 2012 ) . s. datta , b. das , electronic analog of the electro - optic modulator , _ appl . lett . _ * 56 * , 665 ( 1990 ) . c. r. ast , c. r. ast , j. henk , a. ernst , l. moreschini , m. c. falub , d. pacil , p. bruno , k. kern , m. grioni , giant spin splitting through surface alloying , _ phys . * 98 * , 186807 ( 2007 ) . k. sakamoto , t. -h . kim , t. kuzumaki , b. mller , y. yamamoto , m. ohtaka , j. r. osiecki , k. miyamoto , y. takeichi , a. harasawa , s. d. stolwijk , a. b. schmidt , j. fujii , r. i. g. uhrberg , m. donath , h. w. yeom , t. oda , valley spin polarization by using the extraordinary rashba effect on silicon , _ nat . commun . _ * 4 * , 2073 ( 2013 ) . h. yuan , m. s. bahramy , k. morimoto , s. wu , k. nomura , b. -j . yang , h. shimotani , r. suzuki , m. toh , c. kloc , x. xu , r. arita , n. nagaosa , y. iwasa , zeeman - type spin splitting controlled by an electric field , _ nat . phys . _ * 9 * , 563 ( 2013 ) . j. m. riley , f. mazzola , m. dendzik , m. michiardi , t. takayama , l. bawden , c. granerd , m. leandersson , t. balasubramanian , m. hoesch , t. k. kim , h. takagi , w. meevasana , ph . hofmann , m. s. bahramy , j. w. wells , p. d. c. king , direct observation of spin - polarized bulk bands in an inversion - symmetric semiconductor , _ nat . * 10 * , 835 ( 2014 ) . k. ishizaka , m. s. bahramy , h. murakawa , m. sakano , t. shimojima , t. sonobe , k. koizumi , s. shin , h. miyahara , a. kimura , k. miyamoto , t. okuda , h. namatame , m. taniguchi , r. arita , n. nagaosa , k. kobayashi , y. murakami , r. kumai , y. kaneko , y. onose , y. tokura , giant rashba - type spin splitting in bulk bitei , _ nat . _ * 10 * , 521 ( 2011 ) . m. s. bahramy , r. arita , n. nagaosa , origin of giant bulk rashba splitting : application to bitei , _ phys . b _ * 84 * , 041202 ( 2011 ) . h. murakawa , m. s. bahramy , m. tokunaga , y. kohama , c. bell , y. kaneko , n. nagaosa , h. y. hwang , y. tokura , detection of berry s phase in a bulk rashba semiconductor , _ science _ * 342 * , 1490 ( 2013 ) . g. landolt , s. v. eremeev , y. m. koroteev , b. slomski , s. muff , t. neupert , m. kobayashi , v. n. strocov , t. schmitt , z. s. aliev , m. b. babanly , i. r. amiraslanov , e. v. chulkov , j. osterwalder , j. h. dil , disentanglement of surface and bulk rashba spin splittings in noncentrosymmetric bitei , _ phys . lett . _ * 109 * , 116403 ( 2012 ) . a. crepaldi , l. moreschini , g. auts , c. tournier - colletta , s. moser , n. virk , h. berger , ph . bugnon , y. j. chang , k. kern , a. bostwick , e. rotenberg , o. v. yazyev , m. grioni , giant ambipolar rashba effect in the semiconductor bitei , _ phys . * 109 * , 096803 ( 2012 ) . zhu , c. n. veenstra , g. levy , a. ubaldini , p. syers , n. p. butch , j. paglione , m. w. haverkort , i. s. elfimov , a. damascelli , layer - by - layer entangled spin - orbital texture of the topological surface state in bi@xmath45se@xmath46 , _ phys . _ * 110 * , 216401 ( 2013 ) . a. damascelli , z. hussain , z .- x . shen , angle - resolved photoemission studies of the cuprate superconductors , _ rev . phys . _ * 75 * , 473 ( 2003 ) . y. cao , j. a. waugh , x - w . zhang , j - w . luo , q. wang , t. j. reber , s. k. mo , z. xu , a. yang , j. schneeloch , g. d. gu , m. brahlek , n. bansal , s. oh , a. zunger , d. s. dessau , mapping the orbital wavefunction of the surface states in three - dimensional topological insulators , _ nat . phys . _ * 9 * , 499 ( 2013 ) . s. hfner , _ photoelectron spectroscopy _ 82 of _ springer series in solid - state sciences _ ( springer , 1996 ) . m. bahramy , b .- yang , r. arita , n. nagaosa , emergence of non - centrosymmetric topological insulating phase in bitei under pressure , _ nat . commun . _ * 3 * , 679 ( 2012 ) . c. r. ast , g. wittich , p. wahl , r. vogelgesang , d. pacil , m. c. falub , l. moreschini , m. papagno , m. grioni , k. kern , local detection of spin - orbit splitting by scanning tunneling spectroscopy , _ phys . b _ * 75 * , 201401 ( 2007 ) . j. j. yeh , i. lindau , _ atomic data and nuclear data tables _ * 32 * , 1 ( 1985 ) . h. zhang , c .- x . liu , s .- c . zhang , spin - orbital texture in topological insulators , _ phys . lett . _ * 111 * , 066801 ( 2013 ) . y. cao , j. a. waugh , n. c. plumb , t. j. reber , s. parham , g. landolt , z. xu , a. yang , j. schneeloch , g. gu , j. h. dil , d. s. dessau , coupled spin - orbital texture in a prototypical topological insulator , _ arxiv:1211.5998 [ cond - mat ] _ zhu , c. n. veenstra , s. zhdanovich , m. p. schneider , t. okuda , k. miyamoto , s .- y . zhu , h. namatame , m. taniguchi , m. w. haverkort , i. s. elfimov , a. damascelli , photoelectron spin - polarization control in the topological insulator bi@xmath45se@xmath46 , _ phys . lett . _ * 112 * , 076802 ( 2014 ) . h. mirhosseini , j. henk , a. ernst , s. ostanin , c .- t . chiang , p. yu , a. winkelmann , j. kirschner , unconventional spin topology in surface alloys with rashba - type spin splitting , _ phys . rev . b _ * 79 * , 245428 ( 2009 ) . g. bihlmayer , s. blgel , e. v. chulkov , enhanced rashba spin - orbit splitting in bi / ag(111 ) and pb / ag(111 ) surface alloys from first principles , _ phys . b _ * 75 * , 195414 ( 2007 ) . z. zhu , y. cheng , u. schwingenschlgl , orbital - dependent rashba coupling in bulk bitecl and bitei , _ new j. phys . _ * 15 * , 023010 ( 2013 ) . s. r. park , c. h. kim , j. yu , j. h. han , c. kim , orbital - angular - momentum based origin of rashba - type surface band splitting , _ phys . lett . _ * 107 * , 156803 ( 2011 ) . a. crepaldi , f. cilento , m. zacchigna , m. zonno , j. c. johannsen , c. tournier - colletta , l. moreschini , i. vobornik , f. bondino , e. magnano , h. berger , a. magrez , ph . bugnon , g. auts , o. v. yazyev , m. grioni , f. parmigiani , momentum and photon energy dependence of the circular dichroic photoemission in the bulk rashba semiconductors bitex ( x = i , br , cl ) , _ phys . b _ * 89 * , 125408 ( 2014 ) . m. neupane , s. basak , n. alidoust , s .- y . xu , c. liu , i. belopolski , g. bian , j. xiong , h. ji , s. jia , s .- k . mo , m. bissen , m. severson , h. lin , n. p. ong , t. durakiewicz , r. j. cava , a. bansil , and m. z. hasan , oscillatory surface dichroism of the insulating topological insulator bi@xmath45te@xmath45se , _ phys . b _ * 88 * , 165129 ( 2013 ) . y. h. wang , d. hsieh , d. pilon , l. fu , d. r. gardner , y. s. lee , n. gedik , observation of a warped helical spin texture in bi@xmath45se@xmath46 from circular dichroism angle - resolved photoemission spectroscopy , _ phys . lett . _ * 107 * , 207602 ( 2011 ) . w. jung , y. kim , b. kim , y. koh , c. kim , m. matsunami , s .- i . kimura , m. arita , k. shimada , j. h. han , j. kim , b. cho , c. kim , warping effects in the band and angular - momentum structures of the topological insulator bi@xmath45te@xmath46 , _ phys . b _ * 84 * , 245435 ( 2011 ) . http://www.openmx-square.org . t. ozaki , variationally optimized atomic orbitals for large - scale electronic structures , _ phys . b _ * 67 * , 155108 ( 2003 ) . g. theurich , n. a. hill , self - consistent treatment of spin - orbit coupling in solids using relativistic fully separable ab initio pseudopotentials , _ phys . rev . b _ * 64 * , 073106 ( 2001 ) . p d. c. king , t. d. veal , and c. f. mcconville , nonparabolic coupled poisson - schrdinger solutions for quantized electron accumulation layers : band bending , charge profile , and subbands at inn surfaces , _ phys . b _ * 77 * , 125305 ( 2008 ) . s. g. louie , p. thiry , r. pinchaux , y. ptroff , d. chandesris , and j. lecante , periodic oscillations of the frequency - dependent photoelectric cross sections of surface states : theory and experiment , _ phys . lett . _ * 44 * , 549 ( 1980 ) .	the rashba effect is one of the most striking manifestations of spin - orbit coupling in solids , and provides a cornerstone for the burgeoning field of semiconductor spintronics . it is typically assumed to manifest as a momentum - dependent splitting of a single initially spin - degenerate band into two branches with opposite spin polarisation . here , combining polarisation - dependent and resonant angle - resolved photoemission measurements with density - functional theory calculations , we show that the two `` spin - split '' branches of the model giant rashba system bitei additionally develop disparate orbital textures , each of which is coupled to a distinct spin configuration . this necessitates a re - interpretation of spin splitting in rashba - like systems , and opens new possibilities for controlling spin polarisation through the orbital sector .
structural properties of regularly stepped metal surfaces have been the focus of a broad range of theoretical and experimental studies because of the eminent role they play in technologically important phenomena such as thin film growth , epitaxial layer formation , nanostructuring of material , and catalysis @xcite . according to crystallographic notation , these surfaces are denoted by high miller indices and are called vicinals of their low miller index counterparts ( flat surfaces ) . the presence of arrays of atomic steps separated by flat terraces creates regions of differing local coordination and makes the microscopic structure of a vicinal surface distinct from that of a flat surface . according to smoluchowski s idea of charge smoothing @xcite , for example , electronic charge densities are expected to rearrange in the vicinity of the steps , thereby causing the ion cores to relax to new configurations . the modified electronic structure may also be expected to impact the reactivity and the nature of the force fields in the region around the steps . knowledge of atomic relaxations in the equilibrium positions near the step and kink sites is thus a step towards understanding the novel vibrational and electronic properties of vicinal surfaces . fortunately with advances in atomic scale experimental techniques , there has been a surge in investigations of the structure of vicinal surfaces in recent years . the majority of the experimental data have undoubtedly come @xcite from the low - energy electron diffraction ( leed ) technique which is now capable of detecting changes even for interlayer spacings smaller than 1@xmath0@xcite . for some surfaces the x - ray scattering technique has provided much needed complementary structural data @xcite . an impressive number of theoretical calculations of multilayer relaxations @xcite have also helped in bringing several issues related to the characteristics of vicinal surfaces to the forefront . of particular interest here are experimental and theoretical studies of a set of vicinals of cu(100 ) and cu(111 ) which have addressed the question of the impact of local coordination on the structural and dynamical properties of the surface . in an earlier paper @xcite , a comparative study of the local structural and vibrational properties of cu(211 ) , cu(511 ) , and cu(331 ) was performed using empirical potentials from the embedded atom method ( eam ) @xcite . this study found that the first two surfaces displayed similar local characteristics , while the third surface was somewhat different . an explanation provided for this behavior was the similarity in the local environment of the ( 211 ) and ( 511 ) surfaces of fcc metals ( a combination of ( 100 ) and ( 111 ) , terrace geometry , and step face ) , and its consequent difference from that of the ( 331 ) surface ( a ( 111 ) terrace geometry and a ( 111)-microfacetted step face ) . the issue of the impact of the local geometry was further raised in a joint theoretical and experimental study of the vibrational dynamics of cu(211 ) and cu(511 ) together with those of the kinked surface cu(532 ) @xcite . experimental data from electron energy loss spectroscopy ( eels ) found modes above the bulk band on cu(211 ) but not on cu(511 ) ( or on cu(17,1,1 ) which has the same step geometry as cu(511 ) ) , but theoretical calculations based on eam potentials predicted modes ( slightly ) above the bulk phonon spectrum for each of these surfaces . while the similarity between the calculated structural relaxation patterns of cu(211 ) and cu(511 ) argues in favor of a similarity in the local vibrational dynamics of these two surfaces , the disagreement between the experimental and the theoretical results for the high frequency modes on cu(511 ) ( and cu(17,1,1 ) ) remains unreconciled . for cu(211 ) agreement of the eam based results with available structural data from leed @xcite and with _ ab initio _ calculations for both the structure and the dynamics @xcite provides considerable confidence in its predicted properties . the case of cu(511 ) is not as simple because of lack of calculations based on potentials more accurate than eam , and because of conflicting conclusions from the analysis of experimental data from leed @xcite and x - ray scattering measurements @xcite . the most striking difference in these two sets of data is the relaxation pattern for the second layer which is inwards in leed and outwards in the x - ray data . the oscillatory pattern found in the x - ray data is also in disagreement with the conclusion from a series of previous experimental and theoretical findings on stepped surfaces . based on these studies @xcite , there is a definite symmetry in the relaxation patterns of stepped surfaces . all terrace atoms , save for the corner one , display inward relaxations . the eam based calculations @xcite further predict this oscillatory relaxation pattern to continue into the bulk with a damping in the amplitude @xcite . thus the expected relaxation pattern for the ( 211 ) , ( 511 ) , ( 331 ) surfaces , each with 3-atom wide terraces , would be ( - - + - - + ... ) , although questions have been raised whether cu(331 ) follows this rule @xcite . similarly , the patterns for ( 711 ) and ( 911 ) with , respectively , 4 and 5 atoms on the terrace , would be predicted to be ( - - - + - - - + ... ) and ( - - - - + - - - - + ... ) . the leed data on the first three surfaces follow these predicted trend in relaxations , atleast for the top 3 layers . the very recent leed data @xcite for cu(711 ) also displays the pattern ( - - - + ) for the top layers , in good agreement with eam based predictions . however , a small discrepancy in the sign of the relaxation is found for both cu(711 ) and cu(511 ) , for a particular layer separation ( d@xmath1 for cu(511 ) and d@xmath2 for cu(711 ) ) @xcite , beyond that expected from the error bars . arguably the actual numbers involved in these comparisons are small , but the systematic nature of the discrepancies and the fact that it negates the prediction of a periodicity in the oscillatory relaxation pattern @xcite , raise interesting questions about the complexities of the atomic displacements in these systems . given the above uncertainties arising from experimental observations , it is opportune to carry out more accurate calculations of these relaxation patterns using techniques which are capable of revealing the accompanying changes in the surface electronic structure . it is with this goal in mind that we have carried out _ ab initio _ electronic structure calculations of the surface geometry and interlayer spacing for a set of vicinals of cu(100 ) and cu(111 ) . in addition to cu(211 ) , cu(331 ) and cu(511 ) which are included to address the question of the influence of the local geometry on the structure , we have extended the investigation to cu(711 ) and cu(911 ) to examine the influence of increasing terrace width of the relaxation pattern . of course , for all surfaces comparison with available experimental data is of prime concern . the rest of this report is organized as follows . in section ii , the system geometries are presented together with some computational details . section iii contains the results and their discussion . concluding remarks are presented in section iv . vicinal surfaces can easily be constructed by cutting the crystal at an angle slightly away from the lower - index crystal planes ( i.e. ( 100 ) , ( 111 ) , ( 110 ) ) . for reasons discussed above , we are interested here in the vicinals of the ( 100 ) and ( 111 ) surfaces of fcc metals of which the most tightly packed steps are along the @xmath3 direction . in the case of the ( 111 ) surface , however , the @xmath3 direction is not parallel to any plane of symmetry and there are two different ways of generating monoatomic stepped surfaces . in one type of such vicinals , the step edge has a ( 100)-microfacet , while the other has the ( 111)-microfacet ( these are the so called a and b types , respectively ) . in the standard nomenclature , the vicinals of fcc(111 ) surface with monoatomic steps and ( 100 ) step edges are denoted by @xmath4 , while those with ( 111 ) step edges are labeled as @xmath5 , where @xmath6 is the number of atoms on the terrace . the b - type vicinal cu(331 ) considered here is named accordingly , while the a - type vicinal cu(211 ) seems to be a misnomer . similarly , the vicinals of fcc(100 ) surface consisting of monoatomic step edges with ( 111 ) microfacet are labeled @xmath7 . the cu(111 ) vicinals considered here are created by cutting the crystal at an angle of 19.5@xmath8 and 22@xmath8 away from the ( 111 ) plane towards the [ @xmath9 and [ @xmath10 direction to produce the ( 211 ) and ( 331 ) surfaces , respectively , whereas the three vicinals of ( 100 ) , ( 511 ) , ( 711 ) , and ( 911 ) , are constructed by slicing the crystal at angles of 15.8@xmath8 , 11.4@xmath8 and 8.9@xmath8 , respectively , from the ( 100 ) plane towards the [ 011 ] direction . to facilitate the discussion we have also labeled the atoms that play the dominant role in our calculation @xcite . for the three surfaces consisting of three chains of atoms on the terrace we label them as corner - chain ( cc ) , terrace - chain ( tc1 ) , and step - chain ( sc ) . the chain just underneath the step - chain is called a bulk nearest neighbor chain ( bnn ) . the other two surfaces , cu(711 ) and cu(911 ) , contain , respectively , one and two extra chains of terrace atoms , labeled accordingly as ( tc2 ) and ( tc3 ) . we have taken the @xmath11 and @xmath12 axes to lie in the surface plane , the @xmath11-axis being perpendicular to the step and the @xmath12-axis along the step , and the @xmath13-axis is along the surface normal . in fig . 1 , we display a side view of the ( 511 ) surface with the appropriate labeling of the atoms and interlayer spacing . the _ ab initio _ electronic structure calculations are performed within a pseudopotential approach to density - functional theory in the local density approximation @xcite , numerical implementation of the technique is based on a computer code developed by b. meyer _ et al . _ @xcite the local - density approximation is applied using the hedin - lundqvist form of the exchange - correlation functional @xcite . a norm conserving pseudopotential for cu constructed according to a scheme proposed by hamann - schlueter - chiang @xcite has been used which has already been successfully employed for calculations of the structure and the phonons of low index surfaces of cu @xcite . a mixed basis set is applied to represent the valence states consisting of five d - type local functions at each cu site , smoothly cut off at a radius of 2.3 a.u . , and of plane waves with kinetic energy of 11 ry . the brillouin - zone(bz ) integration was carried out using the special point sampling technique @xcite together with a gaussian broadening of the energy levels of 0.2 ev . for simulating surfaces we used the supercell approach with cells containing 21 to 35 atoms ( 1 atom per layer ) , depending on the surface orientation . the z - dimension of all cells was 47.7155 a.u . the distance between the top and the bottom layer of the slabs were thus 31.2761 a.u . for cu(331 ) , 27.8277 a.u . for cu(211 ) , 26.2340 a.u . for cu(511 ) , 25.7715 a.u . for cu(711 ) , and 25.4390 a.u . for cu(911 ) . structure optimization was carried out until forces on all atoms were smaller than @xmath14 ry / a.u . , which is two orders of magnitude smaller than the forces present on the unrelaxed surfaces . with increasing terrace width , the calculations become increasingly tedious since the reduction in interlayer spacing makes it more difficult to achieve geometries converged to 1% of the interlayer spacing . we also find that results for terraces with ( 100 ) geometry are more sensitive to the number of k - points sampled , as compared to those with ( 111 ) geometry . for the latter case 30 points in the bz are sufficient for the determination of the equilibrium structure , while for surfaces with ( 100 ) terraces at least twice as many points are needed to get converged results . our results for the multilayer relaxations of cu(211 ) , cu(331 ) , cu(511 ) , cu(711 ) and cu(911 ) are summarized in table i. as in previous theoretical studies @xcite of relaxations on stepped cu surfaces , we find changes in interlayer separations , from bulk terminated configurations , to persist on all surfaces for a large number of layers . of course , the concept of layers is different here from that on flat surfaces ( see fig . 1 ) . the first _ n _ layers , for example , are all exposed to the vacuum , where _ n _ is the number of atoms on the terrace . correspondingly , the interlayer separations are small and even large percentage changes in interlayer separations correspond to small numbers in distances . it is important to bear this point in mind when comparing the results for a particular surface either with those for flat surfaces or with the results from other methods . a common feature of all surfaces examined in table i is that all terrace atoms except for the ones in the corner chain ( cc ) undergo significantly large inwards relaxations . the corner atoms are always found to relax outwards . additionally , the atoms in the terrace adjacent to cc exhibit comparatively large inward relaxation whose magnitude maybe larger than that of the step atoms ( sc ) . for example , for cu(511 ) , cu(711 ) , and cu(911 ) changes , respectively , in @xmath15 ( involving tc1 ) , @xmath16 ( involving tc2 ) , and @xmath17 ( involving tc3 ) , are larger than that of @xmath18 . thus , in keeping with schmolkowski s @xcite ideas of charge smooting , the maximum relative change in interlayer separation is focussed around the corner atoms . this is particularly true for the vicinals of cu(100 ) . the situation with the more closepacked surface cu(331 ) is somewhat different , as seen in table i the outward relaxation of the corner atom and the inward relaxation of the preceeding atom on the terrace on cu(331 ) are less than half of that for similar atoms on the other surfaces considered here . incidently , this conclusion is in good agreement with results from previous studies which were based on semi - empirical potentials @xcite . there is , however , a disconcerting difference in the results obtained here from _ ab initio _ electronic structure calculations and those from semi - empirical potentials . an intriguing result for multilayer relaxations of the vicinals of cu(100 ) obtained with eam potentials in ref . @xcite was that the pattern of inward and outward relaxations continued well into the bulk with an expontially decreasing amplitude . thus for cu(711 ) , the relaxation pattern predicted by eam was ( - - - + , - - - + , - - - + , ... ) with an eventual dampimg of the relaxations . the pattern for cu(711 ) from table i is instead ( - - - + , - - + + , - - + + ) . that is our present calculations do not predict a periodically oscillatory relaxation pattern with a decaying amplitude as we move into the bulk . as we shall see , this particular feature is more in agreement with experimental data and help remove the slight discrepancy between experiment and theory presented by the eam result pointed by walter et al . again , it should be recalled that the numbers involved are very small and within the limits of accuracy of _ ab initio _ calculations . in particular , the small numbers for the relaxations of the inner layers of cu(911 ) have to be taken with caution as our convergence criteria for this surface were not as good as that for the others because of the demands on computational resources imposed by a system as large as this one . unlike flat surfaces , vicinal surfaces relax in both @xmath11 and @xmath13 directions , since the existence of steps at the surface leads to broken symmetry in both of these directions . while relaxations along the @xmath13-direction yield characteristic interlayer separations that we have discussed above , those along the @xmath11-direction provide new registries of atoms , as compared to those in the bulk . our calculated percentage intralayer registries for five surfaces are summarized in table ii . as in the observations from eam calculations , the changes in the registries of the atoms are small . it is thus not useful to make a one - to - one comparison with results from semiempirical calculations . however , the changes in registries of the atoms are not inconsequential since they affect the changes in the bond lengths between the atoms in these regions of low coordinations . in table iii , we tabulate our results for the total changes in the distances between the step atoms and their nearest neighbors . for comparison we have included in parenthesis the results obtained earlier for the same quantities with eam potentials @xcite . the largest changes in the bond lengths ( from unrelaxed configurations ) are for those between the step atoms and their bulk nearest neighbour ( bnn ) which lies right below them . the bonds between cc and bnn show small enlargement , while all other bonds in table iii are found to undergo shortening . in figs . 2 and 3 , we have drawn the actual displacements of the atoms on the five surfaces obtained from our calculations . while the size of the arrows are exagerated , it is their relative length and direction that is of consequence . as already noted by durukanoglu et al . @xcite , all atoms in low coordinated sites move to enhance their local coordination . the complex displacement pattern that emerges is thus the net outcome of the competition between the different directions in which the various atoms would like to relax to enhance thein own coordination . for readers who are interested in the exact positions of the atoms on the relaxed surfaces , we have summarized them in table iv . the unusual behavior of cu(331 ) terrace atom is more apparent from this table than the earlier one on changes in the bond lengths . the tc1 atom of cu(331 ) undergoes the least displacement among its counterparts . its displacement is also smaller than that of tc2 ( 1.4 , 0.0 , -2.6 ) on cu(711 ) , and of both tc2 ( 0.8 , 0.0 , -2.0 ) and tc3 ( 0.1 , 0.0 , -2.3 ) on cu(911 ) . the coordinates of the displacements above in paranthesis are in the same units as those in table iv . we now turn to comparisons of the results obtained here for individual surfaces with those available from experimental measurements . in table v we show that for cu(211 ) the salient features in the trends in the relaxation patterns predicted by our calculations are observed in the experimental data . apart from the large inward relaxation of the step atoms , the major change occurs at the corner atom and its adjacent terrace atom . our results are in good agreement with previous dft / lda calculations @xcite , based on the pseudopotential approximation and with results from eam based method . theoretical calculations using the full potential linearized augmented plane wave ( flapw ) method @xcite , however , predict much larger relaxation(-28.4% ) of the step atom than any of the previous theoretical or experimental studies . this brings us to the discussion of ionic relaxation on cu(331 ) in table vi for which also we do not get the large relaxation reported in ref[@xcite ] . the present results for cu(331 ) are , however , in good agreement with the leed data @xcite . the discrepancy with the results from leed for @xmath15 should not be taken too seriously , given an error bar of 4% in the analysis of the leed data . with respect to eam based results @xcite , we find a noticeable difference for d@xmath19 , for which the present results agree better with the leed data and also preserve the predicted relaxation pattern ( - - + ) for the terrace atoms . this trend is inkeeping with what was reported in calculations on al(331 ) @xcite . in trying to reconcile our results with those of geng @xcite , we note that the latter predict an outward displacement of the tc1 atomic chain for both cu(211 ) and cu(331 ) , while we find this not to be the case . as already mentioned , while the changes in the bond lengths of the terrace atoms of cu(331 ) are no different from those of the other surfaces , the displacement of tc1 is strikingly smaller than that of the tc s on other surfaces . the case of multilayer relaxations for cu(511 ) is interesting because of the differences in the published data from leed @xcite and x - ray measurements @xcite . these are displayed in table vii . except for the displacement of the step atoms , for which all results point to a large inward relaxation , the results from x - ray scattering measurments are in disagreement with present results and with those from leed , as well as , from eam calculations . we do not understand the reasons for this disagreement but for the notion that x - ray measurements may be very sensitive to the quality of the crystal surface . it should be noted that the differences with the x - ray results are both qualitative and quantitative , beyond the established error bars in the experiments and calculations . because of the controversy in the experimentally determined multilayer relaxations of cu(511 ) , we have carried out an extensive analysis of the dependence of the theoretical results on the approximations necessary to produce computational feasibility : choice of pseudopotentials , maximum kinetic energy of the plane - waves ( e@xmath20 ) , the number of layers in the supercell , and the number of points used to sample the surface brillouin zone . as for the dependence of the results on the type of pseudopotentials and e@xmath20 , we have carried out calculations with three different pseudopotentials ( on three sets of codes ) to find that the results which lie within 3% of each other . we also find our choice of supercell size to be adequate . there is , however , a strong dependence of the results on the number of bz points sampled . for the case of cu(511 ) this dependence is illustrated in table viii . calculations performed with few points could give erroneous relaxations as signified by the case of d@xmath21 in table viii . an inward relaxation of 1.8% is found with 4 points , while the converged result is 10.7% . convergence in the calculated relaxation is reached once the number of points is increased to 24 and beyond . thus , when comparing results from _ ab initio _ calculations , one has to keep these technical points in mind . unless sufficient checks are made for convergence in the reported values , quantities like equilibrium positions of surface atoms may differ in different calculations and lead to disagreement in the calculated relaxations . it would be worthwhile to clarify whether the differences between our results and those from the flapw method for cu(211 ) and cu(331 ) could be attributed to k - points sampling . finally , we come to the comparison of our results for cu(711 ) with those from experiments ( we are not aware of any data on cu(911 ) , so far ) . the leed data @xcite for this surface has been very carefully analyzed and compared to existing calculations . table ix shows that the _ ab initio _ results obtained here are in excellent agreement with the data , and that the small differences with the eam results that the authors@xcite had noted , is removed by the present calculations . as in the case of cu(511 ) , the largest percentage change in the interlayer spacing is not for d@xmath22 . in this case it is for d@xmath19 which separates cc from tc2 . as before , there is outward relaxation of the spacing between cc and bnn . the fact that relaxations near cc persist on being strong even as the terrace width increases , is interesting in itself . this particular argument has not been made in any previous theoretical result . our calculated values for cu(911 ) further support this argument as the largest percentage change is found for d@xmath23 , the interlayer spacing between cc and tc3 ( in this case ) . while these results are intriguing the main outcome of relaxations that ensue when a surface is created is in the actual displacements of the atoms from their bulk terminated positions to the new equilibrium positions . as already stated , these values are summarized in table v and the related patterns presented in figs . 2 and 3 . obviously , for stepped surfaces there is a complex rearrangement of most terrace atoms . our calculations show that despite this complexity , all terrace atoms except for cc move inwards . in summary , we have performed a comparative study of multilayer and atomic relaxations of five stepped cu surfaces which are vicinals of cu(100 ) and cu(111 ) using _ ab initio _ electronic structure calculations based on density functional theory and non - local , norm - conserving pseudopotentials . the set of three of these surfaces : cu(211 ) , cu(331 ) , and cu(511 ) , provides a comparison of structural changes from bulk termination , for vicinals of similar terrace widths but differing local geometry . the other set consisting of cu(511 ) , cu(711 ) and cu(911 ) offers a comparative study of relaxation patterns with changing terrace width . in each case we find the relaxation of the step atoms to be pronounced inwards and that of the corner atom to be outwards . the other terrace atoms and their nearest neighbors also undergo relaxations following a complex displacement pattern . subsequently , the bond lengths between all the surface atoms and their nearest neighbors change from the bulk terminated values while the bond length between cc and bnn atoms experiences an elongation ( about 1% ) all other surface length shrink anywhere from 1% to 4% . most of our findings are in agreement with previous calculations which were based on semiempirical model potentials except that we do not find the pattern of inward relaxations of sc , tc1 , tc2 etc followed by outward relaxation of cc atoms to continue into inner layers . we also find that the percentage contraction of the spacing between the tc and cc atoms is generally larger than that between sc and the tc atoms . while the actual magnitudes of the changes in the spacing considered here are small , there is a systematic trend in the relaxation pattern which points to significant rearrangements in the electronic charge densities near the sc and cc atoms . by and large our results are in good agreement with available structural data on these surfaces , except for the case of cu(511 ) for which we favor the leed results over those from x - ray scattering meaurements . we believe our results will help settle the issues that have emerged on this particular surface . our systematic examination of five surfaces , also helps address the question wether the relaxations on cu(331 ) are anomalous . the only striking difference between this surface and the others is in the relaxation of tc1 which is very small . otherwise the relaxation pattern and the changes in bond lengths are similar to those on the other surfaces . the main message from these observations is that the important quantity to examine is the displacement pattern of the surface atoms as they relax to their equilibrium positions from their bulk terminated configurations . the deeper question , of course , is the nature of the accompanying changes in the surface electronic structure . it will be interesting to examine the characteristics of the local electronic densities of states in the different regions of low symmetry that are present naturally on the stepped surfaces considered here . we leave this as an exercise for the future . the work of tsr and ak was supported in part by the us national science foundation , grant che-9812397 and by the basic energy research division , department of energy , grant de - fg03 - 97er45650 . tsr also acknowledges the alexander von humboldt foundation for the award of a forschungspreis and thanks her colleagues at the fritz haber institut , berlin and at the forschungszentrum , karlsruhe for their warm hospitality . for a review see k. wandelt , surf . sci . * 251/252 * , 387 ( 1991 ) . r. smoluchowski , phys . rev . * 60 * , 661 ( 1941 ) ; see also m. w. finnis and v. heine , j. phys . f * 4 * , l37 ( 1974 ) . d. l. adams and c. s. sorensen , surf . sci . * 166 * , 495 ( 1986 ) ; f. jona and p. m. marcus , in : j . f. van der veen and m. a. van hove ( eds . ) , the structure of surfaces ii , springer , heidelberg , 1988 , p. 90 , and references therein . zhang , p. j. rous , j. m. maclaren , a. gonis , m. a. van hove , and g. a. somorjai , sur . sci . * 239 * , 103 ( 1990 ) . f. jona , p. m. marcus , e. zanazzi , and m. maglietta , sur . * 6 * , 355 ( 1999 ) , and references therein . d. a. walko and i. k. robinson , phys . b * 59 * , 15446 ( 1999 ) . zang , m. a. van hove , g. a. somorjai , p. j. rous , d. tobin , a. gonis , j. m. maclaren , k. heinz , m. michl , h. lindner , k. muller , m. ehsasi , and j. h. block , phys . lett . * 67 * , 1298 ( 1991 ) . j. s. nelson and p. j. feibelman , phys . lett . * 68 * , 2188 ( 1992 ) . z. j. tian and t. s. rahman , phys . b * 47 * 9752 ( 1993 ) . s. durukanolu , a. kara , and t. s. rahman , phys . b * 55 * , 13 894 ( 1997 ) . sklyadneva , g. g. rusina , and e. v. chulkov , surf . sci . * 416 * , 17 ( 1998 ) . s. m. foiles , m. i. baskes , and m. s. daw , phys . b * 33 * , 7983 ( 1986 ) . a. kara , p. staikov , t. s. rahman , j. radnik , r. biagi , and h. j. ernst , phys . b * 61 * , 5714 ( 2000 ) . th . seyller , r. d. diehl , and f. jona , j. vac . a * 17 * , 1635 ( 1999 ) . c. y. wei , s. p. lewis , e. j. mele , and a. m. rappe , phys . b * 57 * , 10062 ( 1998 ) . m. albrecht , r. blome , h. l. meyerheim , w. moritz , and i. k. robinson , unpublished . y. tian , k .- w . lin , and f. jona , phys . rev b * 62 * , 12844 ( 2000 ) s. walter , h. baier , m. weinelt , k. heinz and th . fauster , phys . b * 63 * , 155407 ( 2001 ) . g. allen and m. lannoo , phys . rev . b * 37 * , 2678 ( 1988 ) and references therein . p. hohenberg and w. kohn , phys . rev . b136 , 864 ( 1964 ) . b. meyer , c. elsaesser and m. faehnle , `` fortran90 programm for mixed - basis pseudopotential calculations for crystals '' , max planck institut fuer metallforschung , stuttgart , unpublished . l. hedin and b. i. lundqvist , j.phys.c * 4 * , 2064 ( 1971 ) . d. r. hamann , m. schlueter , and c. chiang , phys.rev.lett . * 43 * , 1494 ( 1979 ) ; g. b. bachelet , d. r. hamann , and m. schlueter , phys.rev.b * 26 * , 4199 ( 1982 ) . monkhorst , and j.d . pack , phys . b * 13 * , 5188 ( 1976 ) . rodach , k .- bohnen , and k. m. ho , surf.sci . * 286 * , 66(1993 ) . w. t. geng and a. j. freeman , phys . b * 64 * , 115401 ( 2001 ) .	we present trends in the multilayer relaxations of several vicinals of cu(100 ) and cu(111 ) of varying terrace widths and geometry . the electronic structure calculations are based on density functional theory in the local density approximation with norm - conserving , non - local pseudopotentials in the mixed basis representation . while relaxations continue for several layers , the major effect concentrates near the step and corner atoms . on all surfaces the step atoms contract inwards , in agreement with experimental findings . additionally , the corner atoms move outwards and the atoms in the adjacent chain undergo large inward relaxation . correspondingly , the largest contraction ( 4% ) is in the bond length between the step atom and its bulk nearest neighbor ( bnn ) , while that between the corner atom and bnn is somewhat enlarged . the surface atoms also display changes in registry of upto 1.5% . our results are in general in good agreement with leed data including the controversial case of cu(511 ) . subtle differences are found with results obtained from semi - empirical potentials . pacs # 61.50.ah , 68.35.bs , 68.47.de
modifying general relativity has been one of the most highlighted fields in gravitational physics in recent years . long distance modification of gravity has been studied extensively so as to explain the current accelerated expansion of the universe ( see , _ e.g. _ , ref . @xcite for a recent review ) . more theoretically , it is interesting to ask the simple question of whether one can consistently modify general relativity to accommodate , _ e.g. _ , massive gravitons @xcite . in many cases , modified theories of gravity can be described , at least effectively , by adding an extra scalar degree of freedom that participates in the dynamics of gravity . the most general lagrangian composed of the metric @xmath3 and a scalar @xmath4 and having second - order field equations will therefore be a powerful tool to study various aspects of modified gravity , and , interestingly , the theory was already constructed forty years ago by horndeski @xcite . recently , the horndeski theory was rediscovered @xcite and shown to be equivalent to the generalized galileon @xcite . while considerable attention has been devoted to cosmological applications of the horndeski theory , black holes in that theory have been less explored so far . in the context of scalar - tensor modification of gravity , one of the central questions to address is whether or not black holes can have scalar hair . it has been proven by hawking that a black hole can not have scalar hair in the brans - dicke theory @xcite . in the traditional scalar - tensor theory where @xmath4 is nonminimally coupled to gravity , the scalar - no - hair theorem was formulated in @xcite ( under the assumption of spherical symmetry ) , and a more general proof was provided recently in @xcite , while the no - hair theorem for a noncanonical scalar field , _ i.e. _ , k - essence , was given in @xcite . it is then natural to ask how those results can be extended to theories whose lagrangian contains second derivatives of @xmath4 . such theories are motivated by the galileon @xcite , for which the equation of motion still remains of second order . for the galileon coupled to gravity , hui and nicolis have shown that static and spherically symmetric black holes can not be surrounded by any nontrivial profile of the scalar field @xcite . the key of the proof of ref . @xcite is shift symmetry of the scalar field , _ i.e. _ , symmetry under @xmath0 , where @xmath5 is a constant , and the regularity of the square of the noether current associated with this symmetry . therefore , the same argument seems to hold for more general scalar - tensor theories with the same symmetry , though there are some loopholes . one loophole can be opened by abandoning the static configuration of @xmath4 and/or relaxing some asymptotic conditions on the metric and @xmath4 , and hairy black hole solutions have thus been constructed explicitly in ref . one can also circumvent the no - hair theorem by tuning the lagrangian @xcite . in light of the former loophole , exact black hole solutions with scalar hair have been found in the theory with nonminimal derivative coupling to the einstein tensor , @xmath6 @xcite . the term @xmath6 has shift symmetry and reflection symmetry , @xmath7 . the goal of this paper is to extend those previous works to go beyond this particular example , giving various exact black hole solutions with scalar hair in a subclass of the horndeski theory possessing shift and reflection symmetries . the theory contains two _ arbitrary _ functions of @xmath2 , and we will provide a variety of solutions without specifying the concrete form of those functions . the paper is organized as follows . in the next section , we present the theory and the black hole ansatz considered in this paper . in secs . 3 and 4 , we give various exact solutions with scalar hair . the regularity of our solutions is discussed in sec . 5 . finally , we conclude in sec . we consider a shift symmetric subclass of the horndeski theory whose lagrangian is given by @xmath8,\label{lagrangian}\end{aligned}\ ] ] where @xmath9 and @xmath10 are arbitrary functions of @xmath11 , and @xmath12 . the most general shift symmetric scalar - tensor theory with second - order field equations can accommodate two more arbitrary functions of @xmath11 , often denoted as @xmath13 and @xmath14 in the literature . however , we restrict ourselves to the theory possessing reflection symmetry as well , @xmath1 , which forbids the @xmath15 and @xmath16 terms . we thus focus on the lagrangian ( [ lagrangian ] ) describing the scalar - tensor theory with shift and reflection symmetries . since @xmath17 up to a total divergence , we notice that the theory considered in refs . @xcite corresponds to the specific case with @xmath18 where @xmath19 , and @xmath20 are constant parameters . ( see ref . @xcite for black hole solutions of the theory ( [ babitheory ] ) in the presence of an electromagnetic field . ) in this paper , we go beyond the specific theory ( [ babitheory ] ) , leaving @xmath21 and @xmath22 arbitrary . as shown in the following , exact black hole solutions with a nontrivial configuration of @xmath4 can still be constructed . variation with respect to the metric yields the gravitational field equations , @xmath23 shift symmetry of the theory allows us to write the scalar - field equation of motion as a current conservation equation , @xmath24 where @xmath25\nabla^\mu\phi \nonumber\\ & & -2g_{4xx}\left(\box\phi\nabla^\mu x-\nabla^\mu\nabla^\nu\phi\nabla_\nu x\right).\end{aligned}\ ] ] the metric we are going to study is of the form @xmath26 where @xmath27 is the metric of a unit two - dimensional sphere , plane , or hyperboloid for @xmath28 , respectively . we take the following @xmath29-dependent ansatz for the scalar field @xcite , @xmath30 for which @xmath31 \label{x - psi}\end{aligned}\ ] ] is , however , independent of @xmath29 . here and hereafter a prime denotes differentiation with respect to @xmath32 . since @xmath9 and @xmath10 are the functions of @xmath11 only , it is more convenient to use @xmath11 rather than @xmath33 when writing the field equations . note in passing that , if @xmath34 , @xmath35 can be chosen arbitrarily by rescaling the time coordinate : @xmath36 , @xmath37 , and @xmath38 with constant @xmath39 . this rescaling keeps the line element and @xmath11 invariant . we will closely follow ref . @xcite to construct black hole solutions dressed with a time - dependent scalar , @xmath34 . an explicit calculation using the ansatz ( [ phi - ansatz ] ) shows that @xmath40f\right\ } , \,\,\,\,\,\,\\ j^t&=&-\frac{qj^r}{fh\psi'}-\frac{2q}{r}\left [ g_{4x}\left(\frac{f}{h}\right)'+2g_{4x}'\frac{f}{h}\right ] , \label{jt } \,\,\,\,\,\,\end{aligned}\ ] ] and @xmath41 , where @xmath42 stands for the other two components of the shift current @xmath43 . since the @xmath44-component of the gravitational field equations can be written as @xmath45 this equation and the scalar - field equation of motion ( [ seom ] ) are satisfied by imposing @xmath46 now one can see from eq . ( [ jt ] ) that @xmath47 does not diverge at the horizon as long as @xmath48 . using eq . ( [ solf ] ) , we find @xmath49,\end{aligned}\ ] ] implying that @xmath50(rh)'=\frac{q^2}{2x}\left[k - r^2\lambda(x)\right ] . \label{dh}\end{aligned}\ ] ] here we have defined the functions of @xmath11 as @xmath51 where @xmath52 combining eqs . ( [ solf ] ) and ( [ dh ] ) , one notices the relation @xmath53h.\label{rel - h - f}\end{aligned}\ ] ] one can exploit the same set of equations , ( [ solf ] ) and ( [ dh ] ) , to write the @xmath54-component of the gravitational field equations simply as @xmath55=0.\label{ett}\end{aligned}\ ] ] this equation can be integrated to give @xmath56=c,\label{int - ett}\end{aligned}\ ] ] where @xmath57 is an integration constant . equation ( [ int - ett ] ) determines @xmath58 _ algebraically_. then , one can integrate eq . ( [ dh ] ) to determine @xmath59 , and finally obtain @xmath60 from eq . ( [ solf ] ) , or , more straightforwardly from eq . ( [ rel - h - f ] ) , for given @xmath21 and @xmath22 . the angular equations are written as @xmath61 using the equations derived above , and hence are fulfilled automatically . a particularly simple solution of eq . ( [ int - ett ] ) is obtained if @xmath57 is chosen in such a way that @xmath63 where @xmath64 is a _ constant _ satisfying @xmath65 in this case , @xmath66 trivially solves eq . ( [ int - ett ] ) . then , assuming that @xmath67 one can integrate eq . ( [ dh ] ) immediately to get @xmath68 where @xmath69 is an integration constant . ( for @xmath70 , eq . ( [ dh ] ) can not be satisfied in general . ) substituting @xmath71 to eq . ( [ rel - h - f ] ) , we arrive at @xmath72 and therefore @xmath64 must be such that @xmath73 for the metric to be lorentzian . we can then rescale the time coordinate to set @xmath74 . the final form of the solution is thus @xmath75 this is identical to the black hole metric in the presence of the ( effective ) cosmological constant @xmath76 , though it exhibits a nontrivial profile of @xmath77 . interestingly , @xmath78 can be nonzero even in the case where the true cosmological constant ( which could be included in @xmath9 ) vanishes . the same metric with an effective cosmological constant has been constructed earlier in ref . @xcite for the theory ( [ babitheory ] ) . the radial profile of the scalar field for the above solution is given by integrating @xmath79 in order for @xmath33 to be real , we must require that @xmath80 . this prohibits negative @xmath78 ( with @xmath81 ) . since @xmath82 diverges as @xmath83 , one would be concerned about the regularity of the scalar field at the horizon . this issue will be discussed in detail in sec . [ sec : regularity ] . however , we would emphasize here that the action depends on the derivatives of the scalar field rather than @xmath4 itself , and we see that the coordinate scalar quantities constructed from derivatives , such as @xmath11 and @xmath84 , are regular everywhere . since the de sitter - schwarzschild metric solves the field equations as shown above , one can consider its extremal limit , _ i.e. _ , the nariai spacetime @xcite . in this nariai limit , the spacetime between the black - hole and cosmological horizons can be described by @xmath85 where @xmath86 and the horizons are located at @xmath87 . the new coordinates @xmath88 and @xmath89 are related to the original ones by @xmath90 and @xmath91 , where the limit @xmath92 is taken keeping @xmath88 and @xmath89 fixed . in these coordinates , the scalar field is written as @xmath93,\label{nariai - scalar}\end{aligned}\ ] ] where @xmath94 is kept fixed in the @xmath92 limit . even though @xmath4 shows a nontrivial profile and in particular @xmath95 as @xmath96 , we see that @xmath97 and @xmath98 everywhere . we will discuss the regularity at the horizons in sec . [ sec : regularity ] . another class of @xmath100 const solutions satisfying eq . ( [ ett ] ) can be obtained by choosing @xmath101 , where @xmath102 . this gives the solution of the form @xmath103 -\frac{\mu}{r},\label{newsol}\end{aligned}\ ] ] where @xmath104 with @xmath105 . we find @xmath106 by using eq . ( [ rel - h - f ] ) , and then @xmath82 by integrating eq . ( [ x - psi ] ) . let us investigate the solution ( [ newsol ] ) closely for the three respective cases , @xmath107 , @xmath108 , and @xmath109 . if @xmath107 , @xmath110 , and @xmath81 , curvature singularities occur at @xmath111 and @xmath112 . however , those singularities can be hidden behind horizons . this is indeed the case if @xmath59 has two roots in the interval @xmath113 , which is realized for @xmath114 , with not too large @xmath115 . the geometry between the two zeros is similar to de sitter - schwartzschild . in the case of @xmath107 , one can also consider the region @xmath116 , with the singularity at @xmath112 being hidden inside the horizon . such solutions with @xmath117 outside the horizon are obtained under either of the following conditions : @xmath118 , @xmath119 , @xmath120 with sufficiently large @xmath69 ; @xmath118 , @xmath121 , @xmath120 with arbitrary @xmath69 ; @xmath122 , @xmath123 , @xmath124 with arbitrary @xmath69 . for @xmath108 and @xmath81 , @xmath125 and @xmath126 are positive outside the horizon provided that @xmath127 and @xmath128 . in this case we are allowed to set @xmath129 and the solution ( [ newsol ] ) reduces simply to @xmath130 if @xmath109 and @xmath131 , a curvature singularity occurs only at @xmath111 . a solution for which @xmath132 everywhere outside the horizon can be obtained if , for example , @xmath133 , @xmath127 , and @xmath128 . in the case of @xmath134 the solution is regular at the origin . for example , for @xmath118 , @xmath135 , and @xmath136 , @xmath125 and @xmath126 are positive only in a finite region around the origin , and the geometry is similar to the static region of the de sitter spacetime . finally , the special case with @xmath81 , @xmath118 , @xmath137 , @xmath124 ( and hence @xmath107 ) reproduces the schwarzschild black hole in an einstein static universe and is discussed in ref . @xcite . let us take a look at the theory with @xmath138 . in this case , we have @xmath139 for any @xmath10 , and hence eq . ( [ int - ett ] ) admits _ only _ the @xmath100 const solutions . taking @xmath140 , from eqs . ( [ dh ] ) and ( [ rel - h - f ] ) we obtain @xmath141 for @xmath142 . therefore , a stealth schwarzschild black hole solution can be obtained for more general @xmath10 than taken in @xcite provided that @xmath138 . for @xmath143 ( @xmath120 ) , ( [ solf ] ) implies that @xmath144 ( @xmath145 ) , and therefore we do not have a sensible solution unless @xmath142 . for @xmath146 , eqs . ( [ solf ] ) and ( [ dh ] ) reduce , respectively , to @xmath147 and @xmath148 where @xmath149 is assumed . the second equation determines @xmath58 algebraically . the @xmath54-component of the field equations for @xmath146 is given by @xmath150=0 . \label{q0ett}\end{aligned}\ ] ] we obtain @xmath60 by integrating eq . ( [ q0ett ] ) . then , eq . ( [ q0f ] ) is used to fix @xmath59 . equation ( [ kr2f ] ) implies that @xmath11 must be dependent on @xmath32 for @xmath67 . in this case , the explicit form of the metric is dependent on the concrete form of @xmath9 and @xmath10 . for example , in the theory ( [ babitheory ] ) , eq . ( [ kr2f ] ) yields a linear equation in @xmath11 , which can be solved to give the solution presented in refs . @xcite . for @xmath143 , however , eq . ( [ kr2f ] ) forces @xmath11 to be constant , @xmath66 . then , from eqs . ( [ q0f ] ) and ( [ q0ett ] ) we obtain @xmath151 where we used @xmath152 . this is a planar anti - de sitter black hole metric ( for @xmath153 ) . the profile of the scalar field is given by @xmath154 , implying that @xmath155 as @xmath156 , though its derivative is finite . let us finally consider the special class of @xmath157 const solutions with @xmath146 , @xmath143 satisfying @xmath158 ( this condition is consistent with @xmath71 . ) in such theories , eq . ( [ q0ett ] ) is trivially fulfilled . one can then take _ any _ @xmath125 and @xmath126 provided that the two functions satisfy eq . ( [ q0f ] ) . for example , eq . ( [ q0f ] ) admits the following solution : @xmath159 where @xmath160 and @xmath161 is arbitrary . a particular case @xmath162 corresponds to the lifshitz black hole solution with the dynamical exponent @xmath89 @xcite , though it was originally constructed in the specific theory ( [ babitheory ] ) . as another example , it is also easy to check that @xmath163 fulfill eq . ( [ q0f ] ) . this solution gives the same geometry as obtained from eq . ( [ nonafsol ] ) . we are now in position to check the regularity of the scalar field at the horizon , @xmath164 . suppose that @xmath125 and @xmath126 are expanded near the horizon as @xmath165 where @xmath166 and @xmath167 are constants . so far we have concentrated mainly on @xmath100 const solutions . it turns out , however , that the regularity at the horizon does not depend on this property . one can determine @xmath11 from eq . ( [ int - ett ] ) ( @xmath34 ) or eq . ( [ kr2f ] ) ( @xmath146 ) , and @xmath11 thus obtained may be constant or may be @xmath32-dependent . in any case , since those algebraic equations do not depend on the metric functions explicitly , nothing special happens to @xmath11 at the horizon . this fact allows us to write @xmath168 where the constant @xmath169 is fixed by @xmath170=c$ ] for @xmath34 and @xmath171 for @xmath146 . substituting eqs . ( [ expandfh ] ) and ( [ expandx ] ) to eq . ( [ jt ] ) , one can see that @xmath84 is also finite at the horizon . thus , all the solutions we have found in this paper are regular in the sense that not only the spacetime but also the coordinate invariants constructed from the derivatives of scalar field , @xmath11 and @xmath84 , are regular at the horizon . let us then examine the behavior of @xmath4 itself near the horizon , though the action depends on the scalar through @xmath172 due to shift symmetry . from eq . ( [ x - psi ] ) , it is found in the @xmath34 case that @xmath173 apparently , @xmath4 diverges logarithmically as @xmath174 . however , as in @xcite , one may introduce the ingoing eddington finkelstein coordinates @xmath175 defined by @xmath176 to write @xmath177 near the horizon , where the plus sign was chosen in eq . ( [ psi - sol ] ) . thus , we see that the scalar field is in fact regular at @xmath164 . note that if the solution exhibits a cosmological horizon in addition to the black hole horizon at @xmath164 then it will correspond to @xmath178 and hence @xmath4 is no longer regular there . in any case , it should be emphasized again that the coordinate invariants constructed from @xmath4 s derivatives are regular , as we noted above . the argument here also applies to the @xmath62 solutions in the nariai limit by replacing @xmath32 with @xmath89 . in the @xmath146 case , it is easy to see @xmath179^{1/2}$ ] near the horizon , and hence @xmath4 is regular at the horizon . as seen in the previous section , @xmath180 as @xmath156 for the solutions in this class . in this paper , we have derived a variety of exact black hole solutions in a subclass of horndeski s scalar - tensor theory having shift and reflection symmetries . assuming the time - dependent ansatz for the scalar field @xcite , @xmath181 , and following the method of constructing solutions of ref . @xcite , we have obtained a solution describing a black hole in the presence of a cosmological constant and other nontrivial solutions without fixing the concrete form of the two arbitrary functions in the theory . this was made possible because the solutions we have explored have the key property : @xmath100 const . our solutions circumvent the no - hair theorem because the scalar field itself is not static or it diverges at infinity . note that the action depends only on the derivatives of @xmath4 due to shift symmetry , and as a consequence the spacetime is static and the coordinate invariants constructed from the derivatives of the scalar field are regular . there are a lot of remaining issues to be addressed . one of the basic questions regarding a black hole solution is whether or not it is stable . it would therefore be interesting to study the stability issue by extending the black hole perturbation theory in horndeski s scalar - tensor gravity formulated in the static background @xcite . it would also be worth trying to generalize our solutions to rotating ones as has been done for a conformal scalar field @xcite , or to black holes with a realistic matter distribution , as in ref . @xcite for typical modified gravity theories with a screening mechanism . we primarily focused on @xmath100 const solutions in this paper , even though there could be many interesting solutions with @xmath32-dependent @xmath11 . it would be useful to clarify the phase space of possible solutions and to discuss their phenomenological properties as in refs . the theory we have studied admits superluminal propagations . it is crucial to study whether such superluminal propagations appear around the present black hole background in order to understand their causal structures and the influence of singularities on the region exterior to the horizon . finally , we would like to remark that the euclidean version of the theory ( [ lagrangian ] ) has been used in constructing a mechanism of emergent lorentz signature @xcite . our black hole solutions may help to give an insight about such a novel scenario . we would like to thank christos charmousis and eugeny babichev for pointing out an erroneous statement in the first version of the paper . we also thank hayato motohashi for a comment on the earlier draft . this work was supported in part by jsps grant - in - aid for young scientists ( b ) no . 24740161 ( t.k . ) . the work of n.t . is supported in part by world premier international research center initiative ( wpi initiative ) , mext , japan , and jsps grant - in - aid for scientific research 25@xmath182755 .	we derive a variety of exact black hole solutions in a subclass of horndeski s scalar - tensor theory possessing shift symmetry , @xmath0 , and reflection symmetry , @xmath1 . the theory admits two arbitrary functions of @xmath2 , and our solutions are constructed without specifying the concrete form of the two functions , implying that black hole solutions in specific scalar - tensor theories found in the literature can be extended to a more general class of theories with shift symmetry . our solutions include a black hole in the presence of an effective cosmological constant , the nariai spacetime , the lifshitz black hole , and other nontrivial solutions , all of which exhibit nonconstant scalar - field profile .
topological insulator @xcite ( ti ) materials host on their boundaries a novel type of topological states of quantum matter , which , unlike the quantum hall state , exist without the breaking of time - reversal symmetry . @xcite theoretical prediction and subsequent experimental demonstration of these topological states in both two- @xcite ( 2d ) and three - dimensional @xcite ( 3d ) systems have given rise to what is now one of the most rapidly developing fields in condensed matter physics . apart from providing a test platform for fundamental concepts , the study of tis holds promise for novel applications in materials science and chemistry , @xcite spintronics @xcite and quantum computation . @xcite however , to be able to fully explore the potential of tis , it is essential to have a detailed knowledge of the nature and properties of topological surface states in real ti materials , @xcite as well as a quantitative understanding of how they respond to external perturbations . @xcite experimentally , these questions are being addressed with advanced surface - sensitive experimental probes , such as spin- and angle - resolved photoemission spectroscopy @xcite [ ( sr)-arpes ] and scanning tunneling microscopy @xcite ( stm ) . along with experimental advances , there is a growing need for atomistic modeling of tis that would enable quantitative predictions and direct comparison with experiment . significant progress has been made in using _ ab initio _ methods to calculate electronic @xcite and magnetic @xcite properties of tis . however , such methods suffer from severe computational limitations , particularly in the case of slab geometry as well as surface supercell calculations , which are employed in studies of impurity - doping effects . in addition , more accurate _ ab initio _ methods often lack the conceptual transparency and flexibility of the model hamiltonian approaches , which have been of fundamental importance for driving progress in this research field . @xcite microscopic tight - binding ( tb ) models , which have already proved successful in quantitative description of electronic and magnetic properties of semiconductors , @xcite may provide a convenient platform to address similar issues in tis . several studies have recently appeared in the literature , in which tb descriptions with different level of complexity have been introduced , ranging from models built on a simplified lattice structure @xcite or a restricted orbital basis set inferred from symmetry arguments @xcite to fully microscopic models , with parameters extracted from density functional theory ( dft ) . @xcite to date , the latter class of models is still the least represented among the model hamiltonian approaches to tis . in this work we employ a microscopic tb model to study the properties of surface states in bi@xmath0se@xmath1 , a prototypical 3d ti , which belongs , along with bi@xmath0te@xmath1 and sb@xmath0te@xmath1 , to the family of binary tetradymite semiconductors with layered structure . @xcite although these materials have been studied for decades due to their excellent thermoelectric properties , @xcite they have recently attracted considerable attention as 3d tis , e.g. materials that exhibit topologically protected conducting surface states with linear ( dirac ) dispersion and helical spin - texture , traversing the bulk insulating gap . @xcite due to a relatively large band gap ( 0.3 ev for bi@xmath0se@xmath1 ) and rather simple surface states , consisting of a single dirac cone , @xcite the bi@xmath0se@xmath1 family of 3d tis is the most studied both experimentally and theoretically . our treatment is based on the _ _ sp__@xmath2 slater - koster hamiltonian . @xcite we use the parametrization developed by kobayashi , @xcite by fitting to dft calculations . throughout this work , our strategy has been to make use of the computational efficiency and simplifications , offered by the tb approach , in order to investigate key features of the surface states in bi@xmath0se@xmath1 3d ti , which are inaccessible by _ ab initio _ methods . importantly , we consider slabs with thicknesses ranging from 1 to 100 quintuple layers ( qls ) , which corresponds to length scales in the range of 1 - 100 nm . in contrast , thicknesses typically investigated in _ ab - initio_based studies do not exceed several quintuple layers . @xcite in agreement with previous reports , @xcite we find a gap due to interaction between opposite surfaces , which decreases with increasing the slab thickness . starting from 5 qls , the size of the gap becomes smaller than @xmath3 ev , and one can identify surface states with linear dispersion and helical spin - texture . for each slab thickness we determine the surface character of bloch states using the procedure put forward in ref . , i.e. based on the contribution of the real - space projected wavefunction onto the two surfaces of the slab . explicit calculations of the atomic- and orbital - projections of the wavefunctions , associated with valence and conduction band extrema in both bulk and slab geometry , allowed us to construct a phenomenological picture of band inversion . the latter effect is induced by spin - orbit interaction and is responsible for the occurrence of topological surface states across the bulk insulating gap . @xcite furthermore , based on a similar analysis , we were able to track the changes in the spatial distribution and the spin character of the surface states wavefunctions at and in the vicinity of the dirac point , for increasing slab thickness . our calculations showed that the states corresponding to top and bottom surfaces become completely decoupled , i.e. spatially separated , only for very thick slabs containing @xmath4 qls . we also calculated the spin - orientation of the surface states in momentum space as a function of thickness . the disturbances in the helical spin - texture , expressed through in - plane and out - of - plane tilting angles of the spin , are shown to be significant for thin slabs up to 10 qls and are manifestations of both inter - surface interaction and the proximity to bulk states . the rest of the paper is organized as follows . in section [ theo ] we discuss the details of the tb model and some computational aspects . the results of the simulations are presented in section [ results ] . we begin in section [ bulk ] by analyzing the effect of so on the bulk bandstructures , where the bulk band inversion is interpreted as a characteristic change in the spatial distributions of the orbital - resolved projections of the wavefunctions at conductance and valence band extrema at the @xmath5 point . the results of surface bandstructure calculations are presented in section [ slab ] , in particular the spatial character of the states emerging within the bulk gap and the opening of the gap at finite thicknesses are described quantitatively in section [ gap ] . we also discuss the procedure to identify the inverted character on conduction and valence bands in the presence of so for a slab geometry . in section [ wf_atg ] , based on the analysis of the wavefunctions of the states at the dirac point , a quantitative criterion of an infinitely thick slab , with no interaction between the surfaces , is formulated . in addition , we comment on the orbital character of the wavefunctions at the dirac point , especially on the non - negligible contribution of the in - plane @xmath6-orbitals . the spin - properties of the dirac - cone states and deviations from the perfect spin - momentum locking due to inter - surface interaction and the presence of bulk states are calculated in section [ spin ] . finally , section [ concl ] contains our conclusions . we begin with a brief description of the crystal structure and chemical bonding of materials belonging to the bi@xmath0se@xmath1 family . bi@xmath0se@xmath1 has a rhombohedral crystal structure with crystallographic @xmath7 [ @xmath8 space group , with five atoms in the unit cell [ see fig . [ fig0](a ) ] . the crystal is formed by stacking of hexagonal monolayers of either bi or se in a close - packed fcc ( abc ) fashion . the sequence of five atomic layers , i.e. se1-bi - se2-bi - se1 , forms a ql [ see fig . [ fig0](b ) ] , with se1 and se2 indicating two nonequivalent positions of se atom . hence it is convenient to describe the crystal structure of bulk bi@xmath0se@xmath1 in terms of qls stacked along the direction perpendicular to atomic planes ( _ c_-axis ) . ( a ) rhombohedral unit cell of the bulk crystal . se1 and se2 refer to two non - equivalent positions of se atoms . ( b ) the structure of 1 ql formed by five alternating atomic layers of bi and se . ( c ) the brillouin zone of the bulk crystal and that of the ( 111 ) surface of bi@xmath0se@xmath1 . ] the chemical bonding between atoms in the atomic layers within a ql is of covalent - ionic type with dominant covalent character . adjacent qls are weakly bound through se1-se1 bonds by van der waals forces , @xcite allowing easy cleavage of bi@xmath0se@xmath1 on the ( 111 ) se surface plane . based on the electronic configurations of bi and se , with the outermost orbitals being of @xmath6-character for both types of atoms , it is reasonable to expect that the bonding within qls is mainly due to interactions between @xmath6-orbitals . in fact , a simple but useful physical picture of the chemical bonding present in qls features @xmath9-chains formed by strongly interacting @xmath6-orbitals of atoms in nearest - neighbors atomic layers . evidence of this coupling can be indirectly inferred from stm topographies of bi - antisites defects in bi@xmath0se@xmath1 . @xcite in sections [ bulk ] and [ slab ] , we will comment on the orbital character of the valence and conduction band extrema in bulk bi@xmath0se@xmath1 and of surface states in bi@xmath0se@xmath1 slab , respectively . for modeling of the electronic structure of bi@xmath0se@xmath1 we employ the _ _ sp__@xmath2 tb model with slater - koster parameters obtained by kobayashi @xcite by fitting to bulk bandstructures calculated with dft . interactions between atoms in the same atomic layer and between atoms in first and second nearest - neighbor layers are included . the spin - orbit interaction ( so ) , which is the key element leading to a non - trivial bulk band structure and metallic surface states , is incorporated in the intra - atomic matrix elements . @xcite thus the hamiltonian of the system reads @xmath10 where @xmath11 is the reciprocal - lattice vector that spans the brillouin zone , @xmath12 is the atomic index , @xmath13 labels atomic orbitals @xmath14 , @xmath15 , @xmath16 , @xmath17 , and @xmath18 denotes the spin . here @xmath19 refers to the atomic positions in the unit cell , while @xmath20@xmath21@xmath19 runs over all neighbors of atom @xmath19 , including atoms in the adject cells , with @xmath22 being the vector connecting the two atoms ( @xmath22=@xmath23 for @xmath19=@xmath20 ) . the coefficients @xmath24 are the slater - koster parameters ( for @xmath19=@xmath20 @xmath25 give the on - site energies ) and @xmath26 is the creation(annihilation ) operator for an electron with spin @xmath27 at the atomic orbital @xmath28 of site @xmath19 . the second term in eq . ( [ eq:1 ] ) represents the on - site so , where @xmath29 are spin- and orbital - resolved atomic orbitals , @xmath30 is the orbital angular momentum operator and @xmath31 is the spin operator ; @xmath32 is the so strength . we refer to ref . for the exact parametrization of @xmath24 and @xmath32 . in bulk - bandstructure calculations we use the rhombohedral unit cell with five nonequivalent atoms , with the cell repeated periodically in @xmath33- , @xmath34- and @xmath35-directions . in calculations involving the bi@xmath0se@xmath1 ( @xmath36 ) surface we consider a slab consisting of @xmath37 quintuple layers , or , equivalently , of @xmath38 atomic layers . since each atomic layer is an equilateral triangular lattice , it is sufficient to assign one atom per each layer , which gives a total of @xmath38 atoms in the slab unit cell . the slab is finite along the @xmath35-direction ( ql - stacking axis ) , with the unit cell repeated periodically in the @xmath33-@xmath34 plane . using this tb model we were able to compute bandstructures of slabs with thicknesses up to @xmath39 qls with a reasonable computational cost . the presence of gapless edge or surface states , robust against time - reversal - invariant perturbations , distinguishes a topological insulating phase from a trivial one . however , in order to elucidate the origin of the topological order in existing tis and to facilitate the search for new ti materials , it is necessary to have a set of criteria that allow us to differentiate between topologically trivial and non - trivial insulators , based on the information about their bulk properties . within the framework of topological band theory , @xcite time - reversal - invariant insulators are classified according to a @xmath40 topological invariant , assigned to their bandstructure . in 2d there is a single @xmath40 invariant , which distinguishes the quantum spin - hall state ( 2d ti ) , as in hgte / cdte quantum wells , from an ordinary insulator . in 3d a set of four @xmath40 invariants leads to a classification based of three classes , namely strong tis , weak tis and ordinary insulators . for systems with inversion symmetry , the @xmath40 invariant is determined by the product of the parity eigenvalues of occupied bands at the time - reversal - invariant momenta in the brillouin zone . @xcite using this scheme , a few strong tis have been predicted , such as strained @xmath28-sn and hgte , bi@xmath41sb@xmath42 and bi@xmath0se@xmath1-like 3d tis . the emergence of topological order can be understood using the concept of _ band inversion_. @xcite clearly , a necessary condition for a ti is the existence of a non - trivial bulk band gap . in fact , in the above mentioned materials , a topological phase transition , or alternatively a change in the value of the @xmath40 invariant which in turn implies the existence of gapless states on the boundary , is accompanied by a visible and well - defined change in the bandstructure . this change occurs precisely in the insulating gap and can be observed as a function of an external or an intrinsic parameter . in hgte , which is a zero - gap semiconductor but acquires a gap due to external potential in a quantum well structure , the bands with @xmath43- and @xmath6-character are inverted with respect to their usual sequence at the @xmath44 point . @xcite a similar mechanism is realized in strained @xmath28-sn , which has recently been shown to exhibit a 3d ti phase , with dirac - like surface states emanating from the second lowest valence band across the strained - induced gap . @xcite in bi@xmath41sb@xmath42 alloy the inverted bandstructure is characterized by a change in the order of the bands with even ( @xmath45 ) and odd ( @xmath46 ) parity at the @xmath47 point of the brillouin zone compared to pure bi . this inversion is induced by sb doping and is due to the strong ti character of the valence band of pure sb . @xcite in bi@xmath0se@xmath1 3d ti a non - trivial bulk band gap owns its existence to so , with the parity of the valence and conduction band inverted at the @xmath44 point . @xcite without so the material would be a trivial insulator . however , in the presence of so the change in the parity of one of the occupied bands leads to a change in the value of the @xmath40 invariant , signaling a topological phase transition . an intuitive way of describing band inversion theoretically is to look at charge density distributions of the inverted bands at the point of the brillouin zone , where the inversion is expected to occur , as functions of a characteristic parameter . @xcite here we apply this procedure to bulk bi@xmath0se@xmath1 . the calculated bandstructures without and with so are shown in fig . [ fig1](a ) and ( b ) , respectively . the effect of so is clearly seen as an increase of the gap and a change in the curvature of the valence band at the @xmath44 point . se@xmath1 ( a ) without and ( b ) with spin - orbit interaction . dashed lines show the position of the fermi level , taken as a reference energy . filled(open ) circles indicate the position of vbm(cbm ) at the @xmath44 point of the bulk brillouin zone . the brillouin zone of the bulk crystal and the path along the high - symmetry directions are depicted in fig . [ fig0](c ) . ] to further quantify the changes in the valence and conduction band character , we calculate the atomic- and orbital - projections of the wavefunctions associated with the eigenvalues at the valence - band minimum ( vbm ) [ filled circles in fig . [ fig1 ] ] , and the conduction - band maximum ( cbm ) [ open circles in fig . [ fig1 ] ] at the @xmath44 point . the orbital - resolved projection of the wavefunction @xmath48 , corresponding to an eigenvalue @xmath49 of the tb hamiltonian in eq . ( [ eq:1 ] ) , onto atomic orbital @xmath50 is calculated as @xmath51 . the total weight of the wavefunction at the atomic site @xmath19 is then given by @xmath52 for all five atoms of the bulk unit cell . point in bulk bi@xmath0se@xmath1 ( a ) without and ( b ) with spin - orbit interaction . ] the most noticeable feature of fig . [ fig2 ] is the _ inversion _ in the spatial distribution of the wavefunctions when so is switched on . without so the cbm wavefunction has a node at se2 , which is in fact the inversion center of the bulk crystal , with the maximum weight on the two bi atoms . in contrast , the maximum weight of the vbm wavefunction is distributed over se2 and two se1 atoms . with so the situation is the opposite : now the vbm wavefunction has a node at se2 while the cbm wavefunction is predominantly localized on se atoms . as expected , both vbm and cbm states are predominantly of @xmath6-character . the @xmath43-orbital contribution to the wavefunctions is approximately @xmath53 for calculations with or without so and the @xmath43-projections are not affected by the inversion . note that in the presence of so , the states at the @xmath44 point are mostly originating from @xmath54 orbitals , however the @xmath55 ( in - plane ) contribution is not negligible ( @xmath56 for vbm and @xmath57 for cbm ) . in the next section , we will discuss the results of bandstructure calculations for bi@xmath0se@xmath1 slabs , namely the emergence of conducting states across the inverted bulk gap in the presence of so . in particular , we will show that the band inversion , characteristic of the bulk gap , can also be identified in the surface bandstructures . the bandstructures of bi@xmath0se@xmath1 slab of varying thickness , calculated using the tb model including so , are shown in fig . [ fig3 ] . a clear and sizable gap is found for 1 and 2 qls ( @xmath58=@xmath59 ev and @xmath58=@xmath60 ev , respectively ) . starting from 3 qls , two surface bands , resembling the dirac states , extend in the range of [ -0.1;0.4 ] ev , with valence and conduction bands beginning to form below and above this range . already for 3 and 4 qls the two bands appear to be almost touching at the @xmath5 point , however the calculated gap is not negligible and is found to be @xmath58=@xmath61 ev for 3ql and @xmath58=@xmath62 ev for 4 qls . the presence of the gap due to finite thickness is a recently discovered feature of 3d ti thin films , which has been attributed to tunneling between surface states localized on the opposite surfaces of the film . @xcite the size of the gap decreases with increasing the film thickness and is almost negligible for 6 qls , which is quantitatively consistent with our numerical observations . such gap - opening mechanism has been considered for possible applications in ti - based mosfet devices . @xcite the value of the gap for increasing slab thickness is plotted in fig . interestingly , we find that the gap initially decreases and reaches the value of @xmath63 ev for 5 qls but then increases up to @xmath64 ev for 6 qls . after this sudden increase , the gap continues to decrease exponentially and we find @xmath58=@xmath65 ev for 10 qls . a similar non - monotonicity in the thickness - dependence of the gap was found in the range of 4 - 6 qls in _ ab initio _ calculations for bi@xmath0te@xmath1 thin films . @xcite this result also resembles the oscillating behavior of the gap found in theoretical work based on effective models . @xcite however , direct comparison with these effective - model results is not straightforward since , in contrast to our atomistic model with slab thickness measured in terms of elementary building blocks of the real material ( qls ) , there the gap is calculated as a continuous function of the size along the @xmath35-direction . since the gap decreases exponentially , we do not display the results for thicknesses greater that 10 qls . note , however , that for 40 qls the size of the gap is smaller than the numerical accuracy of our calculations . for thicknesses larger than this critical value , we identify the crossing of the two branches of dirac states with opposite group velocity as the dirac point , which is found at @xmath66 ev . in section [ wf_atg ] we will investigate in more detail the effect of the coupling between the two surfaces of the slab on the states at the dirac point . in particular , we will look at the spatial character of the wavefunctions associated with these states and their dependence on the slab thickness . se@xmath1 slab of varying thickness . panels ( a)-(e ) correspond to 1 - 5 qls , ( f ) 20 qls and ( g ) 100 qls . the inset of panel ( a ) shows the two - dimensional brillouin zone of ( 111 ) surface of bi@xmath0se@xmath1 . symbols mark the surface states , identified according to the critical weight of the wavefunction projection onto the surfaces of the slab ( see text for details ) . ] point as a function of the slab thickness . panel ( b ) is the zoom in of ( a ) , showing a local minimum in the thickness dependence of the gap occurring at 5 qls . ] we will now comment on another feature of fig . [ fig3 ] , namely on the procedure used to identify the surface states in the bandstructure calculations . the character of each bloch state @xmath67 , where @xmath68 is the band index and @xmath69 is momentum , is determined by the spatial distribution of the corresponding wavefunction : @xcite if the relative weight of the atom - projected wavefunction @xmath70 on the top and bottom qls exceeds a critical value ( critical percentage ) @xmath71 , @xmath72 is identified as a surface state . such criterion is reasonable for slabs with thickness greater than 3 qls , since the typical penetration depth of the surface states in bi@xmath0se@xmath1 is of the order of 1 ql , or 1 nm . @xcite for ultra - thin slabs with thickness below 3qls , the criterion has to be modified , namely we calculate the wavefunction projections onto top two and bottom two atomic layers . the value of the critical percentage is found empirically . starting from a rough estimate of @xmath71 based on the ratio between the penetration depth and the slab thickness , the value should then be optimized in such a way that a small change around this value does not significantly modify the resulting distribution of the surface states . we find , however , that for 1 - 3 qls the identification of such an optimized value of @xmath71 is problematic . to illustrate this point , we present the results of calculations for two different values of @xmath71 . in the case of 1 ql and @xmath71=@xmath73 , for instance , all states in the range of energies considered can be identified as surface states . hence , for such a thin slab it is reasonable to use a stronger criterion . as one can see from fig . [ fig3](a ) , for @xmath71=@xmath74 only the upper band preserves the surface character while the lower band does not satisfy the criterion ( we use terms `` lower '' and `` upper '' for the two bands defining the gap ) . interestingly , this observation is consistent with scanning tunneling spectroscopy measurements on sb@xmath0te@xmath1 ultra - thin films . @xcite a similar situation is found for 2 qls , i.e. for two different values of @xmath71 the lower band is clearly not a surface band . for a stronger criterion with @xmath71=@xmath75 the upper band also partially looses its surface character . in the case of 3 qls , varying @xmath71 in the range of @xmath76-@xmath77% induces some visible changes in the distribution of the surface states , however , the surface character of the dirac - like states is well captured for any value of @xmath71 in this range . for slabs with thicknesses greater than 3 qls , the search for an optimized value of @xmath71 is significantly simplified . already for 4 qls , changing @xmath71 from @xmath75 to @xmath78 does not produce any significant difference . for 5 qls- and 20 qls - thick slabs we use @xmath71=@xmath78 . for a very thick slab of 100 qls , we focus particularly on the states in the vicinity of the dirac point , e.g. in the range of [ 0.0;0.3 ] ev , and do not apply the procedure for the near - continuum of bulk states appearing above and below this range . in fig . [ fig3](g ) we present the result for 100 qls with @xmath71=@xmath56 , where the surface character of the dirac states is clearly confirmed . before concluding this section we repeat the procedure that was used in section [ bulk ] to illustrate the mechanism of band inversion in bulk bi@xmath0se@xmath1 . for a thick slab [ see fig . [ fig3](f ) and ( g ) ] , one can identify a clear gap between valence and conduction bands , which is transversed by conducting surface states . as the thickness of the slab increases , the gap approaches its bulk value , which is found at the @xmath44 point in the bulk bandstructure calculated with so [ fig . [ fig1](b ) ] . vbm(cbm ) at the @xmath5 point can be defined as the first state below(above ) the dirac point . in the case without so , no conducting surfaces states are present and a gap is found at the @xmath5 point for all slab thicknesses ( calculations are not shown here ) . similarly to the so case , as the thickness of the slab increases the value of the gap approaches that found in the bulk calculation without so [ fig . [ fig1](a ) ] . point in a 20 qls - thick slab of bi@xmath0se@xmath1 . top panels show the atomic - layer projections for a selection of layers in the middle of the slab ( a ) without and ( b ) with spin - orbit interaction . @xmath79 is the first ( bottom ) layer , @xmath80 is the last ( top ) layer . ( c ) the total weights of vbm and cbm wavefunctions on five atoms forming a ql ( two se1 , se2 and two bi atoms ) , summed up of the entire slab . ] the atomic - layer projected wavefunctions at vbm and cbm of a 20 qls - thick slab without and with so are shown in fig . [ fig5](a ) and ( b ) , respectively . one immediately notices that the spatial character of vbm and cbm is inverted when so is switched on : similarly to the bulk case , with(without ) so the vbm(cbm ) wavefunction has nodes on se2 atoms . the effect becomes even more appreciable , if we sum the wavefunction projections on se1 , se2 and bi atoms over the entire slab [ fig . [ fig5](c ) ] . the result is clearly similar to what we found in the bulk case , indicating that the inverted character of the valence and conduction bands in the presence of so is also an intrinsic property of the surface , with the difference that in the surface bandstructures one finds the topological surface states , dispersing linearly across the inverted band gap . we will now consider the surface states and their corresponding wavefunctions , found at the @xmath5 point in our surface bandstructure calculations . in a hypothetical situation , when the slab is thick enough so that the two surfaces do not interact with each other , one expects to find four degenerate states at the @xmath5 point , namely one kramers degenerate pair for each of the two surfaces . away from the @xmath5 point , the four - fold degeneracy is lifted and one expects a doubly degenerate state at each momentum @xmath69 , with degeneracy guaranteed by inversion symmetry . in addition , for each state at @xmath69 , there is an identical state with opposite spin at @xmath81 due to time - reversal symmetry . point for ( a ) 5 qls , ( b ) 20 qls and ( c ) 40 qls of bi@xmath0se@xmath1 . the inset of panel ( b ) shows schematically the two pairs of degenerate states at @xmath5 , separated by an energy gap @xmath58 . the inset of panel ( c ) displays the crystal structure of a 40 qls - thick slab of bi@xmath0se@xmath1 with top and bottom surfaces . ] the analysis of the thickness - dependent gap , carried out in the previous section , shows that the limit of an infinitely thick slab with zero interaction between the two surfaces is most likely realized for 40 qls in our system . although the size of the gap already for 6 qls is considerably smaller than the value found in ultra - thin slabs , for 40 qls the gap becomes identically zero within the numerical precision of our calculations ( note that we employ exact diagonalization to calculate the eigen - spectrum of the hamiltonian at each @xmath82 ) . by looking closely at the @xmath5 point , we indeed find four degenerate states @xmath83 ( @xmath19=1, .. ,4 ) in the case of 40 qls , as shown in fig . [ fig6](c ) . within each pair of degenerate states , @xmath84 and @xmath85 , the states have opposite spins and each pair is localized on either top or bottom surface . note that the @xmath86-th cartesian component of the spin of each bloch state is calculated as @xmath87=@xmath88 $ ] , where @xmath89=@xmath90 is the set of pauli matrices and @xmath91 is the @xmath68-th diagonal element of the density matrix constructed from the eigenfunctions of the hamiltonian [ eq . ( [ eq:1 ] ) ] at momentum @xmath82 . from fig . [ fig6](c ) we can also estimate the decay length of the surface states to be approximately 10 atomic layers , or , equivalently , 2 qls in agreement with previous reports . @xcite point for a 20 qls -thick slab of bi@xmath0se@xmath1 . since the wavefunction is symmetric with respect to the center of the slab [ see fig . [ fig6](b ) ] , only atomic layers belonging to the bottom surface are displayed . ] for slabs with thicknesses below 40 qls , we also find four states at the @xmath5 point , however there is a finite gap between degenerate states @xmath92 and @xmath93 [ see the inset in fig . [ fig6](b ) ] . within each pair of states , the bloch states with identical energy have opposite spins but , in contrast to the 40 qls case , their wavefunctions are distributed over both top and bottom sides of the slab , signaling a non - negligible interaction between the the two surfaces . for 20 qls , since the difference in energy between @xmath92 and @xmath93 is finite but small ( less than @xmath94 ev ) , the spatial distributions of the corresponding wavefunctions are nearly identical and are therefore indistinguishable on the scale of the graph [ fig . [ fig6](b ) ] . in the case of 5 qls , when the gap at the @xmath5 point is more appreciable , there is a clear difference between the spatial profiles of the wavefunctions , associated with the two degenerate states [ fig . [ fig6](a ) ] . the difference is the largest in the middle of the slab , where the tails of the wavefunctions of the surface states residing on the opposite sides overlap , creating a mixed state . finally , we comment on the orbital character of surface states at the dirac point . the orbital - resolved atomic - layer projections of the wavefunctions associated with one of the four quasi - degenerate states ( @xmath95 ) at the @xmath5 point in a 20 qls - thick slab are plotted in fig . [ fig7 ] . similarly to the vbm and cbm states at the @xmath44 point of the bulk bandstructure with so [ fig . [ fig1](b ) ] , this state is predominantly of @xmath6-character with @xmath43-orbital contribution less than @xmath53 percent . although the contribution of @xmath54-orbitals is the largest , the relative weight of in - plane ( @xmath55 ) orbitals to the orbital - resolved wavefunction is non - negligible and is of the order of @xmath56 . the importance of the so - induced in - plane orbital contribution to the wavefunction of the dirac states in bi@xmath0se@xmath1 has been recently demonstrated using orbital - selective sr - arpes measurements . @xcite in this section we analyze the surface states away from the @xmath5 point . we describe the spatial distribution and the spin - properties of the wavefunctions , associated with energy states found inside the inverted band gap at non - zero values of momentum @xmath69 . in particular , we quantify the effect of the finite slab thickness on the helical spin - character of the dirac cone states . an equi - energy contour above the dirac point ( at approximately 0.13 ev ) is shown with a dashed line . red circles indicate a doubly degenerate state @xmath92 occurring at @xmath96 and its time - reversal partner at @xmath97 . ( b ) schematic of the two - dimensional brillouin zone with the projection of the equi - energy contour and two doubly degenerate states at @xmath98=@xmath99 . ( c ) atomic - layer projections of the wavefunctions associated with four states @xmath100 . ] figure [ fig8](a ) shows an equi - energy contour located slightly above the dirac point , within the energy range containing the dirac cone , for a 5qls - thick slab . there is a doubly degenerate state at each @xmath69 on the contour and a corresponding doubly degenerate state at @xmath81 with the same energy . we first consider a pair of doubly degenerate states @xmath101 and @xmath102 with @xmath103 as shown in fig . [ fig8](b ) . the atomic - layer projections of the wavefunction for all four states are presented is fig . [ fig8](c ) . each of the degenerate states within a pair @xmath104 is localized on either top or bottom surface . we will refer to the state whose wavefunction is localized predominantly on the top ( bottom ) surface as a top ( bottom ) state . furthermore , we calculate the spins of top and bottom states at several momenta on the equi - energy contour . figures [ fig9](a ) and ( b ) show the projection of the spin on the @xmath33-@xmath34 plane in momentum space for top and bottom states , respectively . the direction of the spin at each @xmath69 appears to be tangential to the equi - energy contour , which is a manifestation of the spin - momentum locking intrinsic to 3d tis . @xcite the direction of the spin is exactly opposite for the states @xmath105 and @xmath106 , hence the top and bottom surface states have opposite helicities . these observations indicate that despite the opening of the gap at the @xmath5 point due to interaction between the two surfaces for a 5qls - thick slab , the states appearing across the bulk band gap at this thickness are of topological character . and ( b ) @xmath107 for several values of momentum @xmath69 on the equi - energy contour [ see fig . [ fig8](a ) ] . ] in order to investigate the effect of the slab thickness on the helical spin - texture of the surface states , we determine the orientation of the spin @xmath108 of the bloch states @xmath101 for thicknesses in the range of 5 to 20 qls . the orientation of vector @xmath108 in momentum space at each @xmath69 can be described by two angles : @xmath109 , or in - plane tilting angle , which is the angle between @xmath108 and the normal to the contour at point @xmath69 , and @xmath110 , or out - of - plane tilting angle , which in the angle between @xmath108 and the direction perpendicular to the plane [ see fig . [ fig10](c ) ] . for a perfect spin - momentum locking @xmath109=@xmath110=@xmath111 . note that the angles @xmath109 and @xmath110 oscillate as functions of @xmath69 along the equi - energy contour due to the three - fold rotational symmetry of the slab crystal structure . @xcite in addition , at each value of momentum the in - plane and out - of - plane tilting angles of the top and bottom states are equal in magnitude and have opposite signs . hence we compute the deviations of @xmath109 and @xmath110 from @xmath111 only for the top state . the resulting absolute - value deviations , averaged over @xmath112 on the equi - energy contour , @xmath113 and @xmath114 , are plotted as functions of the thickness in figs . [ fig10](a ) and ( b ) , respectively . ) , associated with one of the states @xmath92 on the equi - energy contour , ( a ) from direction perpendicular to @xmath96 and ( b ) from in - plane orientation . ( c ) orientation of @xmath108 in momentum space : @xmath109 is the angle between @xmath115 and the normal @xmath116 to the equi - energy contour , @xmath110 is the angle between @xmath108 and the direction perpendicular to the plane . we define @xmath117=@xmath118 ; @xmath119 is the average of @xmath117 over @xmath69 on the equi - energy contour . ] both in - plane and out - of - plane deviations decrease with increasing the slab thickness . the dependence of @xmath113 shows a kink at 6 qls , consistent with the increase of the gap at the @xmath5 point ( see fig . [ fig4 ] ) , while @xmath114 decreases monotonically . importantly , both @xmath120 and @xmath114 do not decrease to zero with increasing the thickness but instead saturate to a constant value . the saturation value is small for @xmath120 ( @xmath121 ) and relatively large for @xmath114 ( @xmath122 ) . the saturation starts at approximately 10 qls for @xmath113 and at 7 qls for @xmath114 . the non - zero residual deviation from the perfect helical spin - texture at large thicknesses is a measure of the proximity of the equi - energy contour to bulk states . the further the contour is from the dirac point , the stronger is the effect of the bulk continuum , with hexagonal wrapping in momentum space due to crystal symmetry , on the surface states located on the contour . for an equi - energy contour located at a higher energy ( @xmath123 ev ) we find considerably larger saturation values , namely @xmath124 for @xmath114 and @xmath125 for @xmath113 . to summarize , both in - plane and out - of plane deviations decrease with increasing the slab thickness and with moving the equi - energy contour closer to the dirac point . however in the vicinity of the dirac point the in - plane tilting angle appears to be more affected by the finite thickness while the out - of - plane tilting angle is more sensitive to the presence of bulk states . we have employed the _ _ sp__@xmath2 tight - binding model , with parameters extracted from _ ab initio _ calculations , to model the electronic structure of bulk and ( 111 ) surface of bi@xmath0se@xmath1 3d ti . we presented a quantitative description of the band inversion mechanism for both bulk and slab geometry , which involves a detectable change in the spatial distribution of @xmath6-orbital projections of the wavefunctions of conduction and valence bands induced by spin - orbit interaction . the surface bandstructures , with the spatial character of bloch states determined by quantitative criteria , were calculated for slabs with thicknesses up to 100 qls . this thickness is well beyond what is accessible by _ ab initio _ approaches . based of the calculated thickness - dependent gap due to inter - surface interaction and the atomic - projections of the wavefunctions for states at the @xmath5 point , we found that the infinite - thickness limit , characterized by zero gap and no interaction between the surfaces , is realized for 40 qls . furthermore , our calculations showed that the disturbances in the helical spin - texture of the dirac - cone states , caused by the finite slab thickness and the proximity to bulk states , persist for thicknesses up to 10 qls even in the vicinity of the dirac point . we would like to comment on our numerical observation that the top and bottom surface states become completely decoupled only for thick slabs containing 40 qls . strictly speaking only states with linear dispersion and identically zero gap at the dirac point can be identified as dirac states . however , in practice our observation does not contradict the commonly accepted results that 6 qls is the critical thickness which determines the topological character of the material . @xcite indeed , we found that the gap for thicknesses greater that 5 qls is very small ( of the order of @xmath126 ev and smaller ) and the surface states appearing inside the bulk insulating gap at these thicknesses have nearly linear dispersion and helical spin - texture . nevertheless , the presence of a weak but finite interaction between the opposite surfaces for slabs of less than 40 qls thick might be crucial for the analysis of subtle effects in the vicinity of the dirac point . preliminary calculations have shown that even a small perturbation preserving the time - reversal symmetry can open a gap for thicknesses as large as 20 qls . @xcite this may further hinder the identification of the gap induced purely by the presence of time - reversal - breaking perturbations , such as magnetic impurities , @xcite in calculations based on tight - binding models similar to ours and certainly in _ ab initio _ studies , where thicknesses are limited to only few qls . effects due to finite thickness can be also expected in landau levels spectroscopy of 3d ti thin films . @xcite on one hand , it has been shown experimentally that in bi@xmath0se@xmath1-like materials the characteristic field - independent ( zero - th ) landau level is absent for slabs with thicknesses smaller than 3 qls but emerges already for 4 qls . @xcite on the other hand , a recent theoretical study suggests a splitting of the zero - th landau level due to hybridization between top and bottom surface states . @xcite based on these considerations , one might expect that a non - negligible inter - surface interaction can lead to more subtle internal structure of the zero - th landau level , persisting even for relatively large thicknesses . we anticipate that microscopic tight - binding models , combined with input from _ ab initio _ calculations , will play an increasingly important role in practical calculations of various properties of ti materials . these include the detailed character of the surfaces states wavefunction , which to some extent can be already probed experimentally , @xcite and the interplay between the surfaces states and external perturbations . @xcite in fact , a finite - cluster tight - binding approach , based on the model used in the present work , has been already employed in the study of native defects in bi@xmath0se@xmath1 , showing good agreement between the calculated local densities of states around the defects and experimental stm topographies . @xcite in connection to this last point , we would like to mention that despite extensive studies of the effect of magnetic doping in 3d tis , a consistent microscopic description of a single magnetic impurity in a ti environment appears to be incomplete , especially when compared to the progress that has been made in investigating similar questions in semiconductors , both theoretically @xcite and experimentally.@xcite a realistic tight - binding approach can be indispensable in providing such a microscopic description for single impurities in 3d tis . we are grateful to a. h. macdonald for illuminating discussions . we acknowledge helpful interactions with m. r. mahani . this work was supported by the faculty of natural sciences at linnaeus university and by the swedish research council under grant number : 621 - 2010 - 3761 . computational resources have been provided by the lunarc center for scientific and technical computing at lund university . 99 m. z. hasan and c. l. kane , rev . * 82 * , 3045 ( 2010 ) . qi and s .- c . zhang , rev . * 83 * , 1057 ( 2011 ) . c. l. kane and e. j. mele , phys . lett , * 95 * , 226801 ( 2005 ) . c. l. kane and e. j. mele , phys . * 95 * , 146802 ( 2005 ) . b. a. bernevig , t. l. hughes , and s .- c . zhang , science * 314 * , 1757 ( 2006 ) . m. knig , s. wiedmann , c. brne , a. roth , h. buhmann , l. w. molenkamp , x .- qi , and s .- c . zhang , science * 318 * , 766 ( 2007 ) . l. fu , c. l. kane , and e. j. mele , phys . lett . * 98 * , 106803 ( 2007 ) . l. fu and c. l. kane , phys . b * 76 * , 045302 ( 2007 ) . j. e. moore and l. balents , phys . b * 75 * , 121306 ( 2007 ) . j. c. y. teo , l. fu , and c. l. kane , phys . b * 78 * , 045426 ( 2008 ) . r. roy , phys . b * 79 * , 195322 ( 2009 ) . zhang , c .- x . qi , x .- y . deng , x. dai , s .- c . zhang , and z. fang , phys . b * 80 * , 085307 ( 2009 ) . h. zhang , c .- x . qi , x. dai , z. fang , and s .- c . zhang , nature physics * 5 * , 438 ( 2009 ) . d. hsieh , d. qian , l. wray . y. xia , y. s. hor , r. j. cava , and m. z. hasan , nature * 452 * , 970 ( 2008 ) . y. xia , d. qian , d. hsieh , l. wray , a. pal , h. lin , a. bansil , d. grauer , y. s. hor , r. j. cava , and m. z. hasan , nature physics * 5 * , 398 ( 2009 ) . d. hsieh , y. xia , d. qian , l. wray1 , j. h. dil , f. meier , j. osterwalder , l. patthey , j. g. checkelsky , n. p. ong , a. v. fedorov , h.lin , a. bansil , d. grauer , y. s. hor , r. j. cava , and m. z. hasan , nature * 460 * , 1101 ( 2009 ) . y. l. chen , j. g. analytis , j .- h . chu , z. k. liu , s .- k . mo , x. l. qi , h. j. zhang , d. h. lu , x. dai , z. fang , s. c. zhang , i. r. fisher , z. hussain , and z .- x . shen , science * 325 * , 178 ( 2009 ) . d. kong and y. cui , nature chemistry * 3 * , 845 ( 2011 ) . d. pesin and a. h. macdonald , nature mater . * 11 * , 409 ( 2012 ) . n. read , phys . today * 65(7 ) , 38 ( 2012 ) . c. nayak , s. h. simon , a. stern , m. freedman , and s. das sarma , rev . phys . 80 * , 1083 ( 2008 ) . y. cao , j. a. waugh , x - w . zhang , j - w . luo , q. wang , t. j. reber , s. k. mo , z. xu , a. yang , j. schneeloch , g. d. gu , m. brahlek , n. bansal , s. oh , a. zunger , and d. s. dessau , nature physics * 9 * , 499 ( 2013 ) . q. liu , c. x. liu , c. xu , x. l. qi , and s .- c . zhang , phys . rev . lett . * 102 * , 156603 ( 2009 ) . l. a. wray , s .- y . xu , y. xia , d. hsieh , a. v. fedorov , y. s. hor , r. j. cava , a. bansil , h. lin , and m. z. hasan , nature physics * 7 * , 32 ( 2010 ) . h. beidenkopf , p. roushan , j. seo , l. gorman , i. drozdov , y. s. hor , r. j. cava , and a. yazdani , nature physics * 7 * , 939 ( 2011 ) . p. roushan , j. seo , c. v. parker , y. s. hor , d. hsieh , d. qian , a. richardella , m. z. hasan , r. j. cava , and a. yazdani , nature * 460 * , 1106 ( 2009 ) . y. s. hor , p. roushan , h. beidenkopf , j. seo , d. qu , j. g. checkelsky , l. a. wray , d. hsieh , y. xia , s .- y . xu , d. qian , m. z. hasan , n. p. ong , a. yazdani , and r. j. cava , phys . rev . b * 81 * , 195203 ( 2010 ) . w. zhang , r. yu , h .- j . zhang , x. dai , and z. fang , new . j. phys . * 12 * , 065013 ( 2010 ) . k. park , j.j . heremans , v.w . scarola , and d. minic , phys . lett . * 105 * , 186801 ( 2010 ) . y. zhao , y. hu , l. liu , y. zhu , and h. guo , nano lett . * 11 * , 2088 ( 2011 ) . c. niu , y. dai , m. guo , w. wei , y. ma , and b. huang , appl . 98 * , 252502 ( 2011 ) . j .- m . zhang , w. zhu , y . zhang , d. xiao , and y. yao , phys . * 109 * , 266405 ( 2012 ) . j. henk , a. ernst , s. v. eremeev , e. v. chulkov , i. v. maznichenko , and i. mertig , phys . * 108 * , 206801 ( 2012 ) . l. b. abdalla , l. seixas , t. m. schmidt , r. h. miwa , and a. fazzio , phys . b * 88 * , 045312 ( 2013 ) . j .- tang and m. e. flatt , phys . 92 * , 047201 ( 2004 ) . t. o. strandberg , c. m. canali , and a. h. macdonald , phys . b * 80 * , 024425 ( 2009 ) . a. m. black - schaffer and a. v. balatsky , phys . rev . b * 85 * , 121103(r ) ( 2012 ) ; a. m. black - schaffer and a. v. balatsky , phys . rev . b * 86 * , 115433 ( 2012 ) . qi , h. j. zhang , x. dai , z. fang , and s .- c . zhang , phys . b * 82 * , 045122 ( 2010 ) . s. mao , a. yamakage , and y. kuramoto , phys . b * 84 * , 115413 ( 2011 ) . k. kobayashi , phys . b * 84 * , 205424 ( 2011 ) . m. s. bahramy , p. d. c. king , a. de la torre , j. chang , m. shi , l. patthey , g. balakrishnan , ph . hofmann , r. arita , n. nagaosa , and f. baumberger , nature communications * 3 * , 1159 ( 2012 ) . a. barfuss , l. dudy , m. r. scholz , h. roth , p. hpfner , c. blumenstein , g. landolt , j. h. dil , n. c. plumb , m. radovic , a. bostwick , e. rotenberg , a. fleszar , g. bihlmayer , d. wortmann , g. li , w. hanke , r. claessen , and j. schfer , phys . * 111 * , 157205 ( 2013 ) . collaboration : authors and editors of the volumes iii/17e-17f-41c : antimony telluride ( sb@xmath0te@xmath1 ) crystal structure , chemical bond , lattice parameters ( including data for bi@xmath0se@xmath1 , bi@xmath0te@xmath1 ) . madelung , o. , rssler , u. , schulz , m. ( ed . ) . springer materials - the landolt - brnstein database . s. k. mishra , s. satpathy , and o. jepsen , j. phys . : condens . matter * 9 * , 461 ( 1997 ) . s. urazhdin , d. bilc , s. h. tessmer , s. d. mahanti , t. kyratsi , and m. g. kanatzidis , phys . b * 66 * , 161306(r ) ( 2002 ) . s. urazhdin , d. bilc , s. d. mahanti , and s. h. tessmer , and theodora kyratsi , and m. g. kanatzidis , phys . b * 69 * , 085313 ( 2004 ) . j. c. koster and g. f. slater , phys . rev . * 94 * , 1498 ( 1954 ) . y. zhang , k. he , c .- z . chang , c .- song , l .- l . wang , x. chen , j .- f . jia , z. fang , x. dai , w .- y . shan , s .- q . shen , q. niu , x .- qi , s .- c . zhang , x .- c . ma , and q .- k . xue , nature physics * 6 * , 584 ( 2010 ) . j. linder , t. yokoyama , and a. sudb , phys . b * 80 * , 205401 ( 2009 ) . lu , w .- y . shan , w. yao , q. niu , and s .- q . shen , phys . b * 81 * , 115407 ( 2010 ) . w. a. harrison , _ elementary electronic structure _ ( world scientific , 2004 ) , p. 199 . l. fu and c. l. kane , phys . b * 74 * , 195312 ( 2006 ) . b. yan , c .- x . liu , h .- j . zhang , c .- y . qi , t. frauenheim , and s .- c . zhang , epl * 90 * , 37002 ( 2010 ) . z. zhu , y. cheng and u. schwingenschlgl , phys . b * 85 * , 235401 ( 2012 ) . j. chang , l. f. register , and s. k. banerjee , j. appl . phys . * 112 * , 124511 ( 2012).y . jiang , y. wang , m. chen , z. li , c. song , k. he , l. wang , x. chen , x. ma , and q .- k . xue , phys . lett . * 108 * , 016401 ( 2012).a . pertsova and c. m. canali ( unpublished ) . t. hanaguri , k. igarashi , m. kawamura , h. takagi , and t. sasagawa , phys . b * 82 * , 081305(r ) ( 2010).z . yang and j. h. han , phys . b * 83 * , 045415 ( 2011 ) . m. r. mahani , a. pertsova , m. f. islam , c. m. canali , a. h. macdonald , theoretical studies of single magnetic impurities on the surface of semiconductors and topological insulators , mrs online proceedings library , 1564 , mrss13 - 1564-gg01 - 01 ( 2013 ) . a. m. yakunin , a. yu . silov , p. m. koenraad , j. h. wolter , w. van roy , j. de boeck , j .- m . tang , and m. e. flatt , phys . * 92 * , 216806 ( 2004 ) . d. kitchen , a. richardella , j .- tang , m. e. flatt , and a. yazdani , nature * 442 * , 436 ( 2006 ) .	we report on microscopic tight - binding modeling of surface states in bi@xmath0se@xmath1 three - dimensional topological insulator , based on a _ _ sp__@xmath2 slater - koster hamiltonian , with parameters calculated from density functional theory . the effect of spin - orbit interaction on the electronic structure of the bulk and of a slab with finite thickness is investigated . in particular , a phenomenological criterion of band inversion is formulated for both bulk and slab , based on the calculated atomic- and orbital - projections of the wavefunctions , associated with valence and conduction band extrema at the center of the brillouin zone . we carry out a thorough analysis of the calculated bandstructures of slabs with varying thickness , where surface states are identified using a quantitative criterion according to their spatial distribution . the thickness - dependent energy gap , attributed to inter - surface interaction , and the emergence of gapless surface states for slabs above a critical thickness are investigated . we map out the transition to the infinite - thickness limit by calculating explicitly the modifications in the spatial distribution and spin - character of the surface states wavefunction with increasing the slab thickness . our numerical analysis shows that the system must be approximately forty quintuple - layers thick to exhibit completely decoupled surface states , localized on the opposite surfaces . these results have implications on the effect of external perturbations on the surface states near the dirac point .
the detection of feii and mgii absorption lines at a redshift of @xmath2 in the optical spectrum of gb970508 ( metzger et al . 1997 ) , provided the first confirmation that @xmath0-ray bursts ( grbs ) originate at cosmological distances . most of the qualitative properties of cosmological grbs are explained by the fireball model ( see e.g. , goodman 1986 ; paczyski 1986 ; mszros & rees 1993 [ mr ] ) . in this model , a compact ( @xmath3 cm ) source releases an energy of @xmath4 erg over a duration @xmath5 seconds with a negligible baryonic contamination ( @xmath6 ) . unsteady activity of the source results in a wind composed of many thin layers ( fireball shells ) of varying energy and baryonic mass . within each shell the high energy - density at the source results in an optically thick @xmath1-pair plasma that expands and accelerates to relativistic speeds . after an initial acceleration phase , the radiation and thermal energy of the fireball plasma is converted into the kinetic energy associated with the radial motion of the protons . collisions between the shells can convert part of that kinetic energy into radiation and yield the primary grb via synchrotron emission and inverse - compton scattering ( paczyski & xu 1994 ; rees & mszros 1994 [ rm ] ; sari & piran 1997 [ sp ] ) . as the wind continues to expand , it impinges on the surrounding medium and eventually drives a relativistic blastwave in it , which heats fresh gas and accelerates electrons to relativistic speeds , thus producing the delayed afterglow radiation observed on time scales of hours to months ( van paradijs et al . 1997 ; bond 1997 ; djorgovski et al . 1997 ; mignoli et al . 1997 ; frail et al . 1997 ) via synchrotron emission ( wijers , rees , & mszros 1997;waxman 1997a , b ; vietri 1997a , b ) . the primary grb emission is more likely caused by internal shocks than the external shocks ( mr ) , since they can occur closer to the source and thus account for the rapid variability observed in many bursts ( rm ; sp ) . unsteady activity of the central source naturally results in faster shells overtaking slower ones in front of them , and hence in energy dissipation by internal shocks . the complex temporal structure observed in grbs then reflects the activity - history of their sources ( sp ; kobayashi , piran , & sari 1997 [ kps ] ) . it is often assumed that behind internal shocks , electrons are fermi accelerated with a near equipartition energy density and magnetic fields acquire nearly equipartition strength . the electrons cool by synchrotron emission and inverse - compton ( ic ) scattering off the synchrotron photons . under typical conditions , the time scale for ic scattering is shorter than the synchrotron cooling time . multiple scattering of the photons boosts a significant fraction of the radiation energy to frequencies above the @xmath1-pair creation threshold . the pairs produced in this process are also relativistic and cool rapidly by ic scattering . since the annihilation time of these pairs is longer than the hydrodynamic time in the comoving frame , they survive in the wind for a long time . although the creation of pairs and their subsequent cooling is likely to leave noticeable imprints on the emergent radiation spectrum , it has not been analyzed before in the grb literature . since the photon and electron densities decline rapidly with radius , the strength of these signatures can serve as a probe of the radius at which the internal shocks occur . the grb spectrum should also depend on the level of baryonic contamination in the wind . the extreme limit of pure energy release with no baryons was ruled out in the past based on the prediction that a point explosion of this type would lead to a roughly thermal spectrum ( goodman 1986 ) . one can place a lower limit on the baryonic mass in the fireball shells by requiring that internal shocks should occur before an external shock does ( this limit depends on the ambient medium density ) . on the other hand , an upper limit can be placed based on the variability time scale of the source and the condition that the shells be optically thin at the radius where internal shocks occur . in this _ letter _ we study in detail the emergent spectra from the collision of two fireball shells . in particular , we quantify the significance of the radiation processes which were previously ignored in the literature , such as multiple compton scattering , @xmath7 creation , and their subsequent cooling . we use the collision kernels , reaction rates , and the computational techniques given in pilla & shaham ( 1997 [ ps ] ) . more details about this calculation and an elaborate study of the spectral characteristics of relativistic shocks will be included in a subsequent publication ( pilla & loeb 1997 ) . in 2 we describe our model and specify the physical conditions in the emission region . in 3 we outline the relevant radiation processes and compute the model spectra . finally , 4 summarizes the main implications of this work . the typical fireball dynamics ( piran , shemi , & narayan 1993 , [ psn ] ; mszros , laguna , & rees 1993 ; sari , narayan , & piran 1996 [ snp ] ; sp ; and kps ) can be illustrated by considering a single shell of total energy @xmath8 erg , rest mass @xmath9 g , and initial radius @xmath10 cm . after a brief acceleration phase , the lorentz factor of the shell reaches a constant value @xmath11 at an observer - frame radius @xmath12 ( the protons are taken to be non - relativistic in the comoving frame , before the collision of shells ) . the energy of the shell is predominantly kinetic beyond this stage . outside the radius @xmath13 ( all radii in the present analysis are measured in the observer s frame ) , the comoving width of the shell increases linearly with radius ( psn ) . the comoving proton density scales as @xmath14 for shell radii @xmath15 and as @xmath16 for @xmath17 . the thomson optical depth of the shell is @xmath18 for @xmath19 , where @xmath20 cm is the radius at which the shell becomes optically thin to thomson scattering by its own electrons . here @xmath21 is thomson cross section and @xmath22 is the proton rest mass . for @xmath17 , the comoving width of the shell is @xmath23 . two photons which are emitted with a proper - time difference @xmath24 , reach the observer with a time separation @xmath25 ( sp ) . assuming that the radiative cooling time is much shorter than the light transit time through the system , one finds that the observed width of the radiation pulse @xmath26 and the radius of the emission region @xmath27 are related through @xmath28 , which in turn can be used to constrain the values of @xmath29 , and @xmath30 from observations . the emission spectrum will be nonthermal only if @xmath31 . now consider a wind of total duration @xmath32 , composed of many thin fireball shells of thickness @xmath33 ( @xmath34 may change from one shell to another ) . we assume that for @xmath17 there are regions in the wind where slower shells precede the faster ones , i.e. @xmath35 , where @xmath36 is the local lorentz factor . the spatial extent of the wind is typically @xmath37 ( in fact , @xmath38 also holds in general ) . it was shown by waxman & piran ( 1994 ) that under these conditions the wind layers are susceptible to rayleigh - taylor instability because a rarefied fluid shell is pushing against a denser one . the resulting turbulent mixing will complicate the shock structure and deform it away from a simple planar geometry . merging of a rarefied shell with a denser one might therefore be accompanied by the formation of `` fingers '' perpendicular to the shell walls , similar to the non - relativistic shock structure in supernova remnants ( e.g. , jun & norman 1996 ) . the combined shells would then break into bubbles of different sizes , and the energy dissipation would take place near the bubble walls , due to collisions among them or instabilities on their surfaces ( kamionkowski & freese 1992 ) . by assuming that the shells as well as the shock fronts remain planar , kps had found that the dissipation efficiency of internal shocks might obtain high values @xmath39 for reasonable wind parameters . we assume that similar efficiencies are achieved in the case of unstable mixing . since the combined area of the bubble walls greatly exceeds that of a planar shock , the electron acceleration efficiency in the present case is likely to be higher . however the temporal and spectral characteristics of the bursts might be different in the two cases . for a planar shock the accelerated electrons populate a thin layer around the shock front since their cooling time is much shorter than the transit time of sound waves across the shells . however , in the unstable mixing case we assume , to a first approximation , that the energy dissipation takes place nearly uniformly and simultaneously throughout the entire volume of the emission region . for the purpose of estimating the physical conditions involved , we take @xmath40 erg , @xmath41 g , @xmath42 cm , and a bulk lorentz factor of the emission region of @xmath43 . these values yield @xmath44 @xmath45 , @xmath46 , a comoving width of the post - shock shell @xmath47 cm , and an average lorentz factor of the protons in the comoving frame ( i.e. , after shock heating ) of @xmath48 . acceleration of electrons to relativistic energies and the presence of strong magnetic fields are essential for converting the energy dissipated by the shock waves into radiation . since the physics of neither of these processes is well understood , we parameterize the corresponding energy densities in units of their equipartition values . a magnetic equipartition parameter @xmath49 , corresponds to a field strength @xmath50^{1/2}10^{4}$ ] g , where @xmath51 ( all quantities are comoving , unless stated otherwise ) . we assume that the electrons are accelerated throughout the emission region . the amount of energy transferred from protons to electrons is uncertain , and we define the acceleration efficiency @xmath52 in such a way that the average lorentz factor of the electrons immediately after they are heated ( via fermi - type acceleration ) is @xmath53 , where @xmath54 is the electron rest mass . because the coulomb collision time @xmath55 , we can safely ignore collisional relaxation in our analysis ( here @xmath56 is the coulomb logarithm ) . collisionless acceleration of electrons can be efficient if they are tightly coupled to the protons and the magnetic field by means of plasma waves ( kirk 1994 ) . the typical alfvn speed in the plasma is @xmath57 . the larmor radius of an electron with a lorentz factor @xmath0 is @xmath58 cm , and that of a proton of equal lorentz factor , @xmath59 , is larger by a factor @xmath60 . the corresponding acceleration time scales ( e.g. , hillas 1984 ) are , @xmath61 sec and @xmath62 , respectively . synchrotron losses limit the maximum value of the electron lorentz factor to @xmath63 . the minimum value is determined by @xmath64 and the shape of the electron distribution . we assume that the fraction of electrons per unit lorentz factor @xmath0 has the form @xmath65 for @xmath66 . thus @xmath67 , where @xmath68 is the average lorentz factor at any given time . the energy density in electrons immediately after their acceleration is @xmath69 . in our model the radiation time scale is much shorter than the hydrodynamic expansion time scale in the comoving frame ( which leads to a radiative efficiency of nearly @xmath70 ) and the radiation density at the end of electron cooling is therefore @xmath71 . we compute the spectra for @xmath72 , @xmath73 , and @xmath74 . we assume that the radiation energy density is initially small , and hence the electrons start losing their energy via synchrotron emission . as the energy density of the emitted radiation builds up , cooling via ic scattering becomes important as well . the typical time scale for synchrotron or ic losses is @xmath75 , where @xmath76 in the synchrotron case and @xmath77 in the ic case . here @xmath78 is the radiation energy density and @xmath79 is a dimensionless constant of order unity . for typical grb conditions @xmath80 , where @xmath81 is the light transit time through the system . therefore electron cooling takes place only locally . we assume that the acceleration proceeds throughout the cooling phase in such a way that it maintains a steady power - law distribution with a constant index @xmath82 , while @xmath68 declines due to radiative losses . if @xmath83 , the synchrotron cooling time is long and multiple ic scatterings become important . each scattering in the thomson regime increases the photon energy by a factor @xmath84 ( loeb , mckee , & lahav 1991 ) , so that some of the photons are eventually boosted into the klein - nishina ( kn ) regime . as @xmath68 decreases , the up - scattered part of the spectrum spreads over many decades in frequency . the electron energy density @xmath85 changes at a rate = ( ) _ sy+ ( ) _ ic , [ energy - rate ] where @xmath86 , @xmath87 , @xmath88 , and @xmath89 is the average value of @xmath90 over the initial electron distribution . the ic cooling rate , @xmath91 , is derived below . the characteristic cooling time @xmath92 is a few times @xmath93 under typical conditions . if @xmath94 is the instantaneous energy density of radiation , then energy conservation implies @xmath95 ( which is constant ) , and by assumption @xmath96 . the instantaneous radiation spectrum can be characterized by @xmath97 so that @xmath98 is the fraction of the radiation energy density within an interval @xmath99 around @xmath100 ; here @xmath101 , @xmath102 is the photon frequency , and @xmath103 is planck s constant . the spectral evolution rate ( see pilla & loeb 1997 ) is derived from the equation u_(t)(,t)- ( , t)=_sy+ _ ic , [ spec - rate ] where we have used the fact that @xmath104 . the first term on the right hand side is the synchrotron emissivity from relativistic electrons ( rybicki & lightman 1979 ) which depends on @xmath105 ; the second term is _ ic= _ -1 ^ 1d(1-)__l^_max df_e(,t ) , [ eq : dvdt ] where @xmath106 is the cosine of the scattering angle , @xmath107 , @xmath108 , @xmath109 , and @xmath110 is the klein - nishina cross - section . by integrating both sides of equation ( [ eq : dvdt ] ) over all values of @xmath100 , we obtain @xmath111 on the left hand side , whereas on the right hand side we use the relation @xmath112 to convert the integral to be over @xmath113 and integrate over all values of @xmath114 , and @xmath106 . finally , we divide the result by 2 to avoid double counting of each scattering event , and obtain ( ) _ ic=-()_ic = _ -1 ^ 1d__min^_max df_e(,t)__min^_maxd_1(_1,t ) ( 1-)^2 , [ ic - rate ] where @xmath115 are the limiting energies of photons in the plasma . in the thomson regime @xmath116 and @xmath117 , and one finds ( ) _ ic==cn_e_tu _ ^2 , in agreement with the well - known result for @xmath118 ( loeb et al . the coupled equations ( [ energy - rate])([ic - rate ] ) are solved numerically for the radiation spectrum at the end of the cooling process ( i.e. , when @xmath119 ) . cooling ends after a relatively short time , @xmath120 , for the parameters of interest here , and the photons decouple from the electrons subsequently . an example for the time evolution of the spectrum is shown in figure 1a . it is evident that the radiation density above the @xmath1-pair creation threshold is substantial . the spectral evolution rate due to pair creation is described by @xmath121_{\pm}&= & u_{\gamma}(t)\frac{\partial}{\partial t}\left[\phi(\ee , t)\right]_{\pm } + \phi(\ee , t)\left(\frac{du_{\gamma}}{dt}\right)_{\pm}\nonumber\\ & & \nonumber\\ & = & \frac{u_{\gamma}}{t^{\prime}}{\cal i}_{1}(\ee , t)\equiv -\frac{u_{\gamma}}{4t^{\prime } } \int d\mu \int d\ee^{\prime } \frac{\phi(\ee , t)}{\ee}\frac{\phi(\ee^{\prime},t ) } { \ee^{\prime}}\frac{\ee}{\overline{\gamma}_{0}}(1-\mu)\frac{\sigma_{\pm } } { \sigma_{t } } , \label{eq : pairc}\end{aligned}\ ] ] where the integration is over the range @xmath122 and @xmath123 subject to the condition @xmath124 . here @xmath125 is the pair creation cross section ( ps ) and @xmath126 . in all examples considered in this _ letter _ , we find that @xmath127 for all relevant photon energies . therefore , pair creation comes into play after the original electrons cool and decouple . the total energy loss per unit time and volume due to pair creation is obtained by integrating equation ( [ eq : pairc ] ) over all values of @xmath100 , ( ) _ = _ _ min^_maxd_1(,t)_1(t ) . the rate at which pairs are created per unit volume is ( ) _ = _ _ min^ _ max d_2(,t)_2(t ) , where @xmath128 is same as @xmath129 except that @xmath130 is absent in its integrand . at the beginning of pair creation , the energy density of radiation is @xmath71 . the average lorentz factor of the newly created pairs is @xmath131 , since @xmath132 . because the pairs are born relativistic , they transfer almost all their energy to the radiation via ic scattering ( and also synchrotron emission if @xmath133 ) on a time scale of order a few @xmath134 . similar cascades of pair creation and cooling occur in active galactic nuclei ( svensson 1987 ) . throughout the pair cascade and cooling process the radiation spectrum evolves continuously but its total energy density remains nearly constant at a value close to @xmath71 . figure 1b shows an example of this evolution up to a time @xmath135 . because the hydrodynamic time scale in the comoving frame @xmath136 , pair processes can not continue to operate for times much longer than @xmath137 because of the decline in the densities due to the expansion of the fireball . we solved the coupled equations given above using the methods described by ps and obtained the model spectra for @xmath138 erg , @xmath139 , @xmath72 , @xmath73 , and @xmath74 . we assume a source redshift @xmath140 and include its effect on photon energy and flux . the total observable fluence for the above parameters is @xmath141 . figure 1 shows the entire spectral evolution ( in the comoving frame ) due to the synchrotron and ic cooling and pair cascade processes . we take @xmath41 g which yields @xmath142 ( which are consistent with empirical constraints , e.g. woods & loeb 1995 ) ; and a dissipation ( or shell collision ) radius of @xmath143 cm . note that the assumed values for @xmath144 and @xmath145 correspond only to a single emitting shell ; they should obviously be higher for the entire burst . for the above parameters , we get @xmath146 cm , @xmath147 , @xmath148 , and @xmath149 g. the characteristic time scale for synchrotron and ic cooling is @xmath150 . we find that after a time @xmath151 , only @xmath152 of the initial energy remains with the electrons . figure 1a illustrates that as a result of electron cooling , the location of the synchrotron peak shifts to lower energies and also becomes broader . later on , another peak develops at a much higher energy due to ic losses , with kn suppression in the very high - energy tail . figure 1b shows the evolution much later due to the pair - cascade process , which depletes the population of high - energy photons . the ic loss by these pairs produces a power law tail at yet higher energies . the low - energy synchrotron part of the spectrum is almost unaltered throughout this phase . the computation was stopped at @xmath153 at which time most of the photons leave the system . figure 2 shows the dependence of the emergent spectrum on the dissipation radius for a fixed value of @xmath11 ( panel [ a ] ) and dependence on @xmath11 for a fixed value of the dissipation radius ( panel [ b ] ) . panel ( a ) shows that the location of synchrotron peak moves up in energy as the radius decreases ( with an opposite trend for the ic peak ) ; hence the overall extent of the emission spectrum gets narrower at smaller radii . since @xmath154 and @xmath155 , as @xmath156 decreases the synchrotron frequency increases and the pair - production depletion of high - energy photons is enhanced . one of the model spectra in figure ( 2a ) is compared with the batse data through the empirical formula for the time - integrated flux given by band et al . ( 1993 [ b93 ] ) . in that case our model predicts @xmath157 more time - integrated flux at 10 kev than observed , which could be due to an oversimplified form of the electron distribution function which we have adopted . in our notation , the number of photons per unit interval of @xmath100 is @xmath158 . therefore , band s formula ( b93 ) reads @xmath159 & = & c_{2}\ee^{-\beta},\qquad\mbox{\rm otherwise . } \label{eq : band}\end{aligned}\ ] ] the constants @xmath160 are fixed by the requirement that this function be continuous at @xmath161 and the normalization @xmath162 . note that we have altered the signs of both indices @xmath163 and @xmath164 relative to the convention of b93 . the observed values of the parameters ( cf . table 4 in b93 ) are in the range @xmath165 ( although in a few cases @xmath163 is zero or negative ) , @xmath166 , and @xmath167(kev)@xmath168 . the majority of the bursts seem to have @xmath169 , @xmath170 , and @xmath171 a few hundred kev . from the model spectra in figure 2 , it is clear that the emission extends over a wide range of photon energies , from the optical to the tev regime . we expect more optical emission when the dissipation takes place at larger radii . it is very likely that internal shocks occur over a wide range of radii , thereby extending the spectrum to longer wavelengths , with simultaneous optical and x - ray emission during a grb . recent reports of nearly simultaneous detection of x - ray emission from gb960720 ( piro et al . 1997 ) , gb970815 ( smith et al . 1997 ) , and gb970828 ( remillard et al . 1997 ) are in qualitative agreement with our expectation for internal shocks . we have shown that the emission spectra from internal shocks are affected by synchrotron emission , multiple compton scatterings , and pair creation ( fig . 1 ) . our model spectra could mimic the observed batse spectra ( b93 ) for reasonable choices of the shell lorentz factor @xmath172@xmath173 and shock radius @xmath174@xmath175 ( fig . the depletion of high energy photons due to pair creation is a sensitive probe of both @xmath11 and @xmath156 . detection of the high energy trough present in figures ( 1b ) and ( 2 ) can therefore be used to constrain these parameters . the potential degeneracy between the values of @xmath11 and @xmath156 can be removed on the basis of variability data , as the characteristic variability time scale depends on a different combination of these parameters , @xmath176 . despite their high - energy cut - off , our model spectra extend all the way up to photon energies as high as 10 gev tev . we acknowledge discussions with m. kamionkowski , t. piran , m. a. ruderman , and e. woods . one of us ( rp ) thanks r. sari for many critical comments and stimulating discussions during the viii marcel grossmann meeting . this research was supported in part by nasa grants nag5 - 618 and -2859 ( for rp ) and nasa atp grant nag 5 - 3085 and the harvard milton fund ( for al ) . band , d. l. et al . 1993 , apj 413 , 281 ( b93 ) bond , h. e. 1997 , iau circ . 6654 djorgovski , s. g. et al . 1997 , nature , 387 , 786 frail , d. a. et al . 1997 , apj , 483 , l91 goodman , j. 1986 , apj 308 , l47 hillas , a. m. 1984 , ann . 22 , 425 jun , b. and norman , m. l. 1996 , apj 472 , 245 kamionkowski , m. and freese , k. 1992 , phys . rev . lett . 69 , 2743 kirk , j. g. , in _ plasma astrophysics _ , edited by a. o. benz & t. j. l. courvoisier , saas fee advanced course 24 ( springer verlag , new york , 1994 ) kobayashi , s. , piran , t. & sari , r. 1997 , astro - ph/970513 ( kps ) loeb , a. , mckee , c. f. , & lahav , o. 1991 , apj 374 , 44 mszros , p. & rees , m. j. 1993 , apj 405 , 278 ( mr ) mszros , p. , laguna , p. , & rees , m. j. 1993 , apj 415 , 181 metzger , m. r. et al . 1997 , iau circ . 6655 mignoli , m. et al . 1997 , iau circ . 6661 paczyski , b. 1986 , apj 308 , l43 paczyski , b. , & xu , g. 1994 , apj , 427 , 708 pilla , r. p. & loeb , a. 1997 , in preparation pilla , r. p. & shaham , j. 1997 , apj 486 , 903 ( ps ) piran , t. , shemi , a. , & narayan , r. 1993 , mnras 263 , 861 ( psn ) piro , l. et al . 1997 , a & a , in press ; astro - ph/9707215 rees , m.j . & mszros , p. 1994 , apj 430 , l93 ( rm ) remillard , r. et al . 1997 , iau circ . 6726 rybicki , g. b. & lightman , a. p. 1979 , _ radiative processes in astrophysics _ ( new york : john wiley ) sari , r. & piran , t. 1997 , mnras 287 , 110 ( sp ) sari , r. , narayan , r. , & piran , t. 1996 , apj 473 , 204 ( snp ) smith , d. a. et al . 1997 , iau circ . 6718 svensson , r. 1987 , mnras 227 , 403 van paradijs , j. , et al . 1997 , nature , 386 , 686 vietri , m. 1997a , apj , 478 , l9 1997b , apj letters , in press ; astro - ph/9706060 waxman , e. 1997a , apj letters , in press ; astro - ph/9704116 1997b , astro - ph/9705229 waxman , e. & piran , t. 1994 , apj 433 , l85 wijers , a. m. , rees , m. j. , & meszaros , p. 1997 , mnras , 288 , l51 woods , e. & loeb , a. 1995 , apj 453 , 583	unsteady activity of @xmath0-ray burst sources leads to internal shocks in their emergent relativistic wind . we study the emission spectra from such shocks , assuming that they produce a power - law distribution of relativistic electrons and posses strong magnetic fields . the synchrotron radiation emitted by the accelerated electrons is compton up - scattered multiple times by the same electrons . a substantial component of the scattered photons acquires high energies and produces @xmath1 pairs . the pairs transfer back their kinetic energy to the radiation through compton scattering . the generic spectral signature from pair creation and multiple compton scattering is highly sensitive to the radius at which the shock dissipation takes place and to the lorentz factor of the wind . the entire emission spectrum extends over a wide range of photon energies , from the optical regime up to tev energies . for reasonable values of the wind parameters , the calculated spectrum is found to be in good agreement with the burst spectra observed by batse . submitted to _ apj letters _ , october 1997
nature offers a wide range of phenomena characterized by power - law distributions : diameter of moon craters , intensity of solar flares , the wealth of the richest people@xcite and intensity of terrorist attacks@xcite , to name a few . these distributions are so - called _ heavy - tailed _ , where the fractional area under the tail of the distribution is larger than that of a gaussian and there is thus more chance for samples drawn from these distributions to contain large fluctuations from the mean . anatomical defects aside , the cosmic ray ( cr ) energy spectrum follows a power - law for over ten orders of magnitude . the predicted abrupt deviation at the very highest energies ( the gzk - cutoff@xcite ) has generated a fury of theoretical and experimental work in the past half century . recently , bahcall@xcite and waxman ( 2003 ) have asserted that the observed spectra ( except agasa ) are consistent with the expected flux suppression above @xmath0ev . however , the incredibly low fluxes combined with as much as @xmath150% uncertainty in the absolute energy determination means that there has yet to be a complete consensus on the existence of the gzk - cutoff energy . with this in mind , we consider statistics which suggest an answer to a different question : _ do the observed cr spectra follow a power - law ? _ specifically , these studies are designed to inquire whether or not there is a flux deviation relative to the power - law form by seeking to minimize the influence of the underlying parameters . the two experimental data sets considered in this study are the agasa@xcite experiment and the preliminary flux result of the pierre auger observatory@xcite . the discussion in @xmath2[sec : data ] uses these spectra to introduce and comment on the power - law form . the first distinct statistical test is applied to this data in @xmath2[sec : dlv ] where we explore the distribution of the largest value of a sample drawn from a power - law . in @xmath2[sec : tp ] we apply the tp - statistic to the cr flux data . this statistic is asymptotically zero for pure power - law samples _ regardless _ of the value power index and therefore offers a ( nearly ) parameter free method of determining deviation from the power - law form . the final section summarizes our results . a random variable @xmath3 is said to follow a power - law distribution if the probability of observing a value between @xmath4 and @xmath5 is @xmath6 where @xmath7 . normalizing this function such that @xmath8 gives , @xmath9 it is convenient to choose @xmath10 , @xmath11 and doing so yields @xmath12 for reference , one minus the cumulative distribution function @xmath13 is given by , @xmath14 taking the log of both sides of equation ( [ eq : pwlpdf ] ) yields @xmath15 where @xmath16 is an overall normalization parameter , and suggests a method of estimating @xmath17 ; the _ power index _ is the slope of the best fit line to the logarithmically binned data ( i.e. bin - centers with equally spaced logarithms ) . in what follows , we refer to the logarithmically binned estimate@xcite of the power index as @xmath18 and assume that the typical @xmath19/ndf is indicative of the goodness of fit . the fitting is done with two free parameters , namely @xmath16 and @xmath17 . this figure displays published agasa@xcite and auger@xcite cr energy spectra . both axis have logarithmic scales to illustrate the power - law behavior . the vertical axis is the flux @xmath20 in ( m@xmath21 sr sec ev)@xmath22 and the horizontal axis is the energy in ev . the best fit lines ( see [ eq : loglogfit ] ) have slope @xmath23 and @xmath24 ( statistical error only).,width=419,height=264 ] the energy flux of two publicly available data sets are shown in fig . [ fig : specs ] . the the red point - down triangles represent the log@xmath25 of the binned agasa flux values in units of ( m@xmath21 sr sec ev)@xmath22 and the blue point - up triangles correspond to the auger flux . the vertical error bars on each bin reflect the poisson error based on the number of events in that bin . the log - binned estimates for each complete cr data set are the slopes of the dashed lines plotted in fig . [ fig : specs ] . to check the stability of @xmath18 we estimate the power index as a function of the minimum energy @xmath26 considered for the agasa and auger cr data sets ; see fig.[fig : specs ] . the left most point is the slope of the best fit lines plotted in fig.[fig : specs ] . the vertical error bars represent @xmath27 deviation.,width=419,height=264 ] in order to check the stability of to bound on our estimate , we compute the estimated power index @xmath18 as a function of the minimum energy @xmath26 considered for each of the two cr data sets . the left - most blue ( red ) point in fig . [ fig : gests ] shows @xmath18 for the auger ( agasa ) data taking into account all of the bin values above @xmath28 ( @xmath29 ) , the next point to the right represents that for all bins above @xmath30 ( @xmath31 ) , and so on . the vertical error bars on these points represent the @xmath27 error of the estimate . to ensure an acceptable chi - squared statistic , we demand that at least five bins be considered , thereby truncating @xmath26 at @xmath32 for the auger and @xmath33 for the agasa data set . the @xmath19/ndf for the left - most points is @xmath34 and it increases to @xmath35 for the right - most for both experiments . we note that these estimates do not vary widely for the lowest @xmath26 s and that the values of @xmath18 from these experiments are consistent . the analyses discussed in @xmath2[sec : dlv ] and @xmath2[sec : tp ] will depend on the total number of events in the data set . since these numbers are not published we use a simple method for estimating them from the cr flux data . if the exposure is a constant function of the energy , then we may take the flux @xmath20 to be proportional to the number of events in the bin and the exposure @xmath36 , namely @xmath37 . the auger exposure is reported to be constant over the energy range reported with @xmath38 ( m@xmath21 sr sec ) . the agasa collaboration report flux data all the way down to @xmath28 but the exposure of the experiment can be considered approximately constant only for energies above @xmath29 ( see fig . 14 of @xcite ) where @xmath39 ( m@xmath21 sr sec ) . using this method we get a total of 3567 events with @xmath40 for the auger flux and 1914 with @xmath41 for the agasa experiment . as evidence suggestive of a gzk - cutoff , an often cited quantity is the flux suppression , or the ratio of the flux one would expect from a power - law to that actually observed above a given maximum , say , @xmath42 . since @xmath43 one may estimate the flux suppression by estimating the number of events @xmath44 out of @xmath45 expected above a given maximum as @xmath46 = n_{tot}z_{max}^{1-\gamma}$ ] . thus , the bin with minimum @xmath47 and maximum @xmath48 would have a height @xmath49 if the data continued to follow a power - law above @xmath42 . as a test statistic for this quantity , one may consider the poissonian probability that the bin height could statistically fluctuate to zero , namely @xmath50 exp@xmath51 $ ] . in this section we derive a similar test statistic based on the distribution of the maximum event from a power - law sample . the statistic discussed here approaches @xmath52 for large @xmath45 and allows us to show that the estimation errors associated with @xmath18 are enough to disallow any significant conclusion about the presence of flux suppression for the highest energy cr s . the form of the power - law distribution allows us to calculate the pdf of the largest value , @xmath53 , out of n events . using the equations ( [ eq : pwlpdf ] ) and ( [ eq : pwlcdf ] ) we can say that the probability that any one value falls between @xmath4 and @xmath54 _ and _ that all of the others are less than it is @xmath55 . there are n ways to choose this event and so the probability for the largest value to be between @xmath4 and @xmath54 is @xmath56 in terms of the ratio @xmath57 , this can be written as @xmath58 fig . [ fig : maxpdf ] contains a plot of this distribution for @xmath59 with three choices of n. the glaring implication of this plot is that even for `` small '' n nearly all of the integral of @xmath60 is above @xmath61 . this implies that the probability of the maximum energy event falling below 10 times the minimum is very small , for a power - law with these parameters . a plot of the probability distribution of the maximum of a sample drawn from a power - law with power index @xmath59 . this is the distribution @xmath60 defined in equation ( [ eq : maxesz ] ) where @xmath57 is the ratio of the maximum to the minimum . the sample sizes are @xmath62 and @xmath63.,width=359,height=226 ] motivated by the location and shape of @xmath60 we consider the probability @xmath64 that the maximum ratio from a given sample @xmath65 is less than or equal to a particular value , , equation ( [ eq : pxc ] ) approaches the poisson probability mentioned above ; @xmath66^{n_{tot } } \rightarrow \exp[-n_{tot}z_{max}^{1-{{\gamma } } } ] = \mathcal{p}(0 , n_{sup})$ ] . ] say @xmath57 , in a convenient form as @xmath67^{n}. \label{eq : pxc}\ ] ] indeed , with @xmath68 ( as in fig . [ fig : maxpdf ] ) , @xmath69 for @xmath70 , @xmath71 for @xmath72 and is @xmath73 for @xmath74 . another way to say this is that if one were to generate @xmath75 _ sets _ of events , each containing 1000 events drawn from a pure power - law with @xmath59 , @xmath76 of these sets would have a maximum element with a value greater than 10 times the minimum . for 500 events / set the fraction decreases to @xmath77 . such simulations were carried out in preparation for this note and the results were consistent with equation ( [ eq : pxc ] ) . to apply this idea to the cr spectrum we consider the following null hypothesis : _ the flux of cr s follow a power - law with index @xmath18 for all energies greater than a given minimum . _ as a test statistic for this hypothesis we use @xmath64 , as defined in equation ( [ eq : pxc ] ) , with the interpretation that if the null hypothesis is true then @xmath64 is the probability that the ratio of the maximum energy to the minimum is less than or equal to the observed ratio . typically , the null hypothesis is rejected at the 5% significance level ( s.l . ) if @xmath78 . to calculate the value of @xmath64 for the observed data sets we need three pieces of information : the ratio of the maximum observed value to the minimum @xmath79 , the number of events @xmath45 with values in the interval @xmath80 $ ] and a reasonable guess for the power index @xmath17 . since a larger @xmath42 will lead to a larger value of @xmath64 we will conservatively take the highest energy agasa ( resp . auger ) event to fall on the upper edge of the highest energy bin . the method of determining the number of events in each bin is described in @xmath2[sec : data ] and here the parameter @xmath45 represents the total number above a given minimum . we will use the logarithmically binned estimates and errors of @xmath18 discussed in @xmath2[sec : data ] . a plot of the probability that the maximum of a sample drawn from a power - law will be less than or equal to the maximum observed by the auger ( @xmath81 , blue point - up ) and agasa ( @xmath82 , red point - down ) experiments as a function of the minimum energy considered . the vertical error bars represent the effect of a @xmath27 deviation and the hatched area shows the 5% significance level.,width=419,height=264 ] the plot in fig . [ fig : pmaxlast ] shows @xmath83 given @xmath45 and @xmath18 as a function of minimum energy considered for each of the cr data sets in fig . [ fig : gests ] . in particular , for each @xmath26 the values of @xmath45 , @xmath79 and @xmath84 are estimated from the cr flux and the resulting @xmath64 are plotted for the auger ( blue ) and agasa ( red ) data . for example , the left - most auger point represents the probability that if @xmath85 events are drawn from a power - law with @xmath86 then there is a @xmath87 chance that the maximum log - ratio @xmath88 would be less than or equal to that reported by the auger experiment , namely @xmath89(ev ) . taken at face value , one may reject the null hypothesis at the 5% s.l . for this data set . the left - most agasa point represents the same probability for the complete set of agasa data , namely @xmath90 for @xmath91 events drawn from a power - law with @xmath92 . thus we can not reject the null hypothesis for the agasa data . the upper ( lower ) vertical error bars depicted in fig . [ fig : pmaxlast ] represent the value of @xmath64 if we have under ( over ) estimated the power index by @xmath93 , that is if @xmath94 , keeping the log - ratio and the total number of events constant . ( the possible errors in the total number of events are on the order of a few percent and are negligible . ) since the fitting scheme considers successively lower energy bins , the points ( and errors ) for each experiment plotted in fig . [ fig : pmaxlast ] are highly correlated . the upper error bars fall above the 5% s.l . for all minimums considered and therefore the statistical error associated with @xmath18 is enough to disallow rejection of the power - law hypothesis . the biggest systematic measurement uncertainty in the cr data is the calibration of the energy . this uncertainty leads to an error in the reported absolute energy values of @xmath95 for the agasa @xcite data and as much as @xmath150% for the highest energy events in the auger data set . since the probability considered here depends only on the ratio of the observed energies , it is independent of any constant systematic uncertainty in the energy determination . however , this probability is sensitive to energy errors which vary over the range considered and will thus cause uncertainty in @xmath79 . for example , if we take the maximum to be 50% higher ( but hold @xmath96 and @xmath97 constant ) the value of @xmath64 represented by the left most auger point in fig . [ fig : pmaxlast ] changes from 1.9% to 17% . thus the large uncertainty in @xmath79 combined with the errors associated with @xmath18 implies that the preliminary auger data set does not suggest sufficient evidence to reject the pure power - law hypothesis for all events above @xmath98(ev ) . considering the error and extra degree of freedom associated with @xmath17 , an analysis of a distribution s adherence to the power - law form without reference to , or regard for , this parameter is could lead to enhanced statistical power . first proposed by v. pisarenko and d. sornette , the so - called _ tp - statistic_@xcite is a function of random variables that ( in the limit of large @xmath99 ) tends to zero for samples drawn from a power - law , regardless of the value of @xmath17 . ( tp stands for _ tail power _ , as oppossed to te , also introduced in @xcite , which stands for _ tail exponential_. ) this section will describe the tp - statistic and apply it to the cr data . the raw moments of the pdf equation ( [ eq : pwlpdf ] ) are @xcite @xmath100 thus power - laws with @xmath101 have a finite mean but an infinite variance ( in the limit of large n ) and sample statistics created from these moments are not particularly helpful . however , taking the natural logarithm of @xmath57 allows the integrals to converge and one may write ( for all @xmath102 and @xmath103 ) , @xmath104 the tp - statistic is calculated by noting that @xmath105 . therefore , if we use the sample analog of these quantities , namely @xmath106 then we can define ( for all @xmath107 ) , @xmath108 by the law of large numbers this sample statistic tends to zero as @xmath109 , independent of the value of @xmath17 . the tp - statistic allows us to test for a power - law like distribution without comment about the value of the power index . furthermore , for any one sample we can vary @xmath110 from the sample minimum @xmath111 to the sample maximum @xmath53 and calculate the tp - statistic over the range of @xmath4 in the sample . given complete event lists one may use equation ( [ eq : tp ] ) to calculate the tp - statistic for the unbinned data . since only the binned cr flux is publicly available we adapt the statistic to a binned analysis and apply it first to an example distribution with a cutoff and then to the cr data sets . in order to build intuition about the tp - statistic and its variance before studying the cr data , we first apply this statistic to simulated event sets drawn from both a pure power - law distribution and a similar distribution with a cut - off . the cut - off pdf is chosen so that it mimics a power - law for the lowest values but has an abrupt ( and smooth ) cut - off at a particular value , say @xmath112 . the functional form we will use here is @xmath113 the normalization of this pdf is @xmath114 , the value of which must be computed numerically . [ fig : fdhisto ] contains a logarithmically binned histogram of 3000 events drawn from a pure power - law ( black circles ) with @xmath115 and @xmath59 , and two pdf s in the from of equation ( [ eq : fdpdf ] ) ; the magenta squares have @xmath116 and the green triangles have @xmath117 . while arbitrary , the values of these parameters are chosen to be similar to the agasa and auger data ( see fig.[fig : specs ] ) . logarithmically binned histogram of 3000 events drawn from a pure power - law with @xmath59 and two power - laws with a cut , see equation ( [ eq : fdpdf ] ) . the magenta squares are drawn from the distribution with @xmath116 and the green triangles have @xmath117 . as noted in the text , while arbitrary , the values of these parameters are chosen to be similar to the agasa and auger data ( see fig.[fig : specs]).,width=359,height=226 ] if we write the sorted ( from least to greatest ) values from a sample as @xmath118@xmath119 @xmath120 , the solid black line in fig . [ fig : fdtp ] is created by calculating @xmath121 for each value of the 3000 events drawn from the pure power - law histogram in fig . [ fig : fdhisto ] . the circles represent the mean of the the statistic within the @xmath122 bin , say @xmath123 , and the vertical error bars show the root - mean - squared deviation of the statistic within the bin . note that the total number of events considered by the statistic decreases quickly from left to right which leads to a bias in and an increasing variance of the statistic . the tp - statistics , defined in equation equation ( [ eq : tp ] ) , as a function of minimum value `` @xmath110 '' for the 3 sets of 3000 events plotted in fig . [ fig : fdhisto ] . also plotted is the mean of the tp - statistic within each of the logarithmically spaced bins which is referred to in the text as @xmath124 . the vertical error bars represent the rms deviation of the statistic within each bin . parenthetically , with increased statistics , say 10,000 events , the distinct characteristics of the tp - statistic for a pure power - law , a power - law with a cutoff @xmath125 or a power - law with a cutoff @xmath126 become more clearly different.,width=419,height=264 ] the jagged magenta line fig . [ fig : fdtp ] shows the most obvious deviation from the power - law form ; it is systematically offset from zero for nearly all minima of the data set . of course , with 3000 events the histograms ( see fig . [ fig : fdhisto ] ) are enough to distinguish between these two distributions . but the tp - statistic allows us to see this deviation by considering the entire data set ( the left most magenta point in fig . [ fig : fdtp ] ) , not just by analyzing the events in the upper most bins . the green line in the figure shows @xmath127 for events drawn from equation ( [ eq : fdpdf ] ) with @xmath117 . the histogram for this set is not as clearly different from the power - law as the magenta points and neither is the tp - statistic ; the left - most green point shows no more deviation from zero than the power - law . however , as the minimum increases ( and nears @xmath112 ) the statistic moves away from zero ( more noise not withstanding ) and suggests that the data above the minimum deviate from the power - law . it is important to note that the tp - statistic is positive for both of the cutoff distributions . recall that for a pure power - law , @xmath105 . the cutoff distribution , however , lacks an extended tail and will therefore have a smaller second log - moment @xmath128 as compared with ( the square of ) the first log - moment @xmath129 and will thus result in a positive tp - statistic . a distribution with an enhancement , rather than a cutoff , in the tail would result in a negative tp - statistic , since it would have a larger second log - moment ( i.e. a larger `` variance '' ) . see the appendix ( @xmath2[sec : app ] ) for a detailed discussion of the tp - statistic applied to the double power - law . to quantify the significance of the tp - statistics deviation from zero , @xmath130 sets of 3000 events were generated for each of the three distributions discussed in this section . for each set we calculate the _ mean _ tp - statistic @xmath124 within each of the logarithmically spaced bins . the resulting distribution of @xmath124 s within each bin is then fitted to a gaussian . the fitted mean and @xmath131 deviation of the @xmath124 s ( see definition in text ) within each bin for the three distributions described in the text . this plot is the result of @xmath132 simulated sets of events , where fig.[fig : fdtp ] is one example , and where each set contains 3000 events.,width=419,height=264 ] the black circles in fig . [ fig : fdtp_skies ] represent the mean of the gaussian fit to the distribution of @xmath124 s within each bin for a power - law and the error bars on the points represent the fitted @xmath131 deviation of the @xmath124 s . we interpret the left - most of these points in the following way : for 3000 events drawn from a power - law the `` expected value '' of @xmath124 in the first bin is effectively indistinguishable from zero , as expected . though the statistic itself does not depend on @xmath17 , the variance on this value does . the reason for this is that the variance of the @xmath124 s depends on the average total number of events greater than a given minimum , which is influenced by @xmath17 . in this case the total number of events per set for minima in the first bin is at least a few thousand and the variance of the @xmath124 s is @xmath133 . these errors increase from left to right since each successively higher bin will contain @xmath124 s based on fewer and fewer events . the magenta squares represent the fitted mean @xmath124 as a function of @xmath134 for sets drawn from a power - law with a cut - off at @xmath116 . they deviate from zero for all but the largest @xmath134 . furthermore , this offset is statistically significant for the lowest few bins of @xmath134 , where the statistic reflects the deviation from power - law considering most of the events in the set . the green triangles show the fitted means for the @xmath117 distribution . they also display some deviation from zero , but they are not as significant since they fall near the @xmath131 errors for the pure power - law distribution . the distribution of the @xmath124 s in the first bin of fig.[fig : fdtp_skies ] ( the bin with minimum @xmath135 ) for the simulated pure power - law ( black , shaded ) and a power - law with a cut - off at @xmath116 ( magenta , hatched ) . for these distributions @xmath136 ( see equation ( [ eq : ptp])).,width=359,height=226 ] indeed , one may inquire as to which of the bins deviate the most from the simulated power - law . this is equivalent to asking , `` above what minimum do the data generated from this cut - off distribution maximally deviate from a pure power - law ? '' to quantify this deviation , we use a p - value given by @xmath137 where @xmath138 @xmath139 is the mean of the fitted gaussian and @xmath140 is the standard deviation . we reject the pure power - law hypothesis ( at the 5% s. l. ) if @xmath141 . the mean of the gaussian fit to the distribution of @xmath124 s for the power - law in the bin with minimum @xmath135 is @xmath142 with a standard deviation @xmath143 . the mean of the fitted gaussian for the @xmath116 distribution in this bin is @xmath144 with a standard deviation @xmath145 . therefore , the significance level of the deviation is @xmath146 and we can reject the pure power - law hypothesis for this distribution . the distribution of the @xmath124 s for this bin is plotted in fig . [ fig : fdthe1tp_100skies ] for the pure power - law ( black , shaded ) and the @xmath116 ( magenta , hatched ) pdf . the maximum deviation for the @xmath117 pdf occurs in the bin with minimum @xmath147 and the corresponding distributions of @xmath124 are plotted in fig . [ fig : fdthe2tp_100skies ] . the significance of this deviation is lower ; @xmath148 . the distribution of the @xmath124 s in the fifth bin of fig.[fig : fdtp_skies ] ( bin with minimum @xmath147 ) for the simulated pure power - law ( black , shaded ) and a power - law with a cut - off at @xmath117 ( green , hatched ) . for these distributions @xmath148 ( see equation ( [ eq : ptp])).,width=359,height=226 ] in order to apply the tp - statistic to the cr data , monte - carlo simulations were conducted and analyzed in a manner similar to that discussed in @xmath2[subsec : tp.anexample ] ; we generate @xmath130 sets of events from the reported flux and the resulting distribution of @xmath124 ( within each bin ) is fitted to a gaussian . since the significance of deviation from zero depends on both the power index and the number of events , we will compare each of the auger and agasa data sets with a unique power - law . we will take the agasa experiment to have @xmath149 events above @xmath29 and we will compare the resulting tp - statistics with those of a power - law with the same minimum and @xmath150 . the auger spectrum has a power - index estimate of @xmath151 considering all of the data above @xmath28 and a total of 3570 events , so we will therefore compare the tp - statistics arising from the auger flux to those of a pure power - law with these parameters . the application of this scheme to the agasa spectrum is plotted in fig . [ fig : tpstatsaga ] in red triangles . the black circles represent average tp - statistic value for data drawn from a pure power - law with @xmath152 . both plots have @xmath153 events per sky . the error bars on each point represent the 1-sigma deviation of the gaussian fit to the distribution of the mean tp - statistic . since the agasa values do not significantly deviate from zero ( or the power - law values ) this plot suggests that the agasa distribution does not significantly deviate from a pure power - law . the most significant deviation occurs in the bin with minimum @xmath154(ev ) and gives @xmath155 , which is consistent with the p - value for this bin discussed in @xmath2[sec : dlv ] . these distributions are plotted in fig [ fig : thetpaga ] . the fitted mean and @xmath131 deviation of the @xmath124 s ( see definition on the text ) within each bin for the agasa spectrum ( red triangles ) and a pure power - law distribution ( black circles ) . this plot is the result of @xmath132 simulated sets of events where each set contains 1916 events and the power - law has index @xmath156.,width=419,height=264 ] the distribution of @xmath124 s in the fifth bin of fig.[fig : tpstatsaga ] ( the bin with minimum @xmath157(ev ) ) for the pure power - law ( black , shaded ) and the agasa spectrum ( red , hatched ) . for these distribution @xmath155 ( see equation ( [ eq : ptp])).,width=359,height=226 ] the simulation results from the auger spectrum are plotted in fig . [ fig : tpstatsaug ] . this plot shows deviation from a power - law for the lowest minimums considered . for the bin with minimum @xmath30 we find @xmath158 . thus we can say that the auger spectrum with energies greater than @xmath159(ev ) deviate from a power - law by @xmath160 , where @xmath161 . the distribution of @xmath124 s for this minimum energy is plotted in fig . [ fig : thetpaug ] . the fitted mean and @xmath131 deviation of the @xmath124 s ( see definition on the text ) within each bin for the auger spectrum ( blue triangles ) and a pure power - law distribution ( black circles ) . this plot is the result of @xmath132 simulated sets of events where each set contains 3570 events and the power - law has index @xmath162.,width=419,height=264 ] the distribution of @xmath124 s in the second bin of fig.[fig : tpstatsaga ] ( the bin with minimum @xmath163(ev ) ) for the pure power - law ( black , shaded ) and the auger spectrum ( blue , hatched ) . for these distribution @xmath164 ( see equation ( [ eq : ptp])).,width=359,height=226 ] since the tp - statistic nearly eliminates the need to estimate @xmath17 , the biggest systematic uncertainty in analyzing the cr data with the tp - statistic is likely to be errors in the event energies . similar to the @xmath64-value discussed in @xmath2[sec : dlv ] , it is only the relative energy errors which can effect the result , since the tp - statistic depends only on the ratio . however , any elongation of the observed spectrum brought about by this relative uncertainty effect the tp - statistic . without further study of the cr energy systematics , we can not draw a conclusion from the @xmath165 deviation in fig . [ fig : thetpaug ] . in @xmath2[sec : data ] we use the reported ( agasa and auger ) cr fluxes to discuss the power - law form and illustrate the logarithmically binned estimates of the power index @xmath17 . the probability @xmath64 that the maximum value of a sample drawn from a power - law is less than or equal to a particular value is defined in equation ( [ eq : pxc ] ) . using reasonable estimates for @xmath17 , @xmath45 and @xmath79 from the cr data sets we calculate @xmath64 in @xmath2[sec : dlv ] . the value of @xmath64 is used to test the null hypothesis that these data sets follow a power - law . the agasa data give no reason to reject the hypothesis ; @xmath166 for the data with @xmath167 . the auger data give more reason to reject the null hypothesis , @xmath168 for the data with @xmath169 . however , consideration of the errors on @xmath18 prevent any solid conclusion . for the purpose of statistical analysis it would be useful to eliminate , or at least minimize , the importance of @xmath17 . the tp - statistic tends ( asymptotically ) to zero regardless of the value of @xmath17 and is the subject of @xmath2[sec : tp ] . we apply the tp - statistic to the cr data sets using a monte - carlo method described in @xmath2[subsec : tp.data ] . the agasa data give a value of @xmath155 for energies greater than @xmath154(ev ) . a value consistent with the @xmath64-value discussed in @xmath2[sec : dlv ] ( fig.[fig : pmaxlast ] ) . the preliminary auger flux results in a tp - statistic with more significant deviation from the power - law form : @xmath164 for @xmath163(ev ) . comparing this value with the @xmath64-value for this bin derived in @xmath2[sec : dlv ] , namely @xmath170 , illustrates the power of the method based on the tp - statistic which is essentially independent of gamma . better understanding of the relative errors on the cr energies should lead to a definitive conclusion on the question of a cut - off in the cr spectrum . 99 m. e. j. newman , contemporary physics * 46 * , 323 - 351 ( 2005 ) . a. clauset , m. young and k.s . gleditsch , * arxiv : physics/0606007 * ( 2006 ) . k. greisen , phys . rev , lett . * 16 * , 748 ( 1966 ) . zatsepin and v.a . kuzmin , jetp . lett . 4 * 78 * , ( 1966 ) . john n. bachall and eli waxman * arxiv : hep - ph/0206217 v5 * ( 2003 ) . auger collaboration , proceedings @xmath171 icrc , pune , india , * 10 * 115 ( 2005 ) , * arxiv : astro - ph/0604114. t. yamamoto and the pierre auger collaboration , arxiv : astro - ph/0601035 v1 * ( 2006 ) . m. takeda _ et al . _ , 81 * 6 ( 1998 ) . unbinned maximum likelihood methods have less error and bias when applied to power - law ( or similar ) distributions than binned methods@xcite . they can also be modified to include energy error and variable acceptance information@xcite . lacking this information , we use the logarithmically binned estimate of @xmath17 where necessary . the minimum variance for _ any _ estimator of @xmath17 is given by the cramer - rao lower bound ; @xmath172 . m. l. goldstein , s. a. morris , and g. g. yen , eur . j. b. * 41 * , 255 - 258 ( 2004 ) . howell , _ statistical properties of maximum likelihood estimators of power law spectra information _ , nasa / tp-2002 - 212020/rev1 , marshall space flight center , dec . , 2002 . v. pisarenko , d. sornette and m rodkin , _ deviations of the distributions of seiesmic energies from the gutenberg - richter law _ , computational seismology * 35 * , 138 - 159 ( 2004 ) , * arxiv : physics/0312020. v. pisarenko , d. sornette , _ new statistic for financial return distributions : power - law or exponential ? _ , arxiv : physics/0403075. in @xmath2[subsec : tp.anexample ] we state that the tp - statistic will be distinctly positive for distributions which contain a tail - suppression and negative for distributions which contain a tail - enhancement ( relative to the pure power - law form ) . in this section we numerically compute the tp - statistic for a `` double power - law '' distribution and describe the parameter space associated with this statistic . consider the following probability distribution function : @xmath173 where @xmath174 and @xmath175 are chosen such that @xmath176 and @xmath177 this distribution follows a power - law with index @xmath17 for @xmath178 , and @xmath179 for @xmath180 . + given the parameter set @xmath181 , we define the tp - statistic for this distribution as @xmath182^{2 } - \frac{1}{2 } \int_{u}^{\infty } \ln^{2 } \left ( \frac{x}{u } \right ) f(x)dx . \label{eq : tpbend}\ ] ] for @xmath183 and/or @xmath184 equation ( [ eq : tpbend ] ) is identically zero since it is equal to @xmath185 ( see equation ( [ eq : lm ] ) ) . however , equation ( [ eq : tpbend ] ) is non - trivial when @xmath186 and @xmath187 . in what follows , we calculate @xmath127 for @xmath186 and various values of @xmath188 and @xmath179 with @xmath189 and @xmath190 fixed . fig.[fig : appxbendpdf ] contains a plot of @xmath191 versus @xmath192 with @xmath193 for several choices of @xmath194 ( namely , for @xmath194 varying from 1 to 2 in steps of 0.2 ) . the red curves correspond to @xmath195 ( tail - suppression ) and the blue curves have @xmath196 ( tail - enhancement ) . the tp - statistic for each of these distributions is shown in fig.[fig : appxbendtp ] as a function of @xmath110 . examination of fig.[fig : appxbendtp ] suggests the following conclusions for a given @xmath17 and @xmath179 : @xmath127 is positive for all values of @xmath110 and @xmath188 if and only if @xmath197 , and it is negative if and only if @xmath198 . * for @xmath188 much greater than @xmath134 , @xmath199 is approximately zero . specifically , as @xmath200 , @xmath201 . * the location of the maximum deviation , say @xmath202 where @xmath203 is highly correlated with the location of the bend @xmath188 . indeed , we have found that there is a linear relationship between @xmath204 and @xmath194 and that this relationship is independent of whether @xmath17 is less than or greater than @xmath179 . * the maximum deviation of the tp - statistic , i.e. @xmath205 , is independent of @xmath194 . a plot of @xmath191 ( see equation ( [ eq : pdfbend ] ) ) versus @xmath192 with @xmath193 for several choices of @xmath194 ( namely , for @xmath194 varying from 1 to 2 in steps of 0.2 ) . the red curves correspond to @xmath195 ( tail - suppression ) and the blue curves have @xmath196 ( tail - enhancement ) . the more black the color of the curve , the larger @xmath194.,width=359,height=226 ] a plot of @xmath127 ( see equation ( [ eq : tpbend ] ) ) for each of the distributions plotted in fig.[fig : appxbendpdf ] . those distributions with tail - suppression ( red ) have @xmath206 and those with tail - enhancement ( blue ) have @xmath207.,width=359,height=226 ] to isolate the effects of power index choice , consider the family of distributions where @xmath208 is fixed but @xmath179 is allowed to vary . since the integrals in equation ( [ eq : tpbend ] ) only converge if @xmath209 , the minimum @xmath179 we can choose is @xmath210 . there is no upper bound on @xmath179 so we vary this parameter over the interval @xmath211 in steps of 0.2 and over the interval @xmath212 in steps of 0.5 . fig.[fig : appdeltapdf ] contains a plot of @xmath191 versus @xmath192 with @xmath208 and @xmath213 . the blue curves have @xmath211 ( i.e. @xmath214 ) and the red curves have @xmath212 ( i.e. @xmath215 ) . the more black the color of these curves , the closer @xmath179 is to @xmath17 . a plot of @xmath191 ( see equation ( [ eq : pdfbend ] ) ) versus @xmath192 with @xmath208 and @xmath213 . the blue curves have @xmath211 ( i.e. @xmath214 ) and the red curves have @xmath212 ( i.e. @xmath215 ) . the more black the color of these curves , the closer @xmath179 is to @xmath17.,width=359,height=226 ] fig.[fig : appdeltatp ] contains a plot of @xmath127 for the distributions plotted in fig.[fig : appdeltapdf ] . as noted earlier , @xmath206 if and only if @xmath216 and @xmath207 if and only if @xmath217 . the colored points on these curves show where each curve maximally deviates from zero ; the coordinates of these points are @xmath218 for each curve ( see equation ( [ eq : u0def ] ) ) . these points show a weak dependence of @xmath204 on @xmath179 , for a given @xmath194 . a plot of @xmath127 ( see equation ( [ eq : tpbend ] ) ) for the distributions plotted in fig.[fig : appdeltapdf ] . the colored points on these curves show where each curve maximally deviates from zero ; the coordinates of these points are @xmath218 for each curve ( see equation ( [ eq : u0def])).,width=359,height=226 ] the value of the maximum deviation @xmath205 also shows dependence on @xmath179 . in fig.[fig : appdeltatpu0 ] we plot @xmath205 versus @xmath219 for each of the points in fig.[fig : appdeltatp ] . these plots suggest the following : * for @xmath220 ( blue and black ) , a small change in @xmath179 will lead to a large change in @xmath205 . * if @xmath221 ( bright red ) , however , a large change in @xmath179 will result in a small change in @xmath205 . this case is of particular interest since a large @xmath179 will mimic the cutoff distribution defined in equation ( [ eq : fdpdf ] ) . * by inspection of fig.[fig : appdeltatpu0 ] we note that @xmath222 for @xmath221 . * comparison with fig.[fig : appxbendtp ] suggests that the limiting value of @xmath205 is roughly independent of @xmath188 . the studies described in this section show that the tp - statistic can distinguish tail - suppressed ( @xmath223 ) from tail - enhanced ( @xmath224 ) distributions , i.e. @xmath206 if and only if @xmath216 and @xmath207 if and only if @xmath217 . furthermore , they show that in the limiting case of @xmath221 the most important parameter in determining @xmath202 is @xmath188 but that the limiting value of @xmath205 is roughly independent of @xmath188 and @xmath225 .	two separate statistical tests are applied to the agasa and preliminary auger cosmic ray energy spectra in an attempt to find deviation from a pure power - law . the first test is constructed from the probability distribution for the maximum event of a sample drawn from a power - law . the second employs the tp - statistic , a function defined to deviate from zero when the sample deviates from the power - law form , regardless of the value of the power index . the agasa data show no significant deviation from a power - law when subjected to both tests . applying these tests to the auger spectrum suggests deviation from a power - law . however , potentially large systematics on the relative energy scale prevent us from drawing definite conclusions at this time . high energy cosmic ray flux power - law tp - statistic
* model parameters . * the single fermion hamiltonian eq . ( [ fh ] ) is obtained by adiabatically integrating out all the electronic excitation states of the atoms in the rotating wave frame . the parameters in eq . ( [ fh ] ) are related to the experimental tunable parameters as @xmath116 , @xmath117 and @xmath118 . here @xmath119 is the pumping field strength , @xmath120 is the pumping laser frequency detuning with respect to electronic transitions of atoms , and @xmath121 is the coupling strength between the cavity mode and the fermions . * mean field equation for fermi superfluids . * when the lattice induced by the pumping field is not strong , we can approximate @xmath122 as a constant . the mean field gap equation becomes @xmath123 together with the number equation @xmath124 , or more explicitly , @xmath125 we can determine @xmath41 and @xmath126 self - consistently for a given pumping strength @xmath89 and given density @xmath7 . * instability condition for superradiant phase transition . * the mean field value of the cavity field @xmath127 satisfies @xcite @xmath128 where @xmath56 is the fermion density order parameter . the introduced decay rate @xmath62 is to model the weak leakage of electromagnetic field from the high-_q _ cavity . in a steady state , @xmath129 ; we have @xmath130 which locks the cavity field to the fermion density order parameter . both @xmath131 and @xmath132 are zero in the normal phase and become nonzero in the superradiant phase . to the second order of @xmath131 , the effective free energy can be obtained as @xmath133 where @xmath134 with a specified @xmath131 . by substituting ( [ mean_alpha ] ) into eq . ( [ fa ] ) , we have @xmath135\eta_0 ^ 2\theta^2,\label{freeenergy}\end{aligned}\ ] ] which determines the superradiant transition when the quadratic coefficient of @xmath132 changes its sign . * explicit expression for density - wave order susceptibility . * the explicit expressions for the density - wave order susceptibility within the bcs theory are @xmath136 in the bsc limit , the factor @xmath137 with @xmath138 the fermi - dirac distribution ; @xmath139 becomes the same as it is for free fermions @xcite . in the bec limit , @xmath140 and @xmath141 , @xmath142 which is the same as it is for condensed noninteracting bosons @xcite . * determination of phase boundary . * the boundary between the non - superradiant and superradiant phases is determined by eq . ( [ cri ] ) . since @xmath143 , @xmath144 , and @xmath87 is a dimensionless function of dimensionless parameters @xmath89 and @xmath98 , we could recast eq . ( [ cri ] ) in the form @xmath145 by introducing @xmath146 . we take typical experimental values @xmath147 and @xmath114 . thus at each given pumping strength @xmath89 we can obtain the critical strengths of the cavity detuning @xmath63 .	in this letter we consider the superradiant phase transition of a two - component fermi gas in a cavity across a feshbach resonance . it is known that quantum statistics plays a crucial role for the superradiant phase transition in atomic gases ; in contrast to bosons , in a fermi gas this transition exhibits strong density dependence . we show that across a feshbach resonance , while the two - component fermi gas passes through the bec - bcs crossover , the superradiant phase transition undergoes a corresponding crossover from a fermionic behavior on the weakly interacting bcs side , to a bosonic behavior on the molecular bec side . this intricate statistics crossover makes the superradiance maximally enhanced either in the unitary regime for low densities , in the bcs regime for moderate densities close to fermi surface nesting , or in the bec regime for high densities . recent experiment has combined atomic bose - einstein condensates and cavity quantum electrodynamics together where atom - light interactions are strongly enhanced @xcite . a superradiant phase transition driven by external pumping field has been observed , across which atoms form a density - wave order @xcite , and roton mode softening has been found in the vicinity of this superradiant phase transition @xcite . theoretical studies have extended to investigate noninteracting fermi gases inside a cavity @xcite . it is shown that the fermi statistics plays a crucial role in the superrandiant phase transition at moderate and high densities @xcite . at moderate densities , fermi surface displays a nesting structure and strongly enhances superradiance , when the momentum of the cavity light field matches the nesting momentum . while at high densities , pauli blocking effect forbids a large part of atom - light scattering processes , and consequently , strongly suppresses superradiance . the strong density dependence marks the major difference between superradiances in ideal fermi gases and bose gases . during the past decade , another important development in cold atom physics is the study of strongly interacting two - component fermi gases and the bec - bcs crossover utilizing feshbach resonance @xcite . the inter - atomic @xmath0-wave scattering length @xmath1 can be continuously changed by a feshbach resonance , and the dimensionless parameter @xmath2 ( @xmath3 is the fermi momentum in the noninteracting limit ) controls the bec - bcs crossover . in the bcs limit of the crossover @xmath4 , fermions form loosely bound cooper pairs and the low - energy response is dominated by fermionic quasi - particles ; the system recovers a noninteracting fermi gas . in the bec limit @xmath5 , cooper pairs transform into tightly bound bosonic molecules , and the system responses to external fields mainly as bosons . in between , when @xmath1 is so large that @xmath6 , the system is in a strongly interacting regime and its response shall exhibit both bosonic and fermionic characters . so far , fermi gases with inter - atomic interactions in a cavity have been barely studied . in this work we consider a two - component fermi gas in a cavity across a feshbach resonance . given that the gas can be continuously tuned between the fermion limit and the boson limit , and that atoms with different statistics have been shown to behave differently in the superradiant phase transition @xcite , the motivation of our study is to address how the statistics crossover manifests itself in the superradiant phase transition across a feshbach resonance , and the physical consequence of this crossover . in experiments , the superradiant phase transition is usually driven by increasing the strength of pumping field . in this work we will reveal nontrivial dependence of the critical pumping strength on the density of fermions @xmath7 and the inter - atomic interaction strength characterized by @xmath2 . our results represent a manifestation of the interplay between strong interactions from feshbach resonance and strong atom - light coupling in a cavity , and will provide insight for future experiments . _ model . _ our system is schematically shown in fig . ( [ setup ] ) . applied on the fermi gas is a pumping field that consists of two laser beams counter - propagating along the @xmath8 direction , with frequency @xmath9 and polarization in the @xmath10 direction . the single - mode cavity field of interest varies in the @xmath11 direction , with frequency @xmath12 close to @xmath9 . the system is described by the hamiltonian @xmath13 , where @xmath14 is the field operator for the cavity mode and @xmath15 is the cavity field detuning . direction shown by the red arrows . the cavity field is represented by the wiggled lines in the @xmath16 direction . fermions of different spins are shown in different colors . , width=226 ] the hamiltonian experienced by the fermions has two parts @xmath17 . the free fermion part is @xcite @xmath18 where @xmath19 are the fermion field operators with ( peudo ) spin index @xmath20 . the pumping field and the cavity field generate respectively the optical potentials @xmath21 and @xmath22 , and the coupling between the pumping field and the cavity field comes from an interference term @xmath23 with @xmath24 , @xmath25 is the wavevector magnitude of both the pumping field and the cavity mode @xcite . the recoil energy @xmath26 is defined for latter use . the inter - atomic interaction nearby a feshbach resonance is described by the hamiltonian @xmath27 the bare attractive inter - fermion interaction coupling @xmath28 is renormalized to the @xmath0-wave scattering length @xmath29 via @xmath30 with @xmath31 the momentum cutoff . this attractive interaction between fermions lead to fermion pairing and a fermi superfluid ground state . _ ground state in non - superradiant phase . _ before entering the suprradiance phase , @xmath32 , fermions only experience a one - dimensional lattice @xmath33 along the direction of the pumping field , and the single - particle eigenstates are the bloch states @xmath34 satisfying @xmath35 . by expanding @xmath36 with @xmath37 and @xmath38 the gas volume , we introduce fermion pairing order parameter @xmath39@xmath40 . here we assume the lattice @xmath33 is weak and we have ignored pairing at non - zero crystal momentum . with this assumption , the order parameter @xmath41 is determined by the gap equation and the number equation @xcite . in this fermi superfluid state , the single - particle green s functions are given by @xmath42 @xmath43 with @xmath44 and the fermionic matsubara frequencies @xmath45 for @xmath46 , and @xmath47 the inverse of temperature . their diagrams are shown in fig . ( [ diagram])(a1 ) . the components @xmath48 and @xmath49 describe the propagation of particles and holes , respectively , while @xmath50 and @xmath51 are the anomalous green s functions proportional to the pairing gap @xmath41 which we take to be real . , title="fig:",width=207 ] + _ condition for superradiant phase transition . _ the superradiant phase transition is determined by the instability of non - superradiance toward developing non - zero @xmath52 . as shown in ref . @xcite , the superradiant phase transition occurs simultaneously with the formation of density - wave order of atoms with momentum @xmath53 , where @xmath54 is the momentum transfer between the cavity field and the pumping field . that is to say , @xmath55 is proportional to the density - wave order parameter @xmath56 with @xmath57 . by integrating out the fermion fields , one can obtain the free - energy of the system in the form @xmath58 @xcite , where @xmath59 changing sign from positive to negative gives the critical pumping field strength @xmath60 for the superradiance transition @xcite @xmath61 here @xmath62 is the cavity mode decay rate , and @xmath63 is the shifted cavity mode detuning @xmath64 , which is assumed to be red - detuned ( @xmath65 ) . the most essential quality determining this transition is the density - wave order susceptibility of the fermi superfluid defined as @xmath66.\end{aligned}\ ] ] here @xmath67 includes the integration of the spatial coordinates and the imaginary times , @xmath68 is the time ordered operator . the expectation value of the fermion operators @xmath69 is taken in the non - superradiant fermi superfluid phase . a larger @xmath70 means that the fermi gas has stronger tendency toward forming a density - wave order at a momentum @xmath53 , and it is easier for the fermi gas to enter the superradiant phase ; in another word , the critical pumping strength shall be smaller . _ density - wave order susceptibility . _ in order to capture both fermionic and bosonic responses of a fermi superfluid , the density - wave order susceptibility @xmath70 should be calculated by the random phase approximation . this approximation maintains conservation laws @xcite and guarantees that one can recover the free fermion and the free boson results in the limits @xmath71 , respectively . within this approximation , we have @xmath72 and @xmath73 where @xmath74 is the mode factor for cooper pair fluctuations . the fermionic response @xmath75 is due to that the cavity field couples to the fermonic excitations of the fermi superfluid by breaking up cooper pairs . the feynman diagrams corresponding to @xmath75 are shown in fig . ( [ diagram])(b ) . the diagrams describe the process that a fermion with momentum @xmath76 is scattered to momentum @xmath77 , where the momentum transfer comes from the photon momentum change from the pumping field to the cavity field , as denoted by the vertex in fig . ( [ diagram])(a2 ) . since all fermions are paired in the fermi superfluid phase , this process must be accompanied by pair breaking . in the bcs limit where the pairing gap @xmath41 is small and pairs are easy to break , @xmath75 is dominant in @xmath70 and could recover the transition for free fermions in the limit of vanishing pairing gap @xcite . while in the bec limit this process is strongly suppressed because of large pairing gap . the bosonic response @xmath78 originates from the process that the cavity field excites nonzero momentum cooper pairs and corresponds to the diagram shown in fig . ( [ diagram])(c ) . in this process , one of the two fermions in the cooper pair , say , the one with momentum @xmath76 , is scattered to momentum @xmath79 by a photon . thus , the cooper pair acquires a finite momentum @xmath80 and propagates with this fixed momentum @xmath80 ( up to a reciprocal lattice vector along @xmath8 ) . after another scattering with a photon , the cooper pair returns to zero - momentum . because of weak lattice @xmath81 we only take into account the contributions from the scattered cooper pairs of momentum @xmath82 . the cooper pair propagator @xmath83 is given in eq . ( [ chib3 ] ) and its diagram in fig . ( [ diagram])(c ) which is a summation of ladder diagrams . there are two ways for a cooper pair to propagate , through multiple scattering and through vacuum fluctuations , respectively . both are included in eq . ( [ chib3 ] ) and in the bottom of fig . ( [ diagram])(c ) . in the bec limit @xmath78 is dominant in @xmath70 and @xmath84 recovering the free boson result . while in the bcs limit , @xmath85 is exponentially suppressed @xcite . , @xmath86 and @xmath87 vs @xmath88 are plotted in ( a ) , ( b ) , ( c ) respectively with the pumping strength @xmath89 fixed at @xmath90 and @xmath91 taking @xmath92 , @xmath93 and @xmath94 . the bottom row is a pictorial representation of the bec - bcs crossover . [ fffbf],width=264 ] we plot in fig . ( [ fffbf ] ) the dimensionless susceptibility @xmath95 , as well as its fermionic and bosonic constituent @xmath96 and @xmath97 , as functions of the bec - bcs crossover controlling parameter @xmath2 , for different filling fractions @xmath98 . first , fig . ( [ fffbf])(a ) shows that @xmath99 exhibits strong density dependence on the bcs side . around a moderate density of @xmath100 , the fermi surface nesting is optimal , and @xmath99 becomes much larger than the low - density limit value @xmath101 @xcite . this is the regime where fermi surface nesting strongly enhances superradiance , as discussed in noninteracting fermi systems @xcite . on the other hand , for high densities , say , @xmath102 in fig . ( [ fffbf])(a ) , @xmath99 is much smaller than @xmath103 on the bcs side . this is the regime where the pauli exclusion principle strongly suppresses superradiance . as approaching the bec side , @xmath99 is strongly suppressed for all densities . second , as shown in fig . ( [ fffbf])(b ) , @xmath104 approaches the value of noninteracting bosons ( also @xmath105 ) in the bec limit , independent of densities . while on the bcs side , for all densities @xmath104 is strongly suppressed . figure ( [ fffbf])(c ) shows the central result of this work . the total @xmath87 exhibits different features for different densities as @xmath88 varies . the most intriguing case is at relatively low - densities , say , @xmath106 , where @xmath87 displays a maximum in the unitary regime ( @xmath107 ) . this maximum is because in this regime , the bosonic contribution already takes off while the fermionic contribution has not damped out . while for moderate densities of fermi surface nesting regime , @xmath87 monotonically increases as @xmath2 increases from the bec limit to the bcs limit , due to the fermi - surface nesting enhancement of @xmath70 on the bcs side . in contrast , for high densities , @xmath87 monotonically decreases , due to the pauli blocking suppression of @xmath70 on the bcs side . the total @xmath87 has strong density dependence on the bcs side where it is dominated by the fermionic behavior , and becomes less and less sensitive to density in the bec limit where it is dominated by the bosonic behavior . this change of @xmath87 with @xmath2 between the two limits is the manifestation of statistics crossover in superradiance . _ phase diagram . _ the boundary separating the normal and the superradiant phases can be obtained by solving eq . ( [ cri ] ) @xcite . in fig . ( [ pd ] ) , we plot the phase diagram in term of @xmath108 and @xmath109 for different densities and interaction strengths . in the bcs region , fig . ( [ pd ] ) ( a ) shows that the moderate density @xmath110 is the easiest to be superradiant . in the unitary region as shown in fig . ( [ pd ] ) ( b ) the low density @xmath111 is the easiest to be superradiant primarily due to the maximum of @xmath87 mentioned above in this part of the parameter space . on the bec side , fig . ( [ pd ] ) ( c ) shows that the density dependence diminishes since it shall be washed out completely in the bec limit . ( a ) , @xmath112(b ) and @xmath113 ( c ) , and with different densities @xmath111 , @xmath93 and @xmath94 . for all cases , we take the typical experimental parameters @xmath114 and @xmath115.,width=302 ] _ conclusion . _ we have presented basic features of the superradiant phase transition of two - component fermi gases across a feshbach resonance . the main results are : i ) on the bcs side of resonance the superradiant phase transition shows strong density dependence , similar as noninteracting fermi gas ; while on the bec side it gradually becomes density independent , similar as noninteracting bosons . ii ) superradiance is mostly enhanced in the unitary regime for low density , in the bcs regime for moderate density , and in the bec regime for high density . in this work , we have only focused on the superradiant phase transition itself . inside the superradiant phase , the additional lattice due to the cavity field will further modify the single - particle dispersion , which will feedback to the fermi superfluid . furthermore , the quantum fluctuation of the cavity field will also generate additional effect on the fermi superfluid . the properties of fermi superfluids in the superradiant phase would be a subject for future studies . _ acknowledgements . _ this work is supported by tsinghua university initiative scientific research program , nsfc under grant no . 11004118 , no . 11174176 , no . 11104157 , no . 11474179 and no . 11204152 , and nkbrsfc under grant no . 2011cb921500 . j. mckeever , a. boca , a.d . boozer , j.r . buck and h.j . kimble , nature * 425 * , 268 ( 2003 ) f. brennecker , t. donner , s. ritter , t. bourdel , m. khl and t. esslinger , nature ( london ) * 450 * , 268 ( 2007 ) . y. colombe , t. steinmetz , g. dubois , f. linke , d. hunger and j. reichel , nature ( london ) * 450 * , 272 ( 2007 ) . a. t. black , h. w. chan and v. vuleti , phys . rev . lett . * 91 * , 203001 ( 2003 ) . k. baumann , c. guerlin , f. brennecke and t. esslinger , nature * 464 * , 1301 ( 2010 ) . k. baumann , r. mottl , f. brennecke and t. esslinger , phys . rev . lett . * 107 * , 140402 ( 2011 ) . r. mottl , f. brennecke , k. baumann , r. landig , t. donner , t. esslinger , science * 336 * , 1570 ( 2012 ) j. larson , g. morigi , and m. lewenstein , phys . rev . a * 78 * , 023815 ( 2008 ) r. kanamote and p. meystre , phys . rev . lett . * 104 * , 063601 ( 2010 ) . m. mller , p. strack , and s. sachdev , phys . rev . a * 86 * , 023604 ( 2012 ) . y. chen , z. yu , and h. zhai , phys . rev . lett . * 112 * , 143004 ( 2014 ) . j. keeling , m.j . bhaseen , and b.d . simons , phys . rev . lett . * 112 * , 143002 ( 2014 ) . f. piazza , and p. strack , phys . rev . lett . * 112 * , 143003 ( 2014 ) . for a review , see s. giorgini , l.p . pitaevskii , and s. stringari , rev . mod . phys . * 80 * , 1215 ( 2008 ) c. chin , r. grimm , p. julienne , and e. tiesinga , rev . mod . phys . * 82 * , 1225 ( 2010 ) . see supplementary material for ( i ) relations between model parameters and experimentally tunable parameters ; ( ii ) details of gap and number equation ; ( iii ) derivation of free - energy in term of order parameter ; ( iv ) asymptotic expression of @xmath75 in the bcs limit ; ( v ) asymptotic expression of @xmath78 in the bec limit ; ( vi ) details of solving the equation for phase boundary . g. baym and l.p . kadanoff , phys . rev . * 124 * , 287 ( 1961 ) ; g. baym , phys . rev . * 127 * , 1391 ( 1962 ) . z. yu and g. baym , phys . rev . a * 73 * , 063601 ( 2006 ) .
in standard @xmath0 cold dark matter ( @xmath0-cdm ) cosmology @xcite galaxies form within dark matter halos that coalesce through repeated mergers . high resolution pure cdm simulations give rise to significantly triaxial galaxy halos , that become progressively less triaxial towards lower sizes and masses due to the increased dynamical age of the lower mass structures @xcite . the inclusion of baryons in galaxy formation simulations alters dramatically the behavior of the central regions of the halo , which added to the effect of merging satellites makes the central halo become rounder @xcite . in the milky way , we have perhaps the best opportunity to constrain the shape of any dark matter distribution . one particularly promising strategy to accomplish this , and which is generally not possible in extragalactic systems , is to use streams of stars to probe the global potential . the streams of low - mass satellites follow fairly closely the centre of mass orbit of their progenitor , and with careful modeling it is possible to find a solution for the stream structure within a galactic mass distribution . this approach has been used in the past to model the stream of the sagittarius ( sgr ) dwarf galaxy @xcite , probably the most significant accretion into the milky way in the last @xmath1 . recently , ( * ? ? ? * hereafter lm10 ) presented a new analysis of the spatial and kinematic structure of m - giant star members of this stream @xcite , and showed that their data could be reproduced if the dark halo of the milky way has a triaxial form with a minor to major axis ratio of @xmath2 and intermediate to major axis ratio of @xmath3 ( in the potential ) . the structure is therefore almost an oblate ellipsoid , but with its minor axis contained within the galactic plane , which presumably induces non - circular orbits at radii where the disk is not dominant . as lm10 point out , such a halo is not natural in @xmath0-cdm . it is also worth noting that their model has an extremely high degree of oblateness in density . these conclusions have been further reinforced by the analysis of ( * ? ? ? * hereafter dw12 ) who in fitting the stream with orbits in a galactic mass model , found a similar oblate dark matter halo also oriented perpendicular to the galactic plane . in this contribution we aim to determine whether it is possible to reproduce the structure and kinematics of the sgr stream , at least as well as lm10 , without invoking a triaxial galactic halo . one means to model the formation of the sgr stream would be to make a mass model of the milky way and an n - body model of the sgr dwarf , and to evolve the latter inside the potential provided by the former using an n - body integrator . through a judicious sampling of orbital and structural parameters of the two bodies it would be possible to assess the triaxiality issue this is essentially the strategy adopted by lm10 , but it has the disadvantage of being computationally extremely costly , which greatly limits the resolution of the parameter space survey . an alternative approach , adopted by dw12 , is to follow orbits in a mass distribution , however due to self - gravity this is not a good approximation for streams of massive satellites @xcite . instead , we have chosen to use a new stream - fitting algorithm presented in @xcite , which searches through the parameter space of the host galaxy mass model and the satellite s orbital properties and mass using a markov chain monte carlo ( mcmc ) scheme . as we have shown in that contribution , the method is able to provide a very close approximation to a stream without having to resort to n - body simulations . the simple insight behind this approximation is that a leading stream is formed from stars that leave the satellite close to the inner l1 lagrange point , while the trailing stars leave close to the l2 lagrange point . in this way , we avoid having to undertake a full n - body simulation , and can feasibly probe millions of model parameter combinations in an automatic way with a mcmc algorithm . following ( * ? ? ? * hereafter db98 ) , we model the milky way as a sum of density components , using the multipole expansion technique to solve the poisson equation . for the bulge , disk , thick disk and interstellar medium component , we have adopted the parameter values of models ` 2b ' and ` 2c ' from db98 : these are realistic density models that fit a wide range of observational constraints . db98 fit these baryonic components within a halo component modeled as a double power law ; the main difference between their models ` 2b ' and ` 2c ' is that ` 2b ' was forced to have an inner power - law exponent @xmath4 , while model ` 2c ' had @xmath5 , similar to that of a universal ( nfw ) cosmological halo model @xcite ( naturally , these constraints on the halo affect the fitted parameters of the other components ) . however , instead of adopting the db98 double power - law dark matter halo model , we allow for a much larger amount of flexibility in the halo radial density profile , as we describe next . the halo is set to be an axisymmetric model , defined by a density at @xmath6 together with an inner power law exponent @xmath7 inside of @xmath6 , a density at @xmath8 together with a power law exponent @xmath9 beyond that radius , and density values at 8 logarithmically - spaced radial locations between 10 and @xmath8 . a spline function is used to interpolate the density between these anchor points . since the dynamics of the sgr stream should not be very sensitive to the mass distribution in the inner galaxy , or the distribution in the outer galaxy beyond its apocenter , we decided to fix @xmath10 and @xmath11 , similar to the inner and outer behavior of an nfw halo model . however , the 10 halo density values are free parameters in our modeling . although our code also allows one to set ( or fit ) the flattening in density ( @xmath12 ) of the axisymmetric halo component , for this particular experiment we chose to probe the case for a spherical halo , and so kept @xmath13 fixed . the remaining parameters that need to be defined are the mass , 3-dimensional distance and velocity of sgr as well as the position and velocity of the sun , in order to correctly project the stream kinematics into line of sight observables . for the initial analysis presented here , we chose @xmath14 @xcite , and following lm10 we adopted @xmath15 , and @xmath16 ; these values were kept fixed . the total space velocity and direction of sgr was allowed to vary , as were the galactocentric radius of the sun @xmath17 and the circular velocity of the local standard of rest @xmath18 . to be consistent with @xcite , we adopted their chosen solar peculiar velocity of @xmath19 . the mcmc algorithm that drives the simulation has been updated from that used in @xcite , and now uses an affine - invariant ensemble - sampler scheme ( with parallel tempering ) , identical to that described in @xcite . the algorithm starts off from an initial guess for the 14 model parameters ( 10 halo density values , 2 velocity parameters for sgr , as well as @xmath17 and @xmath18 ) , to begin its exploration of parameter space . our initial estimations for the mcmc steps were adjusted dynamically in order to attain a target acceptance ratio of 25% . at each iteration , the algorithm makes a stream model that consists of a prediction of the locus of a tidal stream in space . we convert this to a probability density function by convolving the points by gaussian functions that approximate the combined effect of the uncertainties on the observations ( position , distance , velocity ) and the expected intrinsic width of the stream . finally , a likelihood for this set of parameter values is calculated from the product of the probabilities of the data ( m - giant positions , distances and velocities ; and sdss field positions and distances ) and the likelihood of the modelled galactic rotation given the compilation by @xcite . the best - fit model in @xmath20 iterations of the above algorithm adopting the bulge , disk , thick disk and interstellar medium parameters of model ` 2c ' of db98 is shown in figures [ fig : radec]-[fig : xz ] . figure [ fig : radec ] displays the sky positions ( top panels ) and heliocentric distances ( bottom panels ) of the mcmc solution . the locus of the stream in these parameters is displayed in green , these are convolved with the parameter uncertainties and intrinsic stream width estimate to produce the probability density function displayed in the background grayscale image . for comparison we also display the lm10 triaxial n - body model ( small purple dots ) . the left - hand panels show the positions of m - giant stars ( red circles ) from lm10 within @xmath21 of the orbital plane of the sgr dwarf @xcite ; ra ranges close to the disk or close to the main body of the sgr dwarf have been masked out . we discarded the region immediately surrounding sgr as we wanted to avoid being affected by the dynamical properties of the progenitor ( e.g. rotation ) . panel ( b ) shows the sky position of the main detection of the sgr stream in the sdss @xcite , as measured by us directly from sdss data . panel ( d ) shows the distance measurements of @xcite . in both right - hand panels we have masked out ranges in ra not probed by the data . it can be seen that the mcmc solution follows closely the position and distance constraints . figure [ fig : vel ] compares the radial velocity profile of the model to the m - giant data : the kinematic behavior is similar to that of the lm10 model , and reproduces well the trend in the m - giant radial velocities . the three - dimensional structure of the data and models is shown in figure [ fig : xz ] , which gives an edge - on view of the milky way . as before , the red circles are the selection of lm10 m - giants , with sdss red - clump distances from @xcite shown with orange circles . the green dots mark the locus of the mcmc stream solution : the correspondence with the m - giant distances is not perfect in the leading arm , but neither is that of the lm10 n - body simulation ( purple dots ) . note that the m - giant distance uncertainties are considerable ( @xmath22 @xcite ) , and there may also be a systematic difference with respect to the measurements based on the red clump or main sequence . a comparison of the relative likelihood of our model versus that of lm10 is not very useful at this stage , the models are very different , and indeed both models fail to reproduce the data in detail ( such as the stream bifurcation @xcite ) , but it can be seen that in position , distance and velocity the two models are similar , and give a similarly good representation of the youngest parts of the leading and trailing arms . while we use very similar constraints and assumptions to lm10 , we have fit the sgr stream without having to invoke the need for triaxiality . however , the price to pay for this is a peculiar radial mass distribution in the halo . figure [ fig : rotationcurve ] shows the required rotation curve and corresponding density profile of the fitted spherical halo , again using the ( fixed ) non - halo components of model ` 2c ' from db98 . the red dots in panel ( a ) mark the measurements from @xcite . beyond the solar circle there is substantial uncertainty in the rotation curve , as the large scatter reveals , and beyond @xmath23 there are virtually no direct constraints . expectations from @xmath0-cdm would suggest that the rotation curve should decline slowly beyond @xmath6 @xcite , in stark contrast to our model s requirement . nevertheless , we note that hi observations of our neighboring galaxy , andromeda ( black dots in panel ` a ' @xcite ) , show a rotation curve rise beyond @xmath24 that is similar to our model . while it was not our intention to attempt to fit an accurate mass model of the inner galaxy , it is worth noting that the surface mass density of the disk and the vertical force @xmath25 lie within observationally - acceptable ranges ( see db98 ) . in addition , the oort constants for our model ( which are not identical to those of db98 since we have altered the halo ) are @xmath26 and @xmath27 , close to the observed @xmath28 , @xmath29 ( see table 3 of dw12 for comparison to other solutions ) . the detailed structure of the mass distribution in the inner galaxy turns out to be relatively unimportant in fitting the sgr stream : if we use the non - halo components of model ` 2b ' from db98 ( instead of those of ` 2c ' as in figures 1 - 4 ) the mcmc algorithm also converges on similar solutions that require a rising rotation curve . the analysis we have presented here shows that it is possible to reproduce the spatial and kinematic structure of the sgr stream at approximately the same level of precision as lm10 without the need for a triaxial halo , if the galactic rotation curve rises significantly in the range @xmath30 to @xmath31 . interestingly , the break in the density profile occurs at approximately where the outer stellar halo component proposed by @xcite becomes dominant . this rotation curve rise of course places a substantial amount of matter in the outer galaxy , which has the effect of braking the leading arm , bringing its velocity into agreement with that of the m - giants , particularly in the region around @xmath32 . in addition to the mcmc parameter survey discussed above , we have also attempted an interactive investigation , choosing parameter combinations judiciously ; however , we were unable to find another acceptable solution without a strong rotation curve rise . it is unclear at present what the structure of galactic mass dark matter halos actually is in nature , but in @xmath0-cdm , dark matter density profiles decline as @xmath33 only when averaged over a very large number of halos . the profile of fig . [ fig : rotationcurve]b is therefore not inconsistent with @xmath0-cdm , and indeed a rotation curve rise of similar magnitude is observed in m31 over a similar radial range ( fig . [ fig : rotationcurve]a ) . we suspect ( although we have not modeled this ) that there is a trade - off between triaxiality and rotation curve rise , so that a less triaxial model than that of lm10 ( which is actually an oblate spheroid oriented perpendicular to the galactic disk ) can be fitted with a less extreme rotation curve than that presented here . a consequence of this new solution is that the milky way would be considerably more massive than estimated from models with a falling rotation curve within @xmath34 . assuming a virial radius of @xmath35 @xcite , leads to a virial mass of @xmath36@xmath37 for our model . it is interesting to note that the lmc and smc would clearly be bound to the milky way if this reflects the true mass of our galaxy , contrary to conclusions based on hst proper motion measurements that assume a standard galactic mass model ( see.e.g . , @xcite ) . to be consistent with the timing argument limit for the mass of the local group of @xmath38 @xcite , the dark matter would have to be primarily concentrated within the two giant galaxies . while the above mass for the milky way is certainly substantially higher than most recent estimates ( e.g. , @xcite ) , almost all the mass constraints at large distance are based on jeans equations solutions that make the unjustified assumptions of statistical independence of the halo tracers and some form for the orbital anisotropy . in summary , we present this solution as a means to avoid the triaxial halo models of lm10 or dw12 , and in particular their peculiar orientation as oblate structures perpendicular to the galactic plane . we recognize of course that the solution presented here is also strange , being spherical and possessing a very high mass , but we judge that it is not ruled out by observational constraints . the significant shortcomings of both the lm10/dw12 and our models suggest that they do not include sufficient of the actual complexity of ( probably both ) sgr and the milky way halo . future modeling of the more ancient parts of the sgr stream may help discriminate between these models , and pave the way to better solutions . r.a.i . gratefully acknowledges support from the agence nationale de la recherche though the grant pommme ( anr 09-blan-0228 ) . gfl thanks the australian research council for support through his future fellowship ( ft100100268 ) and discovery project ( dp110100678 )	recent analyses of the stellar stream of the sagittarius dwarf galaxy have claimed that the kinematics and three - dimensional location of the m - giant stars in this structure constrain the dark matter halo of our galaxy to possess a triaxial shape that is extremely flattened , being essentially an oblate ellipsoid oriented perpendicular to the galactic disk . using a new stream - fitting algorithm , based on a markov chain monte carlo procedure , we investigate whether this claim remains valid if we allow the density profile of the milky way halo greater freedom . we find stream solutions that fit the leading and trailing arms of this structure even in a spherical halo , although this would need a rising galactic rotation curve at large galactocentric radius . however , the required rotation curve is not ruled out by current constraints . it appears therefore that for the milky way , halo triaxiality , despite its strong theoretical motivation , is not required to explain the sagittarius stream . this degeneracy between triaxiality and the halo density profile suggests that in future endeavors to model this structure , it will be advantageous to relax the strict analytic density profiles that have been used to date .
recent advances in nanoscience have created a vast number of experimentally accessible ways to configure atomic and molecular clusters into different geometries with strongly varying physical properties . specifically , exquisite control of the shape and size of atomic and molecular clusters has made it now possible to investigate the collective electromagnetic response of ultra - small metal and semiconductor particles.@xcite the aim of this study is to model and examine plasmonic excitations in such structures , and thus to gain an understanding of the quantum - to - classical crossover of collective modes with increasing cluster size . there is obvious technological relevance to tunable collective modes in nanostructures . for example , surface plasmon resonances in metallic nanospheres and films have been found to be highly sensitive to nearby microscopic objects , and hence are currently investigated for potential sensing applications.@xcite in this context , it is desirable to design customized nanostructures with specifically tailored resonance properties,@xcite and this study is intended to be a first step into this direction . it is natural to expect that in many cases the electromagnetic response of nanoclusters is considerably different from the bulk . in particular for very small clusters , the quantum properties of electrons confined in the structure need to be taken into account . moreover , unlike in the bulk , the coupling between single - particle excitations and collective modes can be very strongly affected by its system parameters . this exponential sensitivity opens up excellent opportunities to optimize the dielectric response via tuning the cluster geometry and its electron filling . for example , by proper arrangement of atoms on a surface one can design nanostructures with controllable resonances in the near - infrared or visible frequency range.@xcite a possible application of such nanostructures is the creation of metamaterials with negative refractive index at a given frequency . furthermore , since geometry optimization of bulk resonators has demonstrated minimization of losses in metamaterials @xcite , it is also interesting to investigate the effect of the nanostructure shape on the loss function at a given resonance frequency . to approach this problem , in this study we investigate the formation of resonances in generic systems of finite conducting clusters , and examine how their frequency and spatial dielectric response depends on the system size and geometry . in particular , the non - locality of the dielectric response function in these structures is important , and will therefore be properly accounted for . a similar analysis for the case of small _ metallic _ nanostructures was performed recently , using an effective mass approximation.@xcite here we focus on the opposite limit , namely we assume that electrons in the cluster can be effectively described using a tight - binding model.@xcite because of the more localized nature of the electronic wave functions in this model the overall magnitude of the collective modes are expected to be strongly suppressed as compared to metallic clusters . this paper is organized in the following way . in the next section we introduce the model and method . for a more extended discussion , the reader is referred to ref . results for the induced energy as a function of the driving frequency of an externally applied electric field and the corresponding spatial modulations of the charge density distribution function are discussed in the following section . finally , a discussion of possible extentions and applications is given in the conclusion section . the interaction of electromagnetic radiation with nanoscale conducting clusters is conventionally described by semi - classical mie theory.@xcite this is a local , continuum - field model which uses empirical values of the linear optical response of the corresponding bulk material , and has been applied in nanoparticles to describe plasmon resonances . @xcite however , such a semi - empirical continuum description breaks down beyond a certain degree of roughness introduced by atomic length scales , and thus can not be used to describe ultra - small systems . in addition , near - field applications , such as surface - enhanced raman scattering , @xcite are most naturally described using a real - space theory which includes the non - local electronic response of inhomogeneous structures . therefore , we will use a recently developed self - consistent and fully quantum - mechanical model which fully accounts for the non - locality of the dielectric response function.@xcite specifically , to identify the plasmonic modes in small clusters we calculate the total induced energy due to an applied external electric field with driving frequency @xmath0 , and scan for the resonance peaks . the induced energy is determined within the non - local linear response approximation . to keep the computational complexity of this procedure at a minimum , we use a one - band tight - binding model to obtain the electronic energy levels @xmath1 and wave functions @xmath2 as a linear combination of of s - orbitals @xmath3 where @xmath4 is the wave function of a s - orbital around an atom localized at position @xmath5 and @xmath6 are the coeficients of the eigenvector ( with energy @xmath1 ) of the hamiltonian , which has the matrix elements @xmath7 here @xmath8 is the tight - binding hopping parameter . the hamiltonian matrix is diagonalized using the householder method to first obtain a tridiagonal matrix and then a ql algorithm for the final eigenvectors and eigenvalues . @xcite once the electronic wave functions have been obtained , it is possible to calculate the dielectric susceptibility @xmath9 via @xmath10 the induced charge density distribution function is then obtained by @xmath11 where in turn the induced potential is given by @xmath12 we avoid the large memory requirement to store @xmath13 by calculating the induced charge density distribution iteratively via @xmath14 with @xmath15 . the integrals are evaluated using a @xmath16 order formula obtained from a combination of simpson s rule and simpson s 3/8 rule . [ eqn : potential ] and [ eqn : charge ] are solved self - consistently by iterating @xmath17 and @xmath18 . this procedure typically converges in 3 - 8 steps when starting with @xmath19 , depending on the proximity to a resonance and on the value of the damping constant @xmath20 , which throughout this paper is chosen as @xmath21 . a much better performance can be achieved when the initial @xmath17 is taken as the solution of a previously solved nearby frequency . upon its convergence , the frequency and spatial dependence of the induced electric field and the induced energy are obtained using @xmath22 and @xmath23 the observed resonances in the induced energy and charge density distribution at certain driving frequencies of the applied electric field correspond to collective modes of the cluster . in the following , the local induced charge density distribution is used for analyzing the characteristic spatial modulation of a given plasmonic resonance . longitudinal modes in atomic chains . ( a ) decimal logarithm of the total induced energy ( artificially offset ) as a function of the frequency of an external electric field which is applied along the direction of the chain . the resonance peaks correspond to different modes . ( b ) induced charge density distribution for the lowest energy mode at @xmath0 = 0.73@xmath8 in the 5-atom chain . ( c ) induced charge density distribution for the highest - energy mode at @xmath0 = 3.56@xmath8 in the 6-atom chain.,width=302 ] let us first focus on the dielectric response function in linear chains of atoms , with the intent to identify the basic features of their collective excitations . the frequency dependence of the induced energy in such systems , exposed to a driving electric field along the chain direction , is shown in fig.[fig : longitudinal](a ) . it exhibits a series of resonances , which increase in number for chains with increasing length . as observed in the spatial charge density distribution , e.g. shown for the 5-atom chain in fig.[fig : longitudinal](b ) , the lowest peak corresponds to a dipole resonance . when increasing the system size , the dipole peak moves to lower frequencies , which is the expected finite - size scaling behavior . the resonances at higher frequency correspond to higher harmonic charge density distributions . for example , in fig.[fig : longitudinal](c ) , we show the charge density distribution corresponding to the highest frequency resonance of the 6-atom chain . in contrast to the dipole resonance , these modes show a rapidly oscillating charge density distribution , and thus have the potential to provide spatial localization of collective excitations in more sophisticated structures . while an extension to much larger chains is numerically prohibitive within the current method , the finite - size scaling of the observed dielectric response of these clusters is consistent with the 1d bulk expectation of a dominant low - energy plasmon mode , coexisting with a particle - hole continuum of much weaker spectral intensity . transverse modes in coupled chain structures . ( a ) logarithm of the total induced energy ( artificially offset ) as a function of the frequency of an external electric field applied transversely to the chain . the low energy mode is central , the analog of a bulk plasmon , and the high energy mode is located at the surface , the analog of a surface plasmon . ( b ) induced charge density distribution for the mode at @xmath24 in the 3-atom double chain . ( c ) induced charge density distribution for @xmath25 in the 5-atom double chain . ( d ) induced charge density distribution for @xmath26 in the 6-atom double chain . , width=302 ] in order to study the transverse collective modes we apply an external electric field perpendicular to ladder structures made of coupled linear chains of atoms.@xcite fig.[fig : transverse](a ) shows that for every chain size there are two resonance peaks for the total induced energy , the higher energy is an end mode , as shown in figs.[fig : transverse](b ) and ( c ) for the 3 and 5-atom double chains respectively , whereas the lower energy peak corresponds to a central mode , as displayed in fig.[fig : transverse](d ) for the 6-atom double chain . it is also confirmed that as the length of the chain is increased , the central mode gets stronger relatively to the end mode , which is the expected behavior for bulk vs. surface excitations . these results are in agreement with the findings of ref.@xcite . dependence on the direction of the applied electric field . ( a ) logarithm of the total induced energy ( artificially offset ) as a function of frequency of external electric fields applied to a 4@xmath276-rectangle at different incident angles . @xmath28when the field is parallel to the 4-atoms edge , and @xmath29when it is parallel to the 6-atoms edge . ( b ) induced charge distribution for @xmath30and @xmath31 . ( c ) induced charge density distribution for @xmath29and @xmath32.,width=302 ] let us next examine what happens when the direction of the external electric field is varied . fig.[fig : angle ] shows the dielectric response of a 4@xmath276-atom rectangular structure for different angles incidence directions of the applied field . when the field is parallel to one of the edges ( @xmath28 or @xmath29 ) , the response is essentially that of a single chain with the same length , shown in fig.[fig : longitudinal](a ) . also the induced spatial charge density modulations are analogous to those of the correspondent linear chain , which can be seen in figs . [ fig : angle](b ) and ( c ) . at intermediate angles the response is a superposition of the two above cases , changing gradually from one extremum to the other as the angle is changed . notice for instance that as the angle increases , the peak at the same frequency of the 4-atoms dipole resonance diminishes , while simultaneously another resonance is formed at the frequency of the dipole mode of a 6-atoms chain when the angle is tuned from @xmath28 to @xmath29 . for @xmath28 there is only the peak at the frequency of the 4-atom chain dipole resonance , whereas for @xmath29 only the dipole peak corresponding to the 6-atoms dipole frequency is present . the superposition of the responses from each direction is a consequence of the linear response approximation employed , since the response is a linear combination of those obtained from each direction component of the external field . variation of the number of electrons . ( a ) logarithm of the total induced energy as a function of the external electric field frequency . the number of electrons @xmath33 in a 9-atom chain is varied . ( b ) induced charge density distribution for @xmath34 at @xmath35 . ( c ) induced charge density distribution for @xmath36 at @xmath37.,width=302 ] next , let us analyze the dependence of the resonance modes on the number of electrons in the cluster . fig.[fig : filling](a ) shows significant changes in the response of a 9-atom chain with the external field applied along its direction . in particular , it is observed that the response is stronger when there are more electrons in the sample , a quite obvious fact since there are more particles contributing to the collective response . moreover the resonance frequencies of lower modes increase with the number of electrons , which can be understood as a consequence of the one - dimensional tight - binding density of states being smallest at the center of the band . hence the energy levels around the fermi energy are more sparse in the finite system , and therefore the excitations require larger frequencies at half - filling . the same does not hold for higher frequency modes since these correspond to transitions between the lowest and highest levels for any number of electrons in the sample . therefore these modes have the same frequency , independent of the electronic filling . higher filling also allows the induced charge density to concentrate closer to the boundaries of the structure , as a comparison between figs.[fig : filling](b ) and ( c ) demonstrates . fig.[fig : filling](b ) shows that a 9-atoms chain with @xmath34 electron has its induced charge density localized around the center of the chain . in contrast , fig.[fig : filling](c ) displays the induced charge density localized at the boundaries of the same structure with @xmath36 . this concentration closer to the surface happens because higher energy states have a stronger charge density modulation than the lower energy ones . therefore the induced charge density is more localized for higher fillings , because at low fillings the excitations responsible for the induced charge density are between the more homogeneous lower energy levels . this can be interpreted as a finite - size rendition of the fact that by increasing the electronic filling one obtains the classical response with all the induced charge density on the surface of the object . connect between two chains , @xmath33 = 1 . ( a ) logarithm of the total induced energy as a function of the frequency of an external electric field applied to two 8-atom chains with an extra atom connecting them at the center . ( b ) induced charge density distribution for @xmath0 = 2.28@xmath8 . ( c ) induced charge density distribution for @xmath0 = 4.36@xmath8.,width=302 ] access to high energy states is very important for achieving spatial localization of the induced charge density , as the next example shows . in order to find a structure with spatially localized plasmons we consider two parallel 8-atoms chains connected to each other by an extra atom at the center . when an external electric field is applied transversely to the chains , the electrons are stimulated to hop between them , but this is only realizable through the connect , therefore the plasmonic excitation is sharply localized around it . fig.[fig : hshape1](a ) shows the response of two 8-atoms chains , fig.[fig : hshape1](b ) and ( c ) show respectively the induced charge density for the dipole and the highest modes . it is seen that the lowest frequency mode has a much longer spread along the chains , whereas the high frequency plasmon is sharper , since it corresponds to excitations to the highest energy state that has a large charge ondulation as it was pointed out before . variation of the distance between neighbor atoms @xmath38 . logarithm of the total induced energy as a function of the frequency of an external electric field applied to 7-atom chains with different spacing between atoms @xmath38 in units of the bohr radius @xmath39 . the frequency unit is @xmath40 the tight - binding hopping parameter for @xmath38 = 3@xmath39.,width=302 ] let us finish by analyzing the dependence of the various dielectric response modes on the distance between atoms in the chain . the dipole moment of the chain is proportional to its length , and consequently also proportional to the distance between neighboring atoms . hence one can expect that the strength of the dielectric response is also proportional to the atomic spacing . however , higher frequency modes require a larger mobility of electrons since they need to move quickly along the chain in order to produce the charge oscillations of the mode . in other words , the charge - density waves can only have frequencies at most equal to the wavering of the charge carriers . hence the high frequency modes are suppressed for systems where electrons lack the mobility necessary for the execution of their oscillation . in the tight - binding model , the electronic mobility stems from the tunneling of electrons from one atom to another . the tunneling rate depends on the overlap of the atomic orbitals located in different sites , and moreover this overlap decays exponentially with the spacing between atoms . therefore the high frequency modes are weaker for long spaced chains , while the slow modes get stronger . this fact is demonstrated in fig.[fig : spacing ] where the response of 7-atom chains with different atomic spacings are shown . the tight - binding hopping parameter @xmath8 changes with the atomic spacing @xmath38 , and here we considered a generic@xcite power - law dependence @xmath41 . the figure clearly shows that the strength of the slowest mode is reduced for shorter spacings . the opposite is true for the two fastest modes , while the intermediate modes have nearly no change . in conclusion , we have analyzed the evolution of plasmonic resonances in small clusters as a function of the system shape , applied external fields , electron filling and atomic separation . using a fully quantum - mechanical , non - local response theory , we observe that longitudinal and transverse modes are very sensitive to these system parameters . this is reflected in their frequency , oscillator strength and the spatial modulation of the induced charge density . specifically , we identify bulk and surface plasmonic excitations which can be controlled in amplitude and frequency by the cluster size . furthermore , we observe a non - trivial filling dependence , which critically depends on the electronic level spacing in a given structure . we also find that changes in atomic spacings strongly affect the electron mobility in these structures , with a very different impact on low - energy vs. high - energy modes . and we see that changing the position of a single atom in a nanostructure can completely alter its collective dielectric response . this strong sensitivity to small changes is the key to controlling the modes of ultra - small structures , and it can thus become the gateway to a new generation of quantum devices which effectively utilize quantum physics for new functionalities . we would like to thank gene bickers , richard thompson , vitaly kresin , aiichiro nakano amd yung - ching liang for useful conversations , and we acknowledge financial support by the department of energy , grant number de - fg02 - 06er46319 . the numerical computations were carried out on the university of southern california high - performance computer cluster . kresin , phys . rep . * 220 * , 1 ( 1992 ) ; k.d . bonin and v.v . kresin , electric - dipole polarizabilities of atoms , molecules and clusters " ( world scientific , singapore , 1997 ) . nazin , x.h . oiu , and w. ho , science * 302 * , 77 ( 2003 ) ; g.v . nazin , x.h . oiu , and w. ho , phys . lett . * 90 * , 216110 ( 2003 ) ; n. nilius , t.m . wallis , m. persson , and w. ho , phys . lett . * 90 * , 196103 ( 2003 ) ; n. nilius , t.m . wallis , and w. ho , science * 297 * , 1853 ( 2002 ) . j. homola , s.s . yee , and g. gauglitz , sensors & actuators : b * 54 * , 3 ( 1999 ) . c.r . moon , l.s , mattos , b.k . foster , g. zeltzer , w. ko , and h.c . monoharan , science * 319 * , 782 ( 2008 ) . a. k. sarychev and v. m. shalaev , electrodynamics of metamaterials " ( world scientific , singapore , 2007 ) . i. grigorenko , s. haas and a.f.j . levi , phys . 97 * , 036806 ( 2006 ) ; a. cassidy , i. grigorenko , and s. haas , phys . b * 77 * , 245404 ( 2008 ) ; i. grigorenko , s. haas , a.v . balatsky , and a.f.j . levi , new j. phys . * 10 * , 043017 ( 2008 ) . slater and g.f . koster , phys . rev . * 94 * , 1498 ( 1954 ) ; c.m . goringe , d.r . bowler and e. hernndez , prog . phys . * 60 * , 1447 ( 1997 ) ; n. w. ashcroft and n. d. mermin , solid state physics " ( thomson learning , 1976 ) . g. mie , annalen der physik , 25 , 377 ( 1908 ) . d. m. wood and w. r. ashcroft , phys . b 25 , 6255 ( 1982 ) ; m. j. rice , w. r. schneider , and s. strassler , phys . rev . b 8 , 474 ( 1973 ) ; s. das sarma , phys . rev . b 43 , 11768 ( 1991 ) ; d. r. fredkin and l. d. mayergoyz , phys . 91 , 253902 ( 2003 ) . s. nie and s. r. emory , science 275 , 1102 ( 1997 ) . w. h. press , b. p. flanney , s. a. teukolsky , w. t. vetterling , numerical recipes " ( cambridge university press , 1988 ) . within the current approach , at least two coupled chains are necessary to visualize charge redistributions along the transverse direction , since charge fluctuations within the orbitals are not accounted for . j. yan , z. yuan and s. gao , phys . lett . * 98 * , 216602 ( 2007 ) . harrison , electronic structure and the properties of solids " ( w. h. freeman , san francisco , 1980 ) .	we explore the collective electromagnetic response in atomic clusters of various sizes and geometries . our aim is to understand , and hence to control , their dielectric response , based on a fully quantum - mechanical description which captures accurately their relevant collective modes . the electronic energy levels and wave functions , calculated within the tight - binding model , are used to determine the non - local dielectric response function . it is found that the system shape , the electron filling and the driving frequency of the external electric field strongly control the resonance properties of the collective excitations in the frequency and spatial domains . furthermore , it is shown that one can design spatially localized collective excitations by properly tailoring the nanostructure geometry .
understanding the galaxy luminosity function has been a goal of galaxy formation theory for several decades ( e.g. ) . a particularly interesting question is whether the luminosity function is universal or whether it depends on environmental factors such as the mass of the dark halo that hosts a particular galaxy population . considerable attention has been paid to the faint end of the luminosity function which has a much flatter slope than the low mass end of the halo mass function predicted in cold dark matter ( cdm ) models of galaxy formation ( e.g. ) . the early work of showed that the number of faint galaxies must have been strongly affected by feedback processes that prevented most of the gas from cooling in small halos at early times . some likely feedback mechanisms , such as the injection of energy into the interstellar medium in the course of stellar evolution , depend on the internal properties of the galaxy and so their effects may be expected to be independent of the large - scale environment . a number of observational studies , such as a recent analysis of the 2df galaxy redshift survey @xcite , indeed find no significant difference between the luminosity functions of galaxies in rich clusters and in the field . other studies , however , have found the opposite . for example , concluded that galaxies in rich clusters have luminosity functions with considerably steeper faint ends than galaxies in the field . more recently , have claimed that the faint end of the galaxy luminosity function varies systematically with environment , increasing in slope from small , diffuse systems like the local group , to massive , dense systems like the coma cluster . in the cold dark matter model of galaxy formation , dark matter halos retain considerable substructure after they collapse and virialize ( e.g. ) and some of these subhalos are associated with sites of galaxy formation . the mass function of subhalos appears to be relatively independent of the mass of the parent halo . thus , trends such as those inferred by would require processes that either preferentially suppress the formation of dwarf galaxies in low mass systems , or destroy them after they form . an effective mechanism for suppressing the formation of small galaxies is the reheating of the intergalactic medium ( igm ) caused by the reionization of the universe at a redshift @xmath0 . have argued that this process could introduce an environmental dependence in the galaxy luminosity function on the grounds that a higher fraction of the low - mass halos that formed before reionization ( when dwarf galaxy formation proceeded unimpeded by photoionization suppression ) ended up in clusters today than in less massive systems . the effect of reionization on the formation of galaxies has been the subject of several recent studies @xcite , aimed mostly at investigating the discrepancy between the large number of subhalos found in n - body simulations of galactic cdm halos and the small number of satellite galaxies observed in the local group . in this paper , we employ the cdm model of to calculate the luminosity function of galaxy populations residing in dark matter halos of different mass . we find that there are significant differences in these luminosity functions and we then explore how they combine together to build up the global luminosity function , with particular emphasis on the faint end slope . a partial study of luminosity functions in halos of different mass using a semi - analytic model of galaxy formation was carried out by . to calculate galaxy luminosity functions in halos of different mass correctly , it is important to include tidal effects on satellite galaxies , a potential galaxy destruction mechanism . our model treats these effects in considerably more detail than previous models of galaxy formation . we find that tidal effects are important in limiting the formation of massive galaxies at the centre of rich clusters . in this paper , we compare the results of our calculations to the data of and assess whether feedback from reionization is a viable explanation of the trend claimed by these authors . the existing dataset is small , but forthcoming results from the 2df and sloan galaxy surveys will enable much more extensive comparisons with the theory . the remainder of this paper is arranged as follows . in [ sec : model ] we briefly outline our model of galaxy formation , in [ sec : res ] we present results for the environmental dependence of the luminosity function and in [ sec : compobs ] we compare our model with the available observational data . finally , in [ sec : dis ] we present our conclusions . we present , in an appendix , several simple models of photoionization suppression to elucidate how this mechanism works . we employ the semi - analytic model of galaxy formation described in detail by and benson et al . ( 2002a ; hereafter paper i ) to compute the properties of galaxies in a range of environments at @xmath1 . the reader is referred to those papers for a complete and detailed description of the model . briefly , the hierarchical formation of dark matter halos is calculated using the extended press - schechter formalism @xcite . the formation of galaxies in the resulting dark matter halo merger trees is followed by means of simple , physically motivated models of gas cooling , star formation and galaxy merging . recent work has demonstrated that at least in so far as gas cooling is concerned these simplified calculations are in excellent agreement with the results of n - body / hydrodynamical simulations @xcite . applying a stellar population synthesis model gives galaxy luminosities in different passbands . the model includes a prescription for supernovae feedback which drives gas out of galaxies at a rate proportional to the current star formation rate , with a constant of proportionality that is larger for less strongly bound systems . this negative feedback flattens the faint end slope of the luminosity function but , on its own , its effect is not strong enough to account for the measured function ( paper i ) . developed , and incorporated into this model , a detailed , self - consistent treatment of photoionization suppression of galaxy formation ( hereafter abbreviated as ) . by causing further suppression of faint galaxy formation , this mechanism brought the model luminosity function into excellent agreement with observations . in paper i , we also included a much more detailed treatment than was previously possible of the evolution of satellite galaxies as they orbit within larger halos experiencing tidal forces and gravitational shock heating . ( hereafter , we abbreviate tidal limitation as . ) the associated mass loss also affects the shape of the luminosity function , and so we will explore this process in this paper . we adopt essentially the same model parameters as in paper i , with the following differences . while the models of paper i used the press - schechter halo mass function , we choose to adopt the function proposed by which gives a better match to the results of n - body simulations . consequently , to produce a reasonable galaxy luminosity function , we find it necessary to adjust the parameters @xmath2 from @xmath3 to @xmath4 , @xmath5 ( which determines the star formation rate ) from 0.0050 to 0.0067 , and @xmath6 ( which affects mass - to - light ratios ) from 1.38 to 1.03 ( since @xmath6 affects the fraction of mass recycled in star formation , we adjust that fraction accordingly ) . even with these changes , the mass function results in a somewhat worse fit to the bright end of the luminosity function than was obtained in paper i. we defer further study of this aspect of the model to future work . we adopt an escaping fraction of ionizing photons from galaxies , @xmath7 , of 12% . this produces a neutral hydrogen fraction for gas at zero overdensity in the igm at @xmath8 of @xmath9 , in agreement with the lower limit of @xmath10 derived by , while causing reionization to occur at as high a redshift as possible . this maximizes the associated suppression of galaxy formation slightly further without violating the limit of but , since the neutral hydrogen fraction drops very rapidly in our model at these redshifts ( paper i ) , it makes almost no difference to the suppression of galaxy formation . ] . the resulting model is essentially the same as the @xmath11 model presented in paper i. [ sec : res ] we begin by studying the luminosity functions of galaxies in individual dark matter halos of different mass , and then proceed to consider how these combine together to form the global galaxy luminosity function . using the model described in [ sec : model ] , we simulate galaxy formation in a large number of halos of masses @xmath12 , @xmath13 , @xmath14 , @xmath15 and @xmath16 km s@xmath17 mpc@xmath17 . ] . this spans halo masses appropriate to systems ranging from the local group to rich galaxy clusters . ( the @xmath18 bin corresponds to halos less massive than that of the local group but we include it since it shows the most dramatic effects . ) merger trees are constructed with a resolution of @xmath19 of the final halo mass or @xmath20 , whichever is smaller . this is sufficient to follow accurately the formation and evolution of galaxies brighter than @xmath21 except in the most massive clusters ( @xmath16 ) for which resolution begins to affect our results faintwards of @xmath22 ; for these clusters , we show results only down to @xmath23 . we perform calculations for our `` standard model '' ( i.e. including the effects of and of satellite galaxies ) , and repeat them switching off and/or ( resulting in a total of four different sets of results ) , in order to assess the effects of each physical mechanism separately . we simulate a sufficiently large number of halos of each mass so that the average luminosity functions are reasonably smooth . the solid lines in figure [ fig : elf ] show the luminosity functions of galaxies in our simulated halos . the basic form of the standard model luminosity function is similar in all mass halos . it has two components : a near power - law distribution at faint magnitudes which is made up of satellite galaxies ( i.e. those galaxies that do not reside at the centre of the final halo ) , and a strongly peaked distribution at bright magnitudes beyond which no more galaxies are found and which is made up of the central galaxy in each final halo . as we consider halos of increasing mass , both components shift to ever brighter luminosities , but the peaked component decreases in amplitude relative to the power - law component , indicating that , in low mass halos , a single galaxy tends to dominate , while in clusters the light is more equally shared among many galaxies . observed luminosity functions , @xmath24 ( where @xmath25 is the number of galaxies per unit volume and @xmath26 is luminosity ) , are typically described by the schechter function , @xmath27 where @xmath28 , @xmath29 and @xmath30 are parameters . at magnitudes faint compared to the characteristic luminosity , @xmath29 , has the simple form @xmath31 . following normal practice , we will refer to @xmath30 as `` the faint end slope . '' although the luminosity functions in fig . [ fig : elf ] are not well fit by the schechter form , we can still obtain a formal value of @xmath30 by fitting to the data in the first few bins plotted . we find slopes in the range @xmath32 , except in the lowest mass halos ( @xmath18 ) for which a much flatter slope , @xmath33 , is found ( although it is possible that a steeper slope would be obtained also in this case if we probed to even fainter luminosities ) . there is little evidence for a strong variation of slope with halo mass , except perhaps for a weak ( and rather insignificant ) trend for _ flatter _ slopes in higher mass clusters . however , it is important to note that the slope is not constant with galaxy magnitude . this is clearly seen , for example , in the @xmath34 halos for which , in the region @xmath35 , the slope is close to @xmath36 . switching off in our model , and reverting to the simpler calculation of dynamical friction of , results in the dotted curves of fig . [ fig : elf ] . clearly , has little influence on the faint end of the luminosity function , although it does slightly reduce the luminosities of already faint galaxies ( as is particularly noticeable for @xmath34 halos ) . a more interesting effect of is apparent in the massive clusters ( @xmath15 and @xmath16 halos ) . here , switching off results in a new population of very luminous galaxies , visible as a `` bump '' in the luminosity function at bright magnitudes . these highly luminous objects form through multiple mergers at the centres of cluster halos . when is included , the merger rate of massive galaxies in clusters is significantly reduced , and none of these highly luminous objects form . as shown in the left - hand panel of fig . [ fig : tidal ] , cluster central galaxies are about an order of magnitude less massive when our accurate calculation of dynamical friction is used . note that the inclusion of makes much less difference in @xmath34 halos , due to the very different shape of the luminosity function in these systems ( i.e. there is relatively little mass in satellites to merge into the central object ) . this difference between the merger rates in our model and that of can not be fully reproduced by simply lengthening the dynamical friction timescale in the prescription by some factor ( even if this factor is a function of halo mass ) . the shape of the distribution of merger times is significantly different between the models and , in our calculation , varies with halo mass . an effect similar to that shown by the comparison of the solid and dotted lines in fig . [ fig : elf ] was noticed by in cluster luminosity functions calculated using a different semi - analytic model . computed luminosity functions adopting both simple estimates of galaxy merger rates ( similar to those assumed in computing the dotted curves in fig . [ fig : elf ] ) and merger rates taken directly from a high resolution simulation of dark matter . they found that the simple scheme produced highly luminous galaxies through merging that were not present in the n - body calculation . based on the number of massive mergers quoted by , we can conclude that a similar reduction in central galaxy mass occurred when our realistic merger rate calculation was employed . as suggested by , the cause of this difference is that the simple scheme underestimates the merging timescale of rather massive subhalos in clusters , leading to many massive galaxies merging with the central object in the final halo . in our more detailed calculation , as in the n - body simulation , these large subhalos loose a significant amount of their mass as they orbit within the cluster , leading to a reduced dynamical friction force and hence to a reduced merger rate . the right - hand panel of fig . [ fig : tidal ] shows the mass functions of _ surviving _ substructure halos in clusters at @xmath1 . the dotted histogram shows the model with no , while the dashed histogram corresponds to the model with included , but the mass of each substructure is plotted as it was _ before _ it experienced any tidal mass loss . clearly , inclusion of has resulted in fewer mergers since there are more surviving substructures of a given initial mass . the solid histogram gives the distribution of substructure masses after tidal mass loss . comparing the dashed and solid histograms indicates that halos typically lose between 50% to 70% of their mass ( with higher mass halos losing less than lower mass halos ) . the dynamical friction time scales as the mass of the satellite halo and is therefore increased by factors of 2 to 3 when is included , greatly reducing the galaxy merger rates . [ cols="^,^ " , ] for the two most massive systems , our model results are in reasonably good accord with the observational determinations , but for the lower mass systems there are significant discrepancies . this analysis verifies and quantifies the comparison between the shapes of the predicted and observed luminosity functions made in fig . [ fig : elf ] . using a semi - analytic model of galaxy formation , we have calculated the luminosity function of galaxy populations contained in halos of different mass . we find large differences in the shapes of these luminosity functions . in smaller mass halos ( @xmath37 , the luminosity function has a `` hump '' at bright magnitudes and a roughly power - law shape at fainter magnitudes . in more massive systems , the shape approaches the familiar schechter form . these differences reflect the relative contributions of `` central '' and `` satellite '' galaxies . in the smaller mass halos , the bright end of the luminosity function is dominated by a single or at most a few bright galaxies and the rest of the population consists predominantly of much fainter satellites . in the more massive halos , the light is more evenly distributed amongst a large number of galaxies . these differences reflect the complex interplay between the processes that establish the number of halos and subhalos of different mass ( i.e. hierarchical clustering from cdm initial conditions ) and the processes that light them up ( i.e. gas cooling , stellar evolution , feedback , etc ) . the global luminosity function is the weighted sum of the luminosity functions of individual halos . interestingly , the faint end slopes of the individual luminosity functions are always steeper than the faint end slope of the global luminosity function . this can be understood in terms of the relative contributions to the global luminosity function of satellite and central galaxies from individual halos , whose luminosity functions have steep and shallow slopes respectively . the slope of the global function is intermediate between these two . our model of galaxy formation represents an advance over previous models because of its detailed treatment of two processes that are important in establishing the luminosity function : the suppression of galaxy formation in small halos at high redshift ( ) , as a result of the photoionization of the igm by early generations of galaxies and quasars , and tidal mass loss from satellite galaxies ( ) . is particularly relevant at the faint end and has the net effect of reducing the number of faint galaxies , giving rise to a flatter luminosity function than would otherwise be the case . this outcome is far from obvious . for example , a simplistic model of in which galaxy formation is assumed not to occur in low mass halos below the reionization redshift results , in fact , in a steeper luminosity function ( see appendix [ sec : faint ] ) . in reality , the effects of depend on the mass of the halo relative to the time - dependent `` filtering mass '' ( defined as the halo mass which accretes only half the mass that it would have accreted in the absence of photoionization ; ) . galaxies can form even if their halos have mass smaller than the filtering mass , but with reduced luminosity . the lower the mass of the halo relative to the filtering mass , the fainter is the galaxy that forms . the filtering mass is an increasing function of time . thus , although lower mass halos form at higher redshift , the redshift where the filtering mass becomes comparable to their mass is also higher , increasing the fraction of them which experience strong suppression . these two effects combined _ do _ tend to produce a flatter luminosity function but the degree of flattening depends on their detailed balance . as we have shown previously (; see also ) , on its own can not account for the shallow faint end slope of the global luminosity function . other processes , such as feedback from energy released by stellar winds and supernovae , need to be included in order to reduce the number of faint galaxies to the observed level . the other process that we have treated in detail , , is relevant at the bright end of the luminosity function , particularly in massive halos harbouring rich clusters . mass loss from satellite galaxies , as they spiral into the centre of their halo , increases the dynamical friction time , so reducing the merger rate . this prevents the formation of some of the highly luminous and unobserved cluster galaxies that tend to appear in models of galaxy formation that do not take this process into account . keeping in mind the provisos of [ sec : compobs ] , we find that our model fails to reproduce the steepening of the faint end slope of the luminosity function with system richness found by @xcite . the model agrees quite well with the data for the virgo and coma clusters , but it predicts a steeper faint end than is inferred for the local group and ursa minor . photoionization is an effective ( and unavoidable ) mechanism for suppressing galaxy formation in low mass halos , but it does not produce the kind of `` environmental '' variation seen by . at the low mass end , this failure of the model is related to the discrepancy that we ( , see also ) , found previously in our cdm models of the local group . photoionization has a dramatic effect on the overall number of satellites that survive in galactic halos and can reduce it to the levels seen in the local group . however , the satellite luminosity function in the models is considerably steeper than is observed . this discrepancy contrasts with the success of cdm models in reproducing the structural properties of the satellites @xcite . taking at the face value the discrepancy between our models and the data of , it is interesting to speculate about the possible root causes of the problem . leaving aside alternative theories for the nature of the dark matter or drastic changes to the initial power spectrum of density perturbations specifically designed to reduce the amount of small - scale power ( e.g. ) , we focus on our modeling of baryonic processes within the context of the cdm cosmogony . this is undoubtedly simplified . for example , the effects of depend upon both the time variation of the filtering mass and on the assumed functional form of the suppression , both of which are uncertain . as discussed in the appendix , reasonable variations in our assumed mass dependence of do not seem capable of producing a sufficiently flat faint end slope in our models of the local group . however , we can not exclude the possibility that the simplifications made in our model are inappropriate . for example , we neglect the patchy nature of reionization , but if low density regions became ionized first , galaxy formation would be suppressed more effectively in these regions , altering the temporal evolution of the filtering mass . localized triggered , for example , by a nearby galaxy or quasar could also change our model predictions . these processes are poorly understood theoretically , but observational evidence ( e.g. ) suggests that they could be important in determining the thermodynamic state of the igm at high redshift . at a more basic level , there are approximations in our model which might be inappropriate in the regime of low mass halos . thus , although comparisons with n - body / hydrodynamical simulations indicate that our estimates of gas cooling rates are accurate for relatively massive galaxies @xcite , it has not been shown that the same is also true for the low mass galaxies of interest here . cooling times in these objects can become very short and this might conceivably render incorrect the assumption of cooling from an initial , spherically symmetric quasi - hydrostatic equilibrium . further study of gas cooling in low mass systems is necessary to address this issue . similarly , it is far from clear that the simple rules for star formation in our model , which are empirical parameterizations tuned to match observational constraints for relatively bright galaxies at low redshifts , remain valid when extrapolated to low mass galaxies and/or high redshifts . other physical constraints on star formation not currently included in our models may also be important for these low mass galaxies ( e.g. ) . finally , our modeling of feedback is also simplified and , again , its extrapolation to low masses may be unrealistic . possible sources of feedback such as strong outflows or heating by agn which are currently neglected may be important on these scales . in conclusion , we have found substantial differences in the luminosity functions of galaxies residing in dark matter halos of different mass . these result from the interplay of a variety of processes that affect the formation of galaxies of different luminosity . although the exact predictions of our model no doubt depend on its details , the gross differences between the predicted luminosity functions are likely to be generic and to be present in a broad class of cold dark matter models of galaxy formation . comparison with existing data gives mixed results : fair agreement with the luminosity functions in virgo and coma but substantial disagreement with the luminosity functions in the local group and ursa minor cluster . the much larger and better controlled samples that will be forthcoming from cluster analyses of the 2df and sdss galaxy redshift surveys will allow a more comprehensive test of the general class of cold dark matter models of galaxy formation . adelberger k. l. , et al . in preparation balcells m. , peletier r. f. , 1994 , aj , 107 , 135 benson a. j. , pearce f. r. , frenk c. s. , baugh c. m. , jenkins a. , 2001 , mnras , 320 , 261 benson a. j. , lacey c. g. , baugh c. m. , cole s. , frenk c. s. , 2002a , mnras , 333 , 156 benson a. j. , frenk c. s. , lacey c. g. , baugh c. m. , cole s. , 2002b , mnras , 333 , 177 bond j. r. , cole s. , efstathiou g. , kaiser n. , 1991 , apj , 379 , 440 bower r. g. , 1991 , mnras , 248 , 332 bullock j. s. , kravtsov a. v. , weinberg d. h. , 2000 , apj , 539 , 517 cole s. , 1991 , apj , 367 , 45 cole s. , lacey c. g. , baugh c. m. , frenk c. s. , 2000 , mnras , 319 , 168 coln p. , avila - reese v. , valenzuela o. , 2000 , apj , 542 , 622 de propris r. et al . , 2002 , mnras , 329 , 87 diaferio a. , kauffmann g. , colberg j. m. , white s. d. m. , 1999 , mnras , 307 537 eke v. r. et al . , 2002 , in preparation geller m. j. , diaferio a. , kurtz m. j. , 1999 , apj , 517 , 23 gnedin n. y. , 2000 , apj , 542 , 535 helly j. c. , cole s. , frenk c. s. , baugh c. m. , benson a. j. , lacey c. , pearce f. r. , 2002 , submitted to mnras ( astro - ph/0202485 ) klypin a. a. , kravtsov a. v. , valenzuela o. , prada f. , 1999 , apj , 522 , 82 jenkins a. , frenk c. s. , white s. d. m. , colberg j. m. , cole s. , evrard a. e. , couchman h. m. p. , yoshida n. , 2001 , mnras , 321 , 372 kamionkowski m. , liddle a. r. , 2000 , phys.rev.lett . , 74 , 4525 lacey c. g. , cole s. , 1993 , mnras , 262 , 627 lidz a. , hui l. , zaldarriaga m. , scoccimarro r. , 2002 , submitted to apj ( astro - ph/0111346 ) mellier y. , 2002 , space science reviews , 100 , 73 moore b. , ghigna s. , governato f. , lake g. , quinn t. , stadel j. , tozzi p. , 1992 , apj , 524 , 19 norberg p. et al . , 2002 , mnras in press ( astro - ph/0111011 ) phillipps s. , shanks t. , 1987 , mnras , 227 , 115 press w. h. , schechter p. , 1974 , apj , 187 , 425 schindler s. , binggeli b. , bhringer h. , 1999 , a&a , 343 , 420 somerville r. s. , 2002 , submitted to mnras ( astro - ph/0107507 ) springel v. , white s. d. m. , tormen g. , kauffmann g. , 2001 , mnras , 328 , 726 stoehr f. , white s. d. m. , tormen g. , springel v. , 2002 , submitted to mnras ( astro - ph/0203342 ) trentham n. , hodgkin s. , 2002 , submitted to mnras ( astro - ph/0202437 ) tully r. b. , somerville r. s. , trentham n. , verheijen m. a. w. , 2002 , apj , 569 , 573 verde l. , oh s. p. , jimenez r. , 2002 , submitted to mnras ( astro - ph/0202283 ) white s. d. m. , rees m. j. , 1978 , mnras , 183 , 341 white s. d. m. , frenk c. s. , 1991 , apj , 379 , 52 yoshida n. , stoehr f. , springel v. , white s. d. m. , 2002 , submitted to mnras ( astro - ph/0202341 ) the faint end of the luminosity function directly reflects the outcome of feedback effects . the two sources of feedback in our model , supernova - driven winds and photoionization , are both important in determining the number of faint galaxies . as we discussed in paper i , photoionization inhibits the formation of faint galaxies for two reasons : it raises the temperature and pressure of the igm , curtailing the ability of gas to accrete into halos , and it reduces the rate at which gas in halos can cool , inhibiting star formation . the first of these processes is the dominant effect . it can be conveniently described in terms of a `` filtering mass , '' @xmath38 , defined as the mass of a halo which accretes only half the baryonic mass that it would have accreted in the absence of photoionization . the mass of gas accreted by a halo of mass @xmath39 is approximated by the following formula @xcite , based on the results of gas - dynamical simulations : @xmath40 ^ 3},\ ] ] where @xmath41 is the universal baryon fraction and @xmath38 is the filtering mass . 1 . galaxy formation is significantly suppressed in halos less massive than the filtering mass . the filtering mass is an increasing function of time ( at least over the range of redshifts of interest here ) . 3 . for halos formed at a given redshift , the suppression of galaxy formation is greater the smaller the halo mass is relative to the filtering mass . the impact of these three factors will depend on the distribution of formation redshifts of dark matter halos of different mass in different environments . according to the extended press - schechter formalism , it is always true ( for a cdm power spectrum ) that the typical formation redshift of a halo increases as the mass of the halo decreases , and also as the mass of the final host in which the halo resides increases . further understanding of halo formation redshift distributions may be gained by considering eqn . 2.15 of that gives the mass function of progenitor halos at any redshift and which we reproduce in slightly different form below : @xmath42^{3/2 } } \nonumber \\ & & \exp \left ( -{(\delta_{\rm c}^{\rm ( z)}-\delta_{\rm c}^{\rm ( 0)})^2\over 2[\sigma^2(m)-\sigma^2(m_{\rm host } ) ] } \right ) { \d \sigma^2 \over \d m } , \label{eq : ps}\end{aligned}\ ] ] where @xmath43 is the critical value of the linear theory fractional overdensity for collapse at redshift @xmath44 , @xmath45 is the mass of the progenitor halo at that redshift , @xmath46 is the variance of the linear fractional overdensity in a sphere of mass @xmath45 , and @xmath47 is the mass of the halo into which that progenitor has been incorporated by the present day . consider galaxies living in relatively low mass halos at the present day ( in which most of the galaxies that make up the faint end of the luminosity function reside ) . the relative numbers of halos of masses @xmath48 and @xmath49 at some particular redshift depends upon @xmath50 and @xmath51 . as @xmath47 increases , @xmath52 decreases for a cdm power spectrum . thus , for sufficiently large @xmath47 , the relative numbers of halos of masses @xmath48 and @xmath53 tends to a fixed value . for @xmath47 comparable to @xmath48 , the relative numbers of these halos become a very strong function of @xmath47 , and the number of mass @xmath48 halos is exponentially suppressed , as indicated by eqn . ( [ eq : ps ] ) . these theoretical expectations are clearly manifest in fig . [ fig : explan ] , where we plot the positions of galaxies in the halo - mass _ vs. _ formation - redshift plane . all of the galaxies we consider at @xmath1 are satellites in a larger host halo ( of mass @xmath47 ) due to the ranges of luminosity and @xmath47 we have chosen to consider . therefore , we plot the formation redshift and mass for the halo in which the satellite formed . we adopt the definition of formation redshift of , namely a halo is assumed to be newly formed if its mass exceeds twice the mass of each of its progenitors at their own formation time , and plot the halo mass at the formation redshift ( it may have increased afterwards due to continued accretion and merging ) . ( the formation redshifts appear quantized in this figure because of the finite timesteps in our calculations which , however , are sufficiently fine so as not to affect the results . ) the halo mass of the satellite is then set equal to the mass of the halo in which it lived at redshift @xmath54 ) . the position of a halo in this plane , relative to the redshift - dependent filtering mass ( shown as a solid curve ) , determines the degree of it experiences . we examine faint galaxies in two luminosity ranges , @xmath55 ( dots ) and @xmath56 ( small squares ) . consider first a model with no ( left - hand panels ) . in this case , faint galaxies currently hosted in a @xmath57 cluster halo ( lower panel ) have a much more extended range of formation redshifts than galaxies in a @xmath34 milky way - type halo ( upper panel ) . the fainter galaxies located in lower mass halos have a more extended range of formation redshifts than their brighter counterparts , although this difference is rather small . the values of @xmath30 for the two populations , given in each panel , indicate steep slopes in both environments . the filtering mass has no influence on these results , of course , but is shown to indicate which of halos might be expected to still form a galaxy when is switched on . in the right - hand panels we plot results for a model with . the effect of the filtering mass can now be clearly seen : halos below the solid line form galaxies much less efficiently than before ( although there are certainly still some halos that managed to make galaxies below this line ) . the values of @xmath30 given in the panels indicate that some flattening of the luminosity function has occurred , but this effect is small . let us now examine the importance of these various processes using simplified models of . figure [ fig : elftest ] shows luminosity functions in different mass halos constructed using the simplified models described below , together with our standard model of ( heavy solid line ) and a model with no ( thin solid line ) . * model a * : : in the simplest scenario , suppose the filtering mass jumps abruptly from zero to @xmath38 at redshift @xmath58 , and that any halo of smaller mass forming at lower redshift makes no galaxy at all , while galaxy formation in other halos proceeds unchanged . in this case , the luminosity function should be unaffected at bright magnitudes ( since these galaxies form in halos which are more massive than @xmath38 ) , should plummet sharply at a magnitude corresponding to halo mass equal to @xmath38 , and should rise again to fainter magnitudes ( since these galaxies inhabit lower mass halos which typically form at redshifts greater than @xmath59 ) . on the basis of eqn . ( [ eq : ps ] ) , the slope of this rise should be large for low @xmath47 halos since these have the greatest variation in progenitor number with mass at @xmath59 , and should approach some fixed value for large @xmath47 . this behaviour is borne out when we apply this simple prescription to our semi - analytic models , using values of @xmath60 ( comparable to the value of @xmath38 at the redshifts where the majority of the faint galaxies are formed in our standard model ) and @xmath61 ( the redshift of reionization in our standard model ; dot - long - dashed lines in fig . [ fig : elftest])suppression is greatest for small @xmath62 , tends to steepen the faint end of the luminosity function , and causes the most steepening for the lowest @xmath47 . this model produces a discontinuity in the luminosity function and is a poor approximation to the full treatment . * model b * : : at the next level of sophistication , we take into account the variation of the filtering mass with redshift as predicted by our model , but retain the simple prescription wherein any halo forming with mass below the filtering mass forms no galaxy at all . since @xmath38 increases with time , galaxy formation in lower mass halos is suppressed earlier , an effect which should tend to flatten luminosity function slopes . since halos of a given mass form earlier in larger @xmath47 , we still expect the effects of filtering to be weaker for clusters than for the local group . at a given magnitude , this process would simply reduce the number of galaxies . however , the detailed consequences for the faint end slope are now harder to anticipate since they depend on just how @xmath38 varies with redshift . applying this prescription to our model ( long - dashed lines in fig . [ fig : elftest ] ) , we still find a discontinuity in the luminosity function ( although this is much weaker than in model ( i ) since @xmath38 varies smoothly with time ) which is flattened relative to that without any . furthermore , the slopes are significantly flatter than in case ( i ) , indicating that the time variation of @xmath38 largely counteracts the steepening of the luminosity function produced by the variation in halo formation distributions . no environmental variation is introduced by these processes . the resulting luminosity functions are quite similar to those from our full calculations , although the level of suppression is somewhat greater . again , the effects are largest for the lowest @xmath47 . * model c * : : alternatively , we can consider a prescription which has no time variation in @xmath38 , but in which , instead , the degree of suppression varies smoothly with halo mass ( as prescribed by ) constant , resulting in a filtering mass that increased with time . this prescription , also adopted by provides a better match to the results of our standard model . here , however , we are intersted only in exploring simplified models . ] . we find that this model also produces luminosity functions similar to those from our full calculations ( dot - short - dashed lines ) , and with about the right degree of suppression for suitable choices of @xmath38 and @xmath59 . * model d * : : we next consider a model in which @xmath38 varies with time and in which the degree of suppression varies smoothly with halo mass . this then differs from our full calculations only in neglecting suppression due to photoheating by the ionizing background . in this model , photoionization redistributes galaxies which previously had a given magnitude over a range of fainter magnitudes . at a given magnitude , some galaxies are lost due to suppression , but also new ones appear as brighter galaxies are partially suppressed , thus becoming fainter . the net effect is therefore difficult to judge since it depends on the time variation of the filtering mass and on the previous shape of the luminosity function . we can , however , safely say that the effect is still expected to be larger for lower @xmath47 . applying this prescription we reproduce the results of our full calculation rather well ( dotted lines in fig . [ fig : elftest ] ) , finding slightly less suppression overall than in case ( ii ) , but still with no environmental variation . * model e * : : a smoother variation of suppression with halo mass would result in less flattening of the faint end slope . for example , the short - dashed lines in fig . [ fig : elftest ] show results for a suppression of the form @xmath63 with the same variation of filtering mass with redshift as in our standard model . in conclusion , a simple model in which galaxy formation at @xmath64 is entirely suppressed below some fixed mass scale grossly overestimates the effects of , and produces very different luminosity functions from those predicted by our full calculation . including _ either _ a smoothly varying degree of suppression _ or _ a time - varying filtering mass produces better agreement with our full calculations , indicating that these two ingredients are of comparable importance . the degree of suppression ( as characterised by the reduction in amplitude of the lf ) for a given time dependence in @xmath38 depends on the functional form used to characterize the variation of suppression with halo mass . the sharper the transition from weak to strong suppression , the greater the net effect on the luminosity function . smoother transitions tend to produce steeper faint end slopes , but even the sharpest transition possible does not produce slopes as flat as are observed and we never find any significant environmental variation .	using semi - analytic modeling techniques , we calculate the luminosity function of galaxy populations residing in cold dark matter halos of different mass . we pay particular attention to the influence of the reionization of the universe on the number of faint galaxies and to the effects of dynamical friction and tidal limitation of satellites on the number of bright galaxies . we find substantial differences in the shapes of the galaxy luminosity functions in halos of different mass which reflect generic features of the cold dark matter model of galaxy formation and thus offer the opportunity to test it . we then consider how the individual halo luminosity functions combine together to produce the global luminosity function . surprisingly , the global function ends up having a shallower faint end slope than those of the constituent halo luminosity functions . we compare our model predictions with the limited datasets compiled by . we find good agreement with the luminosity functions measured in the virgo and coma clusters but significant disagreement with the luminosity functions measured in the local group and ursa minor cluster . we speculate on possible inadequacies in our modeling and in the existing observational samples . the luminosity functions of galaxies in groups and clusters identified in the 2df and sdss galaxy redshift surveys offer the prospect of testing galaxy formation models in detail .
functional logic programming @xcite joins in a single paradigm the features of functional programming with those of logic programming . logic programming contributes logic variables that are seamlessly integrated in functional computations by narrowing . the usefulness and elegance of programming with narrowing is presented in @xcite . at the semantics level free variables are equivalent to _ non - deterministic functions _ @xcite , i.e. , functions that for some arguments may return any one of many results . thus , at the implementation level variables can be replaced by non - deterministic functions when non - deterministic functions appear simpler , more convenient and/or more efficient to implement @xcite . this paper focuses on a graph transformation recently proposed for the implementation of non - determinism of this kind . this transformation is intended to ensure the completeness of computations without cloning too eagerly a large portion of the context of a non - deterministic step . the hope is that steps following the transformation will create conditions that make cloning the not yet cloned portion of the context unnecessary . non - determinism is certainly the most characterizing and appealing feature of functional logic programming . it enables encoding potentially difficult problems into relatively simpler programs . for example , consider the problem of abstracting the dependencies among the elements of a set such as the functions of a program or the widgets of a graphical user interface . in abstractions of this kind , _ component parts _ `` build '' _ composite objects_. a non - deterministic function , , defines which objects are dependent on each part . the syntax is curry @xcite . @xmath0 - 35pt -.0em a part can build many objects , e.g. : part builds objects and . likewise , an object can be built from several parts , e.g. : object is built by parts and . many - to - many relationships , such as that between objects and parts just sketched , are difficult to abstract and to manipulate in deterministic languages . however , in a functional logic setting , the non - deterministic function is straightforward to define and is sufficient for all other basic functions of the abstraction . for example , a function that non - deterministically computes a part of an object is simply defined by : @xmath1 - 35pt -.0em where is defined using a _ functional pattern _ @xcite . the set of all the parts of an object is computed by , the implicitly defined _ set function _ @xcite of . the simplicity of design and ease of coding offered by functional logic languages through non - determinism do not come for free . the burden unloaded from the programmer is placed on the execution . all the alternatives of a non - deterministic choice must be explored to some degree to ensure that no result of a computation goes missing . doing this efficiently is a subject of active research . below , we summarize the state of the art . there are three main approaches to the execution of non - deterministic steps in a functional logic program . a fourth approach , called _ pull - tabbing _ @xcite , still underdeveloped , is the subject of this paper . pull - tabbing offers some appealing characteristics missing from the other approaches . we borrow from @xcite a simple example to present the existing approaches and understand their characteristics : @xmath2 - 35pt -.0em we want to evaluate the expression @xmath3 - 35pt -.0em we recall that ` ' is a library function , called _ choice _ , that returns either of its arguments , i.e. , it is defined by the rules : @xmath4 - 35pt -.0em and that the clause introduces a _ shared _ expression . every occurrence of in ( [ value ] ) has the same value throughout the entire computation according to the _ call - time choice _ semantics @xcite . by contrast , in each occurrence of is evaluated independently of the other . [ fig : sharing ] highlights the difference between these two expressions when they are depicted as graphs . @1@c=5pt@r=12pt@m=1pt & @-[dl ] @-[dr ] + @-[dr ] & & @-[dl ] + & @1@c=5pt@r=12pt@m=1pt & @-[dl ] @-[dr ] + @-[d ] & & @-[d ] + & & a _ context _ is an expression with a distinguished symbol called _ hole _ denoted ` @xmath5 $ ] ' . if @xmath6 is a context , @xmath7 $ ] is the expression obtained by replacing the hole in @xmath6 with @xmath8 . e.g. , the expression in ( [ value ] ) can be written as @xmath9 $ ] , in which @xmath6 is . the context @xmath5 $ ] is called _ empty _ context . an expression rooted by a node @xmath10 labeled by the choice symbol is informally referred to as _ a choice _ and each argument of the choice symbol , or successor of @xmath10 , is referred to as a choice s _ _ backtracking _ is the most traditional approach to non - deterministic computations in functional logic programming . evaluating a choice in some context , say @xmath11 $ ] , consists in selecting either alternative of the choice , e.g. , @xmath12 ( the criterion for selecting the alternative is not relevant to our discussion ) , replacing the choice with the selected alternative , which gives @xmath13 $ ] , and continuing the computation . in typical interpreters , if and when the computation of @xmath13 $ ] completes , the result is consumed , e.g. , printed , and the user is given the option to either terminate the execution or compute @xmath14 $ ] . backtracking is well - understood and relatively simple to implement . it is employed in successful languages such as prolog @xcite and in language implementations such as pakcs @xcite and @xmath15 @xcite . the major objection to backtracking is its incompleteness . if the computation of @xmath13 $ ] does not terminate , no result of @xmath14 $ ] is ever obtained . + _ copying _ ( or _ cloning _ ) is an approach that fixes the inherent incompleteness of backtracking . evaluating a choice in some context , say @xmath11 $ ] , consists in evaluating simultaneously ( e.g. , by interleaving steps ) and independently both @xmath13 $ ] and @xmath14 $ ] . in typical interpreters , if and when the computation of either completes , the result is consumed , e.g. , printed , and the user is given the option to either terminate the execution or continue with the computation of the other . copying is simpler than backtracking and it is used in some experimental implementations of functional logic languages @xcite . the major objection to copying is the significant investment of time and memory made when a non - deterministic step is executed . if an alternative of a choice eventually fails , cloning the context may have been largely useless . for a contrived example , notice that in an arbitrarily large context is cloned when the choice is evaluated , but for one alternative this context is almost immediately discarded . + _ bubbling _ is an approach proposed to avoid the drawbacks of backtracking and copying @xcite . bubbling is similar to copying , in that it clones a portion of the context of a choice to concurrently compute all its alternatives , but the portion of cloned context is typically smaller than the entire context . we recall that in a rooted graph @xmath16 , a node @xmath17 is a _ dominator _ of a node @xmath10 , _ proper _ when @xmath18 , iff every path from the root of @xmath16 to @xmath10 contains @xmath17 . an expression @xmath11 $ ] can be seen as @xmath19 $ ] in which the root of @xmath20 is a dominator of the hole . a trivial case arises when @xmath21 $ ] and @xmath22 . evaluating a choice in some context , say @xmath11 $ ] , distinguishes whether or not @xmath6 is empty . if @xmath6 is the empty context , @xmath12 and @xmath23 are evaluated simultaneously and independently , as in copying , but there is no context to clone . otherwise , the evaluation consists in finding @xmath24 and @xmath20 such that @xmath11=c_1[c_2[u?v]]$ ] and the root of @xmath20 is a proper dominator of the choice , and evaluating @xmath25?c_2[v]]$ ] . when @xmath24 is the empty context , then bubbling is exactly as copying . otherwise a smaller context , i.e. , @xmath20 instead of @xmath6 , is cloned . bubbling intends to reduce cloning in hopes that some alternative of a choice will quickly fail . r2.45 in an objection to bubbling is the cost of finding a choice s immediate dominator and the risk of paying this cost repeatedly for the same choice . this cost entails traversing a possibly - large portion of the choice s context . traversing the context is more efficient than cloning it , since cloning requires node construction in addition to the traversal , but it is still unappealing , since the cost of a non - deterministic step is not predictable and it may grow with the size of an expression . _ pull - tabbing _ , which is at the core of our work , was first sketched in @xcite . the name `` pull - tab '' originates from the metaphor of pulling the tab of a zipper . for an expression , a choice is a tab and a choice s spine is a zipper . as the tab / choice is pulled up , the zipper / spine opens into two equal strands each of which has a different alternative of the choice at the end . evaluating a choice in some context , say @xmath11 $ ] , distinguishes whether or not @xmath6 is empty . if @xmath6 is empty , @xmath12 and @xmath23 are evaluated simultaneously and independently , as in copying and bubbling , without any context to clone . otherwise , the expression to evaluate is of the form @xmath26 $ ] , for some symbol @xmath27 ( for ease of presentation we assume that @xmath27 is unary , but there are no restrictions on its arity ) and some context @xmath6 . pull - tabbing transforms the expression into @xmath28 $ ] . without some caution , this transformation is unsound . unsoundness may occur when some choice has two predecessors , as in our running example . the choice will be pulled up along two paths creating _ two pairs _ of strands that eventually must be pair - wise combined together . some combinations will contain mutually exclusive alternatives , i.e. , subexpressions impossible to obtain in languages such as curry and @xmath15 that adopt the call - time choice semantics . [ fig : pull - tab ] presents an example of this situation . we will show that the soundness is recovered if the left and right alternative of a choice are _ not _ combined in the same expression . to this aim , we attach an identifier to each choice of an expression . we preserve this identifier when a choice is pulled up . if eventually the choice is reduced to either of its alternatives every other choice with the same identifier must be reduced to the same alternative . a very similar idea in a rather different setting is proposed in @xcite . a pull - tab step clones a single node , a predecessor of the choice being pulled up . if the choice is pulled all the way up to the root of an expression , the choice s entire spine is cloned . but if an alternative of the choice fails before the choice reaches the root , further cloning of the choice s context becomes unnecessary . we define a term graph in the customary way @xcite , but extend the decorations of nodes with choice identifiers . [ def : expression ] let @xmath29 be a _ signature _ , @xmath30 a countable set of _ variables _ , @xmath31 a countable set of _ nodes _ , @xmath32 a countable set of _ choice identifiers_. a _ ( rooted ) _ _ graph _ over @xmath33 is a 5-tuple @xmath34 @xmath35 such that : 1 . @xmath36 is the set of nodes of @xmath16 ; 2 . @xmath37 is the _ labeling _ function mapping each node of @xmath16 to a signature symbol or a variable ; 3 . @xmath38 is the _ successor _ function mapping each node of @xmath16 to a possibly empty string of nodes of @xmath16 such that if @xmath39 , where @xmath40 , and ( for the following condition , we assume that a variable has arity zero ) @xmath41 , then there exist @xmath42 in @xmath43 such that @xmath44 ; 4 . @xmath45 is a subset of nodes of @xmath16 called the _ roots _ of @xmath16 ; 5 . @xmath46 is a partial function mapping nodes labeled by the choice symbol to a choice identifier ; 6 . if @xmath47 and @xmath48 and @xmath49 , then @xmath50 , i.e. , every variable of @xmath16 labels one and only one node of @xmath16 ; and 7 . for each @xmath51 , either @xmath52 or there is a path from @xmath53 to @xmath10 where @xmath54 , i.e. , every node of @xmath16 is reachable from some root of @xmath16 . a graph @xmath16 is called a _ term ( graph ) _ , or more simply an _ expression _ , if @xmath55 is a singleton . typically we will use `` expression '' when talking about programs and `` graph '' when making formal claims . choice identifiers play a role in computations . thus , we will explicitly define the @xmath56 mapping only after formally defining the notion of computation . term graphs can be seen , e.g. , in figs . [ fig : sharing ] and [ fig : bubbling ] . every choice node of every graph of fig . [ fig : pull - tab ] would be decorated with the same choice identifier . choice identifiers are arbitrary and only compared for equality . node names are arbitrary and irrelevant to most purposes and are typically omitted . however , some definitions and proofs of our claims need to explicitly refer to some nodes of a graph . for this purpose , we adopt the _ linear notation for graphs _ ( * ? ? ? * def . 4 ) . with this convention , the left graph of fig . [ fig : sharing ] is denoted , where the node names are the italicized identifiers starting with ` @xmath10 ' . we also make the convention that names of nodes that do not need to be referenced can be omitted , hence . the latter is conceptually identical to ( [ value ] ) . in the linear notation for graphs , infix operators are applied in prefix notation , e.g. , see lemma [ invariance - by - pull - tab ] . this practice eases understanding the correspondence between a node identifier and the label of that node . the definition of graph rewriting @xcite is more laborious than , although conceptually very similar to , that of term rewriting @xcite . sections 2 and 3 of @xcite formalize key concepts of graph rewriting such as _ replacement _ , _ matching _ , _ homomorphism _ , _ rewrite rule _ , _ redex _ , and _ step _ in a form ideal for our discussion . therefore , we adopt entirely these definitions , including their notations , and only discuss the manipulation of choice identifiers , since they are absent from @xcite . we now formalize the class of rewrite systems that we consider in this paper . a _ program _ is a rewrite system in a class called _ limited overlapping inductively sequential _ , abbreviated _ lois_. in _ lois _ systems , the rules are left - linear and constructor - based @xcite . the left - hand sides of the rules are organized in a hierarchical structure called a _ definitional tree _ @xcite that guides the evaluation strategy @xcite . in _ systems , there is a single operation whose rules left - hand sides overlap . this operation is called _ choice _ , is denoted by the infix binary operation `` '' , and is defined by the rules of ( [ binary - choice - rules ] ) . _ lois _ systems have been investigated in some depth . below we highlight informally the key results that justify our choice of _ lois _ systems . 1 . any _ lois _ system admits a complete , sound and optimal evaluation strategy @xcite . any constructor - based conditional rewrite system is semantically equivalent to a _ system @xcite . any _ narrowing _ computation in a _ system is semantically equivalent to a _ rewriting _ computation in another similar _ lois _ system @xcite . for the above reasons , _ lois _ systems are an ideal core language for functional logic programs . informally summarizing , _ lois _ systems are general enough to perform any functional logic computation @xcite and powerful enough to compute by simple rewriting @xcite and without wasting steps @xcite . in our setting , a _ computation _ of @xmath57 is a sequence @xmath58[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } e_1 { \mbox{$\rlap{\raisebox{.15ex}[0pt][0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } \ldots$ ] such that @xmath59[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } e_{i+1}$ ] is a _ step _ , i.e. , is either one of two graph transformations : a rewrite , denoted by `` @xmath60 '' , or a pull - tab , denoted by `` @xmath61[0pt]{\tiny$\xi$}}}$ ] '' . rewrite _ is the replacement in a graph of an instance of a rewrite rule left - hand side ( the _ redex _ ) with the corresponding instance of the rule right - hand side ( the _ replacement _ ) . the pull - tab transformation is formally defined in the next section . in principle , we do not exclude choice reductions , i.e. , non - deterministic steps , but in practice we limit them to the root of an expression . the reason is that reducing a choice makes an irrevocable commitment to one of its alternatives . pull - tab steps are equivalent to non - deterministic steps in the sense , formally stated and proved in the next section , that they produce all and only the same results , but without any irrevocable commitment . a computation can be finite or infinite . a computation is _ successful _ or it _ succeeds _ iff its last element is a _ value _ , i.e. , a constructor normal form . a computation is a _ failure _ or it _ fails _ iff its last element is a normal form with some node labeled by an operation symbol . in non - deterministic programs , such as those considered in this paper , the same expression may have both successful computations and failures . each expression of a computation is also referred to as a _ state _ of the computation . a strategy determines which step(s ) of an expression to execute . essential properties of a strategy , such as to succeed whenever possible , will be recalled in sec . [ sec : correctness ] . as described in section [ computations ] , a computation is a sequence of expressions such that each expression of the sequence , except the first one , is obtained from the preceding expression by either of two transformations . one transformation is an ordinary _ redex replacement_. we defer to ( * ? ? ? 23 ) the precise formulation of this transformation and to the next section the handling of decorations by this transformation . the second transformation is defined below . [ def : pull - tab ] let @xmath16 be an expression , @xmath10 a node of @xmath16 , referred to as the _ target _ , not labeled by the choice symbol and @xmath62 the successors of @xmath10 in @xmath16 . let @xmath63 be an index in @xmath64 such that @xmath65 , referred to as the _ source _ , is labeled by the choice symbol and let @xmath66 and @xmath67 be the successors of @xmath65 in @xmath16 . let @xmath68 , for @xmath69 , be the graph whose root is a fresh node @xmath70 with the same label as @xmath10 and successors @xmath71 . let @xmath72 . the _ pull - tab _ of @xmath16 with source @xmath65 and target @xmath10 is @xmath73 $ ] and we write @xmath74[0pt]{\tiny$\xi$}}}\ , g[n \leftarrow g']$ ] . fig . [ fig : pull - tab ] depicts the result of a pull - tab step . for a trivial textual example , @xmath75 is the pull - tab of @xmath76 . the definition excludes targets labeled by the choice symbol . these targets are not a problem for the pull - tab transformation , but would complicate , without any benefit , our treatment . a pull - tab step is conceptually very similar to an ordinary step in a graph a ( sub)graph is replaced . the difference with respect to a rewrite step is that the replacement is not constructed by a rewrite rule , but according to def . [ def : pull - tab ] . it seems very natural for pull - tab steps too to call _ redex _ the ( sub)graph being replaced . term and graph rewriting are similar formalisms that for many problems are able to model the expressions manipulated by functional logic programs . not surprisingly , expressions are terms in term rewriting and graphs in graph rewriting . a significant difference between these formalisms is the identification of a subexpression of an expression . term rewriting uses positions , i.e. , paths in a tree , whereas graph rewriting uses nodes . nodes are used for defining both rewrite rules and expressions to evaluate . nodes are `` placed in service '' ( 1 ) to define rewrite rules , ( 2 ) when an expression , called _ top - level _ , is defined or created for the purpose of a computation , and ( 3 ) to define or construct the replacement used in a step . we agree that any node is placed in service _ only once _ , i.e. , the same node is never allocated to distinct top - level expressions and/or replacements . however , the same node may be found in distinct graphs related by a step , since a step makes a localized change in a graph . these stipulations are formalized by the following principle , which is a consequence of placing nodes in service _ only once_. [ persistence ] let @xmath77 and @xmath78 be graphs . if @xmath10 is a node in @xmath79 , then there exists a graph @xmath16 such that @xmath80[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } \,$}}g_1 $ ] and @xmath80[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } \,$}}g_2 $ ] . to support pull - tabbing and ensure its correctness we attach additional information to an expression . this additional information is formalized as a decoration of a node similar to other decorations present in graph , e.g. , label and successors . in this section , we rigorously define the function that maps nodes to choice identifiers . [ decorations ] let @xmath81[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } g_1 { \mbox{$\rlap{\raisebox{.15ex}[0pt][0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } \ldots$ ] be a computation . we define the @xmath82 mapping , for each element @xmath83 of @xmath84 , by induction on @xmath63 , as follows : @xmath82 takes a node of @xmath83 labeled by the choice symbol and produces the node s choice identifier . base case : @xmath85 . @xmath86 , where @xmath10 is in @xmath87 and is labeled by the choice symbol , is an arbitrary element of @xmath32 , provided that @xmath88 is one - to - one . ind . case : @xmath89 . by the induction hypothesis , @xmath90 is defined for any choice node . in the step @xmath91[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } g_i$ ] , whether rewrite or pull - tab , a subexpression of @xmath92 rooted by a node @xmath93 is replaced by an expression rooted by a node @xmath94 . let @xmath10 be a node of @xmath83 . 1 . if @xmath10 is a node of @xmath92 labeled by the choice symbol , then @xmath95 . + 2 . otherwise , if @xmath96 an ordinary rewrite and @xmath10 is labeled by the choice symbol , then @xmath97 , for an arbitrary @xmath98 provided that @xmath99 for all @xmath100 and all @xmath101 and @xmath102 for @xmath103 i.e. , @xmath104 is fresh . + 3 . otherwise , if @xmath105[0pt]{\tiny$\xi$}}}\ , g_i$ ] a pull - tab and @xmath106 , then @xmath107 , where @xmath108 is the source node of the pull - tab . the above definition is articulated , but conceptually simple . below , we give an informal account of it . in a typical step @xmath109[0pt]{\kern.95ex\tiny$\xi$}}\to\,$ } } g'$ ] , most nodes of @xmath16 end up in @xmath110 . the choice identifier , for choices , of these nodes remains the same . in a rewrite , some nodes are created . any choice node created in the step gets a fresh choice identifier . in a pull - tab , informally speaking , the source ( a choice ) `` moves '' and the target ( not a choice ) `` splits . '' the choice identifier `` moves '' with its source . split nodes have no choice identifier . each node in the `` universe '' of nodes @xmath31 may belong to several graphs . in @xcite , and accordingly in our extension ( see defs . [ def : expression ] and [ decorations ] ) , the function mapping a node to a decoration depends on each graph to which the node belongs . it turns out that some decorations of a node , e.g. , the label , are _ immutable _ , i.e. , the function mapping a node to such decorations does not depend on any graph . we prove the immutability claim for our extension , the choice identifier . obviously , there is no notion of time when one discusses expressions and considers the decorations of a node . hence immutable decorations `` are set '' with the nodes . in practice , these decorations `` become known '' when a node is `` placed in service '' for the purpose of a computation or is created by a step . in view of this result , we drop the subscript from @xmath56 since this practice simplifies the notation and attests a fundamental invariant . pull - tab steps may produce an expression with distinct choices with the same choice identifier . the same identifier tells us that to some extent these redexes are the `` same '' . therefore , when a computation replaces one such redex with the left , resp . right , alternative , every other `` same '' redex should be replaced with the left , resp . right , alternative , too . if this does not happen , the computation is unacceptable . the notion of consistency of computations introduced next abstracts this idea . [ def : consistent ] a rewrite step that replaces a redex rooted by a node @xmath10 labeled by the choice symbol is called a _ choice step_. a computation @xmath84 is _ consistent _ iff for all @xmath98 , there exists an @xmath63 such that every choice step of @xmath84 at a node identified by @xmath104 applies rule @xmath111 of `` '' defined in _ ( [ binary - choice - rules])_. a _ strategy _ determines which step(s ) of an expression to execute . a strategy is usually defined as a function that takes an expression @xmath57 and returns a set @xmath112 of steps of this expression or , equivalently , the reducts of @xmath57 according to the steps of @xmath112 . we will not define any specific strategy . a major contribution of our work is showing that the correctness of pull - tabbing is strategy - independent . the classic definition of correctness of a strategy @xmath113 is stated as the ability to produce for any expression @xmath57 ( in the domain of the strategy ) all and only the results that would be produced by rewriting @xmath57 . `` all and only '' leads to the following notions . + _ soundness : _ if @xmath114 is a computation of @xmath57 in which each step is according to @xmath113 and @xmath23 is a value ( constructor normal form ) , then @xmath115 . + _ completeness : _ if @xmath115 , where @xmath23 is a value ( constructor normal form ) , then there exists a computation @xmath114 in which each step is according to @xmath113 . + in the definitions of soundness and completeness proposed above , the same expression is evaluated both according to @xmath113 and by rewriting . this is adequate with some conventions . rewriting is not concerned with choice identifiers . this decoration can be simply ignored in rewriting computations . in particular , in rewriting ( as opposed to rewriting and pull - tabbing ) a computation is always consistent . in graph rewriting , _ equality of graphs _ is modulo a renaming of nodes . a precise definition of this concept is in ( * ? ? ? 2.5 ) . typically , the proof of soundness is trivial for strategies that execute only rewrite steps , but our strategy executes also pull - tab steps , hence it creates expressions that can not be produced by rewriting . indeed , some of these expressions will have to be discarded to ensure the soundness . the proof of correctness of pull - tabbing is non - trivial and relies on two additional concepts , _ representation _ and _ invariance _ , which are presented in following sections . r1.8 in [ fig : parallel - moves ] proofs of properties of a computation are often accomplished by `` rearranging '' the computation s steps in some convenient order . a fundamental result in rewriting , known as the parallel moves lemma @xcite , shows that in orthogonal systems the steps of a computation can be rearranged at will . a slightly weaker form of this result carries over to _ lois _ systems . a pictorial representation of this result is provided in fig . the symbol `` @xmath116 '' denotes the reflexive closure of the rewrite relation . the notation `` @xmath117 '' , where @xmath10 is a node and @xmath53 is a rule , denotes either equality or a rewrite step at node @xmath10 with rule @xmath53 . a characteristic of pull - tabbing , similar to bubbling and copying , is that the completeness of computations is obtained by avoiding or delaying a commitment to either alternative of a choice . in pull - tabbing , similar to bubbling , _ both _ the alternatives of a choice are kept or `` represented '' in a _ single _ expression throughout a good part of a computation . the proof of the correctness of pull - tabbing is obtained by reasoning about this concept , which we formalize below . [ representation ] we define a mapping @xmath118 that takes an expression @xmath16 and returns a set @xmath119 called the _ represented set of @xmath16 _ as follows . let @xmath16 be an expression . an expression @xmath57 is in @xmath119 iff there exists a consistent computation @xmath120 modulo a renaming of nodes that makes all and only the choice steps of @xmath16 . in other words , we select either alternative for every choice of an expression . for choices with the same identifier , we select the same alternative . since distinct choice steps occur at distinct nodes , by lemma [ parallel - moves ] the order in which the choice steps are executed to produce any member of the represented set is irrelevant . therefore , the notion of represented set is _ well defined_. the notion of represented set of @xmath16 is a simple syntactic abstraction not to be confused with the notion of set of values of an expression @xmath16 @xcite , which is a semantic abstraction fairly more complicated . the proof of correctness of pull - tabbing is based on two results that informally speaking establish that the notion of represented set is invariant both by pull - tab steps and by non - choice steps . we combine the previous lemmas into computations of any length . theorem [ correctness ] suggests to apply both non - choice and pull - tab steps to an expression . choices pulled up to the root are reduced consistently and without context cloning . of course , by the time a choice is reduced , all its spines have been cloned similar to bubbling and copying . a better option , available to pull - tabbing only , is discussed in the next section . the pull - tab transformation is meant to be used in conjunction with some evaluation strategy . we showed that pull - tabbing is not tied to any particular strategy . however , the strategy should be pull - tab - aware in that : ( 1 ) a choice should be evaluated ( to a head normal form ) only when it is _ needed _ @xcite , ( 2 ) a choice in a root position is reduced ( consistently ) , whereas in a non - root position is pulled , and ( 3 ) before pulling a choice , one of the choice s alternatives should be a head - normal form . the formalization of such a strategy would take us well beyond the scope of this paper . in well - designed , non - deterministic programs , either or both alternatives of most ( but not all ) choices should fail @xcite . under the assumption that a choice is evaluated to a head normal form only when it is _ needed _ @xcite , if an alternative of the choice fails , the choice is no longer non - deterministic the failing alternative can not produce a value . thus , the choice can be reduced to the other alternative without loss of completeness and without context cloning . this is where pull - tabbing is advantageous over copying and bubbling any portion of a choice s context not yet cloned when an alternative fails no longer needs to be cloned . of course , the implementation must still identify the choice , and choice s single remaining strand as either left or right , to ensure consistency . we investigated pull - tabbing , an approach to non - deterministic computations in functional logic programming . section [ approaches ] recalls copying and bubbling , the competitors of pull - tabbing . here , we briefly highlight the key differences between these approaches . pull - tabbing ensures the completeness of computations in the sense that no alternative of a choice is left behind until all the results of some other alternative have been produced . similar to every approach with this property , it must clone portions of the context of a choice . in contrast to copying and bubbling , it clones the context of a choice in minimal increments with the intent and the possibility of stopping cloning the context as soon as an alternative of the choice fails . the idea of identifying choices to avoid combining in some expression the left and right alternatives of the same choice appears in @xcite . the idea is developed in the framework of a natural semantics for the translation of ( flat ) curry programs into haskell . a proof of the correctness of this idea will appear in @xcite which also addresses the similarities between the natural semantics and graph rewriting . this discussion , although informal , is enlightening . we formally defined the pull - tab transformation , characterized the class of programs for which the transformation is intended , extended the computations in these programs to include the transformation , proved the correctness of these extended computations , and described the condition that reduces context cloning . in contrast to its competitors , in pull - tabbing any step is a simple and localized graph transformation . this fact should ease executing the steps in parallel . future work , aims at defining a pull - tab - aware parallel strategy and implementing it to measure the effectiveness of pull - tabbing . , antoy , s. , fischer , s. , and reck , f. 2010 . the pull - tab transformation . in _ proceedinsg of the third international workshop on graph computation models_. enschede , the netherlands , 127133 . available at http://gcm-events.org/gcm2010/pages/gcm2010-preproceedings.pdf . optimal non - deterministic functional logic computations . in _ proceedings of the sixth international conference on algebraic and logic programming ( alp97)_. springer lncs 1298 , southampton , uk , 1630 . , brown , d. , and chiang , s. 2006 . lazy context cloning for non - deterministic graph rewriting . in _ proc . of the 3rd international workshop on term graph rewriting , termgraph06_. vienna , austria , 6170 . . set functions for functional logic programming . in _ proceedings of the 11th acm sigplan international conference on principles and practice of declarative programming ( ppdp 2009)_. lisbon , portugal , 7382 . , hanus , m. , liu , j. , and tolmach , a. 2005 . a virtual machine for functional logic computations . in _ proc . of the 16th international workshop on implementation and application of functional languages ( ifl 2004)_. springer lncs 3474 , lubeck , germany , 108125 . \2007 . on a tighter integration of functional and logic programming . in _ aplas07 : proceedings of the 5th asian conference on programming languages and systems_. springer - verlag , berlin , heidelberg , 122138 . \1997 . on constructor - based graph rewriting systems . tech . rep . 985-i , imag . available at ftp://ftp.imag.fr / pub / labo - leibniz / old - archives / pmp / c - graph - rewriting.p% s.gz[ftp://ftp.imag.fr / pub / labo - leibniz / old - archives / pmp / c - graph - rewriting.p% s.gz ] . , rodrguez - hortal , j. , and snchez - hernndez , j. 2007 . a simple rewrite notion for call - time choice semantics . in _ ppdp 07 : proceedings of the 9th acm sigplan international conference on principles and practice of declarative programming_. acm , new york , ny , usa , 197208 . , rodrguez - hortal , j. , and snchez - hernndez , j. 2008 . rewriting and call - time choice : the ho case . in _ proc . of the 9th international symposium on functional and logic programming ( flops 2008)_. springer lncs 4989 , 147162 . , antoy , s. , and nita , m. 2004 . implementing functional logic languages using multiple threads and stores . in _ proc . of the 2004 international conference on functional programming ( icfp)_. acm , snowbird , utah , usa , 90102 .	pull - tabbing is an evaluation approach for functional logic computations , based on a graph transformation recently proposed , which avoids making irrevocable non - deterministic choices that would jeopardize the completeness of computations . in contrast to other approaches with this property , it does not require an upfront cloning of a possibly large portion of the choice s context . we formally define the pull - tab transformation , characterize the class of programs for which the transformation is intended , extend the computations in these programs to include the transformation , and prove the correctness of the extended computations . functional logic programming , non - determinism , graph rewriting , pull - tabbing to 0pt -4.5 in to 0pt draft we d mar 30 17:21:24 pdt 2011
the study of viscoelastic fluids is of great scientific interest and industrial relevance . viscoelastic fluids are fluids that show not only a viscous flow response to an imposed stress , as do newtonian fluids , but also an elastic response . viscoelastic effects are almost universally observed in polymeric liquids@xcite , where they often dominate the flow behavior . they can also be observed in simple fluids , especially in high frequency testing@xcite or in under - cooled liquids@xcite . because most research into viscoelastic liquids , especially that with an eye toward engineering applications , is directed toward polymeric liquids , the viscoelastic behavior of simple liquids is not as well known among researchers . the fact that the manifestation of viscoelasticity does not require the presence of polymer molecules is at the heart of our approach , as will become clear in the description of the viscoelastic model . although in most practical problems involving polymeric materials the viscosities of the materials involved are so large that the creeping flow approximation is valid , the non - linearity introduced by the viscoelastic response of the liquid makes it difficult to treat any but the most simple cases analytically . in engineering applications the situation is often further complicated by the fact that the system is comprised of several immiscible or partially miscible components with different viscoelastic properties . examples of this include polymer blending , where two immiscible polymers are melted and mixed in an extruder , and the recovery of an oil - and - water mixture from porous bed rock . simulation of these systems is very important , but due to the complexities only few numerical approaches exist to date . boundary element methods have been used to simulate such systems with varying degrees of success , but the allowable complexity of the interface morphology is very limited in such approaches . lattice boltzmann simulations have been shown to be very successful for newtonian two - component systems with complex interfaces@xcite , but for viscoelastic fluids the lattice boltzmann models , derived by giraud et al.@xcite , are limited to one - component systems . ( a ) ( b ) in this article we report the successful combination of both two - component and viscoelastic features into a two - dimensional lattice boltzmann model . we used this model to simulate a bubble rising in a viscoelastic liquid ( see figure 1 ) and in this letter report the first successful simulation of the experimentally observed cusp . we use a two - dimensional lattice boltzmann model on a square lattice with a velocity set of @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 @xmath5 @xmath6 , @xmath7 , @xmath8 and a corresponding set of densities @xmath9 , but following giraud _ et al._@xcite we introduce two densities for each non - zero velocity . we use a bgk lattice boltzmann equation that contains the full collision matrix @xmath10 @xmath11 where the summation rule for repeated indices is implied and the required properties of the equilibrium distributions @xmath12 are discussed below . the local density is given by @xmath13 and the momentum by @xmath14 . in order to simulate a two - component mixture we define a second set of nine densities , @xmath15 , with an appropriate equilibrium distribution , @xmath16 . these densities represent the density difference of the two components a and b as @xmath17 , where the total density introduced earlier is @xmath18 . for the @xmath19s we choose a single relaxation time lattice boltzmann equation @xmath20 where @xmath21 is the relaxation time and @xmath22 is the equilibrium distribution . to use the lattice boltzmann method in order to simulate fluid flow , mass and momentum conservation have to be imposed . mass and momentum conservation are equivalent to constraints on the equilibrium distributions : @xmath23 there will be further constraints on the permissible equilibrium distributions in order for the corresponding macroscopic equations to be isotropic and to simulate the systems in which we are interested . in the next two subsections we will summarize the physics that we want to incorporate and then we will discuss how it imposes constraints on the equilibrium distributions and eigenvalues . to simulate a binary mixture we follow the approach of orlandini _ et al . _ @xcite and begin with a free energy functional @xmath24 that consists of the free energy for two ideal gases and an interaction term as well as a non - local interface term : @xmath25&=&\int_{\bf x } \left[t\rho_a \ln(\rho_a ) + t\rho_b \ln(\rho_b)\nonumber\right.\\ & & \left . + \lambda \rho_a \rho_b+\kappa % ( \|\partial_{\bf x } \rho_a\|^2 + \|\partial_{\bf x}\rho_b\|^2 ) \right ] d{\bf x},\end{aligned}\ ] ] where the densities @xmath26 and @xmath27 are functions of @xmath28 . the repulsion of the two components is introduced in the @xmath29 term and @xmath30 is a measure of the energetic penalty for an interface . when we write this free energy functional in terms of the total density , @xmath31 , and the density difference , @xmath32 , we can derive the chemical potential , @xmath33 , and the pressure tensor , @xmath34 , as@xcite : @xmath35 where @xmath36 indicates a functional derivative and @xmath37 is the kronecker delta . for a two - component model we fix the further moments of the equilibrium distributions@xcite : @xmath38 thus far , the model allows us to simulate a binary mixture that phase separates below a critical temperature of @xmath39 . the surface tension , @xmath40 , can be calculated analytically for a flat equilibrium interface @xmath41 orthogonal to the y direction as @xmath42 where the equilibrium density profile of @xmath32 also depends on @xmath30 . viscoelasticity was first proposed by maxwell in his dynamic theory of gases@xcite . he used the simple argument that in the limit where there are no intermolecular collisions the fluid in a container should behave like a solid : `` then it can easily be shown that the pressures on the sides of the vessel due to the impacts of the molecules are perfectly independent of each other , so that the mass of moving molecules will behave , not like a fluid , but like a solid . '' he goes on to deduce that the observed viscous behavior of fluids is due to binary collisions that randomize the directions of stress in the fluid . since the collisions are fast , but not instantaneous , the elastic properties of the fluid are not completely lost , leading to the maxwell model of viscoelasticity . subsequently , derivations of hydrodynamics from the dynamic theory of gases have made the approximation of a purely viscous behavior because of the difficulties of deriving a continuum approach at the length scales of a mean free path of a molecule . in gases where lengths less than the mean free path are important kinetic theory for rarefied gases . there has recently been much activity in the research of the experimentally observed viscoelastic behavior of simple liquids that are undercooled . in this case , however , viscoelasticity is not obtained because the relevant length scales were of the order of a mean free path , but rather because of the correlations of subsequent collisions as described in the mode coupling theory@xcite . the arguments of maxwell , however , are still valid for describing the behavior of the boltzmann equation , and viscoelastic properties can be derived from the boltzmann equation if the decay of viscous stresses is slow . the approach by giraud _ et al . _ aims not at deriving a convected maxwell fluid , but a convected jeffreys fluid which is a mixture of a maxwell fluid with a newtonian fluid . a double set of densities is introduced allowing two stresses , one of which is chosen to relax quickly and is , therefore , a viscous stress , and the other , which is chosen to decay very slowly , represents a viscoelastic stress . the resulting model is a convected jeffreys model that is often used to describe a polymeric fluid in a solvent . care has to be taken for the choice of the collision matrix and the equilibrium distribution to ensure an isotropic model . the details of this one - component model are described in the publication by giraud _ et al._@xcite a chapman - enskog expansion of the lattice boltzmann equations ( [ feqn ] ) and ( [ geqn ] ) gives the macroscopic equations that our system simulates . mass conservation gives the continuity equation : @xmath43 momentum conservation gives a navier stokes equation : @xmath44 where the viscous stress is given by @xmath45 the viscoelastic stress has the constitutive relation @xmath46 where @xmath47 represents the upper convected derivative of @xmath40 . these equations are equivalent to the navier - stokes and jeffreys equations only in the incompressible limit where @xmath48 . the fully compressible equations can only be simulated when a larger set of velocities is used@xcite . the conservation of the density difference leads to the convection diffusion equation @xmath49 where @xmath50 is a diffusion constant given by @xmath51 . we applied our method to a system similar to the experimentally well - studied system of an air bubble rising in a viscoelastic fluid . in our simulation we represent the bubble using a phase - separated newtonian drop of low viscosity in matrix which is viscoelastic by letting the relaxation time @xmath52 in equation ( [ constitutive ] ) depend smoothly on the density difference @xmath32 between @xmath53 in the drop and @xmath54 in the surrounding fluid . we choose @xmath55 , @xmath56 , and @xmath57 in the drop and @xmath58 outside . for the thermodynamic parameters we select @xmath59 , @xmath60 , and @xmath61 , which corresponds to a surface tension of @xmath62 . all units are in terms of the lattice spacings and the time steps @xmath63 . we introduce a forcing dependent on @xmath32 so that the bubble is forced upward while the surrounding fluid is forced downward . we choose the total change in the momentum due to the forcing to be zero so that no walls are required in the simulations . we start the simulations without forcing and then periodically increase the forcing after @xmath64 to @xmath65 iterations . we observe the change in the velocity @xmath66 and store the distribution of @xmath32 so that we have a way of judging the deformation of the bubble . ( a ) ( b ) ( c ) ( d ) figure [ fig : shapes ] shows the form of the drop for different forcings . at low forcings the drop is elongated in the flow direction . this is in direct contrast to a bubble in a newtonian fluid , which is flattened in the flow direction . at a larger forcing the bubble forms a cusp at the lower tip of the drop . for even larger forcings the drop starts to flatten in the flow direction . this sequence is in agreement with the experimental findings@xcite . the elongation of a rising bubble has been simulated before@xcite , but this is the first time that the formation of a cusp has been simulated . in figure [ fig : shapes](c ) it is shown that the cusp can be fitted to the functional form @xmath67 predicted by joseph _ et al._@xcite for a two - dimensional cusp created by the flow induced by two couter - rotating cylinders . experimentally the formation of a cusp has been observed to coincide with a jump of nearly an order of magnitude in the terminal velocity of the bubble@xcite , although the mechanism remains disputed . on the one hand bird _ _ argue that surface - active impurities tend to immobilize bubble surfaces and hence retard the motion of gas bubbles . this discontinuous change in bubble shape may be responsible for the removal of the impurities , and thus lead to a jump in the final velocity . liu@xcite _ et al . _ alternatively suggest that the change in the shape of the bubble will make it more streamlined , and therefore increase the terminal velocity . we examined the velocity for the rising drop as described above and found no jump of about half an order of magnitude as observed by liu _ _ in their experiment ( see figure [ fig : bubble ] ) . our simulations suggest that the jump in velocity they observe is not connected to a more streamline form of the bubble due to the cusp , but more likely to the presence of surfactants that are absent in our simulations . we introduced a lattice boltzmann model that can simulate viscoelastic two - component flows . we gave an intuitive explanation of the origin of viscoelasticity in our model and the model by giraud _ et al . _ in terms of the original theory of maxwell@xcite . simulations using this method have succeeded in reproducing the cusp at the end of a bubble rising in a viscoelastic medium that have eluded earlier numerical attempts with a more traditional boundary integral approach . the model has been successful in the qualitative simulation of the bubble problem in two dimensions . we intend to extend the model to three dimensions in the future . this will also enable us to compare the results quantitatively with experiment . one of us ( a.w . ) would like to thank brad chamberlain for his help in implementing the algorithm in zpl@xcite . we would also like to thank the scientific computing and visualization center at boston university for a mariner grant . l.g . is grateful for its support by the arc 97/02 - 210 project , communaut franaise de belgique .	we developed a new lattice boltzmann method that allows the simulation of two - phase flow of viscoelastic liquid mixtures . we used this new method to simulate a bubble rising in a viscoelastic fluid and were able to reproduce the experimentally observed cusp at the trailing end of the bubble . + +
in these two decades , many circumstellar disks around young forming / formed stars less than several solar masses are found by direct images in the infrared / visible observations ( e.g. mccaughrean & odell 1996 ; fukagawa et al . 2004 ; fujiwara et al . 2006 ; lagage et al . 2006 ; see also web database ) . radio , infrared , and visible line observations confirmed that some disks have rotating disk kinematics as expected ( simon , dutrey , & guilloteau 2000 ; pontoppidan et al . 2008 ; acke , van den ancker , dullemond 2005 ) . in contrast , the formation scenario for massive stars ( @xmath38m@xmath4 ) is still unclear . since , for such massive stars , the time scale for the kelvin - helmholtz contraction is shorter than that of free - fall or accretion with accretion rate similar to low mass star cases , they start releasing energy through nuclear fusion even during accretion ( palla & stahler 1993 ) . then radiation pressure due to their large luminosities may prevent surrounding material from accreting onto the star , in particular , in the case of very massive stars ( kahn 1974 ; wolfire & cassinelli 1987 ) . several ideas are proposed to overcome the problem : mass accretion through circumstellar disks ( yorke & sonnhalter 2002 ; krumholz et al . 2009 ) , mass accretion under a much larger accretion rate than that usually considered for low mass stars ( mckee & tan 2002 ; krumholz et al . 2009 ) , or merging of low mass stars ( bonnell , vine , & bate 2004 ) . among these ideas , non - isotropic accretion through their circumstellar disks seems most plausible at present . such non - isotropic accretion alleviates effective radiation pressure on the accreting material . supporting evidence for this disk scenario is recent discoveries of rotating gas fragments around possible massive young stellar objects ( ysos ) by interferometric observations in the radio , especially in millimeter and submillimeter wavelength regions ( cesaroni et al . some of them have a velocity gradient in a direction orthogonal to molecular outflow lobes , which suggests that the gas fragments rotate around the central ysos ( zhang , hunter , sridharan 1998 ; cesaroni et al . about a dozen of objects as candidate disks around massive ysos of up to 20m@xmath4 are found so far ( zhang et al . 1998 ; patel et al . 2005 ; cesaroni et al . 2006 ; beltrn et al . 2006 ; cesaroni et al . 2007 ) . typically , their estimated stellar mass , luminosity , and disk radius are 4 to less than about 20m@xmath0 , a few @xmath5(10@xmath610@xmath7)l@xmath0 , and 5002000au , respectively . it is suggested that early b herbig be stars are surrounded by flattened structures from measurement of depolarization across h@xmath8 line although the discussed scale is much smaller ( order of several stellar radii ) than the disk size indicated above ( vink et al . 2002 ) . while radio interferometric observations have so far been the most successful in unveiling the disk existence around massive ysos , the resolution around 1@xmath9 is not sufficient to draw the detailed disk geometry . it is in contrast to the situation that disks around young forming stars less than several solar masses have been well depicted by direct images in visible to infrared wavelengths . for some massive ysos , existence of disks is discussed from the polarization vector distribution of infrared scattered light image of the outflow cavities ( jiang et al . 2005 ) , but their disks themselves are not seen because such objects are still embedded deeply in their envelopes . direct images are strongly required to establish the existence and shape of the disks around massive ysos . our new approach is to search for disks in the mid - infrared around massive ysos that have emerged from their natal clouds . owing to the large luminosities of the central stars , the disk surface can be heated up out to large radii enough to be resolved in the mid - infrared with 8m - class telescopes , which provide 100au resolution for nearby ( @xmath1400pc ) targets . we carry out survey observations for extended emission around herbig be stars and report the discovery of a disk around hd200775 in this paper . hd200775 is a herbig be star located at 430@xmath10pc from the sunpc . their estimate , however , also suggests a much lower stellar mass for hd200775 , which obviously conflicts with the established spectral type . since the original estimate of 430pc agrees with the distance by monnier et al . ( 2006 ) within the error , we adopt 430pc as the distance in this paper . ] ( van den ancker et al . 1997 ) . the spectral type is estimated to be b3(@xmath111)e based on the equivalent width of several optical lines such as + at 4922 and h@xmath12 at 4102 ( hernndez et al . its intrinsic bolometric luminosity is estimated as 5400 to 15000l@xmath0 after corrected for the extinction by assuming @xmath13=3.1 to 5.0 , respectively ( hernndez et al . 2004 ) , where @xmath13=5.0 better agrees with the extinction derived independently from b - v and v - r colors . hd200775 appears above the zero - age main sequence on the hr diagram ( hernndez et al . 2004 ) , which supports the youth of the system as does the co biconical cavity formed by the outflow activity extending in the east - west direction with a size of [email protected] ( fuente et al . 1998a ; ridge et al . no high velocity gas component related to the outflow is currently seen , which suggests that the system is at a stage after its outflow activity . the observed co gas wall corresponds to the ridge of the famous optical reflection nebula ngc7023 ( fuente et al . 1992 ) , which is illuminated by hd200775 . since the gas wall around the east and west cavities has red- and blue - shifted velocity components , the outflow is thought to be observed as an almost edge - on configuration ( fuente et al . 1998a ) . the stellar mass of hd200775 is estimated by various studies . hernndez et al . ( 2004 ) show that hd200775 is located at the evolutionary tracks of a star with 8.9 to 12.5m@xmath0 on the hr diagram , which evolvs into a b3b0 zero age main sequence star . based on the time variations of the h@xmath8 line , miroshnichenko et al . ( 1998 ) point out that the star might have a close companion . recent near - infrared interferometric multi - epoch observations derive the orbital parameters of the close binary , such that the semi - major axis is [email protected] ( corresponding to 6.5au at 430pc ) and the orbital period is 1377@xmath1125days ( monnier et al . 2006 ) . on the basis of these data , monnier et al . ( 2006 ) independently estimated the mass of the sum of the primary ( @xmath14 ) and companion stars ( @xmath15 ) as @xmath16=10.4@xmath17m@[email protected] , which is the weighted average between their astrometric measurement and radial velocity measurement by pogodin et al . they assumed the distance as 430@xmath10pc . ] the mass ratio of the companion to the whole stellar system is derived as [email protected] from the variability and the bisector radial velocity of h@xmath8 emission of the system ( pogodin et al . all these characteristics indicate that the primary star of hd200775 is likely to be a massive yso with the mass larger than 8m@xmath0 or more . alecian et al . ( 2008 ) made monitoring observations of the polarization in h@xmath8 and some photospheric lines of hd200775 . they fit a number of photospheric absorption lines with a combination of the the primary and the secondary stars and estimate the orbital parameters . they obtain @[email protected] , which is similar to those obtained by pogodin et al . ( 2004 ) and monnier et al . they attribute shaper , deeper absorption lines to the primary and broader , shallower absorption lines to the secondary star . they find that the radial velocity of the h@xmath8 line follows that of the secondary . , alecian et al . ( 2008 ) derived the mass ratio @[email protected] . note that the assignment of the primary and the secondary by alecian et al . ( 2008 ) is based not on the measured mass but on the line profiles . as a result , their analysis indicates that the primary star has a larger velocity variation and thus the primary star has a less mass than the secondary . on the other hand , they estimated the mass ratio as @[email protected] from the luminosity analysis described below , which is consistent with the orbital analysis . both of alecian et al . ( 2008 ) and pogodin et al . ( 2004 ) show the time variation plot of the h@xmath8 radial velocities , where that assigned as the secondary s one by alecian et al . ( 2008 ) is very similar to the bisector velocity by pogodin et al . ( 2004 ) who concluded it to follow the velocity of the primary . see alecian et al . ( 2008 ) for the details . ] all lines observed in their spectrum of the secondary are also detected in the spectrum of the primary . they speculate that the higher rotational velocity and slightly fainter luminosity of the secondary star make the lines broader and shallower , concluding that the temperature of the secondary star is similar to that of the primary star . the optical spectrum of hd200775 can be fitted with a combination of two stars better than with a single star . from the fitting , they estimated that the primary and the secondary stars have similar effective temperatures ( 18600@xmath112000k ) and that luminosity ratio as @[email protected] . with the total luminosity of 15000l@xmath21 ( hernndez et al . 2004 ) and the luminosity ratio , the two stars can be plotted on the hr diagram independently , from which the stellar masses are estimated as [email protected] and [email protected]@xmath4 , respectively . the total mass ( 20.0m@xmath4 ) agrees with a simple estimate of the total mass @xmath22m@xmath4 , which is directly derived from the orbital parameters by monnier et al . ( 2006 ) and the distance of 430pc . the orbital parameters , luminosity , spectrum analysis , and spectral type can all be accounted for consistently by the assumed distance of 430pc . thus , it is most likely that the hd200775 system has at least one massive star of @xmath110m@xmath21 and that the 15000l@xmath21 is close to the real total luminosity . from the location on the primary and secondary stars on the hr diagram , alecian et al . ( 2008 ) estimated the system age is [email protected] from the birthline . it is in contrast with fuente et al . ( 1998b ) , who suggest it more likely that hd200775 is a post - main sequence star that left the main sequence 8myr ago , just because a pre - main sequence star has a very short life time . alecian et al . ( 2008 ) also estimated the inclination of the binary orbit and of the stellar rotation axis of the primary star as 48@xmath23degrees and 60@xmath1111degrees , respectively . the latter one is derived from the analysis of the magnetic field of the stars . the imaging of hd200775 with the cooled mid - infrared camera and spectrometer ( comics ; kataza et al . 2000 ; okamoto et al . 2003 ; sako et al 2003 ; ishihara et al . 2006 ) on the 8.2 m subaru telescope was made with four filter bands at 8.75 ( @xmath24=0.752)@xmath2 m , 11.74 ( @xmath24=1.058)@xmath2 m , 18.75 ( @xmath24=0.90)@xmath2 m , and 24.56 ( @xmath24=0.75)@xmath2 m . the observations were made on 2006 august 20 - 21 and 2007 june 5 - 6 ( ut ) for the 8.8 and 11.7@xmath2 m bands , and on 2008 july 19 ( ut ) for the 18.8 and 24.5@xmath2 m bands . the pixel scale in the imaging is 0.13@xmath9/pix and the achieved spatial resolutions are 0.26 , 0.32 , and 0.55 , 0.66@xmath9 as the full width at half maximum of the 8.8 , 11.7 , 18.8 , and 24.5@xmath2 m point spread functions ( psfs ) , respectively . during the observations , the secondary mirror was chopped to cancel the background radiation . for the 8.8 and 11.7@xmath2 m observations , we used an on - chip chopping method with a 0.430.45hz and 12@xmath25 chop throw nearly along the disk major axis . for the 18.8 and 24.5@xmath2 m observations , we used 20 or 30@xmath25 throw with 0.43hz along the disk minor axis . for data reduction , we applied a shift - and - add method to produce the images as described below . for the sky subtraction , we used a standard chop subtraction method . flat fielding was done with self - sky flat . the flux calibration was made based on cohen et al . ( 1999 ) . for both of the 8.8 and 11.7@xmath2 m filter bands , we repeated many short exposures ( 100ms and 80ms , respectively ) for the object and the standard stars ( hd198149 or hd186791 ) as the flux and the psf reference . the standards were observed either before or after hd200775 . because the observations were made on several nights , we have several sets of the object and standard exposures pairs . the numbers of the pairs is 3 and 4 , and the total integration time of the object observations for each pair is 43108sec and 4396sec , for the 8.8 and 11.7@xmath2 m band , respectively . for each exposures pair , we divided the original pixel into 10@xmath510 sub - pixels and applied shift - and - add image stacking to keep the spatial resolution of the stacked images . then we obtained seven sets of final images . fluctuations among these sets are taken into account in the uncertainty in the disk model fit in section 4.1 . finally we stacked all the pair images . the total integration time summed up for all the pairs amounts to 195 and 243sec for the 8.8 and 11.7@xmath2 m bands , respectively . for the 18.8@xmath2 m filter band , the total integration time is 3134sec . the psf reference is hd3712 and the flux references are hd3712 , hd189319 , hd186791 , hd198149 , and hd213310 . reduction similar to the 10@xmath2 m images was made . for the 24.5@xmath2 m filter band , the total integration time is 4412sec . the flux and psf references are hd3712 , hd189319 , and hd213310 . we employed a shift - and - add method , but we did not divide the original comics pixels because the object was faint at 24.5@xmath2 m . nevertheless , accuracy of the peak detection was good enough ( 0.12@xmath25 in rms ) . the obtained images are shown in figures 1 and 2 . the brightness profiles at the 10@xmath2 m bands are shown in figure 3 . we have made low dispersion ( r@xmath1250 ) spectroscopy at the 10@xmath2 m region with the comics on the subaru telescope on 2006 august 2021 ( ut ) . we used the long - slit of 0.33@xmath25@xmath536@xmath25 in size along the disk major axis and 0.45hz chop with 12@xmath25 throw along the slit . we measured the slit position precisely from the slit - viewer images . for the flat - fielding , we used thermal spectra of the cell - cover of the primary mirror . the distortion of the spectra on the detector was corrected based on the standard stars , which were observed with the same instrument rotator angle as used for hd200775 . the wavelength was calibrated based on the atmospheric emission lines . the atmospheric absorption was corrected by the spectra of the standard star hd198149 . the spectra are calibrated to give the absolute flux and the wavelength dependence of the slit efficiency . the standard deviation of the sky background in each wavelength is taken as an error . the obtained spectra at several positions are shown in figure 4 . submillimeter array ( sma ; ho , moran , & lo 2004 ) observations of the continuum emission at 345ghz in the lower side band and at 355ghz in the upper side band were made on 2008 july 9th ( ut ) . the extended array configuration consisting of 7 antennas provided projected baseline lengths between 33k@xmath26 to 212k@xmath26 , and the largest structure to which these observations are sensitive enough ( more than 50% level ) is expected to be 2.7@xmath9 ( wilner & welch 1994 ) . 3c279 was observed to obtain the passband characteristics of the system , while the two quasars 1849 + 670 and 2009 + 724 were observed every 15 minutes to track the time variation of the complex gain . visibility data were calibrated with idl mir package for the sma . after the correction for the visibility amplitude by using the record of system temperature , the gain table constructed from 1849 + 670 data was applied to the visibilities of hd200775 . flux density of 1849 + 670 at 350ghz was estimated to be 1.50 jy from the comparison with uranus . both the data sets from the upper and lower side bands are combined , and natural weighting is applied in the image construction process . the synthesized beam size ( , or psf ) is 0.87@[email protected]@xmath9 ( fwhm ) and its position angle of the major axis is 27.7 degrees . the obtained image is shown in figure 5 . the obtained 8.8 , 11.7 , and 18.8@xmath2 m images ( figs . 1 and 2 ) have enough s / n ratios and consist of unresolved source emission and elliptical diffuse emission of 2@xmath9 in the semi - major axis , whose shape appears as an inclined disk . at 18.8@xmath2 m , additional diffuse emission with a curved tail extending toward the northeast is also seen . the observed total fluxes are [email protected] , [email protected] , and [email protected] at 8.8 , 11.7 , and 18.8@xmath2 m , respectively . the fluxes of the unresolved component are 5.4jy , 3.5jy , and 2.0jy , which are much larger than the expected photospheric emissions 0.2jy , 0.1jy , 0.04jy , at 8.8 , 11.7 , and 18.8@xmath2 m ( acke & van den ancker 2004 ) , respectively . the size of unresolved source emission is estimated from comparison of the azimuthally averaged radial profiles between the unresolved source and the psfs . if we assume a gaussian radial profile for the unresolved source , it is estimated that the 95% of the unresolved flux arises from @xmath2722au and @xmath2729au at the 8.8 and 11.7@xmath2 m bands , respectively . thus the unresolved mid - infrared emission arises mostly from the circumstellar dust in the vicinity of the central star(s ) . the elliptical emission has a position angle of [email protected] degrees and is detected out to 7501000au in radius along the major axis ( fig . it is similar to the radii of disk candidates around massive ysos ( cesaroni et al . 2007 ) and disks around low and intermediate mass stars ( 2001000au ; dutrey , guilloteau , & ho 2007 ) , while it is much smaller than the envelope radius ( 28000au ) estimated from the north - south distance of the co cavity wall at the apex . the major axis of the diffuse elliptical emission is orthogonal to the cavity . also the major axis is in the same direction as the ascending node ( pa= [email protected] degree ) of the binary orbit . all the observed shape , size , and configuration support that the elliptical emission arises from the circumbinary disk around hd200775 , which is in almost the same plane as the binary orbit . the results strongly support that the star formed through the disk accretion and that the disk around a massive star still remains even after the outflow phase , that is , the main accretion phase . the 18.8@xmath2 m tail extends toward the southern edge of the northern part of the co cavity apex . it may indicate a structure that bridges the disk and the outer region . the 24.5@xmath2 m image is relatively similar to that at 18.8@xmath2 m . for the disk emission , the isophotal contours are roughly elliptical . the ellipse center is shifted toward the east from the unresolved flux peak source , where the amount of the shift is larger for the fainter isophotal contours . brightness contribution from the psf skirt of the unresolved source is still not negligible ( cf . figure 3 ) , especially for the inner radius region . however , the shift phenomenon can not be explained only by the diffraction effect . such a shift is characteristic to a flared disk , where the ratio of disk height @xmath28 to radius @xmath29 increases monotonically toward the outer region and only the surface emits the mid - infrared photons ( figure 6 ) . it is consistent with that hd200775 shows a spectral energy distribution ( sed ) steeply arising toward the far - infrared ( acke and van den ancker 2004 ) , suggesting a circumstellar flared disk whose surface is irradiated by the central star even at outer regions ( group i by meeus et al . it is very similar to the geometry confirmed for the intermediate mass young star hd97048 ( herbig b9.5e / a0 star with 2.5m@xmath4 and 40l@xmath4 ; lagage et al . 2006 ) . in the case of hd200775 , the east side surface faces us . to derive the disk geometry , we employed a model image fitting method . we assume a simple axisymmetric disk model , where the emitting surface is flared as @xmath30 from the inner edge ( @xmath31 ) to the outer edge ( @xmath32 ) , to see if the flared disk accounts for our obtained images . it is assumed that only the disk surface emits in the mid infrared , that the disk surface beyond the midplane is completely extinguished , and that the surface brightness of one side per projected midplane area is given by @xmath33 . it corresponds to a passive disk where the disk is mainly heated by the radiation onto the disk surface , which is consistent with the fact that the outflow is not active in hd200775 s case . the unresolved source is not included in the model , but subtracted from the observed images by a clean analysis method . the model disk is inclined toward us by an angle @xmath34 from face - on and is convolved with the psf . it is fitted to the observed ( peak - subtracted ) hd200775 images at 8.8 and 11.7@xmath2 m with free parameters of @xmath31 , @xmath32 , @xmath8 , @xmath35 , @xmath36 , and @xmath37 . we use individual exposure pairs for the fitting because we can not obtain good psf images for the pair - averaged images . the best fit result ( figure 1i ) shows that the observed image and the shift in the centers of the contours are well reproduced by the simple model with @xmath32=679@xmath1148au , @xmath31=159@xmath1125au , @[email protected] , @[email protected] and @xmath28[au]=([email protected] ) @xmath5(@xmath29[au]/280au)@xmath38 . for each observed pair image , the difference between the model and the observed images is less than twice the sky noise level on the average5 subpixels into one pixel at the fitting to reduce calculation time . the difference between the observed image after the clean and the model diffuse disk image is calculated , is squared , and is summed up for the fitted region . the value averaged for the combined pixels is minimized by the fitting . the resulted values for individual pair images are 1.1 to 1.8 times larger than the spatial deviation of the sky area of the observed images . we did not calculate the @xmath39 , because the pixel flux of the observed images are not independent because of the pixel size is smaller than the psf sizes . ] . the inclination of the disk agrees with the inclinations of the binary orbit and the stellar rotation axis derived by alecian et al . ( 2008 ; see @xmath402 ) within errors , suggesting that the axes of the stellar rotation , binary orbit , and the disk are aligned relatively well . the observed brightness profile ( fig . 2a ) is proportional to @xmath41 at @xmath42500au , where the agreement with our disk model is fairly good . in the outer region , the observed profile becomes steeper as @xmath43 in the south or more in the north . the fitted outer radius is similar to the radius where the slope of the observed profile becomes much steeper in the north ( @xmath44 in the 11.7@xmath2 m band and @xmath45 in the 8.8@xmath2 m band ) . most of the fit parameters do not show significant difference between 8.8 and 11.7@xmath2 m , while the fitted beta is slightly larger for 8.8@xmath2 m ( [email protected] ) than for 11.7@xmath2 m ( [email protected] ) . it seems to correspond to the shallower brightness slope for 11.7@xmath2 m and suggests that the fit procedure works well . the real inner radius could be smaller due to possible false removal of the emission in the clean procedure , where we allowed multiple peak detection within 0.33@xmath9 radius for 8.8@xmath2 m and 0.46@xmath9 for 11.7@xmath2 m from the brightest peak . the criteria are adopted since very short - time seeing fluctuation can not be treated perfectly even in the shift - and - add process and could produce spurious sub - peaks . while most detected sub - peaks distribute very close to the brightest peak , some sub - peaks are seen along the major axis . part of the emission coming from the inner disk could have been removed by this process . in general , the simple flared disk model accounts for most of the observed characteristics of hd200775 . our results , the first well - resolved direct infrared images around a massive star , clearly reveal the geometry of the disk around the massive yso . the derived flared geometry of the hd200775 disk is much flatter than the disk around the intermediate mass star hd97048 . the maximal and average opening angle of the hd200775 disk is much smaller than the hd97048 disk where @xmath46[au]=51.3@xmath47(@xmath29[au]/135au)@xmath48 ( lagage et al . then @xmath49 becomes 0.12 ( at 680au radius ) to 0.27 ( if we extend our fit result to 1000au radius ) for hd200775 while @xmath50 is 0.49 ( at 370au radius which is the outer limit discussed by lagage et al . ( 2006 ) ) . in fact , it is suggested that disks around massive stars generally appear to be flatter since @xmath51 estimated from the sed is typically much smaller for early herbig be stars than in herbig ae stars ( e.g. acke & van den ancker 2004 ) . it supports the idea that there is a qualitative difference between disks around massive stars and low mass stars . it may be related to rapid disruption of disks around massive stars due to the large luminosities and/or photoevaporation ( see @xmath404.4 ) . in a hydrostatic disk in keplerian rotation around a star , it is expected that @xmath8 becomes @xmath52 ( e.g. dutrey et al . 2007 ) . here we assume that the hydrostatic structure is determined based on the temperature at the equatorial plane and the surface density of the disk , both of which have power - law dependence on radius . also we assume that the temperature decreases as the radius . owing to the large uncertainty in @xmath8 , we could not conclude from the model fit whether or not the hd200775 disk is flared more steeply than the hd97048 disk ( @xmath8=1.26 ) , which is consistent with a hydrostatic one . in our fitting , we fix @xmath8 between 0 and 4.1 with a step of 0.1 and search for a combination of the free parameters that minimizes the squared sum of difference between the model images and the observed images . we find that @xmath36 is strongly correlated with @xmath8 in our model fitting . thus @xmath36 and @xmath8 can not be determined independently . however , the disk height within @xmath53 tends to become similar even if the combination of @xmath36 and @xmath8 changes . the other parameters ( @xmath54 , @xmath34 , @xmath35 , @xmath37 ) does not change largely . the disk detected here is clearly related with the past outflow activity , binary orbit , and inclination of the star . it strongly suggests that this star is still at a pre - main sequence stage , as discussed by alecian et al . the present observations indicates the presence of a disk around a high - mass pre - main sequence star . figure 7 shows the distribution of color temperature of the hd200775 circumbinary disk . it is estimated by using the brightness ratios of the 11.7 and 18.8@xmath2 m images where the unresolved components are subtracted . the difference of the diffraction size in the two bands is not corrected here . in addition , the 11.7 and 18.8@xmath2 m bands contain contributions from the silicate feature emission , from which the estimated temperature may suffer . the iso / sws spectrum toward hd200775 shows the moderate silicate emission at 10@xmath2 m region , but has no prominent silicate emission at 18.8@xmath2 m region . the 11.7@xmath2 m band is located at the skirt of the 10@xmath2 m silicate feature . thus the estimated temperatures may be overestimated . since it is not necessary that the flux at the two bands comes exactly from the same grains and areas , the actual temperature distribution would be more complicated : grains at the surface has higher temperatures and those at a deeper location in the disk have lower temperatures . figure 7 indicates that the color temperature is @xmath1180330k . black grains becomes only 113k or 149k at 430au and 77k or 97k at 1000au from a 6000l@xmath0 or 15000l@xmath0 star , respectively . it suggests that the grains are not blackbodies but they are small and absorb the stellar photons efficiently in visible wavelengths and radiate in the infrared inefficiently . figure 8 shows the temperature distribution of blackbody and small ( radius @xmath55=0.1 and 1@xmath2 m ) grains that absorb the stellar light completely but have an emissivity of @xmath56 in the infrared . it suggests that grains of @xmath10.1@xmath2 m radius have temperatures similar to that of the observed diffuse disk emission . by assuming amorphous mg - pure pyroxene grains with 0.1@xmath2 m radius ( jger et al . 2003 ; see section 4.5 ) at 300k , we derive the mass of the dust emitting at 10@xmath2 m as 8@xmath57m@xmath4 . the total mass including gas is very small ( 8@xmath58m@xmath4 ) , when the gas - to - dust mass ratio is assumed as 100 . also it is difficult to heat up grains in the midplane warm enough ( @xmath59300k ) to be detected in the 10@xmath2 m bands at the observed large radii in a passive disk . we conclude that the mid - infrared disk emission arises from the disk atmosphere , or the superheated surface layer illuminated directly by the central star as modeled for low to intermediate mass stars ( chiang & goldreich 1997 , chiang et al . 2001 , dullemond & dominik 2004 ) . if the beam filling factor by the diffuse disk emission is assumed as unity , the brightness at 10 and 18.8@xmath2 m suggests the optical depth in the mid - infrared ( @xmath60 ) of @xmath61@xmath62 and @xmath63@xmath61 for the dust temperatures of 300k and 200k , respectively . it corresponds to the optical depth at 0.55@xmath2 m ( @xmath64 ) of @xmath65@xmath61 . it should be noted that the optical depth @xmath64 normal to the disk surface is smaller than most model predictions ( e.g. chiang & goldreich 1997 ; @xmath66 for the hd200775 disk ) where it is assumed that the surface height is proportional to the vertical gas scale height at any radius and that the absorbed flux of the stellar radiation at the disk surface is proportional to the sine of the local grazing angle ( @xmath67 ) . there are several possible reasons for the small optical depth . firstly , the temperatures of the dust emitting in the mid - infrared region might be lower than those assumed ( 300k for 10@xmath2 m region and 200k for 20@xmath2 m region , respectively ) . as we mentioned , the silicate features may cause overestimate of the temperatures . if the dust temperature is 150k , the optical depth in the 10@xmath2 m region becomes 60 to 240 times higher . secondly , the photospheric emission of hd2000775 ( @xmath68=19000k ) peaks around 0.1@xmath2 m in the ultra - violet ( uv ) region . the interstellar extinction at 0.1@xmath2 m is larger than at the v - band as @xmath69/@xmath70 is about 5 in a typical case or higher ( @xmath110 ) in other cases ( ryter 1996 ) . since the thickness or the optical depth of the disk atmosphere is determined so that the most stellar emission is absorbed by the disk surface , the thickness of the hd200775 disk case is likely to be determined by the optical depth at the uv rather than visible wavelengths . thirdly , the ratio of extinction between the mid - infrared and optical wavelengths for grains in the hd200775 disk may not be the same for interstellar grains we assumed ( rieke & lebofsky 1985 ) . in fact , it is suggested that the grains in the hd200775 disk are different from those in the interstellar medium ( ism ) as discussed in @xmath404.5 . if @xmath71 in hd200775 is much larger than the ism , we will have a larger @xmath64 . the ratio @xmath71 becomes larger if the grains are much blacker in the visible to uv wavelengths than in the ism , or if the grains have much lower emissivity in the mid - infrared than in the ism . lastly , the small optical depth in hd200775 might be understood in terms of the thermal instability of protoplanetary disks . watanabe & lin 2008 predict that thermal wave can be excited to make ripples of the disk surface in a disk with a low accretion rate . they predict that @xmath64 normal to the disk is @xmath11 only at the excited ripples ( thermal wave pulse ) and becomes much lower ( a few tenths to less than @xmath63 ) at the other most areas . the ratio of the infrared excess emission ( @xmath72 ) to the total luminosity ( @xmath73 ) is estimated to be 0.07 from the sed ( acke & van den ancker 2004 ) . the flared geometry derived by our disk model fit gives the fraction of the intercepted stellar radiation of 0.120.27 . taking account of the radial optical opacity of the surface layer ( @xmath74 ) , @xmath75 becomes much lower than that from the sed . it can be attributed partly to the small @xmath64 again , and partly to the following fact . in hd200775 , extended far infrared emission that roughly traces the outflow cavity wall ( ngc7023 ) is detected ( kirk et al . 2009 ; we also confirmed it with data of akari satellite ( murakami et al . therefore the far infrared flux of the sed ( iras data and even iso data ) has a significant contribution from the wall . the large @xmath75 of 0.07 indicates that the cavity wall intercepts a large fraction of the stellar flux . the disk itself does not have to absorb a large fraction . from our 350.3ghz sma observations , we detected an unresolved emission source of [email protected] at the star position with the 1@xmath76 noise level of 3.0mjy / beam ( fig . if we assume the mass absorption coefficient in the radio region given by hildebrand ( 1983 ) and the spectrum ( flux @xmath77 ) slope from submillimeter to millimeter wavelengths as @xmath78 , the mass opacity at 350ghz becomes 0.0085@xmath79g@xmath80 . then the observed flux gives a lower limit for the total disk mass ( @xmath81 ) of 0.022 or 0.010m@xmath4 for the dust temperature at the midplane of 50k or 100k ( @xmath82 ) , respectively . the assumed spectrum slope is in agreement with the sed in acke and van den ancker ( 2004 ) . it is also compatible with the upper limit of the flux at 1.3 mm ( @xmath836mjy ) of past bolometer observations of a 10@xmath9 beam ( fuente et al . the slope indicates that a majority of the grains in hd200775 is small in size and not fully grown since large grains have a slower wavelength dependence . the mass above is similar to the estimated minimum mass for the protoplanetary disk around the sun . if it is distributed uniformly in a sphere of the beam radius ( 180au ) , it amounts to visible extinction of @xmath3190mag toward the central star , which is much larger than the observed one ( 1.8 and 3.0mag under @xmath13=3.1 and 5.0 , respectively ; hernndez et al . 2004 ) . here , we assume the optical to mid - infrared extinction ratio given by rieke & lebofsky ( 1985 ) and @xmath84 value in cox ( 2001 ) . thus the mass must be distributed in a flattened structure , which consists of the inner part of the disk . if we assume the brightness of the 2@xmath76 detection limit of our sma observations ( 6mjy / beam ) for the extended emission , the mid - infrared extinction becomes at least 2.0mag ( for 100k dust ) or 4.4mag ( for 50k dust ) at 11.7@xmath2 m even in the outer region . extinction at 8.8@xmath2 m is 1.5 times larger than that at 11.7@xmath2 m ( rieke & lebofsky 1985 ) . extinction at 18.8@xmath2 m is not well known and we assume that it is similar to that at 11.7@xmath2 m ( draine & lee 1984 ; mcclure 2009 ) . then , the mid - infrared emission from the further side of the disk should be extinguished mostly by the midplane before reaching us even in the outer region . the observations of the mid - infrared and 350ghz can be interpreted in terms of the thick disk in the mid - infrared . while the disk shows geometry and structures relatively similar to low mass stars as described above , there are clear differences , too . since the primary star is hot enough , its ionizing photons can photoionize the disk gas . it leads to disk dispersal due to photoevaporation in the region where the sound speed of ionized gas exceeds the escape speed against gravity . the inner radius ( photoevaporation radius @xmath85 ) is estimated as @xmath170au for a 10m@xmath4 star in a weak stellar wind case ( hollenbach et al . the mid - infrared elliptical emission of hd200775 corresponds to the region outside @xmath85 . the distribution of the free - free emission at 3.4 mm detected toward hd200775 ( fuente et al . 2001 ) resembles the elliptical disk and the tail in the mid - infrared well in size and shape . though it was considered to arise from the stellar wind previously ( fuente et al . 2001 ) , the 3.4 mm and mid - infrared emissions are likely to trace the surface of the photoevaporating disk ( fig . 6 ) . when we assume the ionizing flux from a b3 zero - age main sequence star ( @xmath86 photons s@xmath80 ; panagia 1973 ) , the mass loss rate due to the photoevaporation is estimated as 7.7@xmath87m@xmath4yr@xmath80 ( hollenbach et al . 1994 ) . it suggests that a mass of 0.01m@xmath4 to 0.1m@xmath4 will be dispersed within an order of 1000 to 10000yrs . the disk around hd200775 seems to be a short - lived phenomenon . the spatially resolved 10@xmath2 m band spectra of the circumstellar disks have much information on the properties of grains and related processes ( e.g. , okamoto et al . 4 shows the spectra of the unresolved source and at a typical position in the diffuse disk along the major axis . the spectrum of the unresolved source ( fig . 4a ) is dominated by the featureless continuum emission and shows the line at 12.37@xmath2 m of n=7@xmath886 and n=11@xmath888 transitions with a line flux of ( 4.1@xmath89wm@xmath90 . the innermost disk is well heated deep into the midplane and radiates an optically thick blackbody emission . the temperature estimated from the spectrum is 1600k ( @xmath31000k considering the 1 @xmath76 error of the flux calibration ) , which is close to the evaporation temperature of silicate and graphite . black grains illuminated by a 15000l@xmath4 star become 1600k at 3.7au from the star . since it is smaller than the binary orbit , the unresolved source emission arises not from the circumbinary disk but from circumstellar disk(s ) . source . however , artymowicz & lubow ( 1994 ) points out that a circumbinary disk with typical viscous disk parameters around a binary system with a semi - major axis @xmath55 tends to have an inner truncated radius of [email protected]@xmath55 for the binary eccentricity from 0 to 0.25 . they also point out that the minimum separation between the component star and the inner circumbinary disk edge is larger than @xmath55 for the binary eccentricity less than 0.75 . the hd200775 binary orbit has an eccentricity of [email protected] ( monnier et al . thus , it is unlikely that 9.5au is larger than the inner edge of the circumbinary system . it is most likely that the unresolved source emission originates from circumstellar disk(s ) . ] the surface area of the blackbody emission ( @xmath91arcsec@xmath92 ) indicates a dust ring ( or inner wall ) at 3.7au radius with only a 0.06au width ( or height ) . lpine & nguyen - quang - rieu ( 1974 ) detected an oh maser line emission at 1667mhz toward hd200775 and concluded that the line profile may be formed in a rotating gaseous disk - shaped envelope within 8 stellar radii . the detected line emission probably stems from the ionized gas in the dust disk surface or the innermost gas disk where the grains are evaporated . the diffuse emission at 1.3@xmath9 south from the unresolved source ( fig . 4b ) shows the amorphous silicate feature with a triangular shape . the peak is located at 9.2@xmath2 m which is largely shifted from the 9.59.8@xmath2 m amorphous silicate peak often seen toward the interstellar medium ( kemper , vriend , & tielens 2004 ) , evolved mass - losing stars ( dorschner et al . 1995 ) , and disks around low to intermediate mass young stars ( honda et al . it is also shorter than the averaged peak of absorption features of dust grains in outer envelopes of massive protostars and is more compatible with pyroxene ( ( mg , fe)sio@xmath93 ) than olivine ( ( mg , fe)@xmath94sio@xmath95 ) . we compare the continuum subtracted emission with the normalized emission from amorphous silicate grains at 300k based on jger et al . ( 2003 ) ( mg@xmath96sio@xmath97 ) and pyroxene ( mgsio@xmath93 ) grains with 0.1 and 1.0@xmath2 m radii . as can be seen in fig . 4b , the peak wavelength decreases as the sio@xmath94 fraction in silicate increases ( koike & hasegawa 1987 ; jger et al . it suggests the dominance of pyroxene grains or grains with larger sio@xmath94 fraction rather than olivine . the observed peak wavelength is better matched with mg@xmath96sio@xmath97 grains , but mgsio@xmath93 gives a better fit in the feature width . there is a report ( tamanai et al . 2007 ) that sio@xmath94 grains of 0.55@xmath2 m diameter have a peak between 8.99.5@xmath2 m . however , our spectra do not show a feature around 12.5@xmath2 m characteristic to sio@xmath94 . thus the grains are in form of silicate and not pure sio@xmath94 . the presence of a 8.9@xmath2 m feature in regions is reported by peeters et al . ( 2005 ) , but it seems not to coincide with the observed 9.2@xmath2 m feature . grain properties very close to massive stars , especially in disks , may be different from those of disks around low mass stars and even their own envelopes . to understand the characteristic dust properties , the photoevaporation , that is , the ionizing flux or ionized gas might be an important factor . there are many reports of experiments of irradiating ions such as h@xmath98 and he@xmath98 on cosmic grains or model grains . irradiating crystalline olivine with high energy protons of a few mev , typical of cosmic rays , makes no alteration of the crystalline structure at fluences relevant to the ism ( day 1977 ) . in contrast , irradiation of h@xmath99 and he@xmath99 ions of a few to several tens kev can amorphize olivine and pyroxene crystals and change the chemical composition of the grains ( bradley 1994 ; dukes et al . 1999 ; demyk et al.2001 ) . demyk et al . ( 2001 ) made an irradiation experiment on crystalline olivine with he@xmath98 ions at energies of 4 to 10kev and fluences from 5@xmath100 to 10@xmath101 ionscm@xmath90 . it caused amorphyzation of olivine in conjunction with an increase in the porosity of the material due to the formation of bubbles . furthermore , the amorphized layer was deficient in oxygen and magnesium . they found that the o / si and mg / si ratios decreased as the he@xmath98 fluence increased . it suggests the change of the olivine to pyroxene structure in the grains . bradley ( 1994 ) investigated a pyroxene crystal on the surface of an interplanetary dust particle and its rim , where alternation has occurred by the irradiation by solar wind ions . he found that the rim was stoichiometrically depleted in mg and ca and enriched in s , si , and fe relative to pyroxene . he also made irradiation experiments of 20kev protons onto a sample of fine - grained matrix from the allende chondritic meteorite . irradiation of both of olivine and pyroxene crytals in allende again led to the stoichiometric depletion of mg and ca . he concluded that ion irradiation first striped cations such as mg and ca with the weakest bond strengths and then more strongly bound cations such as si . these results indicate that irradiation of energetic ions such as h@xmath98 and he@xmath98 causes a change in the lattice structure of grains from olivine to pyroxenes , then to more sio@xmath94 rich structures . since the hd200775 disk is photoevaporating and a stellar wind activity is suggested ( skinner , brown , & stewart 1993 ) , the grains of the hd200775 disk might be alternated by such a mechanism in plasma of the disk surface or by the stellar wind . further study is required to investigate whether or not the alternation is a common phenomenon in the formation of massive stars . we have made mid - infrared imaging and spectroscopy of the herbig b3e star hd200775 of @xmath110m@xmath4 with subaru / comics . we found elongated emission around the unresolved excess emission in all of the 8.8 , 11.7 , 18.8 , and 24.5@xmath2 m bands . the elongated elliptical emission has a radius of about 1000au and configuration perpendicular to the co cavity wall formed by the past outflow and parallel to the projected major axis of the central close binary orbit . the brightness contours of the elliptical emission are shifted from the stellar position or from the unresolved source position . all of these observed characteristics are well explained as the surface emission of the tilted flared circumbinary disk . the present observations provide the first well - resolved infrared images of a disk around 10m@xmath4 yso and strongly support that hd200775 is formed through the disk accretion . the optical depth and the silicate features of the disk emission support the disk atmosphere configuration , which is well modeled for low to intermediate mass stars . we fit the observed image with a flared disk model . the overall observed structure is well explained by the model . the derived flared geometry is much flatter than the disk around the intermediate mass star hd97048 . the value of @xmath49 is estimated to be 0.12 to 0.27 for the hd200775 disk , while it is 0.49 for the hd97048 disk . it supports the idea that there is a qualitative difference between disks around massive stars and lower mass stars . the 3.4 mm free - free emission image in literature is very similar to the mid - infrared elliptical disk emission in size and shape . it suggests that the disk surface photoevaporates due to the ionizing photons from the central star . the sma 350ghz observations detect the unresolved emission with a 0.8@xmath9 beam . the disk mass concentrated around the star is estimated as around 0.02m@xmath4 , which is similar to that of the minimum mass the solar nebula . the 10@xmath2 m region spectra of the peak unresolved source and at the 1.3@xmath9 south are discussed . the spectrum of the unresolved source shows 1600k featureless emission , which correspond to the innermost circumstellar disk in the vicinity of the central star(s ) . the spectrum of the diffuse emission at 1.3@xmath9 south is dominated by the amorphous silicate emission . the peak at 9.2@xmath2 m is shorter than the usual silicate features and unique to the hd200775 disk . it may suggest alternation of grains due to plasma irradiation . we thank h. nomura for the helpful discussion with her and y. doi for his help in examining the akari data of hd200775 . we thank m. tosa , s. takita , k. sato , k. manabe , and k. tomita for their help at the subaru observations and reduction . we would like to thank all of the staff members of the subaru telescope and sma for their support during the observations and development of the instrument . we are grateful to the anonymous referee for the helpful comments which significantly improved this paper . this work was supported by grant - in - aid for scientific research on priority areas from the ministry of education , culture , sports , science and technology , japan . y. k. o. was also supported by grant - in - aid for young scientists ( b ) and ( a ) , by the ministry of education , culture , sports , science and technology , japan . acke , b. & van den ancker , m. e. 2004 , , 426 , 151 acke , b. & van den ancker , m. e. , dullemond , c. p. , 2005 , , 436 , 209 alecian , e. et al . , 2008 , mnras , 385 , 391 artymowicz , p. & lubow , s. h. , 1994 , , 421 , 651 beltrn , m. t. , cesaroni , r. , codella , c. , testi , l. , furuya , r. s. , olmi , l. , 2006 , nature , 443 , 427 bonnell , i. a. , vine , s. g. , & bate , m. r. , 2004 , , 349 , 735 bradley , j. p. , 1994 , science , 265 , 925 cesaroni , r. , neri , r. , olmi , l. , testi , l. , walmsley , c. m. , hofner , p. , 2005 , , 434 , 1039 cesaroni , r. , galli , d. , lodato , g. , walmsley , c. m. , & zhang , q. , 2006 , nature , 444 , 703 cesaroni , r. , galli , d. , lodato , g. , walmsley , c. m. , & zhang , q. , 2007 , in protostars and planets v , ed . reipurth , b. , jewitt , d. , & keil , k. ( univ . arizona press , tucson , 2007 ) , 197 chiang , e. i. & goldreich , p. , 1997 , apj , 490 , 368 chiang , e. i. et al . 2001 , apj , 547 , 1077 cohen , m. , walker , r. g. , carter , b. , hammersley , p. , kidger , m. , noguchi , k. , 1999 , apj , 117 , 1864 cox , a. n. , 2001 , allen s astrophysical quantities , fourth edtition ( 4th ed . ; new york : springer ) day , k. l. , 1977 , mnras , 178,49 demyk , k. , et al . , 2001 , a&a , 368 , l38 dorschner , j. , begemann , b. , henning , th , jger , c. , & mutschke , h. , 1995 , a&a , 300 , 503 draine , b. t. & lee , h. m. 1984 , apj , 285 , 89 dukes , c. a. , baragiola , r. a. , mcfadden , l. a. , 1999 , jgr , 104 , 1865 dullemond , c. p. & dominik , c. , 2004 , a&a , 417 , 159 dutrey , a. , guilloteau , s. , & ho , p. , 2007 , in protostars and planets v , reipurth , b. , fuente , a. , et al . , 1998b , a&a , 334 , 253 fuente , a. , martin - pintado , j. , cernicharo , j. , brouillet , n. , duvert , g. , 1992 , a&a , 260 , 341 fuente , a. , martin - pintado , j. , rodriguez - franco , a. , & moriarty - schieven , 1998a , g. d. a&a , 339 , 575 fuente , a. , neri , r. , martin - pintado , j. , bachiller , r. , rodriguez - franco , a. , & palla , f. 2001 , a&a , 366 , 873 fujiwara , h. , et al .. 2006 , apj , 644 , l133 fukagawa , m. , et al . 2004 , apj , 605 , l53 hernndez , j. , calvet , n. , riceo , c. , hartmann , l. , berling , p. , 2004 , aj , 127 , 1682 hildebrand , r. h. , 1983 , roy . soc . quart . 24 , 267 ho , p. t. p. , moran , j. m. , lo , k. y. , 2004 , apj , 616 , l1 hollenbach , d. , johnstone , d. , lizano , s. , shu , f. , 1994 , apj , 428 , 654 honda , m. et al . , 2006 , apj , 646 , 1024 ishihara , d. , et al . 2006 , apj , 131 , 1074 jger , c. , dorschner , j. , mutschke , h. , posch , th . , & henning , th . , 2003 , a&a , 408 , 193 jiang , z. , tamura , m. , fukagawa , m. , hough , j. , lucas , p. , suto , h. , ishii , m. , yang , j. , 2005 , nature , 437 , 112 kahn , f. d. , 1974 , a&a , 37 , 149 kataza , h. , okamoto , y. , takubo , s. , onaka , t. , sako , s. , nakamura , k. , miyata , t. , yamashita , t. et al . 2000 , proc . spie , 4008 , 1144 kemper , f. vriend , w. j. , & tielens , a. g. g. m. , 2004 , apj , 609 , 826 kirk , j. m. , et al . 2009 , astro - ph arxiv:0903.4063 krumholz , m. r. et al . 2009 , science , 323 , 754 koike , c. & hasegawa , h. 1987 , ap&ss , 134 , 361 lagage , p - o et al . , 2006 , science , 314 , 621 lpine , j. r. d. & nguyen - quang - rieu , 1974 , a&a , 36 , 469 mccaughrean , m. j. & odell , c. r. , 1996 , aj , 111 , 1977 mcclure , m. , 2009 , apj , 693 , l81 mckee , c. f. & tan , j. c. , 2002 , nature , 416 , 59 meeus , g. , waters , l. b. f. m. , bouwman , j. , van den ancker , m. e. , waelkens , c. , malfait , k. , 2001 , a&a , 365 , 476 monnier , j. d. et al . 2006 , apj , 647 , 444 miroshnichenko , a. s. , mulliss , c. l. , bjorkman , k. s. , morrison , n. d. , glagolevskij , y. v. , chountonov , g. a. , 1998 , pasp , 110 , 883 murakami , h. , et al . 2007 , pasj , 59 , 369 okamoto , y. k. , et al . 2004 , nature , 431 , 660 okamoto , y. k. , kataza , h. , yamashita , t. , miyata , t. , sako , s. , takubo , s. , honda , m. , onaka , t. et al . , 2003 , proc . spie , 4841 , 169 palla , f. & stahler , s. w. , 1993 , apj , 418 , 414 panagia , n. , 1973 , aj , 78 , 929 patel , n. a. et al . , 2005 , nature , 437 , 109 peeters , e. , tielens , a. g. g. m. , boogert , a. c. a. , hayward , t. l. , and allamandra , l. j. , 2005 , apj , 620 , 774 pogodin , m. a. et al . 2004 , a&a , 417 , 715 pontoppidan , k. m. , blake , g. z. , van dishoeck , e. f. , smette , a. , ireland , m. j. , brown . j. , 2008 , apj , 684 , 1323 ridge , n. a. , wilson , t. l. , megeath , s. t. , allen , l. e. , myers , p. c. , 2003 , , 126 , 286 rieke g. h. & lebofsky , m. j. , 1985 , apj , 288 , 618 ryter , c. e. 1996 , , 236 , 285 sako , s. et al . 2003 , pasp , 115 , 1407 simon , m. , dutrey , a. , guilloteau , s. , 2000 , apj , 545 , 1034 skinner , s. l. , brown , a. , steward , r. t. , 1993 , apjs , 87 , 217 tamanai , a. mutschke , h. , blum , j. , & kurirov , a. , et al . 2007 , preprint of astrophysical institute and university observatory , friedrich - schiller - university jena van den ancker , m. e. , the , p. s. , tjin a djie , h. r. e. , catala , c. , de winter , d. , blondel , p. f. c. , waters , l. b. f. m. et al . , 1997 , a&a , 324 , l33 vink , j. s. , drew , j. e. , harried , t. j. , oudmaijer , r. d. , 2002 , , 337 , 356 watanabe , s. & lin , d. n. c. , 2008 , apj , 672 , 1183 wilner , d. j. , & welch , w. j. 1994 , , 427 , 898 wolfire , m. g. & cassinelli , j. p. , 1987 , apj , 319 , 850 yorke , h. w. & sonnhalter , 2002 , apj . 569 , 846 zhang , q. , hunter , t. r. , & sridharan t. k. , 1998 , apj , 505 , l151	we made mid - infrared observations of the 10m@xmath0 herbig be star hd200775 with the cooled mid - infrared camera and spectrometer ( comics ) on the 8.2 m subaru telescope . we discovered diffuse emission of an elliptical shape extended in the north - south direction in @xmath11000au radius around unresolved excess emission . the diffuse emission is perpendicular to the cavity wall formed by the past outflow activity and is parallel to the projected major axis of the central close binary orbit . the centers of the ellipse contours of the diffuse emission are shifted from the stellar position and the amount of the shift increases as the contour brightness level decreases . the diffuse emission is well explained in all of geometry ( the shape and the shift ) , size , and configuration by an inclined flared disk where only its surface emits the mid - infrared photons . our results give the first well - resolved infrared disk images around a massive star and strongly support that hd200775 is formed through the disk accretion . the disk survives the main accretion phase and shows a structure similar to that around lower - mass stars with disk atmosphere. at the same time , the disk also shows properties characteristic to massive stars such as photoevaporation traced by the 3.4 mm free - free emission and unusual silicate emission with a peak at 9.2@xmath2 m , which is shorter than that of many astronomical objects . it provides a good place to compare the disk properties between massive and lower - mass stars .
methods for protein structure prediction , simulation and design rely on an energy function that represents the protein s free energy landscape ; a protein s native state typically corresponds to the state with minimum free energy @xcite . so - called knowledge based potentials ( kbp ) are parametrized functions for free energy calculations that are commonly used for modeling protein structures @xcite . these potentials are obtained from databases of known protein structures and lie at the heart of some of the best protein structure prediction methods . the use of kbps originates from the work of tanaka and scheraga @xcite who were the first to extract effective interactions from the frequency of contacts in x - ray structures of native proteins . miyazawa and jernigan formalized the theory for contact interactions by means of the quasi - chemical approximation @xcite . many different approaches for developing kbps exist , but the most successful methods to date build upon a seminal paper by sippl published two decades ago which introduced kbps based on probability distributions of pairwise distances in proteins and reference states @xcite . these kbps were called `` potentials of mean force '' , and seen as approximations of free energy functions . sippl s work was inspired by the statistical physics of liquids , where a `` potential of mean force '' has a very precise and undisputed definition and meaning @xcite . however , the validity of the application to biological macromolecules is vigorously disputed in the literature @xcite . nonetheless , pmfs are widely used with considerable success ; not only for protein structure prediction @xcite , but also for quality assessment and identification of errors @xcite , fold recognition and threading @xcite , molecular dynamics @xcite , protein - ligand interactions @xcite , protein design and engineering @xcite , and the prediction of binding affinity @xcite . in this article , the abbreviation `` pmf '' will refer to the pairwise distance dependent kbps following sippl @xcite , and the generalization that we introduce in this article ; we will write `` potentials of mean force '' in full when we refer to the real , physically valid potentials as used in liquid systems @xcite . at the end of the article , we will propose a new name for these statistical quantities , to set them apart from true potentials of mean force with a firm physical basis . despite the progress in methodology and theory , and the dramatic increase in the number of experimentally determined protein structures , the accuracy of the energy functions still remains the main obstacle to accurate protein structure prediction @xcite . recently , several groups demonstrated that it is the quality of the coarse grained energy functions @xcite , rather than inadequate sampling , that impairs the successful prediction of the native state @xcite . the insights presented in this article point towards a new , theoretically well - founded way to construct and refine energy functions , and thus address a timely problem . we start with an informal outline of the general ideas presented in this article , and then analyze two notable attempts in the literature to justify pmfs . we point out their shortcomings , and subsequently present a rigorous probabilistic explanation of the strengths and shortcomings of traditional pairwise distance pmfs . this explanation sheds a surprising new light on the nature of the reference state , and allows the generalization of pmfs beyond pairwise distances in a statistically valid way . finally , we demonstrate our method in two applications involving protein compactness and hydrogen bonding . in the latter case , we also show that pmfs can be iteratively optimized , thereby effectively sculpting an energy funnel @xcite . in order to emphasize the practical implications of the theoretical insights that we present here , we start with a very concrete example that illustrates the essential concepts ( see fig . [ fig : simple ] ) . currently , protein structure prediction methods often make use of fragment libraries : collections of short fragments derived from known protein structures in the protein data bank ( pdb ) . by assembling a suitable set of fragments , one obtains conformations that are protein - like on a local length scale . that is , these conformations typically lack non - local features that characterize real proteins , such as a well - packed hydrophobic core or an extensive hydrogen bond network . such aspects of protein structure are not , or only partly , captured by fragment libraries . formally , a fragment library specifies a probability distribution @xmath0 , where @xmath1 is for example a vector of dihedral angles . in order to obtain conformations that also possess the desired non - local features , @xmath0 needs to be complemented with another probability distribution @xmath2 , with @xmath3 being for example a vector of pairwise distances , the radius of gyration , the hydrogen bonding network , or any combination of non - local features . typically , @xmath3 is a deterministic function of @xmath1 ; we use the notation @xmath4 when necessary . for the sake of argument , we will focus on the radius of gyration ( @xmath5 ) at this point ; in this case @xmath4 becomes @xmath6 . we assume that a suitable @xmath7 was derived from the set of known protein structures ; without loss of generality , we leave out the dependency on the amino acid sequence for simplicity . the problem that we address in this article can be illustrated with the following question : how can we combine @xmath7 and @xmath0 in a rigorous , meaningful way ? in other words , we want to use the fragment library to sample conformations whose radii of gyration @xmath8 are distributed according to @xmath7 . these conformations should display a realistic _ local _ structure as well , reflecting the use of the fragment library . simply multiplying @xmath9 and @xmath0 does not lead to the desired result , as @xmath1 and @xmath10 are not independent ; the resulting conformations will not be distributed according to @xmath11 . the solution is given in fig . [ fig : simple ] ; it involves the probability distribution @xmath12 , the probability distribution over the radius of gyration for conformations sampled solely from the fragment library . the subscript @xmath13 stands for _ reference state _ as will be explained below . the solution generates conformations whose radii of gyration are distributed according to @xmath7 . the influence of @xmath0 is apparent in the fact that for conformations with a given @xmath8 , their local structure @xmath1 will be distributed according to @xmath14 . the latter distribution has a clear interpretation : it corresponds to sampling an infinite amount of conformations from a fragment library , and retaining only those with the desired @xmath5 . note that even if we chose the uniform distribution for @xmath0 , the resulting @xmath12 will _ not _ ( necessarily ) be uniform . intuitively , @xmath11 provides correct information about the radius of gyration , but no information about local structure ; @xmath0 provides approximately correct information about the structure of proteins on a local length scale , but is incorrect on a global scale ( leading to an incorrect probability distribution for the radius of gyration ) ; finally , the formula shown in fig . [ fig : simple ] merges these two complementary sources of information together . another viewpoint is that @xmath11 and @xmath15 are used to correct the shortcomings of @xmath0 . this construction is statistically rigorous , provided that @xmath7 and @xmath0 are proper probability distributions . after this illustrative example , we now review the use of pmfs in protein structure prediction , and discuss how pmfs can be understood and generalized in the theoretical framework that we briefly outlined here . many textbooks present pmfs as a simple consequence of the boltzmann distribution , as applied to pairwise distances between amino acids . this distribution , applied to a specific pair of amino acids , is given by : @xmath16 where @xmath17 is the distance , @xmath18 is boltzmann s constant , @xmath19 is the temperature and @xmath20 is the partition function , with @xmath21 . the quantity @xmath22 is the free energy assigned to the pairwise system . simple rearrangement results in the _ inverse boltzmann formula _ , which expresses the free energy @xmath22 as a function of @xmath23 : @xmath24 to construct a pmf , one then introduces a so - called _ reference state _ with a corresponding distribution @xmath25 and partition function @xmath26 , and calculates the following free energy difference : @xmath27 the reference state typically results from a hypothetical system in which the specific interactions between the amino acids are absent @xcite . the second term involving @xmath20 and @xmath26 can be ignored , as it is a constant . in practice , @xmath23 is estimated from the database of known protein structures , while @xmath28 typically results from calculations or simulations . for example , @xmath23 could be the conditional probability of finding the @xmath29 atoms of a valine and a serine at a given distance @xmath17 from each other , giving rise to the free energy difference @xmath30 . the total free energy difference of a protein , @xmath31 , is then claimed to be the sum of all the pairwise free energies:@xmath32 where the sum runs over all amino acid pairs @xmath33 ( with @xmath34 ) and @xmath35 is their corresponding distance . it should be noted that in many studies @xmath25 does not depend on the amino acid sequence @xcite . intuitively , it is clear that a low free energy difference indicates that the set of distances in a structure is more likely in proteins than in the reference state . however , the physical meaning of these pmfs have been widely disputed since their introduction @xcite . indeed , why is it at all necessary to subtract a reference state energy ? what is the optimal reference state ? can pmfs be generalized and justified beyond pairwise distances , and if so , how ? before we discuss and clarify these issues , we discuss two qualitative justifications that were previously reported in the literature : the first based on a physical analogy , and the second using a statistical argument . the first , qualitative justification of pmfs is due to sippl , and based on an analogy with the statistical physics of liquids @xcite . for liquids @xcite , the potential of mean force is related to the _ pair correlation function _ @xmath36 , which is given by : @xmath37 where @xmath23 and @xmath28 are the respective probabilities of finding two particles at a distance @xmath17 from each other in the liquid and in the reference state__. _ _ for liquids , the reference state is clearly defined ; it corresponds to the ideal gas , consisting of non - interacting particles . the two - particle potential of mean force @xmath38 is related to @xmath36 by : @xmath39 according to the _ reversible work theorem _ , the two - particle potential of mean force @xmath38 is the reversible work required to bring two particles in the liquid from infinite separation to a distance @xmath17 from each other @xcite . sippl justified the use of pmfs a few years after he introduced them for use in protein structure prediction @xcite by appealing to the analogy with the reversible work theorem for liquids @xcite . for liquids , @xmath36 can be experimentally measured using small angle x - ray scattering ; for proteins , @xmath23 is obtained from the set of known protein structures , as explained in the previous section . the analogy described above might provide some physical insight , but , as ben - naim writes in a seminal publication @xcite : `` the quantities , referred to as ` statistical potentials , ' ` structure based potentials , ' or ` pair potentials of mean force ' , as derived from the protein data bank , are neither ` potentials ' nor ` potentials of mean force , ' in the ordinary sense as used in the literature on liquids and solutions . '' another issue is that the analogy does not specify a suitable reference state for proteins . this is also reflected in the literature on statistical potentials ; the construction of a suitable reference state continues to be an active research topic @xcite . in the next section , we discuss a second , more recent justification that is based on probabilistic reasoning . baker and co - workers _ _ @xcite justified pmfs from a bayesian point of view and used these insights in the construction of the coarse grained rosetta energy function ; samudrala and moult used similar reasoning for the rapdf potential @xcite . according to bayesian probability calculus , the conditional probability @xmath40 of a structure @xmath1 , given the amino acid sequence @xmath41 , can be written as : @xmath42 @xmath43 is proportional to the product of the likelihood @xmath44 times the prior @xmath45 . by assuming that the likelihood can be approximated as a product of pairwise probabilities , and applying bayes theorem , the likelihood can be written as : @xmath46 where the product runs over all amino acid pairs @xmath33 ( with @xmath34 ) , and @xmath35 is the distance between amino acids @xmath47 and @xmath48 . obviously , the negative of the logarithm of expression ( [ eq_rosettapairs ] ) has the same functional form as the classic pairwise distance pmfs , with the denominator playing the role of the reference state in eq . [ eq : classic ] . the merit of this explanation is the qualitative demonstration that the functional form of a pmf can be obtained from probabilistic reasoning . although this view is insightful it rightfully drew the attention to the application of bayesian methods to protein structure prediction there is a more quantitative explanation , which does not rely on the incorrect assumption of pairwise decomposability @xcite , and leads to a different , _ quantitative _ conclusion regarding the nature of the reference state . this explanation is given in the next section . expressions that resemble pmfs naturally result from the application of probability theory to solve a fundamental problem that arises in protein structure prediction : how to improve an imperfect probability distribution @xmath0 over a first variable @xmath1 using a probability distribution @xmath2 over a second variable @xmath3 ( see fig . [ fig : full ] , fig . [ fig : simple ] and materials and methods ) . we assume that @xmath3 is a deterministic function of @xmath1 ; we write @xmath4 when necessary . in that case , @xmath1 and @xmath3 are called _ fine _ and _ coarse grained variables _ , respectively . when @xmath3 is a function of @xmath1 , the probability distribution @xmath0 automatically implies a probability distribution @xmath49 . this distribution has some unusual properties : @xmath50 ; and if @xmath51 , it follows that @xmath52 . typically , @xmath1 represents _ local _ features of protein structure ( such as backbone dihedral angles ) , while @xmath3 represents _ nonlocal _ features ( such as hydrogen bonding , compactness or pairwise distances ) . however , the same reasoning also applies to other cases ; for example , @xmath2 could represent information coming from experimental data , and @xmath0 could be embodied in an empirical force field as used in molecular mechanics @xcite ( see fig . [ fig : full ] ) . typically , the distribution @xmath0 in itself is not sufficient for protein structure prediction : it does not consider important nonlocal features such as hydrogen bonding , compactness or favorable amino acid interactions . as a result , @xmath0 is incorrect with respect to @xmath3 , and needs to be supplemented with a probability distribution @xmath2 that provides additional information . by construction , @xmath2 is assumed to be correct ( or at least useful ) . the above situation arises naturally in protein structure prediction . for example , @xmath2 could be a probability distribution over the radius of gyration , hydrogen bond geometry or the set of pairwise distances , and @xmath0 could be a fragment library @xcite or a probabilistic model of local structure @xcite . in fig.[fig : simple ] , we used the example of a distribution over the radius of gyration for @xmath2 and a fragment library for @xmath0 . obviously , sampling from a fragment library and retaining structures with the desired nonlocal structure ( radius of gyration , hydrogen bonding , etc . ) is in principle possible , but in practice extremely inefficient . how can @xmath0 be combined with @xmath2 in a meaningful way ? as mentioned previously , simply multiplying the two distributions resulting in @xmath53 does not lead to the desired result as the two variables are obviously not independent . the correct solution follows from simple statistical considerations ( see materials and methods ) , and is given by the following expression : @xmath54 we use the notation @xmath55 , as this distribution implies the desired distribution @xmath2 for @xmath4 . the distribution @xmath56 in the denominator is the probability distribution that is implied by @xmath0 over the coarse grained variable @xmath3 . conceptually , dividing by @xmath56 takes care of the signal in @xmath0 with respect to the coarse grained variable @xmath3 . the ratio in this expression corresponds to the probabilistic formulation of a pmf , and @xmath56 corresponds to the reference state ( see materials and methods ) . in practice , @xmath0 is typically not evaluated directly , but brought in through conformational monte carlo sampling ( see materials and methods ) ; often sampling is based on a fragment library @xcite , although other methods are possible , including sampling from a probabilistic model @xcite or a suitable energy function @xcite . the ratio @xmath57 , which corresponds to the probabilistic formulation of a pmf , also naturally arises in the markov chain monte carlo ( mcmc ) procedure ( see materials and methods ) . an important insight is that , in this case , the conformational sampling method uniquely defines the reference state . thus , in the case of a fragment library , the reference distribution @xmath56 is the probability distribution over @xmath3 that is obtained by sampling conformations solely using the fragment library . as the method we have introduced here invariably relies on the ratio of two probability distributions one regarding protein structure and the other regarding a well - defined reference state we refer to it as the _ reference ratio method_. in the next section , we show that the standard pairwise distance pmfs can be seen as an approximation of the reference ratio method . in this section , we apply the reference ratio method to the standard , pairwise distance case . in the classic pmf approach , one considers the vector of pairwise distances @xmath13 between the amino acids . in this case , it is usually assumed that we can write@xmath58 where the product runs over all amino acid pairs @xmath33 ( with @xmath34 ) , and @xmath35 is their matching distance . clearly , the assumption that the joint probability can be written as a product of pairwise probabilities is not justified @xcite , but in practice this assumption often provides useful results @xcite . in order to obtain protein - like conformations , @xmath59 needs to be combined with an appropriate probability distribution @xmath60 that addresses the local features of the polypeptide chain . applying eq . [ eq : ratio ] to this case results in the following expression:@xmath61 where the denominator @xmath62 is the probability distribution over the pairwise distances as induced by the distribution @xmath60 . the ratio in this expression corresponds to the probabilistic expression of a pmf . the reference state is thus determined by @xmath60 : it reflects the probability of generating a set of pairwise distances using local structure information alone . obviously , as @xmath60 is conditional upon the amino acid sequence @xmath41 , the reference state becomes sequence dependent as well . we again emphasize that the assumption of pairwise decomposability in eq . [ eq : pairwise ] is incorrect @xcite . therefore , the application of the reference ratio method results in a useful approximation , at best . as a result , the optimal definition of the reference state also needs to compensate for the errors implied by the invalid assumption . as is it well established that distance dependent pmfs perform well with a suitable definition of the reference state @xcite , and the incorrect pairwise decomposability assumption impairs a rigorous statistical analysis , we do not discuss this type of pmfs further . indeed , for pairwise distance pmfs , the main challenge lies in developing better probabilistic models of sets of pairwise distances @xcite . the pairwise distance pmfs currently used in protein structure prediction are thus not statistically rigorous , because they do not make use of a proper joint probability distribution over the pairwise distances , which are strongly intercorrelated due to the connectivity of molecules . a rigorous application of the reference ratio method would require the construction of a proper joint probability distribution over pairwise distances . this is certainly possible in principle , but currently , as far as we know , a challenging open problem and beyond the scope of this article . however , we have clarified that the idea of using a reference state is correct and valid , and that this state has a very precise definition . therefore , in the next two sections , we show instead how statistically valid quantities , similar to pmfs , can be obtained for very different coarse grained variables . as a first application of the reference ratio method , we consider the task of sampling protein conformations with a given probability distribution @xmath7 for the radius of gyration @xmath8 . for @xmath7 , we chose a gaussian distribution with mean @xmath63 and standard deviation @xmath64 . this choice is completely arbitrary ; it simply serves to illustrate that the reference ratio method allows imposing an exact probability distribution over a certain feature of interest . applying eq . [ eq : ratio ] results in:@xmath65 for @xmath60 , we used torusdbn a graphical model that allows sampling of plausible backbone angles @xcite and sampled conditional on the amino acid sequence @xmath41 of ubiquitin ( see materials and methods ) . @xmath66 is the probability distribution of the radius of gyration for structures sampled solely from torusdbn , which was determined using generalized multihistogram mcmc sampling ( see materials and methods ) . in fig . [ fig : rg_plot ] , we contrast sampling from eq . [ eq : rg ] with sampling from @xmath67 . in the latter case , the reference state is not properly taken into account , which results in a significant shift towards higher radii of gyration . in contrast , the distribution of @xmath8 for the correct distribution @xmath55 , given by eq . [ eq : rg ] , is indistinguishable from the target distribution . this qualitative result is confirmed by the kullback - leibler divergence @xcite a natural distance measure for probability distributions expressed in bits between the target distribution and the resulting marginal distributions of @xmath8 . adding @xmath68 to the denominator diminishes the distance from 0.08 to 0.001 bits . for this particular pmf , the effect of using the correct reference state is significant , but relatively modest ; in the next section , we discuss an application where its effect is much more pronounced . here , we demonstrate that pmfs can be optimized iteratively , which is particularly useful if the reference probability distribution @xmath69 is difficult to estimate . we illustrate the method with a target distribution that models the hydrogen bonding network using a multinomial distribution . we describe the hydrogen bonding network ( @xmath70 ) with eight integers ( for details , see materials and methods ) . three integers @xmath71 represent the number of residues that do not partake in hydrogen bonds in @xmath72-helices , @xmath73-sheets and coils , respectively . the five remaining integers @xmath74 represent the number of hydrogen bonds within @xmath72-helices , within @xmath73-strands , within coils , between @xmath72-helices and coils , and between @xmath73-strands and coils , respectively . as target distribution @xmath75 over these eight integers , we chose a multinomial distribution whose parameters were derived from the native structure of protein g ( see materials and methods ) . @xmath75 provides information , regarding protein g , on the number of hydrogen bonds and the secondary structure elements involved , but does not specify _ where _ the hydrogen bonds or secondary elements occur . as in the previous section , we use torusdbn as the sampling distribution @xmath60 ; we sample backbone angles conditional on the amino acid sequence @xmath41 of protein g. native secondary structure information was _ not _ used in sampling from torusdbn . the reference distribution @xmath76 , due to torusdbn , is very difficult to estimate correctly for several reasons : its shape is unknown and presumably complex ; its dimensionality is high ; and the data is very sparse with respect to @xmath73-sheet content . therefore , @xmath76 can only be approximated , which results in a suboptimal pmf . a key insight is that one can apply the method iteratively until a satisfactory pmf is obtained ( see fig . [ fig : full ] , dashed line ) . in each iteration , the ( complex ) reference distribution is approximated using a simple probability distribution ; we illustrate the method by using a multinomial distribution , whose parameters are estimated by maximum likelihood estimation in each iteration , using the conformations generated in the previous iteration . in the first iteration , we simply set the reference distribution equal to the uniform distribution . formally , the procedure works as follows . in iteration @xmath77 , the distribution @xmath78 is improved using the samples generated in iteration @xmath47 : @xmath79 where @xmath80 is the reference distribution estimated from the samples generated in the @xmath47-th iteration , @xmath81 stems from torusdbn , and @xmath82 is the uniform distribution . after each iteration , the set of samples is enriched in hydrogen bonds , and the reference distribution @xmath83 can be progressively estimated more precisely . note that in the first iteration , we simply use the product of the target and the sampling distribution ; no reference state is involved . [ fig : hbond_counts ] shows the evolution of the fractions versus the iteration number for the eight hydrogen bond categories ; the structures with minimum energy for all six iterations are shown in fig . [ fig : hbond_structures ] . in the first iteration , the structure with minimum energy ( highest probability ) consists of a single @xmath72-helix ; @xmath73-sheets are entirely absent ( see fig . [ fig : hbond_structures ] , structure 1 ) . already in the second iteration , @xmath73-strands start to pair , and in the third and higher iterations complete sheets are readily formed . the iterative optimization of the pmf quickly leads to a dramatic enrichment in @xmath73-sheet structures , as desired , and the fractions of the eight categories become very close to the native values ( fig.[fig : hbond_counts ] ) . the strengths and weaknesses of pmfs can be rigorously explained based on simple probabilistic considerations , which leads to some surprising new insights of direct practical relevance . first , we have made clear that pmfs naturally arise when two probability distributions need to be combined in a meaningful way . one of these distributions typically addresses local structure , and its contribution often arises from conformational sampling . each conformational sampling method thus requires its own reference state and corresponding reference distribution ; this is likely the main reason behind the large number of different reference states reported in the literature @xcite . if the sampling method is conditional upon the amino acid sequence , the reference state necessarily also depends on the amino acid sequence . second , conventional applications of pairwise distance pmfs usually lack two necessary features to make them fully rigorous : the use of a proper probability distribution over pairwise distances in proteins for @xmath84 , and the recognition that the reference state is rigorously defined by the conformational sampling scheme used , that is , @xmath85 . usually , the reference state is derived from external physical considerations @xcite . third , pmfs are not tied to pairwise distances , but generalize to any coarse grained variable . attempts to develop similar quantities that , for example , consider solvent exposure @xcite , relative side chain orientations @xcite , backbone dihedral angles @xcite or hydrogen bonds @xcite are thus , in principle , entirely justified . hence , our probabilistic interpretation opens up a wide range of possibilities for advanced , well - justified energy functions based on sound probabilistic reasoning ; the main challenge is to develop proper probabilistic models of the features of interest and the estimation of their parameters @xcite . strikingly , the example applications involving radius of gyration and hydrogen bonding that we presented in this article _ are _ statistically valid and rigorous , in contrast to the traditional pairwise distance pmfs . finally , our results reveal a straightforward way to optimize pmfs . often , it is difficult to estimate the probability distribution that describes the reference state . in that case , one can start with an approximate pmf , and apply the method iteratively . in each iteration , a new reference state is estimated , with a matching probability distribution . in that way , one iteratively attempts to sculpt an energy funnel @xcite . we illustrated this approach with a probabilistic model of the hydrogen bond network . although iterative application of the inverse boltzmann formula has been described before @xcite , its theoretical justification , optimal definition of the reference state and scope remained unclear . as the traditional pairwise distance pmfs used in protein structure prediction arise from the imperfect application of a statistically valid and rigorous procedure with a much wider scope , we consider it highly desirable that the name `` potential of mean force '' should be reserved for true , physically valid quantities @xcite . because the statistical quantities we discussed invariably rely on the use of a ratio of two probability distributions , one concerning protein structure and the other concerning the ( now well defined ) reference state , we suggest the name `` reference ratio distribution '' deriving from the application of the `` reference ratio method '' . pairwise distance pmfs , as used in protein structure prediction , are not physically justified potentials of mean force or free energies @xcite and the reference state does not depend on external physical considerations ; the same is of course true for our generalization . however , these pmfs are approximations of statistically valid and rigorous quantities , and these quantities can be generalized beyond pairwise distances to other aspects of protein structure . the fact that these quantities are not potentials of mean force or free energies is of no consequence for their statistical rigor or practical importance both of which are considerable . our results thus vindicate , formalize and generalize sippl s original and seminal idea @xcite . after about twenty years of controversy , pmfs or rather the statistical quantities that we have introduced in this article are ready for new challenges . we consider a joint probability distribution @xmath86 and a probability distribution @xmath2 over two variables of interest , @xmath1 and @xmath3 , where @xmath3 is a deterministic function of @xmath1 ; we write @xmath4 when relevant . note that because @xmath3 is a function of @xmath1 , it follows that @xmath87 ; and if @xmath51 , then @xmath52 . we assume that @xmath2 is a meaningful and informative distribution for @xmath3 . next , we note that @xmath86 implies a matching marginal probability distribution @xmath56 ( where the subscript @xmath13 refers to the fact that @xmath56 corresponds to the reference state , as we will show below):@xmath88 we consider the case where @xmath56 differs substantially from @xmath2 ; hence , @xmath56 can be considered as incorrect . on the other hand , we also assume that the conditional distribution @xmath89 is indeed meaningful and informative ( see next section ) . this distribution is given by : @xmath90 where @xmath91 is the delta function . the question is now how to combine the two distributions @xmath2 and @xmath0 each of which provide useful information on @xmath1 and @xmath3 in a meaningful way . before we provide the solution , we illustrate how this problem naturally arises in protein structure prediction . in protein structure prediction , @xmath86 is often embodied in a fragment library ; in that case , @xmath1 is a set of atomic coordinates obtained from assembling a set of polypeptide fragments . of course , @xmath86 could also arise from a probabilistic model , a pool of known protein structures , or any other conformational sampling method . the variable @xmath3 could , for example , be the radius of gyration , the hydrogen bond network or the set of pairwise distances . if @xmath3 is a deterministic function of @xmath1 , the two variables are called _ coarse grained _ and _ fine grained _ variables , respectively . for example , sampling a set of dihedral angles for the protein backbone uniquely defines the hydrogen bond geometry between any of the backbone atoms . above , we assumed that @xmath89 is a meaningful distribution . this is often a reasonable assumption ; fragment libraries , for example , originate from real protein structures , and conditioning on protein - like compactness or hydrogen bonding will thus result in a meaningful distribution . of course , sampling solely from @xmath86 is not an efficient strategy to obtain hydrogen bonded or compact conformations , as they will be exceedingly rare . we now provide the solution of the problem outlined in the previous section , and discuss its relevance to the construction of pmfs . a first step on the way to the solution is to note that the product rule of probability theory allows us to write : @xmath92 as only @xmath2 is given , we need to make a reasonable choice for @xmath93 . we assume , as discussed before , that @xmath89 is a meaningful choice , which leads to:@xmath94 in the next step , we apply the product formula of probability theory to the second factor @xmath89 , and obtain:@xmath95 the distribution @xmath96 has the correct marginal distribution @xmath2 . in the next two sections , we discuss how this straightforward result can be used to great advantage for understanding and generalizing pmfs . first , we show that the joint distribution specified by eq . [ eq : bh ] can be reduced to a surprisingly simple functional form . second , we discuss how this result can be used in mcmc sampling . in both cases , expressions that correspond to a pmf arise naturally . using the product rule of probability theory , eq . [ eq : bh ] can be written as:@xmath97 because the coarse grained variable @xmath3 is a deterministic function of the fine grained variable @xmath1 , @xmath98 is the delta function : @xmath99 finally , we integrate out the , now redundant , coarse grained variable @xmath3 from the expression:@xmath100 and obtain our central result ( eq . [ eq : ratio ] ) . sampling from @xmath55 will result in the desired marginal probability distribution @xmath2 . the influence of the fine grained distribution @xmath86 is apparent in the fact that @xmath93 is equal to @xmath89 . the ratio in this expression corresponds to the usual probabilistic formulation of a pmf ; the distribution @xmath56 corresponds to the reference state . in the next section , we show that pmfs also naturally arise when @xmath2 and @xmath86 are used together in metropolis - hastings sampling . here , we show that metropolis - hastings sampling from the distribution specified by eq . [ eq : bh ] , using @xmath86 as a proposal distribution , naturally results in expressions that are equivalent to pmfs . the derivation is also valid if the proposal distribution depends on the previous state , provided @xmath86 satisfies the detailed balance condition . according to the standard metropolis - hastings method @xcite , one can sample from a probability distribution @xmath101 by generating a markov chain where each state @xmath102 depends only on the previous state @xmath103 . the new state @xmath102 is generated using a proposal distribution @xmath104 , which includes @xmath105 as a special case . according to the metropolis - hastings method , the proposal @xmath102 is accepted with a probability @xmath72:@xmath106 @xmath107 where @xmath108 is the starting state , and @xmath109 is the next proposed state . we assume that the proposal distribution @xmath110 satisfies the detailed balance condition:@xmath111 as a result , we can always write eq . [ eq : methas ] as:@xmath112 the metropolis - hastings expression ( eq . [ eq : methas ] ) , applied to the distribution specified by eq . [ eq : bh ] and using @xmath113 or @xmath114 as the proposal distribution , results in : @xmath115 which reduces to:@xmath116 hence , we see that the metropolis - hastings method requires the evaluation of ratios of the form @xmath57 when @xmath113 or @xmath114 is used as the proposal distribution ; these ratios correspond to the usual probabilistic formulation of a pmf . finally , when @xmath3 is a deterministic function of @xmath1 , the proposal distribution reduces to @xmath117 or @xmath118 , and eq . [ eq : mhratio ] becomes:@xmath119 conformational sampling from a suitable @xmath120 was done using torusdbn @xcite as implemented in phaistos @xcite ; backbone angles ( @xmath121 and @xmath122 ) were sampled conditional on the amino acid sequence . we used standard fixed bond lengths and bond angles in constructing the backbone coordinates from the angles , and represented all side chains ( except glycine and alanine ) with one dummy atom with a fixed position @xcite . for the radius of gyration application , we first determined @xmath66 using the multi - canonical mcmc method to find the sampling weights @xmath123 that yield a flat histogram @xcite . sampling from the resulting joint distribution ( eq . [ eq : rg ] ) was done using the same method . in both cases , we used 50 million iterations ; the @xmath8 bin size was 0.08 @xmath124 . sampling from torusdbn was done conditional on the amino acid sequence @xmath41 of ubiquitin ( 76 residues , pdb code 1ubq ) . for the hydrogen bond application , sampling from the pmfs was done in the @xmath125-ensemble @xcite , using the metropolis - hastings algorithm and the generalized multihistogram method for updating the weights @xcite . in each iteration @xmath47 , 50,000 samples ( out of 50 million metropolis - hastings steps ) were generated , and the parameters of the multinomial distribution @xmath126 were subsequently obtained using maximum likelihood estimation . hydrogen bonds were defined as follows : the @xmath127 distance is below 3.5 @xmath124 , and the angles formed by @xmath128 and @xmath129 are both greater than 100@xmath130 . each carbonyl group was assumed to be involved in at most one hydrogen bond ; in case of multiple hydrogen bond partners , the one with the lowest @xmath131 distance was selected . each residue was assigned to one of the eight possible hydrogen bond categories @xmath132 based on the presence of hydrogen bonding at its carbonyl group and the secondary structure assignments ( for both bond partners ) by torusdbn . the target distribution the multinomial distribution @xmath75 used in eq . [ eq : hbond_iterative ] was obtained by maximum likelihood estimation using the number of hydrogen bonds , for all eight categories , in the native structure of protein g ( 56 residues , pdb code 2gb1 ) . sampling from torusdbn was done conditional on the amino acid sequence of protein g ; native secondary structure information was _ not _ used . t.h . , m.b . and m.p . are joint first authors . t.h . and j.f.b . developed the theory . m.b . and m.p . performed the simulations . the remaining authors contributed new tools . t.h . wrote the paper . we acknowledge funding by the _ danish program commission on nanoscience , biotechnology and it _ ( nabiit , project : `` simulating proteins on a millisecond time - scale '' , 2106 - 06 - 0009 ) , the _ danish research council for technology and production sciences _ ( ftp , project : `` protein structure ensembles from mathematical models '' , 274 - 09 - 0184 ) and the _ danish council for independent research _ ( fnu , project : `` a bayesian approach to protein structure determination '' , 272 - 08 - 0315 ) . 10 [ 1]`#1 ` urlstyle [ 1]doi:#1 [ 1 ] [ 2 ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ key : # 1 + annotation : # 2 _ _ _ _ _ _ _ _ _ _ _ _ _ _ anfinsen cb ( 1973 ) principles that govern the folding of protein chains . science 181 : 223230 . moult j ( 1997 ) comparison of database potentials and molecular mechanics force fields . curr opin struct biol 7 : 194199 . shen my , sali a ( 2006 ) statistical potential for assessment and prediction of protein structures . protein sci 15 : 25072524 . tanaka s , scheraga ha ( 1976 ) medium- and long - range interaction parameters between amino acids for predicting three - dimensional structures of proteins . macromolecules 9 : 945950 . miyazawa s , jernigan r ( 1985 ) estimation of effective interresidue contact energies from protein crystal structures : quasi - chemical approximation . macromolecules 18 : 534552 . miyazawa s , jernigan r ( 1999 ) an empirical energy potential with a reference state for protein fold and sequence recognition . proteins 36 : 357369 . sippl mj ( 1990 ) calculation of conformational ensembles from potentials of mean force . an approach to the knowledge - based prediction of local structures in globular proteins . j mol biol 213 : 859883 . chandler d ( 1987 ) introduction to modern statistical mechanics . oxford university press , usa . mcquarrie d ( 2000 ) statistical mechanics . university science books , usa . finkelstein a , badretdinov a , gutin a ( 1995 ) why do protein architectures have boltzmann - like statistics ? proteins struct func gen 23 : 142150 . rooman m , wodak s ( 1995 ) are database - derived potentials valid for scoring both forward and inverted protein folding ? protein eng 8 : 849 - 858 . thomas pd , dill ka ( 1996 ) statistical potentials extracted from protein structures : how accurate are they ? j mol biol 257 : 457469 . ben - naim a ( 1997 ) statistical potentials extracted from protein structures : are these meaningful potentials ? j chem phys 107 : 3698 - 3706 . koppensteiner wa , sippl mj ( 1998 ) knowledge - based potentials back to the roots . biochemistry mosc 63 : 247252 . shortle d ( 2003 ) propensities , probabilities , and the boltzmann hypothesis . protein sci 12 : 12981302 . kirtay c , mitchell j , lumley j ( 2005 ) knowledge based potentials : the reverse boltzmann methodology , virtual screening and molecular weight dependence . qsar & combinatorial sci 24 : 527536 . muegge i ( 2006 ) pmf scoring revisited . j med chem 49 : 58955902 . simons kt , kooperberg c , huang e , baker d ( 1997 ) assembly of protein tertiary structures from fragments with similar local sequences using simulated annealing and bayesian scoring functions . j mol biol 268 : 209225 . colubri a , jha a , shen m , sali a , berry r , et al . ( 2006 ) minimalist representations and the importance of nearest neighbor effects in protein folding simulations . j mol biol 363 : 835857 . sippl mj ( 1993 ) recognition of errors in three - dimensional structures of proteins . proteins 17 : 355362 . eramian d , shen m , devos d , melo f , sali a , et al . ( 2006 ) a composite score for predicting errors in protein structure models . protein sci 15 : 16531666 . rykunov d , fiser a ( 2010 ) new statistical potential for quality assessment of protein models and a survey of energy functions . bmc bioinformatics 11 : 128 . jones dt , taylor wr , thornton jm ( 1992 ) a new approach to protein fold recognition . nature 358 : 8689 . mjek p , elber r ( 2009 ) a coarse - grained potential for fold recognition and molecular dynamics simulations of proteins . proteins 76 : 822836 . gohlke h , hendlich m , klebe g ( 2000 ) knowledge - based scoring function to predict protein - ligand interactions1 . j mol biol 295 : 337356 . gilis d , rooman m ( 1997 ) predicting protein stability changes upon mutation using database - derived potentials : solvent accessibility determines the importance of local versus non - local interactions along the sequence1 . j mol biol 272 : 276290 . gilis d , rooman m ( 2000 ) popmusic , an algorithm for predicting protein mutant stability changes . application to prion proteins . protein eng 13 : 849856 . su y , zhou a , xia x , li w , sun z ( 2009 ) quantitative prediction of protein - protein binding affinity with a potential of mean force considering volume correction . protein sci 18 : 25502558 . chandler d ( 2005 ) interfaces and the driving force of hydrophobic assembly . nature 437 : 640647 . bowman gr , pande vs ( 2009 ) simulated tempering yields insight into the low - resolution rosetta scoring functions . proteins 74 : 777788 . shmygelska a , levitt m ( 2009 ) generalized ensemble methods for de novo structure prediction . proc natl acad sci u s a 106 : 14151420 . bryngelson j , wolynes p ( 1987 ) spin glasses and the statistical mechanics of protein folding . proc natl acad sci u s a 84 : 75247528 . leopold p , montal m , onuchic j ( 1992 ) protein folding funnels : a kinetic approach to the sequence - structure relationship . proc natl acad sci u s a 89 : 87218725 . dill k , chan h ( 1997 ) from levinthal to pathways to funnels . nat struct biol 4 : 1019 . reith d , ptz m , mller - plathe f ( 2003 ) deriving effective mesoscale potentials from atomistic simulations . j comput chem 24 : 16241636 . fain b , levitt m ( 2003 ) funnel sculpting for in silico assembly of secondary structure elements of proteins . proc natl acad sci u s a 100 : 1070010705 . sippl mj , ortner m , jaritz m , lackner p , flockner h ( 1996 ) helmholtz free energies of atom pair interactions in proteins . fold des 1 : 28998 . zhang c , liu s , zhou h , zhou y ( 2004 ) an accurate , residue - level , pair potential of mean force for folding and binding based on the distance - scaled , ideal - gas reference state . protein sci 13 : 400411 . cheng j , pei j , lai l ( 2007 ) a free - rotating and self - avoiding chain model for deriving statistical potentials based on protein structures . biophys j 92 : 38683877 . rykunov d , fiser a ( 2007 ) effects of amino acid composition , finite size of proteins , and sparse statistics on distance - dependent statistical pair potentials . proteins 67 : 559568 . bernard b , samudrala r ( 2008 ) a generalized knowledge - based discriminatory function for biomolecular interactions . proteins 76 : 115128 . samudrala r , moult j ( 1998 ) an all - atom distance - dependent conditional probability discriminatory function for protein structure prediction . j mol biol 275 : 895916 . pearl j ( 1988 ) probabilistic reasoning in intelligent systems , morgan kaufmann , san francisco , usa , chapter 3 . pp . 108115 . lazaridis t , karplus m ( 2000 ) effective energy functions for protein structure prediction . curr opin struct biol 10 : 139145 . boomsma w , mardia kv , taylor cc , ferkinghoff - borg j , krogh a , et al . ( 2008 ) a generative , probabilistic model of local protein structure . proc natl acad sci u s a 105 : 89328937 . sippl m , hendlich m , lackner p ( 1992 ) assembly of polypeptide and protein backbone conformations from low energy ensembles of short fragments : development of strategies and construction of models for myoglobin , lysozyme , and thymosin @xmath734 . protein sci 1 : 625640 . hamelryck t , kent j , krogh a ( 2006 ) sampling realistic protein conformations using local structural bias . plos comput biol 2 : e131 . zhao f , peng j , debartolo j , freed k , sosnick t , et al . ( 2010 ) a probabilistic and continuous model of protein conformational space for template - free modeling . j comput biol 17 : 783 - 798 . hamelryck t ( 2009 ) probabilistic models and machine learning in structural bioinformatics . stat methods med res 18 : 505526 . kullback s , leibler r ( 1951 ) on information and sufficiency . annals math stat 22 : 7986 . zhou h , zhou y ( 2002 ) distance - scaled , finite ideal - gas reference state improves structure - derived potentials of mean force for structure selection and stability prediction . protein sci 11 : 27142726 . bowie j , luthy r , eisenberg d ( 1991 ) a method to identify protein sequences that fold into a known three - dimensional structure . science 253 : 164164 . liithy r , bowie j , eisenberg d ( 1992 ) assessment of protein models with three - dimensional profiles . nature 356 : 8385 . buchete nv , straub je , thirumalai d ( 2004 ) development of novel statistical potentials for protein fold recognition . curr opin struct biol 14 : 225232 . rooman m , kocher j , wodak s ( 1991 ) prediction of protein backbone conformation based on seven structure assignments : influence of local interactions . j mol biol 221 : 961979 . kocher j , rooman m , wodak s ( 1994 ) factors influencing the ability of knowledge - based potentials to identify native sequence - structure matches . j mol biol 235 : 15981613 . simons kt , ruczinski i , kooperberg c , fox ba , bystroff c , et al . ( 1999 ) improved recognition of native - like protein structures using a combination of sequence - dependent and sequence - independent features of proteins . proteins 34 : 8295 . thomas p , dill k ( 1996 ) an iterative method for extracting energy - like quantities from protein structures . proc natl acad sci u s a 93 : 1162811633 . huang s , zou x ( 2006 ) an iterative knowledge - based scoring function to predict protein - ligand interactions : i. derivation of interaction potentials . j comp chem 27 : 18661875 . gilks w , richardson s , spiegelhalter d ( 1996 ) markov chain monte carlo in practice . chapman & hall / crc , usa . borg m , mardia k , boomsma w , frellsen j , harder t , et al . ( 2009 ) a probabilistic approach to protein structure prediction : phaistos in casp9 . in : gusnanto a , mardia k , fallaize c , editors , lasr2009 - statistical tools for challenges in bioinformatics . leeds university press , leeds , uk , pp . ferkinghoff - borg j ( 2002 ) optimized monte carlo analysis for generalized ensembles . eur phys j b 29 : 481 - 484 . hesselbo b , stinchcombe r ( 1995 ) monte carlo simulation and global optimization without parameters . phys rev lett 74 : 21512155 . delano wl ( 2002 ) the pymol molecular graphics system . palo alto , ca , usa : delano scientific . * illustration of the central idea presented in this article . * in this example , the goal is to sample conformations with a given distribution @xmath7 for the radius of gyration @xmath8 , and a plausible local structure . @xmath7 could , for example , be derived from known structures in the protein data bank ( pdb , left box ) . @xmath0 is a probability distribution over local structure @xmath1 , typically embodied in fragment library ( right box ) . in order to combine @xmath0 and @xmath7 in a meaningful way ( see text ) , the two distributions are multiplied and divided by @xmath12 ( formula at the bottom ) ; @xmath12 is the probability distribution over the radius of gyration for conformations sampled solely from the fragment library ( that is , @xmath0 ) . the probability distribution @xmath55 will generate conformations with plausible local structures ( due to @xmath0 ) , while their radii of gyration will be distributed according to @xmath7 , as desired . this simple idea lies at the theoretical heart of the pmf expressions used in protein structure prediction . ] * general statistical justification of pmfs . * the goal is to combine a distribution @xmath0 over a fine grained variable @xmath1 ( top right ) , with a probability distribution @xmath2 over a coarse grained variable @xmath4 ( top left ) . @xmath0 could be , for example , embodied in a fragment library ( @xmath133 ) , a probabilistic model of local structure ( @xmath134 ) or an energy function ( @xmath135 ) ; @xmath3 could be , for example , the radius of gyration , the hydrogen bond network , or the set of pairwise distances . @xmath2 usually reflects the distribution of @xmath3 in known protein structures ( pdb ) , but could also stem from experimental data ( @xmath136 ) . sampling from @xmath0 results in a distribution @xmath56 that differs from @xmath2 . multiplying @xmath2 and @xmath0 does not result in the desired distribution for @xmath3 either ( red box ) ; the correct result requires dividing out the signal with respect to @xmath3 due to @xmath0 ( green box ) . the _ reference _ distribution @xmath56 in the denominator corresponds to the contribution of the reference state in a pmf . if @xmath56 is only approximately known , the method can be applied iteratively ( dashed arrow ) . in that case , one attempts to iteratively sculpt an energy funnel . the procedure is statistically rigorous provided @xmath0 and @xmath2 are proper probability distributions ; this is usually not the case for conventional pairwise distance pmfs . ] * a pmf based on the radius of gyration . * the goal is to adapt a distribution @xmath60 which allows sampling of local structures such that a given target distribution @xmath7 is obtained . for @xmath41 , we used the amino acid sequence of ubiquitin . sampling from @xmath60 alone results in a distribution with an average @xmath8 of about 27 @xmath124 ( triangles ) . sampling using the correct expression ( open circles ) , given by eq . [ eq : rg ] , results in a distribution that coincides with the target distribution ( solid line ) . not taking the reference state into account results in a significant shift towards higher @xmath8 ( black circles ) . ] * iterative estimation of a pmf . * for each of the eight hydrogen bond categories ( see text ) , the black bar to the right denotes the fraction of occurrence @xmath137 in the native structure of protein g. the gray bars denote the fractions of the eight categories in samples from each iteration ; the first iteration is shown to the left in light gray . in the last iteration ( iteration 6 ; dark gray bars , right ) the values are very close to the native values for all eight categories . note that hydrogen bonds between @xmath73-strands are nearly absent in the first iteration ( category @xmath138 ) . ]	understanding protein structure is of crucial importance in science , medicine and biotechnology . for about two decades , knowledge based potentials based on pairwise distances so - called `` potentials of mean force '' ( pmfs ) have been center stage in the prediction and design of protein structure and the simulation of protein folding . however , the validity , scope and limitations of these potentials are still vigorously debated and disputed , and the optimal choice of the reference state a necessary component of these potentials is an unsolved problem . pmfs are loosely justified by analogy to the reversible work theorem in statistical physics , or by a statistical argument based on a likelihood function . both justifications are insightful but leave many questions unanswered . here , we show for the first time that pmfs can be seen as approximations to quantities that do have a rigorous probabilistic justification : they naturally arise when probability distributions over different features of proteins need to be combined . we call these quantities `` reference ratio distributions '' deriving from the application of the `` reference ratio method '' . this new view is not only of theoretical relevance , but leads to many insights that are of direct practical use : the reference state is uniquely defined and does not require external physical insights ; the approach can be generalized beyond pairwise distances to arbitrary features of protein structure ; and it becomes clear for which purposes the use of these quantities is justified . we illustrate these insights with two applications , involving the radius of gyration and hydrogen bonding . in the latter case , we also show how the reference ratio method can be iteratively applied to sculpt an energy funnel . our results considerably increase the understanding and scope of energy functions derived from known biomolecular structures .
since the discovery of ultra - compact dwarf galaxies ( ucds ; @xcite ; @xcite ) , it is still a complicated puzzle in extragalactic astronomy how such compact and luminous objects may have formed . they are brighter and larger than globular clusters ( gcs ) @xcite and much smaller than early - type dwarf galaxies ( des ) in both size and luminosity . a number of studies targeting various ucd samples in different galaxy clusters also revealed the diverse nature of ucds : fornax ucds are slightly redder on average than virgo ucds @xcite . on the other hand , it is still a matter of debate whether or not the ucds contain dark matter @xcite . this makes them very special objects to study in extragalactic astronomy , suggesting that the presence of dark matter or not can be directly related to whether ucds have a galactic origin or not . overall it has been already noted that virgo ucds contain fairly old ( age : @xmath0 8 - 10 gyr ) and metal poor ( @xmath10.5 dex ) stellar populations ( * ? ? ? * hereafter e07 ) . therefore , it is also proposed that they could be very luminous intra - cluster gcs @xcite . another popular formation scenario is the threshing of nucleated des @xcite . in this picture , ucds are the remnants of galaxies that have been significantly stripped in the cluster environment . numerical simulations @xcite have generally confirmed that the remnant nuclei resemble ucds in their structural parameters . stellar population studies of des provide evidence that the nuclei have intermediate ages and moderately metal - enriched stellar populations @xcite . in addition to this , since ucds show slightly super solar [ @xmath2/fe ] abundances , @xcite argued that the stellar population properties rather support the view that ucds are luminous globular clusters than being nuclei of des . in this letter , we present a stellar population analysis based on absorption - line strengths ( lick indices , @xcite ; @xcite ; @xcite ) of a fairly large sample of 34 nucleated des and 10 ucds in the virgo cluster . so far , studies comparing stellar population parameters derived from spectra used rather low numbers of objects . moreover , the extraction of nuclear spectra has been made without subtracting the underlying galactic light , which can still contribute significantly at the photometric center of the des . we therefore apply a simple method to subtract most of this light ( see section [ reduc ] ) , thus expecting that our measurements are representative for the stellar population properties of the nuclei themselves . finally , we present the distributions of the stellar population parameters of de nuclei and ucds with respect to local galaxy density and to their luminosity , and we try to constrain possible formation scenarios of virgo ucds . to values brighter than 21 mag , and the `` nucleus strength '' , i.e. the value of @xmath3 to less than -1 . note that the nucleus sample is not complete within this parameter region , i.e. not all virgo de nuclei with these parameters have been observed.,width=302 ] our de sample comprises 34 nucleated des in the virgo cluster ( virgo cluster catalog , vcc , * ? ? ? * ; * ? ? ? * ) , selected to have a relatively high `` nucleus strength '' ( details of the sample see , * ? ? ? * ) , which we define as the difference between the nucleus magnitude and the host galaxy effective surface brightness , @xmath4 , measured in sdss @xmath5 ( see below ) . thereby , @xmath6 is a measure for the brightness of a unit area of the galaxy , determining the `` contrast '' between galaxy and nucleus ( also see @xcite , their fig . 1 ) . we select nuclei with @xmath7 and @xmath8 mag ( see fig . [ his ] ) . our ucd sample selection ( see table 1 ) , is based on @xcite and @xcite ; our numbering follows the latter . three of the nine virgo ucds of jones et al . were not included in the lick index study of evstigneeva et al . ( vucd2 , 8 , and 9 ) , so they were selected by us as targets . three further ucds were selected , since they fell in the same field - of - view as de targets of our study . due to the multi - slit observations , they could be easily included . we also targeted a new virgo ucd candidate , which we now indeed confirm as virgo cluster member ; it is named vucd10 in table 1 . it was identified through a simple multiparameter selection procedure . from sdss dr5 pipeline photometry , we obtained @xmath9 magnitudes and colors for all nine virgo ucds in @xcite . when excluding vucd7 , which is clearly brighter than the others and appears to be an extended object in the sdss images ( also see @xcite ) , the _ r_-band magnitudes ( sdss `` modelmag '' values ) lie within 18.0 to 19.1 mag . their petrosian radii in @xmath5 , again excluding vucd7 , are below 2.2 arcsec . their _ u - r _ colors , when excluding the much redder vucd3 , cover the range 1.8 to 2.4 mag ( which includes vucd7 ) . their _ i - z _ colors , again excluding vucd3 , lie between 0.1 and 0.25 mag ( which again includes vucd7 ) . the right ascension and declination of all objects except vucd3 and vucd7 ranges from 187.5@xmath10 to 188.1@xmath10 and 11.9@xmath10 to 12.7@xmath10 , respectively . when querying the sdss database for all objects fulfilling the above criteria of magnitude , radius , color , and position , 20 objects were identified that the sdss classified as stars , among them vucd1 , 2 , and 5 . the same query , but for objects classified as galaxies , yielded only five objects : vucd4 , 6 , 8 , 9 , and the new vucd10 , which we therefore included in our target sample . with its radial velocity of 2425 km / s that we now measured from its spectrum , it is consistent with being a virgo cluster member : in velocity space , virgo member galaxies in the central cluster region reach velocities of 2600 km / s ( @xcite ) . we therefore consider vucd10 a new virgo cluster ucd , and include it in the following analysis . to obtain nucleus and ucd magnitudes , we use sdss dr5 @xmath5-band images to which we applied our own sky subtraction procedure ( see * ? ? ? we also correct them for galactic extinction , following @xcite , and use the flux calibration provided directly by the sdss . a virgo distance modulus of @xmath11 mag @xcite , corresponding to @xmath12 mpc and an sdss pixel scale of 32 pc ( 80 pc/ " ) , is adopted here and in the further analysis . nucleus magnitudes are derived as follows . a srsic fit to the radial profile of the galaxy is done , measured with elliptical annuli and using only the radial interval from 2 `` to 1/3 of the half - light radius , or extending it if the interval subtends less than 2.5 '' ( 200 pc ) . from this fit , a two - dimensional elliptical model image , taking into account the median sdss psf of 1.4 " fwhm , is created and subtracted from the original image , leaving only the nucleus in the center . the nucleus magnitude is then measured by circular aperture photometry with @xmath13 . we estimate the magnitude error to be 0.2 mag , which is consistent with the rms scatter of a comparison to the _ hubble space telescope ( hst ) _ acs virgo cluster survey @xcite for the 10 objects in common . magnitudes were converted into @xmath14 by adding the @xmath15 color value of the nucleated de s center ( @xmath16 ) , and were then compared to acs @xmath14 magnitudes , showing no systematic offset . ] the same aperture photometry is done to obtain ucd magnitudes . to be comparable to the nucleus measurements , this aperture is also applied to vucd7 , even though it is known to be extended , as mentioned previously . for the ucd magnitudes we estimate an error of 0.1 mag . the spectroscopic observations were carried out at eso very large telescope ( vlt ) with fors2 . the 1 " slit and v300 grism provide an instrumental resolution @xmath1711 fwhm . details of the observational layout and data reduction are described in ( * ? ? ? * hereafter p10 ) . our data was obtained in two observing runs , using the same instrumental setup . the first set of des ( 26 objects , see p10 ) was observed in the semester 2007a , and the second set ( 8 des ) in 2010a . in total our sample covers the full range of local projected galaxy density that is populated by nucleated des in virgo , from less than 10 to more than 100 galaxies per square degree . the ucds were observed in 2007a together with the first set of des , using the same instrumental configuration . we provide the basic properties such as position and radial velocity in table [ ucds ] and the detailed ucd sample selection criteria in section [ scucd ] . while the nucleus typically contributes the majority of light at the photometric center of the des , still a considerable amount of underlying galactic light of the host galaxy could alter the actual properties of the nuclei . therefore , we attempt to subtract the galactic light from the nuclear spectra . this is done by inward extrapolation of the exponentially - fitted light profile of the host galaxy along the slit , yielding the amount of galaxy light contained in the nucleus extraction aperture of 0``.75 ( central 3 pixels ) . the galaxy spectrum is extracted in a radial interval from 3 '' to 8 " , scaled accordingly , and subtracted from the centrally extracted spectrum . further details of this process are given in paudel et al . ( 2010b , submitted to mnras ) . while certainly not being perfect , we expect that this approach removes most of the galactic light contamination and thus provides spectra that are fairly well representative for the stellar population properties of de nuclei . llllllllll ucds & ra & dec & rv & age & [ z / h ] & [ @xmath2/fe ] & m@xmath18 & remark + name & ( h : m : s ) & ( @xmath19 : : ") & ( kms ) & gyr & dex & dex & mag & + vucd1 & 12:30:07.61 & 12:36:31.10 & 1227.8 @xmath20 1.7 & 11.9@xmath21 & -0.90 @xmath20 0.05 & 0.38 @xmath20 0.09 & 18.6 & e07 + vucd2 & 12:30:48.24 & 12:35:11.10 & 911 @xmath20 6 & 5.9@xmath22 & -0.77 @xmath20 0.09 & -0.14 @xmath20 0.08 & 18.6 & f09 + vucd3 & 12:30:57.40 & 12:25:44.8 & 710.6 @xmath20 3.5 & 15.0@xmath23 & 0.13 @xmath20 0.03 & 0.38 @xmath20 0.03 & 18.1 & e07 + vucd4 & 12:31:04.51 & 11:56:36.8 & 919.7 @xmath20 1.7 & 11.9@xmath24 & -1.24 @xmath20 0.05 & -0.00 @xmath20 0.12 & 18.6 & e07 + vucd5 & 12:31:11.90 & 12:41:01.20 & 1290 @xmath20 3 & 10.7@xmath25 & -0.40 @xmath20 0.03 & 0.31 @xmath20 0.02 & 18.5 & e07 + vucd6 & 12:31:28.41 & 12:25:03.30 & 2101 @xmath20 4 & 8.3@xmath26 & -1.03 @xmath20 0.12 & 0.41 @xmath20 0.09 & 18.7 & e07 + vucd7 & 12:31:52.93 & 12:15:59.50 & 985 @xmath20 5 & 10.7@xmath25 & -0.54 @xmath20 0.03 & 0.13 @xmath20 0.04 & 18.0 & e07 + vucd8 & 12:32:04.33 & 12:20:30.62 & 1531 @xmath20 19 & 11.6@xmath27 & -1.24 @xmath20 0.03 & -0.09 @xmath20 0.09 & 19.1 & f09 + vucd9 & 12:32:14.61 & 12:03:05.40 & 1325 @xmath20 5 & 11.6@xmath28 & -1.03 @xmath20 0.06 & 0.07 @xmath20 0.11 & 18.8 & f09 + vucd10 & 12:30:30.86 & 12:18:41.66 & 2425 @xmath20 103 & 9.8@xmath29 & -0.89 @xmath20 0.09 & 0.50 @xmath20 0.04 & 19.1 & new + /fe]-abundance versus local projected galaxy density are shown for de nuclei ( black ) and ucds ( green).,width=302 ] in order to derive stellar population parameters from the measured line strengths , we translate our lick index measurements into ssp - equivalent ages , metallicities , and @xmath2-element abundance ratios by comparing them to the model of @xcite by @xmath30minimization , following @xcite . we use nine well - measured indices ( h@xmath31 , h@xmath32 , fe4383 , h@xmath33 , fe5015 , mgb , fe5270 , fe5335 & fe5406 ) in case of the de nuclei . for the ucds we use only h@xmath33 , mgb , fe5270 and fe5335 , because we only find these four indices available in the literature for those three ucds that we did not target ourselves . moreover , due to the slit mask placement , our observations only cover the full range of indices ( i.e. , @xmath174000 to @xmath175600 ) for three of the observed ucds . for completeness , we also checked whether the use of different sets of indices produces any significant discrepancy in the estimated ssp parameters . we find that the estimated ssp parameters agree well within the errors , although the use of a larger set of indices naturally leads to smaller errors in the ssp parameters . therefore , we keep using all nine indices for the de nuclei in the further analysis . /fe]-abundance versus nucleus / ucd magnitude , for de nuclei ( black ) and ucds ( green).,width=302 ] the relation between the stellar population parameters and the local projected galaxy density is plotted in fig . the local projected density has been calculated from a circular projected area enclosing the 10th neighbor . it seems that there is a weak correlation between the local projected density and the ages of the nuclei . but , more prominent than the correlation , an age break is seen at a projected density @xmath3440 sq.deg.@xmath35 . we thus use this value to divide our de sample in two groups , located in cluster regions of lower / higher density . we find median ages of 2.6 gyr for the low density group and 11.6 gyr at high densities . almost all ucds lie in the high density region as defined above . their ages have a median of 11.7 gyr , strikingly consistent with the nuclei in the same density regime . as a statistical comparison , we performed a kolmogorov - smirov ( k - s ) test with the null hypothesis that there is no difference in the distribution of ssp parameters between the ucds and the de nuclei of the high density group . we find that there is a probability of 88% , 97% and 73% for having the same age , metallicity , and [ @xmath2/fe]-abundance distribution , respectively . the metallicities of de nuclei also show a difference between the low and high density regions : the median values differ by more than 0.5 dex , although the scatter is fairly large . the ucds also have a fairly large scatter in metallicity , ranging from + 0.1 to @xmath361.4 dex . however , it is remarkable that only two ucds ( vucd3 and 4 ) have a metallicity above @xmath360.5 dex . the [ @xmath2/fe]-abundances are more or less consistent with solar in case of the de nuclei , but also a slight increase with density can be recognized for the nuclei and ( more clearly ) for the ucds . although the scatter is fairly large , the spearman s rank correlation coefficients are 0.3 and 0.7 for de nuclei and ucds , with the probabilities for the null hypothesis that there is no correlation being 8.5% and 1.5% , respectively . in case of the ucds , 7 out 10 have super - solar values of [ @xmath2/fe ] . the relation between the stellar population parameters and the @xmath5-band magnitudes of ucds and nuclei is presented in fig . we do not see a relation of age and [ @xmath2/fe]-abundance with luminosity , both for nuclei and ucds . unlike that , the metallicity of both ucds and de nuclei tends to increase with increasing luminosity , following the well known metallicity - luminosity relation . however , bimodal peaks of the age and metallicity distributions of de nuclei can be seen as a striking feature in fig [ env ] ( right panel ) . this indicates the inhomogeneous nature of de nuclei , which can be divided into two different groups of stellar population characteristics : a young metal - rich group and an old metal - poor group . brighter nuclei predominantly have metal - rich stellar populations consistent with solar values . the metal - poor nuclei ( metallicity @xmath37 - 0.5 dex ) have an average of @xmath361.0 dex , and most of them are less luminous than @xmath38 mag ( @xmath39 mag ) . we have derived stellar population parameters of virgo cluster ucds and nuclei of early - type dwarf galaxies , through fitting measured lick absorption line strengths to models of @xcite . since we use simple stellar population models , it must be noted that our estimated parameters are interpreted as _ ssp - equivalent . _ indications that the ucds in the virgo cluster more likely have formed via de destruction in dense cluster regions than in the fornax cluster have already been presented in the literature @xcite . an explanation could be the higher mass of the virgo cluster , which provides a deeper potential well with stronger tidal forces than in the fornax cluster . virgo ucds were also found to be significantly different from gcs in colors and sizes , as they are bluer and larger than gcs @xcite . e07 showed that virgo ucds have structural properties similar to the fornax ucds . furthermore , @xcite showed with the _ hst _ that fornax ucds are more extended and have higher surface brightnesses than typical dwarf nuclei . that is not necessarily inconsistent with the stripping scenarios , because the stripping process can alter structural parameters of the embedded nuclei . therefore , a direct comparison of structural parameters may not be able to constrain the formation scenarios of ucds properly . in the light of our new age and metallicity estimates in the previous section , we want to address the possibility of formation of virgo ucds via tidal destruction of nucleated des . we observe that ucds exhibit significantly older and more metal deficient stellar populations as compared to de nuclei _ on average_. however , if we compare the ages of ucds and nuclei only in high density regions , where most of the ucds are located , we do not find any significant difference in their age and metallicity distribution . only two de nuclei ( vcc0965 and vcc1122 ) out of 13 from the high density group have ages less than 5 gyr and metallicities above -0.2 dex . note , however , that due to projection effects , a number of objects that are actually located in the outskirts of the cluster must fall into regions of high _ projected _ density , lying in front or behind the center along the line of sight . if we consider the low density group of our sample to be representative for nuclei of des outside of the ( threedimensional ) cluster center , then we actually _ expect _ to have a few objects with young ages and high metallicities in the high density group , consistent with what we observe . moreover , even without taking this effect into account , our statistical tests already show that the age distributions of ucds and de nuclei are similar when considering only the group of nuclei in the high density . this supports the idea of the threshing of des in high density environments to form the ucds . at low densities , a certain fraction of des probably still retained some central gas and transformed it into stars ( some des are still doing this , see @xcite ) . consequently , their nuclei today would indeed show a younger stellar population than a ucd , the latter being an `` old nucleus '' . for such des , not only the orbital average should be located in the outskirts , but they should also not have passed right through the center ( e.g. on an eccentric orbit ) , otherwise the gas would have been stripped by ram pressure and tidal forces . this would explain the fact that we do not see young ucds : a de destruction is only possible if the galaxy really goes through the center and experiences the strongest tidal forces . if we assume that the _ full _ destruction , i.e. the complete removal of the de s main body , takes significantly more orbital time than the gas stripping , then nucleated des with an orbit leading through the center first lost their gas , halting any star formation , and then became destroyed . therefore , we have no young ucds . in any case , young and bright de nuclei have either late or prolonged star formation activity . we assess this by modeling spectra of composite stellar populations through superposition of a young and an old component , using the model spectra of @xcite . we fix the old population at an age of 11 gyr and a metallicity of -1.2 dex , and add a component with solar metallicity and exponentially declining star formation rate , starting from an age of 11 gyr . we find that spectra like those of the young nuclei ( with an ssp - equivalent age of @xmath173 gyr ) can be achieved when the exponentially declining component formed stars until 2 gyr ago , at which point it is truncated . this means , if we assume our model to represent the observed young nuclei , their star formation activity would have stopped 9 gyr later than that of the old nuclei . on the one hand , the estimated ages and metallicities of ucds nicely agree with the ages and metallicities of nuclei of des from the dense cluster regions . on the other hand , most of the old nuclei are fainter than the ucds . however , the ucd discoveries might suffer a selection effect : if a _ faint _ nucleus was stripped , it would now be automatically counted into the globular cluster system of m87 . we can hardly estimate how many nucleated des have already been destroyed and transformed into ucds during the lifetime of the cluster . thus we can not rule out or confirm any scenario _ just by counting _ the number of bright nuclei or that of ucds . only detailed future simulations can resolve this issue . t.l . thanks anna pasquali and eva grebel for fruitful discussions concerning ucds , and katharina glatt for support with the 2007 observing run . the authors thank g. hensler , s. kim , r. kotulla , h. kuntschner , c. mastropietro , s .- c . rey , and s. weinmann for their valuable input to the 2009/10 observing proposal and the data analysis . and t.l . are supported within the framework of the excellence initiative by the german research foundation ( dfg ) through the heidelberg graduate school of fundamental physics ( grant number gsc 129/1 ) . j.j . acknowledges support by the gottlieb daimler and karl benz foundation . this study is based on the sloan digital sky survey ( http://www.sdss.org ) .	to address the question of whether the so - called ultra compact dwarf galaxies ( ucds ) are the remnant nuclei of destroyed early - type dwarf galaxies ( des ) , we analyze the stellar population parameters of the nuclei of 34 virgo des , as well as ten virgo ucds , including one that we discovered and which we report on here . based on absorption line strength ( lick index ) measurements , we find that nuclei of virgo des have younger stellar population ages than ucds , with averages of 5 gyr and @xmath010 gyr , respectively . in addition to this , the metallicity also differs : de nuclei are on average more metal - rich than ucds . on the other hand , comparing the stellar population parameters at the same local galaxy density , with ucds being located in the high density cluster regions , we do not find any difference in the stellar populations of de nuclei and ucds . in those regions , the de nuclei are as old and as metal poor as ucds . this evidence suggests that the virgo ucds may have formed through the stripping of de nuclei .
over the last two decades , significant progress in the development of low temperature detectors for particle astrophysics applications has been made . thermal detectors with dielectric crystal absorbers have been employed in many rare event experiments in deep underground laboratories around the world . these detectors can be designed to have high energy resolution , achieved by using sensitive temperature sensors such as neutron transmutation doped ( ntd ) ge thermistors , transition edge sensors ( tess ) , and metallic magnetic calorimeters ( mmcs)@xcite . when semiconductor or scintillating crystals are used as the absorber , charge or light signals provide additional measurement channels along with temperature measurement ( phonon channel ) . simultaneous measurement of two different channels enables to reduce background events by different ionization or quenching factors for different particle types . although any dielectric crystal can be used as an absorber of a thermal detector , the appropriate measurement of phonon signals is a non - trivial task . at low temperatures , the phonon transport through the interface between two solids becomes highly inefficient due to thermal boundary resistance . moreover , if the temperature sensor measures the electron temperature , the energy of phonons must be transfered to the electron system . in the temperature sensor described above , weak electron - phonon interactions can be the major thermal impedance of phonon measurement at low temperatures . for example , the signal of an ntd ge thermistor is slow to measure the instant temperature increase of a crystal . the rise - time can be extended as tens of milliseconds at a temperature of 10 mk . the direct measurement of athermal phonons using tes sensors combined with superconducting phonon collectors on the crystal surface @xcite may be an efficient way to overcome thermal impedances . however , the required fabrication processes for high performance tess limit the choice of crystals . we employed a mmc as the temperature sensor of a camoo@xmath1 crystal . camoo@xmath1 is a scintillating crystal that has the highest light yield among the studied molybdate crystals@xcite . @xmath2mo within the crystal is a candidate for neutrinoless double beta decay , and its q value is 3034.40(17 ) kev@xcite . since the resolving power of a typical mmc used in x - ray measurement can be better than 2000 , and the response time can be faster than 1 @xmath3s , mmcs can provide high performance in time and energy measurements in general@xcite . mmcs can also present several advantages in detector development studies based on a given crystal . many detector parameters can be adjusted to optimize the sensitivity under the experimental conditions during the design and operation stages . the present study aims to understand and manipulate the heat flow mechanism in camoo@xmath1 thermal detectors suitable for neutrinoless double beta decay experiments@xcite . a camoo@xmath1 crystal with a diameter and height of 4 cm was used as an energy absorber . the crystal was held by teflon - coated phosphor - bronze springs attached to a cylindrically structured copper sample holder , as shown in fig . [ fig : detector_design ] . a micro - fabricated mmc chip@xcite produced in the kirchhoff - institute for physics at heidelberg university was employed as the temperature sensor . a gold film with an area and thickness of 10 mm@xmath4 and 200 nm , respectively , was evaporated on the crystal surface close to the edge and was used for the thermal connection between the crystal and mmc sensor . five 25 @xmath5 m gold wires were bonded to the gold film and au : er layer of the mmc chip to make a thermal connection . one advantage of the proposed design is that the heat flow between the absorber and the sensor can be controlled by changing the dimensions and location of the gold film and the number of gold bonding wires . considerations and dimensions of the first designs were reported in detail elsewhere@xcite . we established a thermal model for the detector as shown in the schematic diagram illustrated in fig . [ fig : thermal_model ] . the detector system is divided into three main components , the camoo@xmath0 crystal , gold film , and mmc sensor . after particle absorption in the crystal , athermal phonons are initially generated in the crystal and travel around the crystal ballistically within a certain period of time@xcite . a small portion of athermal phonons can be transmitted into the gold film . they can also be scattered by conduction electrons in the gold film and deposit their energy to conduction electrons . the remaining athermal phonons are down - converted to a thermal phonon distribution over time . subsequently , the energy transfer between the thermal phonons in the crystal and conduction electrons in the gold film follows thermal heat flow processes at low temperatures . phonon transmission through the interface between camoo@xmath1 and the gold film can be understood by acoustic mismatch model@xcite . thermal phonons and electrons in the gold film interact with each other via electron - phonon interactions@xcite . heat flow along the gold wires between the gold film and mmc sensor , in which electrons in the au : er layer are strongly coupled to er spins , occurs by electronic heat transfer according to the weidemann - franz law@xcite . @xmath6 , @xmath7 and @xmath8 are the heat capacities of the camoo@xmath1 , the gold film , and the mmc , respectively . @xmath9 , @xmath10 , @xmath11 and @xmath12 are the thermal conductances between the crystal and gold film , the gold film and mmc , the crystal and heat bath , and the mmc and heat bath , respectively . athermal and thermal processes in the thermal model are represented by long - curved and short - straight arrows in fig . [ fig : thermal_model ] . the differential equations of heat flow in the thermal model can be written as : @xmath13 where @xmath14 , @xmath15 , @xmath16 and @xmath17 are the total input energy into the crystal , fraction of energy thermalized in the gold film by athermal phonon absorption , thermalization time of the crystal , and the temperature of the heat bath , respectively . the three simultaneous equations are equivalent to a third - order differential equation that has a complicated analytical solution . for simplicity , we assumed that the gold film did not have a heat capacity but only mediated heat flow . this is a valid approximation when @xmath18 . thus , eq . ( [ thermal_eq ] ) can be reduced to second - order differential equations , and the solutions can be written as : @xmath19 where @xmath20 , @xmath21 , @xmath22 and @xmath23 are functions of the thermal parameters . the temperature change of the mmc sensor , @xmath24 , can be considered as the sum of the two pulses that represent athermal and thermal contributions of the signal@xcite . we attempted to fit @xmath24 to the measured temperature signal by varying thermal parameters in the model . @xmath6 and @xmath8 were set to the calculated heat capacity of the camoo@xmath1 crystal and mmc sensor , respectively , while @xmath25 , @xmath15 , and @xmath16 were used as fitting parameters in the analysis . the energy input @xmath14 in eq . ( [ thermal_eq ] ) corresponds to the phonon energy initiated by particle absorption . we set @xmath14 as 90% of the total energy due to energy losses to scintillation lights , lattice damage , tunneling systems that their relaxation time is much longer than our signal decay - time , and unexpected thermal systems@xcite . [ fig : old_fit ] shows the measured signal of 5.3 mev alpha particles @xcite and the fitted results of a non - linear least - square fitting method . both the rise and decay regions of the experimental and fitted signals were in good agreement . a similar analysis was repeated for measured pulses at 44 mk , 55 mk , and 60 mk . consequently , @xmath15 and @xmath16 were approximately 2% and 1.5 ms at 44 mk , respectively , and both @xmath15 and @xmath16 decreased slightly as the temperature increased . the values of @xmath25 s found in the analysis were in reasonable agreements and tendencies with expected values at different temperatures , and are not discussed further herein . the collection efficiency of athermal phonons depends on both the area and thickness of the gold film . larger areas increase the chance that the initial athermal phonons hit the gold film . thicker layers also increase the absorption probability of transmitted phonons to deposit their energy to electrons in the film . however , the signal size eventually becomes smaller with the larger heat capacity ( or volume ) of the film increased over a set of optimal dimensions . to investigate such dimensions of the gold film that maximize the signal size , the numerical solution of eq . ( [ thermal_eq ] ) was calculated with different area and thicknesses of the gold film based on the parameters mentioned in the above analysis . [ fig : optimization ] shows the signal size and rise - time of the numerical solution for different film sizes . the time taken by a signal to change from 10% to 90% of the pulse height was considered the rise - time . based on the calculations of the expected pulses , we evaporated a new gold film in the middle of the surface of the camoo@xmath1 crystal , as shown in fig . [ fig : new_design ] . we chose a round film with a diameter and thickness of 2 cm and 200 nm , respectively . an additional pattern with the same thickness and 1/3 of the film area was evaporated to increase lateral thermal conductance of the gold film , and the expected value of @xmath15 was 39% . the detector setup with the revised gold film showed larger and faster signals than the first design . the measured pulses in the two cases are compared in fig . [ fig : old_new ] . the energy of signals were normalized to 1 mev for comparison because the source used in the revised design had different energy with the first design . the normalized pulse height of the revised detector was about three times larger than that of the first design . the rise - time was 0.6 ms at 42 mk , which is close to the expected values shown in fig . [ fig : rise_goldfilm ] . even though the heat capacity of the gold film increased by a factor of 31 , the efficient collection of athermal phonons resulted in a larger and faster signal . the heat capacity of the revised design becomes comparable to that of the mmc . the approximation used for negligible heat capacity of the small gold film should not be applied in the thermal model of the revised detector . further pulse shape studies and related details using the original thermal equations will be reported in future work . the background spectrum measured with the revised setup showed an energy resolution of [email protected] kev fwhm for a full absorption line of 2.6 mev(@xmath27tl ) gammas . moreover , due to the rapid rise - time , alpha events could be discriminated from electron events using pulse shape analysis . details on the measurement and analysis of the results will be discussed in a future report . we established a thermal model of a large crystal detector composed of a camoo@xmath1 crystal and a mmc sensor for cryogenic particle detection . heat flow to a gold film from the camoo@xmath1 crystal was described by athermal and thermal processes . based on the thermal model , an optimization simulation was performed for different sizes and thicknesses of the gold film . significant improvements in the pulse size and rise - time were obtained in an experiment using an improved design . however , the thermal model discussed herein assumes instant generation of athermal phonons after particle absorption in the crystal . although the majority of absorbed energy was converted to phonons even in the scintillating crystals , some fraction of the phonons should be generated on a time scale related to the scintillation decay - time . a scintillation decay - time of about 0.4 ms was reported in a measurement with a camoo@xmath1 crystal for alpha particles at 7 k@xcite . the lifetime of athermal phonons ( or thermalization time , @xmath16 ) derived here may be partially affected by this generation of phonon . moreover , the scintillation decay - time might be different for alpha and electron events@xcite . the present analysis was performed for surface alpha signals ; however , bulk electron events in camoo@xmath1 may have different characteristics in phonon generation . although the current model does not cover all of the physical phenomena in the crystal , improvements were achived by understanding heat flow mechanisms in the detector . the detector performance with respect to energy and time resolutions in the present work demonstrated that cryogenic detection schemes of using camoo@xmath1 and mmc are promising in the search for neutrinoless double beta decay . this work was supported by the national research foundation of korea grant funded by the korean government ( nrf-2011 - 220-c00006 ) .	we established a simple thermal model of the heat flow in a large crystal detector designed for a neutrinoless double beta decay experiment . the detector is composed of a camoo@xmath0 crystal and a metallic magnetic calorimeter ( mmc ) . the thermal connection between the absorber and the sensor consists of a gold film evaporated on the crystal surface and gold bonding wires attached to this film and the mmc sensor . the model describes athermal and thermal processes of heat flow to the gold film . a successive experiment based on optimization calculations of the area and thickness of the gold film showed a significant improvement in the size and rise - time of the measured signals .
in the early 1990s , kontsevich @xcite defined the kontsevich invariant ( the universal finite - type invariant ) of knots by the integral on the configuration space of finite distinct points in @xmath0 . all rational - valued vassiliev invariants are recovered from the kontsevich invariant through weight systems . in the late 1990s , ohtsuki @xcite showed that one can consider an arithmetic expansion of the quantum @xmath1-invariant of rational homology spheres . the result of this expansion is called the perturbative @xmath1-invariant . ohtsuki @xcite also introduced integer - valued finite - type invariants of integral homology spheres . kricker and spence @xcite proved that the coefficients of the perturbative @xmath1-invariant are of finite - type . on the other hand , the perturbative @xmath1-invariant was extended to the perturbative @xmath2-invariant for any simply connected simple lie group @xmath3 , where @xmath2 is the quotient lie group of @xmath3 by its center . moreover , using the kontsevich invariant , le , murakami and ohtsuki @xcite introduced the lmo invariant of connected oriented closed 3-manifolds . it is known that the lmo invariant is universal among perturbative invariants of rational homology spheres . bar - natan , garoufalidis , rozansky and thurston @xcite gave an alternative construction of the lmo invariant of rational homology spheres by introducing the rhus integral that is also called the formal gaussian integral . in these papers , it is suggested that the rhus integral can be extended to an invariant of tangles in a rational homology sphere , which is called the kontsevich - lmo invariant in @xcite and @xcite . using the kontsevich - lmo invariant , cheptea , habiro and massuyeau @xcite defined the lmo functor as a functorial extension of the lmo invariant . in fact , the value for a rational homology cube @xmath4 ( in which case the boundary of @xmath4 is @xmath5 ) coincides with the lmo invariant of the closed 3-manifold obtained from @xmath6 ^ 3 $ ] and @xmath4 by gluing their boundaries , see ( * ? ? ? * section 3.5 ) . one of the advantage of the lmo functor is that we can use its functoriality to calculate its values and to prove its properties . the lmo functor is defined for a connected , oriented , compact 3-manifold regarded as a certain cobordism between two surfaces . here , these surfaces are assumed to be with at most one boundary component . the purpose of this paper is to construct an extension of the lmo functor to the case of _ any _ number of boundary components ( compare two figures in figure [ fig : oldnewcob ] ) . it is expected that this extension enables us to introduce many _ categorical _ operations on cobordisms , for instance , which corresponds to the pairing or shelling product defined in @xcite . we define the braided non - strict monoidal category @xmath7 to be lagrangian @xmath8-cobordisms extended as above ( see definition [ subsec : qcob ] and remark [ rem : braided ] ) . the main result is the following theorem . there is a tensor - preserving functor @xmath9 between two monoidal categories , which is an extension of the lmo functor . a generating set of @xmath7 as a monoidal category is determined in proposition [ prop : gen ] and the values on them are listed in table [ tab : value ] . therefore , the functoriality and tensor - preservingness of @xmath10 enable us to compute the value on a lagrangian @xmath8-cobordism by decomposing it into the generators . it should be emphasized that there are diagrams colored by both @xmath11 ( or @xmath12 ) and @xmath13 in table [ tab : value ] , which implies that our extension is non - trivial . habiro @xcite and goussarov @xcite introduced claspers and clovers respectively that play a crucial role in a theory of finite - type invariants of arbitrary 3-manifolds . in @xcite , using claspers , it was shown that the lmo functor is universal among rational - valued finite - type invariants . we prove that our lmo functor @xmath10 has the same property . @xmath10 is universal among rational - valued finite - type invariants . in @xcite , milnor defined an invariant of links , which is called the milnor @xmath14-invariant . habegger and lin @xcite showed that this invariant is essentially an invariant of string links . in @xcite , it was proven that the tree reduction of the lmo functor is related to milnor invariants of string links . by extending the milnor - johnson correspondence ( see , for example , @xcite ) suitably , we show that the same is true for @xmath10 . let @xmath15 be a string link in an integral homology sphere @xmath16 . then the first non - trivial term of @xmath17 is determined by the first non - vanishing milnor invariant of @xmath15 , and vice versa . finally , in @xcite , @xcite and @xcite , one can find related researches from different points of view from this paper . in section [ sec : cobbttangle ] , we define cobordisms and bottom - top tangles that are main object in this paper . section [ sec : jacobidiagop ] is devoted to reviewing jacobi diagrams and the formal gaussian integral . in section [ sec : kontsevichlmoinv ] , the kontsevich - lmo invariant of tangles in a homology cube is explained , which plays a key role in the subsequent sections . the main part of this paper is section [ sec : extlmofunc ] , where we construct an extension of the lmo functor . in section [ sec : genvalue ] , we shall give a generating set of the category @xmath18 and calculate the values on them . these values will be used later . section [ sec : universal ] is devoted to reviewing clasper calculus and proving the universality among finite - type invariants . finally , in section [ sec : knot ] , we apply our lmo functor @xmath10 to some cobordisms arising from knots or string links . in particular , the relationship between @xmath10 and milnor invariants of string links will be discussed . throughout this paper , we denote by @xmath19 the homology groups of a topological space @xmath20 with coefficients in @xmath21 . ( however , arguments in sections [ sec : cobbttangle][sec : extlmofunc ] are valid for the case of coefficients in @xmath22 , see remark [ rem : rational ] . ) let @xmath23 denote the set @xmath24 , where @xmath25 is @xmath26 or we omit @xmath25 . let @xmath20 and @xmath27 be geometric objects equipped with the `` top '' and `` bottom '' , for example , cobordisms or tangles . then , the composition of @xmath20 and @xmath27 is always defined by stacking the bottom of @xmath27 on the top of @xmath20 . we use almost the same notation and terminology as in @xcite . however , their definitions are suitably extended . the author would like to thank takuya sakasai , who provided him with helpful comments and suggestions . he also feels thankful to kazuo habiro , who helped him to prove proposition [ prop : gen ] . he would also like to thank gwnal massuyeau for pointing out many crucial mistakes in his first draft and recommending him to introduce the functor @xmath28 in remark [ rem : kill ] . finally , this work was supported by the program for leading graduate schools , mext , japan . in this section , we give the definition of cobordisms and a way to express them as certain tangles . we first introduce some notation . let @xmath29 denote the free monoid generated by letters @xmath30 . similarly , we denote by @xmath31 the free magma generated by @xmath30 . we call an element of @xmath29 or @xmath31 a _ word_. let @xmath32 be a word . we denote by @xmath33 the element obtained from @xmath32 by replacing all letters except for @xmath34 with the empty word @xmath35 . let @xmath36 denote the word length of @xmath32 . for example , if @xmath37 , then @xmath38 . next , we prepare two kinds of surfaces . let @xmath39 and let @xmath40 , @xmath41 . we define the oriented compact surface @xmath42 in @xmath43 as illustrated in figure [ fig : f ] , which has handles and inner boundary components corresponding to letters @xmath44 and @xmath45 in @xmath32 , respectively . moreover , let @xmath46 , @xmath47 and @xmath48 denote the oriented simple closed curves based at @xmath25 as drawn in figure [ fig : f ] . we often regard the above closed curves as free loops . with @xmath49 let @xmath50 such that @xmath51 , and let @xmath52 be an element of the @xmath53th symmetric group @xmath54 . we define the _ reference surface _ @xmath55 as illustrated in figure [ fig : r ] , which is an oriented closed surface consisting of four kinds of surfaces : the _ top surface _ @xmath56 , the _ bottom surface _ @xmath57 , @xmath53 tubes and four sides , where @xmath57 is the surface @xmath58 with the opposite orientation . the boundary of the tubes are attached to the inner boundary of the top surface and bottom surface according to the permutation @xmath52 . with @xmath59 , @xmath60 let @xmath61 such that @xmath62 . a _ cobordism _ from @xmath56 to @xmath58 is an equivalence class of triples @xmath63 , where * @xmath4 is a connected , oriented , compact 3-manifold , * @xmath52 is an element of the @xmath53th symmetric group @xmath64 , * @xmath65 is an orientation - preserving homeomorphism . here , @xmath63 is equivalent to @xmath66 if @xmath67 and there is an orientation - preserving homeomorphism @xmath68 satisfying @xmath69 . let @xmath70 denote the restriction of @xmath71 to @xmath72 . we define the category @xmath73 of cobordisms as follows : the set of objects is the monoid @xmath74 . a morphism from @xmath75 to @xmath76 is a cobordism from @xmath56 to @xmath58 . the composition of @xmath77 and @xmath78 is defined by @xmath79 namely , the composition is obtained by stacking the bottom of @xmath80 on the top of @xmath4 . the identity of @xmath32 is the cobordism @xmath81,{\mathrm{id}}_{{\mathfrak{s}}_{\|w^\circ\|}},{\mathrm{id}})$ ] , where the last @xmath82 means the obvious parametrization . moreover , @xmath73 is a strict monoidal category , where the tensor product of two cobordisms is their horizontal juxtaposition and the unit is the empty word @xmath35 . [ rem : kill ] we denote by @xmath83 the full subcategory whose objects belong to the monoid @xmath84 , namely , @xmath83 is the category @xmath73 defined in @xcite . the full functor @xmath85 is defined to be killing the extra boundary , that is , for @xmath86 , @xmath87 is obtained by attaching @xmath88 tubes to @xmath4 . let @xmath39 such that @xmath89 , and let @xmath90 be points on each inner boundary component . we recall that @xmath25 is the point on the outer boundary as drawn in figure [ fig : f ] . the _ mapping class group _ of @xmath42 is defined by @xmath91 where an isotopy fixes the set @xmath92 pointwise . for any @xmath93 , we denote by @xmath94 the _ mapping cylinder _ of @xmath95 , that is , the cobordism @xmath96 , \pi(h ) , { \mathrm{id}}\times(-1)\cup h\times 1),\ ] ] where @xmath97 is defined by @xmath98 . note that an isotopy does not necessarily fix @xmath99 , however , by definition , @xmath94 is well - defined . one can show that the map @xmath100 is an injective monoid homomorphism . it is obvious that the image of @xmath101 is contained in the group consisting of invertible elements of @xmath102 . in fact , the following proposition holds . [ prop : invertelem ] let @xmath39 and @xmath103 . then @xmath4 is left - invertible if and only if it is right - invertible . moreover , the group @xmath104 consisting of invertible elements of @xmath102 is equal to @xmath105 . the definition of the mapping class group is different from @xcite , although the same proof as ( * ? ? ? * proposition 2.4 ) is valid for our monoid homomorphism @xmath101 . the mapping class group of @xmath42 is usually defined by @xmath106 which is isomorphic to the kernel of the homomorphism @xmath107 . if @xmath108 , then @xmath109 and @xmath110 are called the _ framed braid group _ and the _ framed pure braid group _ on @xmath53 strands , respectively . we should restrict cobordisms to the good ones in order to define the kontsevich - lmo invariant ( see section [ subsec : defkontsevichlmoinv ] ) and to let it take value in the space of top - substantial jacobi diagrams ( see section [ subsec : topsubstantial ] ) . let @xmath39 . we define the subgroups of @xmath111 by @xmath112 a cobordism @xmath113 is _ lagrangian _ if the following two conditions are satisfied : 1 . @xmath114 , 2 . @xmath115 . it follows from a mayer - vietoris argument that the composition of two lagrangian cobordisms is again lagrangian . ( one can get an alternative proof by lemma [ lem : lagrangian ] and ( * ? ? ? * claim 3.18 ) or the fact that @xmath4 is lagrangian if and only if @xmath87 is lagrangian in the sense of @xcite . ) thus , we obtain the wide subcategory @xmath18 whose morphisms are lagrangian cobordisms . actually , we need a non - strict monoidal category @xmath7 defined in section [ subsec : qcob ] . [ 1 ] under the condition , is equivalent to the following condition : the map @xmath116 is an isomorphism . it is easy to see that ( 1 ) implies ( 1 ) . let us prove the converse . it follows from ( 1 ) and ( 2 ) that @xmath117 , where @xmath118 is the first betti number of @xmath4 . since @xmath4 is a compact odd - dimensional manifold , we have @xmath119 , and thus @xmath120 . hence , the above inequality is an equality , that is , @xmath121 is an isomorphism . [ rem : homcube ] by the proof of the above lemma , we have @xmath122 . since @xmath123 is free , we conclude @xmath124 , namely @xmath4 is a _ homology handlebody_. in particular , a homology handlebody of genus zero is called a _ homology cube_. the mapping cylinder of @xmath93 is lagrangian if and only if @xmath125 . indeed , if @xmath94 is lagrangian , then we have @xmath126 . since @xmath127 is an isomorphism , so is @xmath128 . the converse is obvious . in this subsection , we translate cobordisms to certain tangles in a homology cube . let @xmath39 . we prepare the reference surface @xmath129 with the pairs of points @xmath130 and the points @xmath131 corresponding to the letters @xmath44 s and @xmath45 s in @xmath32 , respectively . these points are on the @xmath132-axis according to @xmath32 as drawn in figure [ fig : ponitedreferencesurface ] . such that @xmath133 . ] let @xmath134 such that @xmath135 and let @xmath136 . a _ bottom - top tangle _ of _ type _ @xmath137 is an equivalence class of pairs @xmath138 , where * @xmath139 is a cobordism from @xmath129 to @xmath129 , * @xmath140 is a framed oriented tangle in @xmath16 , which consists of the top components @xmath141 , the bottom components @xmath142 and the vertical components @xmath143 , where * * each @xmath144 runs from @xmath145 to @xmath146 , * * each @xmath147 runs from @xmath148 to @xmath149 , * * each @xmath150 runs from @xmath151 to @xmath152 . here , @xmath138 and @xmath153 are equivalent if they are of the same type and @xmath16 is equivalent to @xmath154 as cobordisms by a homeomorphism that sends @xmath140 to @xmath155 and respects their framings . we define the category @xmath156 as follows : an object of @xmath156 is an element of @xmath74 . a morphism from @xmath75 to @xmath76 is a bottom - top tangle of type @xmath137 for some @xmath136 . the composition of @xmath157 and @xmath158 is the bottom - top tangle obtained by inserting @xmath159 ^ 3 , t_v)$ ] between them and performing the surgery along @xmath160-components link @xmath161 as illustrated in figure [ fig : tvidw ] . the identity of @xmath32 is the bottom - top tangle as depicted in figure [ fig : tvidw ] . indeed , one can check that @xmath162 using the kirby move ii and the useful fact that if a disk @xmath163 in a 3-manifold @xmath4 intersects a framed knot @xmath164 once , then the result of surgery along @xmath164 and @xmath165 with 0-framing is again @xmath4 ( see figure [ fig : composition ] ) . ^3 , t_v)$ ] and the identity @xmath166 ] moreover , @xmath156 is a strict monoidal category , where the tensor product of two bottom - top tangles is their horizontal juxtaposition and the unit is the empty word @xmath35 . the next lemma is used in the following theorem and so on . let @xmath138 be a bottom - top tangle . then there exist a framed link @xmath167 and a tangle @xmath155 in @xmath168 ^ 3 $ ] such that @xmath169 ^ 3_l,\gamma')$ ] , where @xmath168 ^ 3_l$ ] is the -manifold obtained by surgery along @xmath167 . the pair @xmath170 is called a _ surgery presentation _ of @xmath138 . the operation as drawn in figure gives an isomorphism @xmath171 between two strict monoidal categories . precisely , @xmath172 is defined by digging a bottom - top tangle @xmath173 and parametrizing the boundary of the resulting manifold in accordance with the homeomorphism @xmath174 and the framings of @xmath140 . it is easy to see that @xmath172 satisfies the definition of a monoidal functor except @xmath175 to prove this equality , we take a surgery presentation @xmath170 of @xmath138 . moreover , using an additional surgery link @xmath176 , @xmath138 is expressed as the bottom - top tangle obtained from a bottom - top tangle @xmath159 ^ 3,\widetilde{\gamma})$ ] such that @xmath177 is trivial by surgery along @xmath178 . each @xmath179 bounds a half - disk @xmath180 that could intersect @xmath181 or @xmath178 we now gather the intersections on @xmath180 , and we regard the components intersecting @xmath180 as a bundle in a neighborhood of @xmath180 . since @xmath180 intersects the bundle once , we can apply the argument in the proof of @xmath182 , and thus the desired equality is obtained . next , let @xmath63 be a cobordism . the inverse of @xmath172 can be constructed by attaching 2-handles @xmath183 ^ 2\times[0,1]$ ] to the boundary of @xmath4 . ( we recall that the way of attaching a 3-dimensional 2-handle is uniquely determined by the way one glues the attaching sphere of the 2-handle with @xmath184 . ) we obtain the 3-manifold @xmath16 by attaching 2-handles along @xmath185 s , @xmath186 s and @xmath187 s and define @xmath155 by the co - cores of the 2-handles . by construction , this correspondence gives the inverse of @xmath172 . let @xmath188 be a simple closed curve on the surface @xmath189 that is the interior of @xmath42 . we denote by @xmath190 the dehn twist along @xmath188 . by the definition of a dehn twist and surgery , the cobordisms @xmath191 , @xmath192 and @xmath193 are sent by @xmath194 to respectively . @xmath191 , @xmath193 are lagrangian while @xmath192 is not . there is a natural question when a bottom - top tangle is sent to a lagrangian cobordism by @xmath172 . we introduce an extension of the linking matrix defined in ( * ? ? ? * definition 2.11 ) to answer this question . let @xmath173 be a bottom - top tangle of type @xmath137 such that @xmath16 is a homology cube , and let @xmath195 , @xmath196 . prepare the manifold @xmath6 ^ 3 $ ] containing four kinds of arcs : @xmath197 copies of @xmath198-like arcs , @xmath199 @xmath200-like arcs , @xmath53 parallel arcs and a braid @xmath201 such that @xmath202 , as illustrated in figure [ fig : lk ] . then the _ linking matrix _ of @xmath140 in @xmath16 is defined by @xmath203 where @xmath204 is the usual linking matrix of the link @xmath205 ^ 3$})\ ] ] in the homology sphere @xmath206 ^ 3)$ ] . here , @xmath207 is the matrix whose @xmath208-entry is half the number of positive crossings of the @xmath209th strand and the @xmath210th strand minus the number of negative ones , and @xmath211 is the matrix obtained from @xmath207 by permuting its rows and columns by @xmath212 . and additions , that is , @xmath213 moreover , we set @xmath214 for a cobordism @xmath215 , where @xmath16 is a homology cube . the above map @xmath216 represents a non - trivial class of the first cohomology group @xmath217 , where @xmath218 is the group obtained from @xmath219 by reversing the order of multiplication . indeed , @xmath218 acts on @xmath220 , via the anti - homomorphism @xmath221 , such that @xmath222 . then @xmath223 is a crossed homomorphism , that is , it satisfies @xmath224 furthermore , @xmath225 for any @xmath226 . while , @xmath227 . the next lemma is proved in much the same way as ( * ? ? ? * lemma 2.12 ) . [ lem : lagrangian ] let @xmath138 be a bottom - top tangle . then @xmath228 is lagrangian if and only if @xmath16 is a homology cube and @xmath229 holds . let @xmath163 be a uni - trivalent graph , which is depicted by dashed lines . a univalent ( resp . trivalent ) vertex of @xmath163 is called an _ internal _ ( resp . _ external _ ) _ vertex_. we denote by @xmath230 ( resp . @xmath231 ) the number of internal ( resp . external ) vertices of @xmath163 and we set @xmath232 . let @xmath20 be a compact oriented 1-manifold and let @xmath233 be a finite set . a _ jacobi diagram _ based on @xmath234 is a vertex - oriented uni - trivalent graph whose vertices are either embedded into @xmath20 or are colored by elements of @xmath233 . when identifications of @xmath235 with @xmath236 and @xmath233 with @xmath237 are given , two jacobi diagrams @xmath163 based on @xmath234 and @xmath238 based on @xmath239 are identified if there exists a homeomorphism @xmath240 respecting the identification , orientations of @xmath20 and @xmath241 and the vertex - orientations of @xmath163 and @xmath238 . a jacobi diagram is drawn in the plane whose vertex - orientation agrees with the counter - clockwise order . we denote by @xmath242 the quotient @xmath22-vector space spanned by jacobi diagrams based on @xmath234 subject to the as , ihx and stu relations depicted in figure [ fig : relations ] . since these relations preserve the degree of a jacobi diagram , @xmath242 is a graded @xmath22-vector space . thus , we can take the degree completion of @xmath242 and denote it by @xmath242 again . note that the internal degree is not well - defined for jacobi diagrams of @xmath242 . a jacobi diagram @xmath243 is said to have _ -filter at least _ @xmath53 if @xmath163 is written as a series of jacobi diagrams whose internal degrees are at least @xmath53 , and then we use the notation @xmath244 . if @xmath245 ( resp . @xmath246 ) , then we simply write @xmath247 ( resp . @xmath248 ) for @xmath242 when no confusion can arise . it is well known that @xmath247 is naturally equipped with a connected , commutative , cocommutative , graded hopf algebra structure . there are two important subsets of @xmath247 . we first set @xmath249 where @xmath250 denotes the empty diagram . @xmath251 is a group whose multiplication is disjoint union @xmath252 , and its element is said to be _ group - like_. let @xmath253 denote the subspace of @xmath247 spanned by non - empty connected jacobi diagrams , which is identified with the quotient by the subspace generated by @xmath250 and disconnected jacobi diagrams . the image of @xmath254 by the quotient map is denoted by @xmath255 . moreover , it is well known that the map @xmath256 defined by @xmath257 is a group isomorphism . we further use two subalgebras of @xmath247 which play an important role in this paper . the subalgebra generated by jacobi diagrams only with struts ( resp . without struts ) is denoted by @xmath258 ( resp . @xmath259 ) that is identified with the quotient by the ideal generated by diagrams with @xmath260 ( resp . the ideal generated by struts ) . the quotient maps are called the _ @xmath261-reduction _ ( resp . _ @xmath27-reduction _ ) , the images of @xmath262 is denoted by @xmath263 ( resp . @xmath264 ) . [ lem : group - like ] the group @xmath251 coincides with the set @xmath265 the subspace @xmath259 can also be regarded as the result of the internal degree completion , since @xmath266 holds for all @xmath267 . we denote by @xmath268 the subspace of all homogeneous elements of internal degree @xmath209 . next , we review an analogue of the poincar - birkhoff - witt isomorphism . let @xmath269 ( resp . @xmath270 ) denote @xmath271 intervals ( resp . circles ) indexed by elements of a finite set @xmath272 . the graded linear map @xmath273 is defined for a jacobi diagram @xmath274 to be the average of all possible ways of attaching the @xmath261-colored vertices in @xmath163 to the @xmath261-indexed interval , for all @xmath275 , as depicted in figure [ fig : pbw ] . with @xmath276 , @xmath277 and @xmath278 it is well known that @xmath279 is a graded coalgebra isomorphism . similarly , an isomorphism @xmath280 is defined , where @xmath281 denotes the @xmath272-link relation defined in ( * ? ? ? * section 5.2 ) ( see also ( * ? ? ? * remark 3.7 ) ) . we need an extension of @xmath279 to construct the composition law of the category @xmath282 defined later . let @xmath283 be a copy of the above set @xmath272 . the graded linear map @xmath284 is defined for a jacobi diagram @xmath285 to be stacking of the @xmath286-indexed interval on the @xmath261-indexed interval in @xmath287 , for all @xmath275 , as illustrated in figure [ fig : pbw2 ] . with @xmath288 , @xmath289 and @xmath290 similar ideas appear , for example , in ( * ? ? ? * claim 5.6 ) , ( * ? ? ? * proposition 5.4 ) and ( * ? ? ? * ( 3.3 ) ) . before finishing this subsection , we recall the following lemma that is used several times . [ lem : filter ] let @xmath163 be a jacobi diagram based on @xmath291 . then we have @xmath292 where @xmath293 is obtained from @xmath163 by deleting the @xmath272-indexed intervals . we first review @xmath272-substantial jacobi diagrams according to @xcite . a jacobi diagram @xmath294 is _ @xmath272-substantial _ if @xmath163 contains no strut both of whose vertices are colored by the elements of @xmath272 , and an element of @xmath295 which can be written as a series of such diagrams is also called @xmath272-substantial . let @xmath163 and @xmath296 be jacobi diagrams based on @xmath297 such that at least one of them is @xmath272-substantial . if @xmath163 and @xmath296 have the same number of @xmath261-colored vertices , for all @xmath275 , then we define @xmath298 by @xmath299 otherwise , we set @xmath300 . if @xmath163 and @xmath296 have no @xmath272-colored vertex , we naturally interpret @xmath301 as their disjoint union . from now on , we identify a symmetric matrix @xmath302 with a linear combination of struts @xmath303{strut.pdf}\scriptsize \put ( 35,80){$j$ } \put ( 35,0){$i$ } \end{overpic}}\hspace{1em}}$ ] , and let @xmath304:=\exp_\sqcup(x)$ ] for any @xmath254 . [ def : gaussianint ] an element @xmath305 is _ gaussian _ in the _ variable @xmath272 _ if there are a symmetric matrix @xmath306 and an @xmath272-substantial element @xmath307 such that @xmath308\sqcup p$ ] . here , if @xmath309 , then @xmath3 is said to be _ non - degenerate _ , and we set @xmath310,p}\right\rangle}_s\ \in { \mathcal{a}}(x , c)\ ] ] that is called the _ formal gaussian integral _ of @xmath3 along @xmath272 . if @xmath167 and @xmath311 exist , they are unique . @xmath167 is called the _ covariance matrix _ of @xmath3 in @xcite . in this section , we review the domains and codomains of the kontsevich invariant and kontsevich - lmo invariant in accordance with @xcite . in order to define the kontsevich invariant and the kontsevich - lmo invariant as functors , we use the categories @xmath312 , @xmath313 and @xmath314 . we define the non - strict monoidal category @xmath313 as follows : the set of objects is the magma @xmath315 . a morphism from @xmath75 to @xmath76 is a @xmath8-tangle in a homology cube of type @xmath316 , that is , a properly embedded compact 1-manifold whose boundary corresponds to letters in @xmath317 . the composition of @xmath318 and @xmath319 is stacking @xmath233 on the top of @xmath16 . the identity of @xmath32 is the @xmath8-tangle @xmath159 ^ 3,\downarrow^w)$ ] , where @xmath320 is the union of @xmath321 s and @xmath322 s and each @xmath321 ( resp . @xmath322 ) corresponds to a letter @xmath11 ( resp . @xmath12 ) in @xmath32 . the tensor product of @xmath8-tangles is their juxtaposition . the unit is the empty word @xmath35 . moreover , we denote by @xmath312 the wide subcategory whose morphisms are @xmath8-tangles in @xmath168 ^ 3 $ ] . next , we define the strict monoidal category @xmath314 as follows : the set of objects is @xmath323 . a morphism from @xmath32 to @xmath324 is an element of @xmath325 , the union is taken over all oriented compact 1-manifolds @xmath20 such that each start ( resp . end ) point corresponds to a letter @xmath11 in @xmath32 or @xmath12 in @xmath324 ( resp . @xmath12 in @xmath324 or @xmath11 in @xmath32 ) . roughly speaking , we assign @xmath11 to the boundary of an interval which faces downward . the composition of two elements @xmath326 and @xmath327 is defined to be stacking @xmath296 on @xmath163 according to the letters in @xmath324 . we often denote by @xmath328 the composite @xmath329 . the identity of @xmath32 is the intervals @xmath320 . the tensor product of two morphisms is defined to be their disjoint union . the unit is the empty word @xmath35 . here , we prepare three operations in accordance with @xcite . let @xmath330 and @xmath272 denote the doubling map and orientation reversal map defined as usual : @xmath331 let @xmath332 such that @xmath333 . we define the linear map @xmath334 by applying @xmath330 as much as @xmath335 times to the interval indexed by the @xmath209th letter in @xmath32 for @xmath336 , and applying @xmath272 to the intervals where the two letters differ . we first fix an ( rational ) associator @xmath337 to define the kontsevich invariant of a @xmath8-tangle in @xmath168 ^ 3 $ ] . the _ kontsevich invariant _ @xmath338 is defined by @xmath339 for a @xmath8-tangle @xmath140 , where @xmath340 is the kontsevich integral in @xcite . by definition , @xmath338 defines a tensor - preserving functor @xmath341 , that is , it satisfies the following conditions : 1 . @xmath342 , 2 . @xmath343 , 3 . @xmath344 , 4 . @xmath345 . next , we review the kontsevich - lmo invariant of a @xmath8-tangle in a homology cube @xmath16 . we define @xmath346 by @xmath347 where @xmath25 is the label of the @xmath348-framed unknot , and @xmath349 denotes the kontsevich invariant of the 0-framed unknot , and @xmath350 means the multiplication induced by the connected sum of two circles . the _ kontsevich - lmo invariant _ @xmath351 is defined by @xmath352 for a @xmath8-tangle @xmath138 , where @xmath353 is a surgery presentation of @xmath138 , @xmath354 is the number of positive / negative eigenvalues of the linking matrix of @xmath167 , and let @xmath355 denote @xmath356 . one can check that @xmath351 defines a tensor - preserving functor @xmath357 . here , a bottom - top @xmath8-tangle in a homology cube @xmath16 is regarded as a @xmath8-tangle in @xmath16 via the embedding @xmath358 , where words @xmath359 are obtained from words @xmath360 by the rule @xmath361 then we denote by @xmath362 the kontsevich - lmo invariant of a bottom - top @xmath8-tangle @xmath138 . the next lemma is used in section [ subsec : constextlmofunc ] and proved similarly to ( * ? ? ? * lemma 3.17 ) . the only difference is that the definition of @xmath363 is extended . [ lem : bottom - top tangle ] let @xmath138 be a bottom - top @xmath8-tangle . then @xmath364 is group - like and its @xmath261-reduction is equal to @xmath365 $ ] . in this section , we construct categories @xmath7 , @xmath282 and an extension of the lmo functor @xmath9 . we refine cobordisms by replacing the monoid @xmath74 with the magma @xmath366 . let @xmath367 . we denote by @xmath42 the compact surface @xmath368 defined in section [ subsec : cob ] , where @xmath369 is the word in @xmath74 obtained from @xmath32 by forgetting its parenthesization . in the same manner , we use the notation @xmath55 . [ def : qcob ] let @xmath370 with @xmath135 . a _ @xmath8-cobordism _ from @xmath56 to @xmath58 is a cobordism from @xmath56 to @xmath58 equipped with the parenthesizations of @xmath75 and @xmath76 . we define the category @xmath371 of @xmath8-cobordisms by the same way as the category @xmath73 in section [ subsec : cob ] . similarly , the wide subcategory @xmath7 of _ lagrangian @xmath8-cobordisms _ is obtained , which is the domain of an extension of the lmo functor . furthermore , we set @xmath372 for a @xmath8-cobordism @xmath373 , where the functor @xmath172 has been defined in section [ subsec : bttangle ] , and @xmath374 is called the _ unnormalized lmo invariant _ of @xmath4 in @xcite . an element @xmath375 is said to be _ top - substantial _ if @xmath132 is @xmath376-substantial . recall that the term @xmath272-substantial is defined in section [ subsec : gaussianint ] . we define the strict monoidal category @xmath282 as follows : the set of objects is the set @xmath377 of pairs of non - negative integers . a morphism from @xmath378 to @xmath379 is a pair @xmath380 , where @xmath375 is top - substantial and @xmath381 . there is no morphism from @xmath378 to @xmath382 if @xmath383 , and we simply denote @xmath384 by @xmath385 . the composition of @xmath386 and @xmath387 is defined to be @xmath388 where two kinds of @xmath389 s have been defined in section [ subsec : jacobidiag ] . the identity of @xmath378 is @xmath390{strut.pdf}\scriptsize \put ( 35,80){$i^+$ } \put ( 35,0){$i^-$ } \end{overpic}}\hspace{1em}}\right]$ ] . the tensor product of @xmath386 and @xmath391 is defined to be @xmath392 where @xmath393 maps @xmath209 to @xmath394 ( resp . @xmath395 ) if @xmath396 ( resp . @xmath397 ) . finally , @xmath398 and the unit is @xmath399 . note that we omit @xmath52 from @xmath380 when no confusion can arise . let us describe the above composition law concretely . [ lem : precomposition ] let @xmath400 , @xmath401 and let @xmath402 be a @xmath403 matrix that is regarded as a linear map @xmath404 . then @xmath405\sqcup x)\circ y$ ] is equal to @xmath406 where @xmath407 . similarly , let @xmath163 be a @xmath408 matrix . then @xmath409\sqcup y)$ ] is equal to @xmath410 the next lemma is proven by applying the previous lemma repeatedly . [ lem : composition ] let @xmath411 be as above and suppose that they are decomposed as @xmath412\sqcup x^y$ ] and @xmath413\sqcup y^y$ ] , where @xmath414 then @xmath415 is equal to @xmath416 \\ & \quad \sqcup\chi_{{\lfloor{n}\rceil}^0}^{-1}\chi_{{\lfloor{n}\rceil}^0,{\lfloor{n}\rceil}^{0'}}\left([\tau^{-1}(x^{00}/2+x^{-0})+y^{00}/2+y^{+0}]\sqcup{\left\langle{l , r}\right\rangle}_{{\lfloor{g}\rceil}^\ast}\right),\end{aligned}\ ] ] where @xmath417\sqcup x^y / i^{+ } \mapsto i^\ast + x^{-+}y^{--}i^{- } + y^{+-}i^{- } , i^0 \mapsto \tau^{-1}(i)^0 \right ) , \\ r & : = \left([y^{--}/2]/i^{- } \mapsto i^\ast\right)\sqcup\left([y^{-0}]\sqcup y^y / i^{- } \mapsto i^\ast + x^{-+}i^{+ } , j^0 \mapsto j^{0'}\right).\end{aligned}\ ] ] [ cor : composition ] in the above lemma , if @xmath418 and @xmath419 , then @xmath420 $ ] . by the previous lemma , we have @xmath421\sqcup\chi_{{\lfloor{n}\rceil}^0}^{-1}\chi_{{\lfloor{n}\rceil}^0,{\lfloor{n}\rceil}^{0'}}\bigl([x^{00}/2+x^{-0}/2 ] \\ & \quad \sqcup{\left\langle{([x^{+0}]\sqcup x^y / i^{+}\mapsto i^\ast+i^{+}),(y^y / i^{-}\mapsto i^\ast + x^{-+}i^{+})}\right\rangle}_{{\lfloor{g}\rceil}^\ast}\bigr).\end{aligned}\ ] ] hence , using lemma [ lem : filter ] , @xmath422 is equal to @xmath423\sqcup\chi_{{\lfloor{n}\rceil}^0}^{-1}\chi_{{\lfloor{n}\rceil}^0,{\lfloor{n}\rceil}^{0'}}\left([x^{00}/2+x^{-0}/2]\sqcup[x^{+0}]\right ) = [ x/2].\ ] ] in @xcite , a certain element @xmath424 is introduced , which roughly corresponds to the bottom - top tangle @xmath425 defined in section [ subsec : bttangle ] . we first set @xmath426{bch.pdf}\scriptsize \put ( 33,69){$x$ } \put ( 33,31){$y$ } \put ( 18,3){$r$ } \end{overpic}}\hspace{0.2em}}\right ) \in { \mathcal{a}}(\emptyset,\{x , y , r\}),\ ] ] where is the map @xmath427 . next , @xmath428 is defined to be @xmath429 finally , @xmath430 is defined by @xmath431 since @xmath432 is group - like and its @xmath261-reduction is equal to @xmath433{strut.pdf}\scriptsize \put ( 35,80){$x_+$ } \put ( 35,0){$x_-$ } \end{overpic}}\hspace{1em}}\right]$ ] ( ( * ? ? ? * lemma 4.9 ) ) , @xmath430 is also group - like and its @xmath261-reduction is equal to @xmath434 $ ] . the proof of the following key lemma is similar to ( * ? ? ? * lemma 4.10 ) , however , we must pay attention to the additional map @xmath435 in the definition of the composition law in the category @xmath282 . [ lem : key ] let @xmath436 , @xmath437 be bottom - top @xmath8-tangles of type @xmath438 , @xmath439 respectively and suppose that @xmath440 , @xmath441 are lagrangian . then we have @xmath442 where @xmath443 is regarded as an element of @xmath444 . let @xmath445 , @xmath446 be surgery presentations of @xmath16 , @xmath233 respectively . let @xmath447 ( resp . @xmath448 ) be the framed link in @xmath449 obtained from @xmath450 and the bottom components of @xmath451 ( resp . @xmath452 and the top components of @xmath451 ) by gluing their boundaries . finally , let @xmath453 and @xmath454 . then @xmath455 is a surgery presentation of @xmath449 , and thus @xmath456 here , the above integration is computed as follows : @xmath457 where @xmath458 , and @xmath459 means the composition in the category @xmath314 . it follows from the identity @xmath460 ( ( * ? ? ? * ( 4 - 2 ) ) ) that @xmath461 therefore , @xmath462 is equal to @xmath463 let us introduce notation to calculate the above integration : @xmath464 where @xmath465 , @xmath466 . then , the above integrand is computed as follows : @xmath467{bch.pdf}\scriptsize \put ( 33,69){$i^\vartriangle$ } \put ( 33,31){$i^\cup$ } \put ( 18,3){$i^\bot$ } \end{overpic}}\hspace{0.2em } } \sqcup \bigsqcup_{i=1}^g { \hspace{-0.2em}\raisebox{-1.7em } { \begin{overpic}[height=3.8em]{bch.pdf}\scriptsize \put ( 33,69){$i^\cap$ } \put ( 33,31){$i^\triangledown$ } \put ( 18,3){$i^\top$ } \end{overpic}}\hspace{0.2em}}\ , \chi_{\pi_0 t_{w^\bullet}}^{-1}z(t_{w^\bullet}^\nu ) \sqcup \chi_{\pi_0 \gamma^\pm}^{-1}z(b ) \natural \chi_{\pi_0 \omega^\pm}^{-1}z(c ) \biggr\rangle_{\parbox{2em}{\scriptsize $ \mathord{\vartriangle}\mathord{\cap } \\[-4pt ] \mathord{\cup}\mathord{\triangledown}$ } } \\ & = \chi_{\pi_0(\gamma^0\omega^0)}^{-1 } { \left\langle { { \left\langle { \lambda , \chi_{\pi_0 t_{w^\bullet}}^{-1}z(t_{w^\bullet}^\nu ) } \right\rangle}_{\vartriangle \triangledown } , \chi_{\pi_0 \gamma^\pm}^{-1}z(b ) \natural \chi_{\pi_0 \omega^\pm}^{-1}z(c ) } \right\rangle}_{\cup \cap } \\ & = \chi_{\pi_0(\gamma^0\omega^0)}^{-1 } { \left\langle { { \left\langle { \lambda , \chi_{\pi_0 t_{w^\bullet}}^{-1}z(t_{w^\bullet}^\nu ) } \right\rangle}_{\vartriangle \triangledown } , \chi_{\pi_0 \gamma^0 , \pi_0 \omega^0 } \left(\chi_{\pi_0 \gamma}^{-1}z(b ) \sqcup \chi_{\pi_0 \omega}^{-1}z(c)\right ) } \right\rangle}_{\cup \cap},\end{aligned}\ ] ] where the @xmath209th component of @xmath468 corresponds to the @xmath469th component of @xmath470 , and @xmath471 is defined by @xmath472 therefore , the above integration is written as follows : @xmath473 here , @xmath474 therefore , the left - hand side of this lemma is equal to @xmath475 on the other hand , the right - hand side of this lemma is obtained as follows : connect the @xmath476-colored vertices and @xmath477-colored vertices in @xmath430 with the @xmath478-colored vertices in @xmath479 and @xmath480-colored vertices in @xmath481 respectively and apply the composite map @xmath482 . we can now define @xmath10 , which is the main purpose of this paper . let @xmath483 with @xmath484 and let @xmath485 , @xmath486 . the _ normalized lmo invariant _ @xmath10 of @xmath487 is defined by @xmath488 let @xmath489 ( resp . @xmath490 ) denote the @xmath261- ( resp . @xmath27- ) reduction of @xmath491 . the next lemma follows from lemma [ lem : bottom - top tangle ] and corollary [ cor : composition ] ( 1 ) . @xmath492 is group - like and @xmath493 $ ] . [ thm : zt ] the normalized lmo invariant defines a tensor - preserving functor @xmath494 which is called _ _ ( _ _ an extension of the _ lmo functor _ , that is , @xmath10 satisfies the following conditions : 1 . @xmath495 , 2 . @xmath496 , 3 . @xmath497 , 4 . @xmath498 . \(3 ) and ( 4 ) follows from these of @xmath338 ( see section [ subsec : defkontsevichlmoinv ] ) . ( 2 ) is due to lemma [ lem : key ] . let us prove ( 1 ) . by the previous lemma , @xmath499 is group - like , thus it is written as @xmath500 . we assume that the higher - degree terms are not zero , and write @xmath501 it follows from ( 2 ) that @xmath502 . by comparing both side of this equality , @xmath503 thus we have @xmath504 , and it is a contradiction . therefore , @xmath505 , and we conclude ( 4 ) . by definition , if @xmath506 , then @xmath10 coincides with the original lmo functor defined in @xcite . moreover , the diagram @xmath507 ^ -{{\mathrm{k } } } \ar[d]_-{{\widetilde{z } } } & { \mathcal{lc}\mathit{ob}}^\bullet_q \ar[d]^-{\text{original $ { \widetilde{z}}$ } } \\ { { } ^{\mathit{ts}}\!\mathcal{a}}\ar[r]^-{(-/i^0\mapsto0 ) } & { { } ^{\mathit{ts}}\!\mathcal{a}}}\ ] ] commutes , where @xmath28 is the functor defined in the same manner as @xmath28 in remark [ rem : kill ] . in contrast , if @xmath508 , then we have @xmath509 ^ 3,-\sigma ) ) = \left(\chi_{\pi_0 \sigma}^{-1 } { z^{\mathrm{k}}}(-\sigma ) , { \mathrm{id}}_{{\mathfrak{s}}_n}\right)\ ] ] for any ( not necessarily pure ) string link @xmath52 in @xmath168 ^ 3 $ ] on @xmath53 strands . [ rem : rational ] by the same manner , one can define the category @xmath510 of _ rational lagrangian @xmath8-cobordisms _ and the functor @xmath511 , see ( * ? ? ? * remark 2.8 , 3.19 , 4.14 ) . indeed , the formal gaussian integral is originally defined under rational condition and linking matrices in this paper are defined homologically . in @xcite , it was proven that the category @xmath83 , see remark [ rem : kill ] , is generated as a monoidal category by the morphisms @xmath512 , @xmath513 , @xmath330 , @xmath514 , @xmath515 , @xmath516 , @xmath517 and @xmath27 listed in table [ tab : gen ] , and the values on them was calculated up to internal degree 2 . we shall add some morphisms and calculate the values on them . we first define some bottom - top tangles as table [ tab : gen ] . by lemma [ lem : lagrangian ] , the corresponding cobordisms are lagrangian , and we use the same notation to represent the cobordisms . the bottom - top tangles except @xmath518 , @xmath519 , @xmath520 , @xmath521 and @xmath522 are same as in @xcite . @xmath523{eta.pdf } } } & \varepsilon:={\raisebox{-0.8em}{\includegraphics[height=2.5em]{epsilon.pdf } } } & v_{+}:={\raisebox{-1em}{\includegraphics[height=2.5em]{v+.pdf } } } & v_{-}:={\raisebox{-1em}{\includegraphics[height=2.5em]{v-.pdf } } } \\ \mu:={\raisebox{-2em}{\includegraphics[height=5em]{mu.pdf } } } & \delta:={\raisebox{-2em}{\includegraphics[height=5em]{delta.pdf } } } & \psi_{\bullet,\bullet}:={\raisebox{-2em}{\includegraphics[height=5em]{psibb.pdf } } } & \psi_{\circ,\circ}:={\raisebox{-2em}{\includegraphics[height=5em]{psiww.pdf } } } \\ \tau:={\raisebox{-2em}{\includegraphics[height=5em]{tau.pdf } } } & \beta:={\raisebox{-2em}{\includegraphics[height=5em]{beta.pdf } } } & \psi_{\bullet,\circ}:={\raisebox{-2em}{\includegraphics[height=5em]{psibw.pdf } } } & \psi_{\circ,\bullet}:={\raisebox{-2em}{\includegraphics[height=5em]{psiwb.pdf } } } \\ s:={\raisebox{-2em}{\includegraphics[height=5em]{s.pdf } } } & \multicolumn{3}{l}{y:={\raisebox{-2em}{\includegraphics[height=5em]{ycob.pdf } } } } \end{array } $ ] when we write @xmath27 as a morphism of @xmath7 , we always regard it as a morphism from @xmath524 to @xmath35 . in general , when we regard a morphism of @xmath73 as one of @xmath371 , we use the left - handed parenthesization @xmath525 unless otherwise stated . [ rem : braided ] four kinds of @xmath526 s and their inverses induce braidings @xmath527 , which make the category @xmath18 into a braided strict monoidal category . it is easy to see that the cobordisms , , and are the ( two - sided ) inverses of @xmath528 , @xmath518 , @xmath522 and @xmath272 in the category @xmath7 . in the same manner , one finds the inverses of @xmath520 , @xmath519 and @xmath521 . ( by proposition [ prop : invertelem ] , it suffices to check that they are left or right inverses . ) [ prop : gen ] the category @xmath18 is generated as monoidal category by the morphisms @xmath512 , @xmath513 , @xmath330 , @xmath514 , @xmath529 , @xmath516 , @xmath517 , @xmath27 , @xmath530 , @xmath531 , @xmath532 and @xmath521 . moreover , @xmath7 is generated by the morphism @xmath533{puvw.pdf}}}\ ] ] for @xmath534 and their inverses together with the above morphisms . the latter follows from the former . we prove that every @xmath535 belongs to the category @xmath536 generated by the above morphisms . using the braidings , it suffices to consider the case of @xmath537 . namely , the bottom - top tangle @xmath538 is represented as follows : . ( this correspondence is essentially same as the preferred bijection @xmath539 in ( * ? ? ? * section 13 ) . ) since @xmath229 , it suffice to show that the bottom - top tangle @xmath540 at the lower right in the right - hand side of the above equality belongs to @xmath536 . here , one can show the following equalities @xmath541 therefore , the proof is completed . one can check the following decomposition @xmath542{stringlink.pdf } } } = ( ( \varepsilon\otimes{\mathrm{id}}_\circ)\check{\tau}\otimes{\mathrm{id}}_\circ)({\mathrm{id}}_\bullet\otimes\psi_{\circ,\circ}^2)(\psi_{\bullet,\circ}^{-1}\otimes{\mathrm{id}}_\circ)(\tau\otimes{\mathrm{id}}_\circ)({\mathrm{id}}_\circ \otimes v_{-}\otimes{\mathrm{id}}_\circ),\ ] ] where the composition @xmath45 is omitted and the morphism @xmath543 is defined by @xmath544{ttau.pdf } } } = \psi_{\bullet,\circ}^{-1 } \circ \tau \circ \psi_{\bullet,\circ}\ .\ ] ] from now on , we assume an associator @xmath545 is the one derived from a rational even drinfeld series @xmath546 , which was introduced in ( * ? ? ? * section 3 ) . by the pentagon and hexagon relations and the evenness , @xmath547 must be of the form @xmath548 + ( \deg>3).\ ] ] thus , under the above assumption , we have @xmath549{ass23.pdf}}},{\raisebox{-1em}{\includegraphics[height=2.5em]{ass12.pdf}}}\right ) = { \raisebox{-1em}{\includegraphics[height=2.5em]{ass0.pdf } } } + \frac{1}{24}{\raisebox{-1em}{\includegraphics[height=2.5em]{ass123.pdf } } } + ( \deg>3).\ ] ] we can calculate the values on @xmath530 , @xmath531 , @xmath532 and @xmath522 using ( * ? ? ? * lemma 5.7 ) , and the results are written in table [ tab : value ] . @xmath550{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } + { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$2^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & -\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{y.pdf}\scriptsize \put ( 0,105){$1^+$ } \put ( 90,105){$2^+$ } \put ( 60,0){$1 ^ -$ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{yen.pdf}\scriptsize \put ( -5,105){$1^+$ } \put ( 45,105){$1^+$ } \put ( 90,105){$2^+$ } \put ( -5,-30){$1 ^ -$ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{myen.pdf}\scriptsize \put ( -5,105){$1^+$ } \put ( 45,105){$2^+$ } \put ( 90,105){$2^+$ } \put ( 90,-30){$1 ^ -$ } \end{overpic } } } \rule[-1.5em]{0em}{3.7em } \\ \hline \eta & 0 & 0 \rule[-0.7em]{0em}{2em } \\ \hline \delta & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } + { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & \frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{dy.pdf}\scriptsize \put ( 60,80){$1^+$ } \put ( 0,-30){$1 ^ -$ } \put ( 90,-30){$2 ^ -$ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{mlambda.pdf}\scriptsize \put ( 90,105){$1^+$ } \put ( -10,-30){$1 ^ -$ } \put ( 40,-30){$2 ^ -$ } \put ( 90,-30){$2 ^ -$ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lambda.pdf}\scriptsize \put ( -10,105){$1^+$ } \put ( -10,-30){$1 ^ -$ } \put ( 40,-30){$1 ^ -$ } \put ( 90,-30){$2 ^ -$ } \end{overpic } } } - \frac{1}{4}{\hspace{0.8em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{h.pdf}\scriptsize \put ( -50,80){$1^+$ } \put ( 60,80){$1^+$ } \put ( -50,0){$1 ^ -$ } \put ( 60,0){$2 ^ -$ } \end{overpic}}\hspace{0.6em } } \rule[-1.5em]{0em}{3.7em } \\ \hline \varepsilon & 0 & 0 \rule[-0.7em]{0em}{2em } \\ \hline s^{\pm1 } & -{\hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & \mp\frac{1}{4 } { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{phi.pdf}\scriptsize \put ( 40,80){$1^+$ } \put ( 40,0){$1 ^ -$ } \end{overpic}}\hspace{0.5em } } \mp\frac{1}{4}{\hspace{0.8em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{h.pdf}\scriptsize \put ( -50,80){$1^+$ } \put ( 60,80){$1^+$ } \put ( -50,0){$1 ^ -$ } \put ( 60,0){$1 ^ -$ } \end{overpic}}\hspace{0.6em } } \rule[-1.5em]{0em}{3.7em } \\ \hline \psi_{\bullet,\bullet}^{\pm1 } & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$2 ^ -$ } \end{overpic}}\hspace{1em } } + { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$2^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & \mp\frac{1}{2}{\hspace{0.8em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{h.pdf}\scriptsize \put ( -50,80){$1^+$ } \put ( 60,80){$2^+$ } \put ( -50,0){$2 ^ -$ } \put ( 60,0){$1 ^ -$ } \end{overpic}}\hspace{0.6em } } \rule[-1.2em]{0em}{3em } \\ \hline v_\pm & \mp\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{dstrut.pdf}\scriptsize \put ( -10,-5){$1 ^ -$ } \put ( 85,-5){$1 ^ -$ } \end{overpic}}\hspace{0.5em } } & \frac{1}{48 } { \raisebox{-0.7em } { \begin{overpic}[height=2em]{dphi.pdf}\scriptsize \put ( -10,-5){$1 ^ -$ } \put ( 80,-5){$1 ^ -$ } \end{overpic}}\hspace{0.3em } } \rule[-1em]{0em}{3em } \\ \hline y & 0 & -{\hspace{0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{uy.pdf}\scriptsize \put ( -10,80){$1^+$ } \put ( 40,80){$2^+$ } \put ( 90,80){$3^+$ } \end{overpic } } } + \frac{1}{2}{\raisebox{-0.9em } { \begin{overpic}[height=2em]{uh.pdf}\scriptsize \put ( -5,70){$1^+$ } \put ( 25,70){$1^+$ } \put ( 60,70){$2^+$ } \put ( 90,70){$3^+$ } \end{overpic } } } + \frac{1}{2}{\raisebox{-0.9em } { \begin{overpic}[height=2em]{uh.pdf}\scriptsize \put ( -5,70){$2^+$ } \put ( 25,70){$2^+$ } \put ( 60,70){$3^+$ } \put ( 90,70){$1^+$ } \end{overpic } } } + \frac{1}{2}{\raisebox{-0.9em } { \begin{overpic}[height=2em]{uh.pdf}\scriptsize \put ( -5,70){$3^+$ } \put ( 25,70){$3^+$ } \put ( 60,70){$1^+$ } \put ( 90,70){$2^+$ } \end{overpic } } } \rule[-1.5em]{0em}{3.7em } \\ \hline p_{u , v , w } & \sum\limits_{i=1}^{\|u^\bullet\|+\|v^\bullet\|+\|w^\bullet\| } \!{\hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$i^+$ } \put ( 35,0){$i^-$ } \end{overpic}}\hspace{1em } } & 0 \rule[-1.1em]{0em}{3em } \\ \hline \psi_{\bullet,\circ}^{\pm1 } & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & \pm\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{rt.pdf}\scriptsize \put ( 10,85){$1^+$ } \put ( 10,0){$1 ^ -$ } \put ( 45,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } + \frac{1}{8}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{rpi.pdf}\scriptsize \put ( 5,100){$1^+$ } \put ( 5,-20){$1 ^ -$ } \put ( 55,70){$1 ^ 0 $ } \put ( 55,10){$1 ^ 0 $ } \end{overpic}}\hspace{0.7em } } \rule[-1.5em]{0em}{3.7em } \\ \hline \psi_{\circ,\bullet}^{\pm1 } & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } & \pm\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{rt.pdf}\scriptsize \put ( 10,85){$1^+$ } \put ( 10,0){$1 ^ -$ } \put ( 45,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } + \frac{1}{8}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{rpi.pdf}\scriptsize \put ( 5,100){$1^+$ } \put ( 5,-20){$1 ^ -$ } \put ( 55,70){$1 ^ 0 $ } \put ( 55,10){$1 ^ 0 $ } \end{overpic}}\hspace{0.7em } } \rule[-1.5em]{0em}{3.7em } \\ \hline \psi_{\circ,\circ}^{\pm1 } & \pm\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{dstrut.pdf}\scriptsize \put ( -10,-5){$1 ^ 0 $ } \put ( 85,-5){$2 ^ 0 $ } \end{overpic}}\hspace{0.5em } } & -\frac{1}{32}{\hspace{-0.1em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{lrphi.pdf}\scriptsize \put ( -5,55){$1 ^ 0 $ } \put ( 85,55){$2 ^ 0 $ } \end{overpic}}\hspace{0.3em } } \pm\frac{1}{24}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lrh.pdf}\scriptsize \put ( -5,80){$1 ^ 0 $ } \put ( 80,80){$2 ^ 0 $ } \put ( -5,-15){$1 ^ 0 $ } \put ( 80,-15){$2 ^ 0 $ } \end{overpic } } } \rule[-1.2em]{0em}{3em}\\ \hline \tau & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } + { \raisebox{-0.7em } { \begin{overpic}[height=2em]{lustrut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( -5,-15){$1 ^ 0 $ } \end{overpic } } } & -\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lt.pdf}\scriptsize \put ( 55,80){$1^+$ } \put ( 55,0){$1 ^ -$ } \put ( 0,65){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } + \frac{1}{4}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{uchair.pdf}\scriptsize \put ( 40,105){$1^+$ } \put ( 90,105){$1^+$ } \put ( 90,-25){$1 ^ -$ } \put ( 0,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.7em } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{chair.pdf}\scriptsize \put ( 80,105){$1^+$ } \put ( 30,-25){$1 ^ -$ } \put ( 80,-25){$1 ^ -$ } \put ( 0,60){$1 ^ 0 $ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lpi.pdf}\scriptsize \put ( 60,100){$1^+$ } \put ( 60,-25){$1 ^ -$ } \put ( 0,85){$1 ^ 0 $ } \put ( 0,-10){$1 ^ 0 $ } \end{overpic}}\hspace{0.8em } } -\frac{1}{4}{\hspace{-0.1em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{luphi.pdf}\scriptsize \put ( 60,90){$1^+$ } \put ( -5,60){$1 ^ 0 $ } \end{overpic } } } + \frac{1}{4}{\hspace{0.4em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{luh.pdf}\scriptsize \put ( 30,105){$1^+$ } \put ( 80,105){$1^+$ } \put ( -30,45){$1 ^ 0 $ } \put ( -30,0){$1 ^ 0 $ } \end{overpic } } } \rule[-1.5em]{0em}{3.7em}\\ \hline \tau^{-1 } & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } -{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lustrut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( -5,-15){$1 ^ 0 $ } \end{overpic } } } & \frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lt.pdf}\scriptsize \put ( 55,80){$1^+$ } \put ( 55,0){$1 ^ -$ } \put ( 0,65){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } -\frac{1}{4}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{uchair.pdf}\scriptsize \put ( 40,105){$1^+$ } \put ( 90,105){$1^+$ } \put ( 90,-25){$1 ^ -$ } \put ( 0,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.7em } } -\frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{chair.pdf}\scriptsize \put ( 80,105){$1^+$ } \put ( 30,-25){$1 ^ -$ } \put ( 80,-25){$1 ^ -$ } \put ( 0,60){$1 ^ 0 $ } \end{overpic } } } + \frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lpi.pdf}\scriptsize \put ( 60,100){$1^+$ } \put ( 60,-25){$1 ^ -$ } \put ( 0,85){$1 ^ 0 $ } \put ( 0,-10){$1 ^ 0 $ } \end{overpic}}\hspace{0.8em } } \rule[-1.5em]{0em}{3.7em}\\ \hline \beta^{\pm1 } & { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$1^+$ } \put ( 35,0){$1 ^ -$ } \end{overpic}}\hspace{1em } } \pm{\raisebox{-0.7em } { \begin{overpic}[height=2em]{ldstrut.pdf}\scriptsize \put ( 35,-5){$1 ^ -$ } \put ( -5,90){$1 ^ 0 $ } \end{overpic } } } & \mp\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lt.pdf}\scriptsize \put ( 55,80){$1^+$ } \put ( 55,0){$1 ^ -$ } \put ( 0,65){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } -\frac{1}{8}{\hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{ldphi.pdf}\scriptsize \put ( 60,-10){$1 ^ -$ } \put ( -5,65){$1 ^ 0 $ } \end{overpic } } } \pm\frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{uchair.pdf}\scriptsize \put ( 40,105){$1^+$ } \put ( 90,105){$1^+$ } \put ( 90,-25){$1 ^ -$ } \put ( 0,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.7em } } -\frac{1}{12}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{lpi.pdf}\scriptsize \put ( 60,100){$1^+$ } \put ( 60,-25){$1 ^ -$ } \put ( 0,85){$1 ^ 0 $ } \put ( 0,-10){$1 ^ 0 $ } \end{overpic}}\hspace{0.8em } } \pm\frac{1}{8}{\hspace{0.4em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{ldh.pdf}\scriptsize \put ( 30,-30){$1 ^ -$ } \put ( 80,-30){$1 ^ -$ } \put ( -30,90){$1 ^ 0 $ } \put ( -30,45){$1 ^ 0 $ } \end{overpic } } } \rule[-1.5em]{0em}{3.7em}\\ \end{array } $ ] on the other hand , to calculate the value on @xmath551 , we need lemma [ lem : updown ] that is an upside - down version of ( * ? ? ? * lemma 5.5 ) . one can prove lemma [ lem : updown ] in a similar way . [ lem : updown ] let @xmath552 . suppose @xmath553 is the composition of the @xmath8-tangle @xmath554{u.pdf}}}\ ] ] and some @xmath8-tangle @xmath167 in @xmath168 ^ 3 $ ] . then @xmath555 is equal to the image by @xmath556 of the composition of the series of jacobi diagrams @xmath557{v.pdf}}}\ ] ] and @xmath558 in the category @xmath314 . here , @xmath559 and a directed rectangle is same as in ( * ? ? ? * lemma 5.5 ) . the lower degree terms of @xmath560 and @xmath561 are given in ( * ? ? ? * lemma 5.7 ) and its proof . le @xcite proved that the lmo invariant is universal among rational - valued finite - type invariants of rational homology spheres ( see ( * ? ? ? * , remark 1 ) ) . as mentioned before , cheptea , habiro and massuyeau showed that the lmo functor is universal among rational - valued finite - type invariants of 3-manifolds . we prove that the extension @xmath10 has the same property . in this subsection , we prepare terminology of clasper calculus based on @xcite . consider pairs @xmath562 where @xmath4 is a compact oriented 3-manifold and @xmath140 is a framed oriented tangle in @xmath4 . if @xmath4 has a boundary , we take into account a parametrization of @xmath184 . moreover , if @xmath140 has a boundary , it must be attached to assigned points in @xmath184 . a _ graph clasper _ is an oriented compact surface with a certain decomposition into three kinds of pieces : disks , bands and annuli , which are called _ nodes _ , _ edges _ and _ leaves _ respectively . each leaf should be connected to a node and no leaf by a band . each node should be connected to nodes or leaves by exactly three bands . the surface on the left of figure [ fig : ygraph ] is one of the simplest example of a graph clasper . for simplicity , it is represented as the graph on the right of figure [ fig : ygraph ] , that is , leaves , nodes and bands of a graph clasper @xmath3 are replaced with circles , points and arcs respectively , here we should remember information required for recovering @xmath3 . next , consider surgery along a graph clasper @xmath3 in a 3-manifold @xmath4 . let @xmath563 denotes the graph clasper obtained from @xmath3 by applying the _ fission rule _ illustrated in figure [ fig : fissionrule ] to each edge connecting two nodes . the resulting graph is the disjoint union of @xmath564 copies of @xmath27-graphs , where @xmath564 is the _ internal degree _ of @xmath3 , that is , the number of nodes of @xmath3 . the pair @xmath565 obtained from @xmath562 by surgery along @xmath3 is defined as follows : consider the case in which @xmath3 is a @xmath27-graph . let @xmath566 be a @xmath27-graph in @xmath43 and @xmath567 be its neighborhood , namely a genus three handlebody . then there is an orientation - preserving homeomorphism @xmath568 sending @xmath566 to @xmath3 . here , we replace @xmath566 with the six - components framed link drawn in figure [ fig : ygraphsurgery ] . we denote by @xmath569 the manifold obtained from @xmath3 by surgery along the framed link in @xmath570 corresponding to the above link , and let @xmath571 denote the image of @xmath140 in @xmath569 . in the general case , we perform surgery along each @xmath27-graph of @xmath563 . -graph , its neighborhood and the corresponding link ] let @xmath572 . a pair @xmath562 is _ @xmath573-equivalent _ to @xmath574 if there exists a graph clasper @xmath3 such that the internal degree of each connected component is equal to @xmath575 and @xmath3 satisfies @xmath576 . it is known that the @xmath573-equivalence is an equivalence relation ( see , for example , ( * ? ? ? * theorem 3.2 ) ) . let us fix a @xmath577-equivalence class @xmath578 of 3-manifolds with tangles ( see ( * ? ? ? * theorem 7.6 ) ) . we define @xmath579 to be the @xmath22-vector space spanned by @xmath580 \mid m \in { \mathcal{m}}^0,\ { \operatorname{i - deg}}g = k\}$ ] . here , @xmath581 $ ] is defined by @xmath582 : = \sum_{g ' \subset g}(-1)^{\|g'\|}m_{g ' } \in { \mathbb{q}}{\mathcal{m}}^0,\ ] ] where @xmath583 runs over all connected components of @xmath3 , and @xmath584 is the number of connected components of @xmath583 . ( the above sum consists of @xmath585 terms . ) we have the _ @xmath27-filtration _ @xmath586 and we set @xmath587 . moreover , the associated graded vector space @xmath588 is defined by @xmath589 . let @xmath590 be a @xmath22-vector space and let @xmath591 . a map @xmath592 is a ( @xmath590-valued ) _ finite - type invariant _ of _ degree _ @xmath593 if the linear extension @xmath594 is non - trivial on @xmath595 and trivial on @xmath596 . to state theorem [ thm : gar02 ] below , we review the surgery map @xmath597 according to @xcite . fix a @xmath577-equivalence class @xmath578 of @xmath598 and let @xmath486 , @xmath485 , @xmath599 . let @xmath163 be a jacobi diagram in the vector space @xmath600 . we first construct an oriented surface @xmath601 by replacing the internal vertices , external vertices and edges to disks , annuli and bands respectively , where the vertex orientation of @xmath163 induces the orientation of disks , and the bands connect the disks so that the orientations are compatible . for each annulus , @xmath602 , the core of @xmath602 is defined to be the push - off into @xmath603 of the outer boundary of @xmath602 that is the boundary connecting with an edge . namely , the core is an oriented simple closed curve on @xmath602 . next , we take a @xmath8-cobordism @xmath487 . the graph clasper @xmath604 is defined to be the image of an embedding @xmath601 into @xmath4 such that each annulus corresponding to the @xmath605- ( resp . @xmath606- , @xmath607- ) colored vertices is the push - off into @xmath608 of a ( unoriented ) neighborhood of @xmath609 ( resp . @xmath610 , @xmath611 ) and its core corresponds to the oriented curve @xmath612 ( resp . @xmath610 , @xmath611 ) . here , @xmath604 is called a _ topological realization _ of a jacobi diagram @xmath163 . the manifold @xmath613 depends on the choice of the cobordism @xmath63 and the topological realization @xmath604 . however , it is known that @xmath614 \in { \mathcal{g}}_i({\mathcal{m}}^0)$ ] is independent of the choice , and the following theorem holds . [ thm : gar02 ] the graded linear map @xmath615 defined by @xmath616 $ ] is well - defined and surjective . using this theorem , we prove the next theorem saying that @xmath10 is universal among rational - valued finite - type invariants . the following proof is an obvious extension of the proof of ( * ? ? ? * theorem 7.11 ) . [ thm : universal ] let @xmath617 . the map @xmath618 is a finite - type invariant of degree at most @xmath209 . moreover , the induced map @xmath619 satisfies @xmath620 where @xmath621 , @xmath622 and @xmath35 is the number of internal edges of @xmath163 . consequently , @xmath597 and @xmath623 are isomorphisms , and @xmath624 is a finite - type invariant of degree @xmath209 . let us prove that @xmath625 is a finite - type invariant of degree at most @xmath626 . we first take a graph clasper @xmath3 in a cobordism @xmath627 such that @xmath581 \in{\mathcal{f}}_i({\mathcal{m}}^0)$ ] . let @xmath628 , @xmath629 and @xmath630 . then we have @xmath631 & = \sum_{j \subset{\lfloor{i}\rceil } } ( -1)^{\|j\| } m_{\bigsqcup_{j\in j } g_j } \\ & = \sum_{g'\subset g } ( -1)^{\|g\setminus g'\|+\|y(g\setminus g')\| } ( -1)^{\|y(g')\| } m_{g ' } \\ & = \sum_{g'\subset g } ( -1)^{c-\|g'\|+i } m_{g ' } = ( -1)^{i+c}[m , g],\end{aligned}\ ] ] where the second equality follows from the equality @xmath632 and the fact that surgery along a part of a connected component is same as doing nothing . using the transformation depicted in figure [ fig : inserthopflink ] , the bottom - top tangle @xmath633 can be expressed as figure [ fig : universal ] , where @xmath634 is some cobordism of @xmath635 . moreover , @xmath636 are the bottom - top tangles illustrated in figure [ fig : ybyclasper ] , which are same as the cobordism @xmath27 listed in table [ tab : gen ] . by using a @xmath27-graph ] by using a @xmath27-graph ] then we have , by the functoriality of @xmath10 , @xmath637 ) = { \widetilde{z}}([m , y(g ) ] ) \\ & = \left({\widetilde{z}}({\mathrm{id}}_v ) \otimes \bigotimes_{j=1}^i \left ( { \widetilde{z}}((\varepsilon\otimes\varepsilon)\otimes\varepsilon ) -{\widetilde{z}}(y_j ) \right ) \right ) \circ { \widetilde{z}}(r ) \notag \\ & = \left ( \left[\sum_{k=1}^f { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$k^+$ } \put ( 35,0){$k^-$ } \end{overpic}}\hspace{1em}}\right ] \sqcup \bigsqcup_{j=1}^i \left(-{\widetilde{z}}_1(y_j ) + ( { \operatorname{i - deg}}>1)\right ) \right ) \circ { \widetilde{z}}(r ) . \notag \ ] ] since the composition in @xmath282 preserves the filtration @xmath638 , we conclude @xmath639)=0 $ ] . next , we prove the equality in the second statement . we take a jacobi diagram @xmath640 and define @xmath3 to be a topological realization @xmath604 in some cobordism @xmath4 . the goal is to show @xmath641)=(-1)^{i+c+e}d$ ] . we first define three sets by @xmath642 since the leaf corresponding to @xmath643 arises from some external vertex of @xmath163 , we define @xmath644 to be its color . on the other hand , @xmath645 is divided into @xmath646 and @xmath647 , where @xmath648 and @xmath649 . we now have @xmath650 where @xmath651 is the matrix unit . it follows from this equality , and lemma [ lem : composition ] that @xmath652 ) \\ & = { \left\langle { \left ( \bigsqcup_{j=1}^i -{\widetilde{z}}_1(y_j ) \middle/ \parbox{7.8em}{$k^+ \mapsto k^\ast + b^{--}k^{- } \\\hspace{2.5em } + b^{+-}k^{-}$ } \right ) , \left ( [ b^{--}/2+b^{-0 } ] \middle/ k^{-}\mapsto k^\ast \right ) } \right\rangle}_{{\lfloor{f+3i}\rceil}^\ast } \\ & = { \left\langle { \left ( \bigsqcup_{j=1}^i -{\widetilde{z}}_1(y_j ) \middle/ \parbox{11em}{$l^+ \mapsto \begin{cases } c(l ) & \text{if $ l\in{\mathcal{o}}$ } , \\ l^\ast & \text{otherwise . } \end{cases}$ } \right ) , \bigsqcup_{(t_1,t_2)\in{\mathcal{t } } _ < } -{\hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$t_2^\ast$ } \put ( 35,0){$t_1^\ast$ } \end{overpic}}\hspace{1em } } } \right\rangle}_{\{l^\ast \mid l\in{\mathcal{l}}\setminus{\mathcal{o}}\ } } \\ & = ( -1)^e d,\end{aligned}\ ] ] where the second equality follows from the following argument : since the linear combination on the left in the angular brackets contains no @xmath653-colored vertex , @xmath654 $ ] can be moved to the left as lemma [ lem : precomposition ] . we focus on cobordisms derived from knots with additional information . we first review a natural construction of a cobordism from a knot according to ( ? ? ? * section 2.2 ) . let @xmath27 be a homology sphere and let @xmath164 be an oriented framed knot in @xmath27 . the boundary of @xmath655 is a torus with the oriented longitude determined by the orientation and framing of @xmath164 . here , let @xmath656 be a homeomorphism that maps the oriented simple closed curves @xmath657 and @xmath658 depicted in figure [ fig : cfk11 ] to the meridian and longitude of @xmath164 , respectively . then we obtain @xmath659 . let @xmath664 . then the closure @xmath665 defined in section [ subsec : bttangle ] coincides with the pair @xmath666 . therefore , @xmath667 ^ 3,\gamma ) \\ & = { z^{\mathrm{k\text{-}lmo}}}(y\setminus{\operatorname{int}}[-1,1]^3,k ) \sharp \nu^{-1}.\end{aligned}\ ] ] the proof is completed . since @xmath10 is a functor , we have @xmath668 here , @xmath669 can be interpreted as a certain connected sum defined as follows : consider an embedding between pairs @xmath670 ^ 3,0\times0\times[-1,1 ] ) \to ( y_i , k_i)$ ] that preserves the orientations and framing . then the above composition is equal to the pair @xmath671 ^ 3 \sqcup ( y_2,k_2)\setminus{\operatorname{int}}[-1,1]^3 \right ) / \iota_1(x , y , z)\sim\iota_2(x , y ,- z).\ ] ] let @xmath27 be an integral homology sphere and let @xmath167 be an oriented link in @xmath27 . one can take a seifert surface @xmath672 of @xmath167 , and let @xmath673 be its genus . we fix an orientation - preserving homeomorphism @xmath674 , where @xmath39 satisfies @xmath133 , @xmath675 . using the above information , we construct the cobordism @xmath676 ( that is not necessarily lagrangian ) as follows : define @xmath4 to be the manifold obtained from @xmath677 by cutting along @xmath672 . now , @xmath184 is the union of the surface @xmath678 to which the normal vectors point , @xmath679 and the @xmath168\times l$ ] . therefore , we decide @xmath680 and @xmath681 by @xmath95 , and the rest is uniquely determined by the meridians of @xmath167 . let @xmath167 be the negative hopf link in @xmath682 . since we need a seifert surface that is easy to see , we represent the pair @xmath683 by surgery along the @xmath684-framed unknot depicted in figure [ fig : hopflinkseifert ] . one can now easily construct a seifert surface by avoiding the surgery knot , and we give the parametrization indicated by the line drawn in the middle of figure [ fig : hopflinkseifert ] . then we have the lagrangian cobordism @xmath4 illustrated in figure [ fig : hopflinkseifert ] , where the oriented closed curve represents @xmath685 . hence , @xmath686 is exactly equal to @xmath687{arrowarc.pdf}}}\right)$ ] . it follows from ( * ? ? ? * exercise 5.4 ) that @xmath688{dstrut.pdf}\scriptsize \put ( -10,-5){$1 ^ 0 $ } \put ( 85,-5){$1 ^ 0 $ } \end{overpic}}\hspace{0.5em } } \right ) \right ) = \left[\frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{dstrut.pdf}\scriptsize \put ( -10,-5){$1 ^ 0 $ } \put ( 85,-5){$1 ^ 0 $ } \end{overpic}}\hspace{0.5em } } + \frac{1}{48}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{dphi.pdf}\scriptsize \put ( -10,-5){$1 ^ 0 $ } \put ( 80,-5){$1 ^ 0 $ } \end{overpic}}\hspace{0.3em } } + ( { \operatorname{i - deg}}>2)\right].\end{aligned}\ ] ] habegger and masbaum @xcite proved that the first non - vanishing milnor invariant of a string link appears in the tree reduction of its kontsevich invariant . the same is true for the kontsevich - lmo invariant of string links in a homology cube , which was shown in @xcite . moreover , the same holds for the lmo functor of the cobordisms derived from string links in a homology cube , which was proven in @xcite . we prove that the same is true for the extension of the lmo functor . a _ string link _ on @xmath689 strands is an orientation - reversed bottom - top tangle of type @xmath690 , where @xmath32 is a word in @xmath691 of length @xmath689 . namely , each strand runs from @xmath692 to @xmath151 . actually , we are interested in the monoid @xmath693 let @xmath694 . then @xmath695 is a cobordism from @xmath42 to @xmath42 . however , in this subsection , we use the surface @xmath696 with the oriented simple closed curves @xmath697 as in figure [ fig : d ] instead of @xmath42 with the oriented simple closed curves @xmath698 note that the orientation of @xmath34 differs from the one of @xmath699 . let @xmath701 denote the fundamental group @xmath702 and @xmath703 is its lower central series defined by @xmath704 $ ] , @xmath705 . since the continuous maps @xmath706 ) \xrightarrow{s } s\ ] ] induce isomorphisms on their first homologies and surjections on their second homologies ( ( * ? ? ? * theorem 5.1 ) ) , they induce isomorphisms @xmath707 for all @xmath572 . the @xmath575th _ artin representation _ is the monoid anti - homomorphism @xmath708 defined by @xmath709 . we set @xmath710:={\operatorname{ker}}a_k$ ] , and the filtration @xmath711 \supset { \mathcal{s}}_l[2 ] \supset \dotsb\ ] ] is obtained . from now on , we suppose that each strand of @xmath52 is 0-framed , and let @xmath712 be the preferred longitude of the @xmath209th strand of @xmath52 . it is well known that @xmath694 belongs to @xmath710 $ ] if and only if @xmath713 for all @xmath209 ( see , for example , ( * ? ? ? * remark 5.1 ) ) . next , we regard @xmath719 as a linear combination of connected tree jacobi diagrams . the subalgebra of @xmath247 generated by tree jacobi diagrams is denoted by @xmath720 , which is identified with the quotient by the ideal generated by looped jacobi diagrams , and the image of @xmath254 by the quotient map is denoted by @xmath721 . moreover , the subspace of @xmath720 spanned by connected jacobi diagrams is denoted by @xmath722 . the free lie algebra over @xmath22 generated by a set @xmath233 is denoted by @xmath723 and its degree @xmath53 part is denoted by @xmath724 . we define the linear map @xmath725 by @xmath726 where @xmath324 runs over all external vertices in @xmath163 , and @xmath727 is the tree rooted at @xmath324 . here , @xmath728 is defined as follows : @xmath729{tree - crop.pdf } } } \right ) = [ c_1,[[c_2,c_3],c_4]].\ ] ] moreover , it is well known that the sequence @xmath730 } { \operatorname{lie}}_{k+1}(c ) \to 0\end{aligned}\ ] ] is exact ( see , for example , ( * ? ? ? * theorem 1 ) ) . let @xmath3 be the free group generated by @xmath733 and let @xmath734 be its lower central series . recall the well - known fact that there exists a natural isomorphism @xmath735 . in particular , if @xmath736 , then one can show that @xmath737 \circ \alpha_k \circ \mu_k = 0 $ ] . the exact sequence enable us to define @xmath738 to be the unique element sent to @xmath739 by @xmath740 . finally , we discuss a relation between cobordisms and string links . let @xmath39 and let @xmath486 , @xmath599 , @xmath741 . figure [ fig : milnorjohnson ] represents a map from @xmath742 to @xmath743 , which sends @xmath744 to the trivial string link on @xmath689 strands , which is an extension of the original one defined in @xcite , @xcite . the _ milnor - johnson correspondence _ @xmath745 is the bijection from the @xmath577-equivalence class of @xmath744 in @xmath746 to the @xmath577-equivalence class of the trivial string link in @xmath743 , depicted in figure [ fig : milnorjohnson ] . let us introduce a map @xmath747 for @xmath39 to state the main theorem in this section . we first define the bijection @xmath748 by sending the @xmath210th color in the sequence @xmath749 to @xmath750 . next , @xmath751 is defined by @xmath752 for example , if @xmath753 , then we have @xmath754 [ lem : psi ] there exists a linear map @xmath755{180}{$\curvearrowleft$}}^{{\lfloor{g}\rceil}^+ } \curvearrowleft^{{\lfloor{g}\rceil}^- } \downarrow^{{\lfloor{n}\rceil}^0 } ) \to { \mathcal{a}}(\downarrow^{{\lfloor{2g+n}\rceil}^\ast})\ ] ] such that @xmath756 . moreover , @xmath757 satisfies @xmath758{strut.pdf}\scriptsize \put ( 35,80){$\rho_w(i^+)$ } \put ( 35,0){$-\rho_w(i^-)$ } \end{overpic}}\hspace{1em } } \qquad \right ] \sqcup r_w(\widetilde{e } ) + ( { \operatorname{i - deg}}>{\operatorname{i - deg}}e),\ ] ] for any jacobi diagram @xmath759{180}{$\curvearrowleft$}}^{{\lfloor{g}\rceil}^+ } \curvearrowleft^{{\lfloor{g}\rceil}^- } \downarrow^{{\lfloor{n}\rceil}^0})$ ] . the string link @xmath760 can be transformed as illustrated in figure [ fig : psi ] . therefore , we set @xmath761 where @xmath762 is the element obtained from @xmath296 by applying the orientation reversal map @xmath272 to the @xmath763-indexed components of @xmath296 . then , @xmath757 has the desired property . [ thm : milnorinv ] let @xmath764 , @xmath765 , @xmath716 $ ] and let @xmath39 such that @xmath766 . then @xmath767 is equal to @xmath768 conversely , if @xmath769 $ ] and @xmath770 is of the form @xmath771 then @xmath15 belongs to @xmath710 $ ] , and @xmath772 is equal to @xmath773 . let @xmath716 $ ] and let @xmath40 , @xmath41 . we set @xmath774 . let us prove the former by induction on @xmath575 . we first assume that it is true up to @xmath775 , where @xmath776 . ( the case @xmath777 is mentioned later . ) it follows from lemma [ lem : mu ] that @xmath686 is written as @xmath778 \sqcup \left(\varnothing + x + ( \text{${\operatorname{i - deg}}\ge k$ or looped } ) \right),\ ] ] where @xmath779 . hence , lemma [ lem : composition ] implies @xmath780 \sqcup { \left\langle{\left ( x + ( \text{${\operatorname{i - deg}}\ge k$ or looped } ) \middle/ \parbox{3em}{$i^{+}\mapsto \\ i^\ast + i^{+}$}\right ) , \left ( ( { \mathsf{t}}_g^{-1})^y \middle/ \parbox{3em}{$i^{-}\mapsto \\ i^\ast + i^{-}$ } \right)}\right\rangle}_{{\lfloor{g}\rceil}^\ast } \\ & = { \mathsf{t}}_g^{-1 } + [ i_g^{+- } ] \sqcup x + ( \text{${\operatorname{i - deg}}\ge k$ or looped}).\end{aligned}\ ] ] by applying the composite map @xmath781 , we have @xmath782{strut.pdf}\scriptsize \put ( 35,80){$\rho_w(i^+)$ } \put ( 35,0){$\rho_w(i^-)$ } \end{overpic}}\hspace{1em } } \qquad \right ] \sqcup r_w\left ( \widetilde{\chi([i_g^{+- } ] \sqcup x ) } \right ) + \left(\parbox{4.3em}{${\operatorname{i - deg}}\ge k$ \\ \text{or looped}}\right ) \\ & = \chi^{-1}\psi(z({\mathrm{id}}_w ) ) + \left[\sum_{i=1}^g { \hspace{-0.2em}\raisebox{-0.7em } { \begin{overpic}[height=2em]{strut.pdf}\scriptsize \put ( 35,80){$\rho_w(i^+)$ } \put ( 35,0){$\rho_w(i^-)$ } \end{overpic}}\hspace{1em } } \qquad \right ] \sqcup r_w\left ( [ i_g^{+- } ] \sqcup x \right ) + \left(\parbox{4.3em}{${\operatorname{i - deg}}\ge k$ \\ \text{or looped}}\right ) \\ & = \varnothing + r_w(x ) + ( \text{${\operatorname{i - deg}}\ge k$ or looped}),\end{aligned}\ ] ] where we use lemma [ lem : psi ] and the fact that @xmath783 . on the other hand , by ( * theorem 3 ) , @xmath784 comparing the internal degree @xmath575 parts of each tree reduction , we conclude @xmath785 . first , one can check that @xmath792 , where @xmath519 and @xmath518 are the cobordisms define in table [ tab : gen ] . it follows from the functoriality of @xmath10 , lemma [ lem : composition ] and table [ tab : value ] that @xmath793{rt.pdf}\scriptsize \put ( 10,85){$1^+$ } \put ( 10,0){$1 ^ -$ } \put ( 45,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } + \frac{1}{2}{\raisebox{-0.7em } { \begin{overpic}[height=2em]{rt.pdf}\scriptsize \put ( 10,85){$1^+$ } \put ( 10,0){$1 ^ -$ } \put ( 45,60){$1^{0'}$ } \end{overpic}}\hspace{0.4em } } + ( { \operatorname{i - deg}}>1 ) \right ) \\ & = \varnothing + { \raisebox{-0.7em } { \begin{overpic}[height=2em]{rt.pdf}\scriptsize \put ( 10,85){$1^+$ } \put ( 10,0){$1 ^ -$ } \put ( 45,60){$1 ^ 0 $ } \end{overpic}}\hspace{0.4em } } + ( { \operatorname{i - deg}}>1).\end{aligned}\ ] ] thus the previous theorem implies @xmath794{rt.pdf}\scriptsize \put ( 10,85){$2^\ast$ } \put ( 10,0){$1^\ast$ } \put ( 45,60){$3^\ast$ } \end{overpic}}\hspace{0.4em}}$ ] . therefore , @xmath795 + x_2\otimes[x_3,x_1 ] + x_3\otimes[x_1,x_2 ] \in \varpi/\varpi_2 \otimes \varpi_2/\varpi_3.\ ] ] k. habiro and g. massuyeau . from mapping class groups to monoids of homology cobordisms : a survey . in _ handbook of teichmller theory . volume iii _ , volume 17 of _ irma lect . math . _ , pages 465529 . soc . , zrich , 2012 . t. t. q. le . an invariant of integral homology @xmath797-spheres which is universal for all finite type invariants . in _ solitons , geometry , and topology : on the crossroad _ , volume 179 of _ amer . math . ser . 2 _ , pages 75100 . soc . , providence , ri , 1997 . j. levine . addendum and correction to : `` homology cylinders : an enlargement of the mapping class group '' [ algebr . geom . topol . * 1 * ( 2001 ) , 243270 ; mr1823501 ( 2002m:57020 ) ] . , 2:11971204 ( electronic ) , 2002 .	cheptea , habiro and massuyeau constructed the lmo functor , which is defined on a certain category of cobordisms between two surfaces with at most one boundary component . in this paper , we extend the lmo functor to the case of any number of boundary components , and our functor reflects relations among the parts corresponding to the genera and boundary components of surfaces . we also discuss a relationship with finite - type invariants and milnor invariants .
the pierre auger observatory is situated in the argentinian province mendoza , close to the city of malarge . it consists of a 3000 km@xmath1 surface detector array and a set of 24(+3 ) fluorescence telescopes . the surface detector stations are water cherenkov tanks each equipped with 3 photomultipliers measuring light induced by electromagnetic particles and muons . six fluorescence telescopes occupy one fluorescence detector building . in total four of these buildings are located on the array border on small hills and thus overlook the interior of the array measuring the longitudinal profile of electromagnetic shower @xcite . already in the year 2008 the observatory was fully completed with successful operation of all four fluorescence detector buildings and by fulfilling the original aim of 1600 deployed and working surface detector stations . during the next years several observatory enhancements were built such as 3 high elevation atmospheric telescopes ( heat ) and amiga infill both used to to lower the energy threshold below 10@xmath0 ev . the pierre auger collaboration reports many results in the field of ultra - high energy cosmic rays ( uhecr ) . the energy spectrum with the observation of the rapid flux suppression above 5@xmath210@xmath3 ev @xcite , various studies of expected anisotropy of ultra - high energy events ( eg . @xcite ) or the estimates of the upper limits on the cosmic - ray photon and the diffuse neutrino fluxes @xcite are among the most important scientific outcomes of this project . in this contribution we rather focus on the results that are related to particle physics . namely , we concentrate on the estimation of the proton - air cross - section , the connection between the analysis of chemical composition of cosmic rays ( cr ) and the hadronic interaction models . we focus also on the muon deficit in current hadronic interaction models . the number of particles in an extensive air shower ( eas ) as a function of the atmospheric slant depth ( the amount of atmosphere traversed from its upper edge in g / cm@xmath1 ) is called the shower longitudinal profile . most of the eas energy is released via the electromagnetic sub - shower . therefore , as in the electromagnetic calorimeter the shower size increases until the average energy of the e@xmath5 in the eas is about the critical energy . the slant depth at which the longitudinal profile of a shower reaches its maximum is denoted as x@xmath6 and it is one of the primary observables measured by the fluorescence detector . measurements of x@xmath6 can be used both to get insight into the composition of primary cosmic ray particles ( showers induced by light particles penetrate deeper in the atmosphere than the showers originating from heavier nuclei ) and also to measure the proton - air cross - section . the differences in x@xmath6 between showers of the same primary energy and same primary particle type are due to fluctuations in hadronic interactions . for purely proton primaries , the x@xmath6 distribution is a convolution of the fluctuations in the shower development from the point of the first interaction to the shower maximum ( dependent on the hadronic interactions ) and the exponential distribution of the depth of the first interaction . for this reason the fitted slope @xmath7 of the tail of the x@xmath6 distribution is ( inversely ) proportional to the inelastic cross - section of proton - air ( see fig . [ crossair ] - left ) . the data were selected in the energy range 10@xmath0 @xmath8 10@xmath9 ev where the proton component is expected to dominate cr chemical composition . to avoid biases in the measured x@xmath6 distribution fiducial volume cuts based on the shower geometry were applied . the largest source of systematic uncertainties is the lack of knowledge of the helium component . uncertainties due to model assumptions were also addressed . for details see @xcite . the value @xmath10 mb of inelastic proton - air cross - section was finally obtained at laboratory energy e@xmath11=10@xmath12 ev and compared to results of previous cr experiments ( fig . [ crossair ] - right ) . ) of the x@xmath13 distribution tail and the inelastic cross - section for several hadronic interaction models . the estimated slope is obtained by fitting the tail of the x@xmath13 distributions by @xmath14 . right : resulting inelastic proton - air cross - section compared to other measurements and several hadronic interaction models.,title="fig:",scaledwidth=43.0% ] ) of the x@xmath13 distribution tail and the inelastic cross - section for several hadronic interaction models . the estimated slope is obtained by fitting the tail of the x@xmath13 distributions by @xmath14 . right : resulting inelastic proton - air cross - section compared to other measurements and several hadronic interaction models.,title="fig:",scaledwidth=45.0% ] using the glauber model ( with intermediate inelastic states included ) the proton - air cross - section can be converted to proton - proton inelastic cross - section ( @xmath15 ) at @xmath16 tev . the conversion is illustrated in fig . [ glaub ] ( left ) in the @xmath15 , @xmath17 plane where b@xmath18 is the elastic slope . the obtained value of @xmath15 is compared to results of lhc experiments and model predictions in fig . [ glaub ] ( right ) . , and the inelastic proton - proton cross section in the glauber model . the solid line shows combinations of the parameters that yield the observed proton - air production cross section , and the dotted lines are the statistical uncertainties . the hatched area corresponds to the predictions by sibyll , qgsjet , qgsjetii , and epos . right : derived @xmath15 together with model predictions and accelerator data.,title="fig:",scaledwidth=45.0% ] , and the inelastic proton - proton cross section in the glauber model . the solid line shows combinations of the parameters that yield the observed proton - air production cross section , and the dotted lines are the statistical uncertainties . the hatched area corresponds to the predictions by sibyll , qgsjet , qgsjetii , and epos . right : derived @xmath15 together with model predictions and accelerator data.,title="fig:",scaledwidth=45.0% ] quantitative estimates of the mass composition of cosmic rays deduced from measured mass sensitive observables can be obtained only when distributions of these quantities are compared to predictions of hadronic interaction models for different primary particle types . hadronic interaction models , when used to describe uhecr showers , naturally rely on uncertain extrapolations to an order of magnitude larger centre - of - mass energies than are the energies currently achievable at accelerators . even now when hadronic interaction models are extensively tested and tuned at the lhc it is not obvious at which point the models reliability would be such that they can be universally used to unambiguously interpret uhecr shower data in terms of cr composition . for this reason the statements concerning the mass composition of cosmic rays have large uncertainties . the behavior of the x@xmath6 distributions ( e.g. their main characteristics @xmath19 and @xmath20(x@xmath6 ) ) strongly suggests that at energy @xmath21 10@xmath22 ev the mean mass of cr components starts to increase @xcite ( see fig . [ xmax ] ) . this conclusion is valid whatever model of hadronic interactions is compared to data of the pierre auger observatory . in other words the only alternative explanation other than a sudden change of chemical composition deduced from the behavior of x@xmath6 is that the hadronic interactions change at that energy and that the important parameters inside the hadronic interaction models ( such as the cross - section ) must behave very differently than expected . ( left ) and @xmath20(x@xmath6 ) ( right ) as a function of energy @xcite . measurements are from the hybrid data set . data ( points ) are shown with the predictions for protons and iron nuclei as primary particles for several hadronic interaction models . , scaledwidth=90.0% ] the problems of the measurement of uhecr chemical composition and studies of hadronic interaction models are thus coupled subjects which shall be addressed at the same time . the cross - study of both problems using the first two moments of the x@xmath6 distribution ( @xmath23 and @xmath20(x@xmath6 ) ) is presented in @xcite . the situation gets even more complicated when other observables related to chemical composition are included . namely , it was shown @xcite that current models of hadronic interactions predict fewer muons at ground than what is actually observed at the pierre auger observatory . figure [ muons ] shows an estimation of the relative number of muons in data normalized to predictions of qgsjetii-03 ( fig . [ muons ] left ) and qgsjetii-04 ( fig . [ muons ] right ) for proton primary particles at 10@xmath3 ev . the data are compared to predictions of two post - lhc interaction models qgsjetii-04 and epos - lhc for proton and iron nuclei as primary particles . the figures summarize the results of independent studies , one done for inclined showers , with zenith angles larger than 62@xmath24 @xcite , and another for showers with zenith smaller than 60@xmath24 @xcite and using two different methods based on the time - structure of the surface detector signals to estimate the muon content of the shower . the mentioned independent methods together with the analysis of hybrid events @xcite consistently indicate that the hadronic interaction models predict smaller size of the muon component than what is observed in the data unless pure iron composition is assumed . such a heavy composition would be in contradiction with the x@xmath6 data when interpreted using the same models . this leads to the conclusion that shower models do not correctly describe the muonic ground signal . ev ) r@xmath25/(e@xmath26/10@xmath3 ev ) relative to prediction of qgsjetii-03 ( protons ) as a function of shower energy . theoretical curves for proton and iron showers simulated with qgsjettii-04 and epos - lhc are shown for comparison . open circles indicate the result if the fd energy scale is varied by its systematic uncertainty . the gray thick error bars indicate the systematic uncertainty of r@xmath25 . right : the measured muon signal rescaling at e = 10@xmath3 ev and at 1000 m from the shower axis vs. zenith angle , with respect to qgsjetii.04 proton as baseline . the rectangles represent the systematic uncertainties , and the error bars represent the statistical uncertainties added to the systematic uncertainties . the points for auger data are artificially shifted by @xmath27 0.5 for visibility.,title="fig:",scaledwidth=32.0% ] ev ) r@xmath25/(e@xmath26/10@xmath3 ev ) relative to prediction of qgsjetii-03 ( protons ) as a function of shower energy . theoretical curves for proton and iron showers simulated with qgsjettii-04 and epos - lhc are shown for comparison . open circles indicate the result if the fd energy scale is varied by its systematic uncertainty . the gray thick error bars indicate the systematic uncertainty of r@xmath25 . right : the measured muon signal rescaling at e = 10@xmath3 ev and at 1000 m from the shower axis vs. zenith angle , with respect to qgsjetii.04 proton as baseline . the rectangles represent the systematic uncertainties , and the error bars represent the statistical uncertainties added to the systematic uncertainties . the points for auger data are artificially shifted by @xmath27 0.5 for visibility.,title="fig:",scaledwidth=32.0% ] the data from extensive air shower measurements can be used to address problems of hadronic interactions at energies far from the reach of current accelerators . namely the inelastic proton - air cross - section at e@xmath11=10@xmath28 ev was estimated and converted to inelastic p - p cross - section at @xmath29 tev . the estimation of the chemical composition of cosmic rays and the behavior of the hadronic interactions at ultra - high energies are coupled problems . however , using any current model of hadronic interactions to interpret the evolution of the measured x@xmath6 distributions with energy , an increase of the mean mass of cr species is needed above @xmath21 10@xmath22 ev . a muon deficit in the predictions based on current hadronic interaction models is observed when compared to the data of the pierre auger observatory . the author thanks the organizers of eds - blois2013 and the colleagues from the pierre auger observatory for opportunity to participate in the conference . this contribution was supported by the ministry of education , youth and sports of the czech republic within the project lg13007 . 99 j. abraham _ et al . _ [ pierre auger collaboration ] , nucl . instrum . a * 620 * ( 2010 ) 227 . j. abraham _ et al . _ [ pierre auger collaboration ] , phys . b * 685 * ( 2010 ) 239 . p. abreu _ et al . _ [ pierre auger collaboration ] , astropart . * 34 * ( 2010 ) 314 . j. abraham _ et al . _ [ pierre auger collaboration ] , astropart . * 31 * ( 2009 ) 399 . p. abreu _ et al . _ [ pierre auger collaboration ] , phys . rev . d * 84 * ( 2011 ) 122005 . p. abreu _ et al . _ [ pierre auger collaboration ] , phys . rev . * 109 * ( 2012 ) 062002 . j. abraham _ et al . _ [ pierre auger collaboration ] , phys . * 104 * ( 2010 ) 091101 . a. letessier - selvon [ pierre auger collaboration ] , highlights from the pierre auger observatory , 33rd international cosmic ray conference ( icrc 2013 ) , arxiv:1310.4620 [ astro-ph.he ] . p. abreu _ et al . _ [ pierre auger collaboration ] , jcap * 1302 * ( 2013 ) 026 . i. valio [ pierre auger collaboration ] , a measurement of the muon number in showers using inclined events recorded at the pierre auger observatory , 33rd international cosmic ray conference ( icrc 2013 ) , arxiv:1307.5059 [ astro-ph.he ] , page 44 . b. kegl [ pierre auger collaboration ] , measurement of the muon signal using the temporal and spectral structure of the signals in surface detectors of the pierre auger observatory , 33rd international cosmic ray conference ( icrc 2013 ) , arxiv:1307.5059 [ astro-ph.he ] , page 48 . g. farrar [ pierre auger collaboration ] , the muon content of hybrid events recorded at the pierre auger observatory , 33rd international cosmic ray conference ( icrc 2013 ) , arxiv:1307.5059 [ astro-ph.he ] , page 52 .	the pierre auger observatory in argentina provides the largest data sample of the cosmic ray events with energy above 10@xmath0 ev . these high energy events can be used to test our understanding of the hadronic interactions at energies beyond the reach of colliders and to probe the basic properties of these interactions such as the inelastic cross - section of proton - air collisions . the combination of an array of surface detectors and the fluorescence telescopes of the pierre auger observatory reduces significantly the dependency of the shower energy estimation on mc simulations . despite that , the interpretation of mass sensitive quantities such as the shower maximum in terms of chemical composition of cosmic rays still depends on the hadronic interaction models . this contribution describes the main results of the observatory concerning the chemical composition of the cosmic rays and focuses on the problem of muon deficit in hadronic interaction models and on the estimation of proton - air cross - section from air - shower data . report - no : edsblois/2013/56
a consistent description of interaction of fundamental fields both at large and short distances is one of the most important problems of qcd . the strength of the interaction is defined by invariant charge @xmath0 ( the running coupling constant ) , which satisfies the renormalization group equation . the purpose of the present work consists in a formulation of a model for @xmath0 for all @xmath17 , which is appropriate for description both of perturbative and of nonperturbative phenomena and needs minimal number of parameters . we assume that there exists some value @xmath18 , which characterizes the nonperturbative effects scale and the corresponding value @xmath8 @xmath19 such , that for @xmath20 and thus @xmath21 the finite loop perturbation theory is applicable and sufficient , while nonperturbative effects prevails in region @xmath22 , which contains also nonphysical singularities of the perturbation theory and so here this theory essentially needs an extension of definition . in general , the invariant charge and the beta - function as well are depending on a renormalization scheme @xcite . for definiteness while performing calculations at @xmath23 we use @xmath24 scheme . it is well - known , that the @xmath3-function in the perturbative qcd is of the form @xmath25 for @xmath26 3 we have values of coefficients @xmath27 9 , @xmath28 32 , @xmath29 1287.67 , @xmath30 12090.38 ( coefficients @xmath31 , @xmath32 do not depend on renormalization scheme while values @xmath33 , @xmath34 correspond to a choice of @xmath24-scheme ) . expressions obtained by solution of gell - mann low equation @xmath35 with the use of ( [ 1 ] ) , are widely used for sufficiently large momenta transfer , however they can not be applied in the infrared region . as a matter of fact the behavior of @xmath4 at small momenta till now is an open question . lattice methods and sd equations give no an ultimate answer . for behavior of the invariant charge @xmath0 at @xmath36 a number of variants are considered ( see , e.g. ref . now the most popular variants for @xmath37 behavior are : @xmath38 , @xmath39 , @xmath4 is strongly enhanced . we consider the last possibility , in particular the well - known singular infrared asymptotic behavior @xmath40 for a review see , e.g. ref . @xcite and more recent papers refs . . such behavior of the invariant charge @xmath41 in so - called v - scheme corresponds to a linear confining quark - antiquark static potential at that @xmath42 , where @xmath43 is the string tension . results of some works @xcite on the lattice study of the three - gluon vertex demonstrate a necessity of taking into account of nonperturbative contributions to the running coupling constant being of the form ( [ 3 ] ) . in the framework of the continuous qft additional arguments in favor of behavior ( [ 3 ] ) are also presented in paper @xcite . highly similar singular infrared behavior has recently developed model for the qcd analytic invariant charge @xcite . asymptotic behavior ( [ 3 ] ) occurs provided @xmath44 for @xmath45 . we consider a possibility of behavior ( [ 3 ] ) and assume the following form of the infrared @xmath3-function @xmath46 where @xmath47 is a constant and @xmath48 corresponds to the boundary between perturbative and nonperturbative regions . for @xmath49 we shall use @xmath3-function ( [ 1 ] ) with a finite number of loops taken into account . our recipe for construction of @xmath3-function for all @xmath50 consists in a smooth matching of expressions ( [ 4 ] ) and ( [ 1 ] ) at point @xmath51 in approximations of the perturbation theory up to four loops . the demand of the @xmath3-function and its derivative to be continuous uniquely fix free parameters @xmath47 and @xmath48 of the global @xmath3-function ( the matched one ) . note , that the presence of parameter @xmath47 in eq . ( [ 4 ] ) which corresponds to coulomb contribution in invariant charge gives a possibility of smooth matching . we are willing to build the model for the invariant charge , which precisely coincides the perturbation theory in the perturbative region @xmath52 , while in the nonperturbative infrared region it provides simple description of main nonperturbative parameters . the work is organized in the following way . in section 2 we obtain the matched solutions for the cases of 1 4 loops . dimensionless parameters of the model are uniquely defined and further the solutions are normalized at the scale of the @xmath53-lepton mass , that leads to definite values of dimensional parameters @xmath54 , @xmath18 , @xmath43 . in section 3 the gluon condensate is calculated . in doing this we consider first a possibility of freezing of the perturbative component of @xmath4 in the infrared region and secondly a possibility of analytic behaviour of this component in the infrared region . section 4 contains concluding remarks . for an illustration let us consider the most simple one - loop case . conditions of matching give two equations @xmath56 @xmath57 the solution of set ( [ 5 ] ) reads @xmath58 we shall normalize perturbative solution @xmath59 by value @xmath60 , that gives @xmath61 where @xmath62 . imposing on @xmath0 the natural condition to be continuous at @xmath63 , we may normalize nonperturbative solution of equation ( [ 2 ] ) @xmath64 by @xmath60 as well . as a result we obtain @xmath65 equations ( [ 10 ] ) are correct for one four loops , for one - loop case @xmath66 . for final fixation of the solution for all @xmath5 we need to define @xmath6 by normalizing the solution , say , at point @xmath67 , where @xmath68 is the mass of the @xmath53-lepton . for one - loop case we have following simple formulae for quantities under consideration @xmath69 @xmath70 the gluon condensate @xmath71 for frozen perturbative constituent which is given here for completeness will be calculated below . for @xmath72 one has @xmath73 , @xmath74 @xmath75 @xmath76 , @xmath18 @xmath75 @xmath77 , @xmath43 @xmath75 @xmath78 , @xmath71 @xmath75 + @xmath79 . for normalization condition , e.g. @xmath80 0.32 one obtains @xmath81 0.201 gev and we have only qualitative agreement for quantities considered . as we shall see further on , rather poor results in the one - loop case are marginal , while results for many - loop cases turn to be much more promising . let us consider multi - loop cases . solution @xmath82 of equation ( [ 2 ] ) for @xmath83 reads as follows @xmath84\right.\ ] ] @xmath85 @xmath86+o\left(\frac{1}{l^4}\right)\right\}. \label{15}\ ] ] keeping in the expression terms with powers of logarithms in denominators up to the first , the second , the third and the fourth , we fix the 1 4-loop approximations of the perturbation theory for running coupling constant . it may be written in the form @xmath87 @xmath88\ ] ] @xmath89 . \label{16}\ ] ] here @xmath90 , and coefficient are defined as follows @xmath91 coefficients @xmath92 , @xmath93 , @xmath94 depend on @xmath95.with @xmath26 3 we have @xmath96 0.7901 , @xmath97 0.4147 , @xmath98 . in the case of the two - loop approximation for perturbative @xmath4 we have the following set of equations @xmath99 @xmath100 @xmath101 @xmath102 the set for three loops reads @xmath103 @xmath104 @xmath105 = b_0h_0,\ ] ] @xmath106 the set for four loops reads @xmath107 @xmath108 @xmath109\ ] ] @xmath110=b_0h_0,\ ] ] @xmath111 from sets of equations ( [ 18 ] ) ( [ 20 ] ) we find values of @xmath48 , @xmath47 , @xmath112 and @xmath113 . they are presented in table [ tab1 ] . taking into account the existing data @xcite we finally fix the momentum dependence of solutions . the dimensionless parameters @xmath48 , @xmath47 , @xmath114 , @xmath115 on the number of loops , @xmath26 3 . [ cols="^,^,^,^,^",options="header " , ] [ tab8 ] for cases from one loop up to four loops the behavior of the running coupling constant @xmath0 for all @xmath5 is shown in fig . [ fig1 ] . here the behavior of the analytic coupling constant @xmath116 for @xmath117 , which defines the perturbative part of @xmath0 in this region , is also shown . in fig . [ fig2 ] in addition to @xmath0 its nonperturbative part @xmath118 is shown , which turns to be zero for @xmath52 . normalization condition for all curves in fig . [ fig1 ] , fig . [ fig2 ] is @xmath119 . starting from the well - known perturbative expression ( [ 1 ] ) for the @xmath3-function for small values of the coupling constant @xmath120 and the behavior ( [ 4 ] ) for large values of the coupling constant , which corresponds to the linear confinement , we have constructed the model @xmath3-function for all @xmath121 . to control dependence of the results on the number of loops we simultaneously consider cases corresponding to perturbative @xmath3-function for 1 4 loops . while constructing the matched @xmath3-function we assume it to be smooth at the matching point , that leads to the invariant charge being smooth together with its derivative for all @xmath122 . the normalization of the invariant charge , e.g. at @xmath123 fix it thoroughly . then value @xmath124 corresponds to value of the parameter @xmath125 gev of our model which fits the string tension parameter of the string model @xcite . the obtained invariant charge is applied to study of the important physical quantity , the gluon condensate . in doing this we consider two variants of extracting of the nonperturbative contributions from the overall expression for the invariant charge . the first variant assumes freezing " of perturbative part of the charge in the nonperturbative region @xmath126 , ( average for 3-loop and 4-loop cases ) which is very close to critical value @xmath127 of ref . @xcite . ] while for the second one we choose the analytic behavior of the perturbative part in this region . as we see from tables [ tab4 ] , [ tab8 ] the first variant leads to values of the gluon condensate being somewhat larger , than that for the second variant . emphasize , that for @xmath128 values of the gluon condensate for both variants are quite satisfactory . namely the value for the second variant practically coincides the conventional value @xcite , while for the first variant it is only slightly higher , @xmath129 @xmath130 gev . note , that other important parameter , the nonperturbative scale @xmath18 for the same normalization condition also turns to be of a reasonable magnitude , @xmath131 gev ( see table [ tab3 ] ) . we may conclude , that the present model consistently describes the most important nonperturbative parameters . from this point of view the results support a possibility of singular infrared behaviour ( [ 3 ] ) of the invariant charge which contains both perturbative and nonperturbative contributions . the authors express deep gratitude to v.a . petrov , v.e . rochev and d.v . shirkov for valuable discussions . moreover , we thank a.l . kataev and a.a . pivovarov for stimulating discussions and comments . the work has been partially supported by rfbr under grant no . 02 - 01 - 00601 . 99 a. i. alekseev and b.a . arbuzov , _ beta - function in qcd and gluon condensate _ , arxiv : hep - ph/0407056 . n. n. bogoliubov and d. v. shirkov , _ introduction to the theory of quantized fields _ ( interscience , new york , 1980 ) . p. m. stevenson , _ phys . * d23 * , 2916 ( 1981 ) ; _ phys . rev . _ * d33 * , 3130 ( 1986 ) . d. v. shirkov , _ teor . fiz . _ * 132 * , 484 ( 2002 ) [ _ theor . math . phys . _ * 132 * , 1309 ( 2002 ) ] . b. a. arbuzov , _ fiz . chast . atom . * 19 * , 5 ( 1988 ) [ _ phys . part . * 19 * , 1 ( 1988 ) ] . a. i. alekseev and b.a . arbuzov , _ yad . fiz . _ * 61 * , 314 ( 1998 ) [ _ phys . nucl . _ * 61 * , 264 ( 1998 ) ] ; a. i. alekseev and b. a. arbuzov , _ mod . _ * a13 * , 1447 ( 1998 ) ; a. i. alekseev , in _ proc . of the workshop on nonperturbative methods in quantum field theory _ , adelaide , australia , 1998 , eds . a. v. schreiber , a. g. williams and a.w . thomas ( world scientific , singapore , 1998 ) , p. 104 ; arxiv : hep - ph/9808206 . g. burgio , f. di renzo , g. marchesini and e. onofri , _ phys . _ * b422 * , 219 ( 1998 ) , _ nucl . suppl . _ * 63 * , 805 ( 1998 ) ; g. burgio , f. di renzo , c. parrinello and c. pittori , _ nucl . suppl . _ * 73 * , 623 ( 1999 ) , _ nucl . . suppl . _ * 74 * , 388 ( 1999 ) ; ph . _ , _ j. high energy phys . _ * 0004 * , 006 ( 2000 ) . v. gogohia , _ phys . * b584 * , 225 ( 2004 ) . a. v. nesterenko , _ int . phys . _ * a18 * , 5475 ( 2003 ) . particle data group , s. eidelman _ et al . _ , _ phys . lett . _ * b592 * , 1 ( 2004 ) . s. bethke , _ j. phys . _ * g26 * , r27 ( 2000 ) , arxiv : hep - ex/0004021 . a. a. pivovarov , institute for nuclear research , preprint no . inr - th-03 - 3 , moscow , 2003 , arxiv : hep - ph/0301074 . a. simonov , _ yad . fiz . _ * 58 * , 113 ( 1995 ) [ _ phys . nucl . _ * 58,107 ( 1995 ] ; _ yad . 65 * , 140 ( 2002 ) ; yu . a. simonov , _ yad . fiz . _ * 66 * , 796 ( 2003 ) [ _ phys . nucl . _ * 66 * , 764 ( 2003 ) ] . m. a. shifman , a. i. vainshtein and v. i. zakharov , _ nucl . phys . _ * b147 * , 385 , 448 ( 1979 ) . p. j. redmond , _ phys . _ * 112 * , 1404 ( 1958 ) . n. n. bogolubov , a. a. logunov and d. v. shirkov , _ zh . ksp . fiz . _ * 37 * , 805 ( 1959 ) [ _ sov . phys . jetp _ * 10 * , 574 ( 1960 ) ] . d. v. shirkov and i. l. solovtsov , _ jinr rapid comm . _ * 2[76]-96 * , 5 ( 1996 ) ; d. v. shirkov and i. l. solovtsov , _ phys . rev . * 79 * , 1209 ( 1997 ) ; i. l. solovtsov and d. v. shirkov , _ teor . fiz . _ * 120 * , 482 ( 1999 ) [ _ theor . _ * 120 * , 1210 ( 1999 ) ] ; n. v. krasnikov and a. a. pivovarov , _ phys . lett . _ * b116 * , 168 ( 1982 ) ; a. v. radyushkin , joint institute for nuclear research , report no . jinr - e2 - 82 - 159 , 1982 ; _ jinr rapid comm . _ * 4[78]-96 * , 9 ( 1996 ) ; arxiv : hep - ph/9907228 . a. i. alekseev , _ phys . rev . _ * d61 * , 114005 ( 2000 ) . a. i. alekseev , in _ proceedings of the xvth workshop on high energy physics and quantum field theory _ , tver , russia , 2000 , edited by m.n . dubinin and v.i . savrin ( inp moscow state university , moscow , 2001 ) , arxiv : hep - ph/0011178 ; a. i. alekseev , _ yad . fiz . _ * 65 * , 1722 ( 2002 ) [ _ phys . at . nucl . _ * 65,1678 ( 2002 ) ] . a. i. alekseev , _ j. phys . _ g27 * , l117 ( 2001 ) ; a. i. alekseev , _ few - body systems _ * 32 * , 193 ( 2003 ) . l. d. soloviev , _ phys . * d58 * , 035005 ( 1998 ) ; _ phys _ * d61 * , 015009 ( 2000 ) . l. dokshitzer , d. e. kharzeev , arxiv : hep - ph/0404216 .	under assumption of singular behavior of invariant charge @xmath0 at @xmath1 and of large @xmath2 behavior , corresponding to the perturbation theory up to four loops , a procedure is considered of smooth matching the @xmath3-function at a boundary of perturbative and nonperturbative regions . the procedure results in a model for @xmath4 for all @xmath5 with dimensionless parameters being fixed and dimensional parameters being expressed in terms of only one quantity @xmath6 . the gluon condensate which is defined by the nonperturbative part of the invariant charge is calculated for two variants of true perturbative " invariant charge , corresponding to freezing option and to analytic one in nonperturbative region . dimensional parameters are fixed by varying normalization condition @xmath7 . it is obtained that on the boundary of perturbative region @xmath8 @xmath9 0.44 , the procedure results in nonperturbative coulomb component @xmath10 0.25 , the nonperturbative region scale @xmath11 1 gev , the model parameter @xmath12 ( 0.42 gev)@xmath13 which suits as string tension parameter , the gluon condensate appears to be close for two variants considered , @xmath14 ( 0.33 0.36 gev)@xmath15 ( for @xmath16 ) . + _ keywords : _ nonperturbative qcd ; gluon condensate ; running coupling constant ; infrared region . + pacs nos . : 12.38.-t , 12.38.aw , 11.15.tk
there has been much work over the past decade applying the techniques of effective field theory ( eft ) and the renormalization group ( rg ) to low energy nuclear systems such as the nucleon - nucleon force , finite nuclei , and nuclear matter @xcite . conventional nuclear force models such as the paris , bonn , and argonne potentials incorporate the same asymptotic tail generated by one pion exchange , as the long wavelength structure of the interaction is unambiguously resolved from fits to low energy phase shifts and deuteron properties . the short wavelength part of the interaction is then generated by assuming a specific dynamical model based on heavy meson exchanges , combined with phenomenological treatments at very small distances . such approaches are necessarily model dependent , as the low energy two - nucleon properties are insufficient to resolve the short distance structure . such model dependence often appears in many - body calculations , e.g. the coester band in nuclear matter , when highly virtual nucleons probe the short distance structure of the interaction . the eft approach eliminates the unsatisfactory model dependence of conventional force models and provides an effective description that is consistent with the low energy symmetries of the underlying strong interactions ( qcd ) . this is accomplished by keeping only nucleons and pions as explicit degrees of freedom , as dictated by the spontaneously broken chiral symmetry of qcd . all other heavy mesons and nucleon resonances are considered to be integrated out of the theory , their effects contained inside the renormalized pion exchange and scale dependent coupling constants that multiply model independent delta functions and their derivatives @xcite . no underlying dynamics are assumed for the heavy mesons and nucleon resonances , as they simply can not be resolved from low energy data . since rg decimations generate all possible interactions consistent with the symmetries of the underlying theory , it is sufficient to consider all interactions mandated by chiral symmetry and then tune the couplings to the low energy data . power counting arguments are then used to truncate the number of couplings that need to be fit to experiment , thus endowing the eft with predictive power and the ability to estimate errors resulting from the truncation . moreover , the breakdown of the eft at a given scale signals the presence of new relevant degrees of freedom that must be considered explicitly to properly describe phenomena at that scale . similar concepts of integrating out the high energy modes have long been used to derive effective interactions in nuclei within a truncated model space , e.g. the sd - shell for the two valence nucleons in @xmath2o . starting from the vacuum two - body force in the full many - body hilbert space , one can construct an effective theory for the low - lying excitations provided the effective interaction encodes the effects of the integrated high energy modes . although the traditional model space methods apparently share similarities to the modern rg - eft approaches , these have not been exploited until recently @xcite and not sufficiently in any realistic nuclear many - body calculation . in the traditional approaches , there has been little success in predicting how the effective interaction changes with the model space size . in rg language , no beta functions have been derived that would allow one to calculate an effective interaction in one convenient choice of model space , and then evolve the effective theory to any other scale by following the flow of the beta function . moreover , one could push the analogy with eft further by projecting the effective interaction onto a few leading operators , with the ability to reliably estimate the errors on calculated nuclear spectra resulting from this truncation . therefore , it is of greatest interest to address the issue of calculating beta functions within the framework of model space effective interaction methods and to exploit these powerful similarities with the rg - eft approach . two well known methods for deriving _ energy independent _ model space interactions are the kuo - lee - ratcliff ( klr ) folded diagram theory @xcite and the related similarity transformation method of lee and suzuki ( ls ) @xcite . the authors have applied these methods to the nucleon - nucleon problem in vacuum where the model space was taken to be plane wave states with relative momenta @xmath3 . the resulting unique low momentum potential @xmath0 preserves the deuteron binding energy and the low energy half - on - shell @xmath1 matrix , but with all intermediate state summations cut off at @xmath4 @xcite . in this paper , we restrict our analysis to the two - body problem in free space . we show that the model space interaction @xmath0 scales with @xmath4 in the same way one would expect from a exact rg treatment of the scattering problem . in this way , we show that the methods originally used to derive model space interactions in nuclei can be interpreted in modern language as renormalization group decimations , at least for two - body correlations in the nucleus . this work is a step towards reformulating traditional nuclear many - body methods in a manner consistent with the more systematic and controlled rg - eft approaches . to the best of our knowledge , this is a genuinely new result as previous rg studies have dealt with energy dependent effective potentials @xcite . it is also the first rg flow study of realistic nucleon - nucleon interactions . from a practical perspective , it is simpler to use energy independent effective interactions in many - body calculations , as one does not have to recalculate the interaction vertex depending on the energy variable of the propagator it is linked to in a particular diagram . let us begin with a rg treatment of the scattering problem . working in a given partial wave we denote the bare ( bonn , paris , argonne , etc . ) two - body potential as @xmath5 , although the following developments are general and apply to any nonrelativistic scattering problem . employing principle value propagators we have the half - on - shell ( hos ) @xmath1 matrix equation @xcite @xmath6 in a rg treatment of two - body scattering , we impose a cutoff @xmath4 on all the loop integrals in the @xmath1 matrix equation and replace the bare @xmath5 with an effective potential @xmath0 . the effective potential is independent of the scattering energy and is allowed to depend on @xmath4 so that @xmath7 for @xmath8 is preserved . clearly this is sufficient to ensure that low energy observables , which are fully on - shell quantities are independent of @xmath4 . explicitly , we have @xmath9 by construction , the low momentum hos @xmath1 matrices calculated from eq . ( [ fulltmat ] ) and eq . ( [ lowtmat ] ) are identical . consequently , the low momentum components of the low energy scattering states are preserved . we denote the standing wave scattering states of the effective theory by @xmath10 , which can be written in the standard fashion as @xmath11 we therefore take @xmath12 in eq . ( [ lowtmat ] ) to obtain a flow equation for @xmath0 @xmath13 by writing @xmath14 and using the completeness of the scattering states in the model space . therefore , one must use the bi - orthogonal complement @xmath15 in the completeness relation . in the cutoff range of interest , the non - hermiticity for the nucleon - nucleon problem is generally small and can be transformed away @xcite . ] , we obtain @xmath16 & = \frac{2}{\pi } \ , \lambda^{2 } \ , v_{\text{low k}}(k',\lambda ) \int_{0}^{\lambda } v_{\text { low k}}(\lambda , p ) \ : \mathcal{g}(p , p';\lambda^{2 } ) \ : p^{2 } dp . \label{gflow}\end{aligned}\ ] ] where @xmath17 denotes the interacting green s function in the effective theory . writing the interacting green s function in terms of the @xmath1 matrix in the low momentum sector we obtain the exact rg equation @xmath18 the derivation of eqs . ( [ gflow],[tflow ] ) is similar to the work of birse _ et al . _ @xcite , although they consider the fully off - shell @xmath1 matrix and an energy dependent effective potential @xmath19 . the rg equation of birse _ et al . _ is exact to second order in @xmath19 . it reads @xmath20 where @xmath21 denotes the scattering energy . our rg treatment is based on hos @xmath1 matrix preservation , as this is the minimally off - shell quantity needed to preserve the low energy phase shifts as well as the low momentum components of the wave functions , while allowing one to consider an energy independent effective interaction . our exact rg equation for energy independent effective interactions , eq . ( [ tflow ] ) , agrees with the result of birse _ et al . _ @xcite at the one - loop level . as we shall see below , the starting point of energy independent effective interaction methods is the above off - shell effective potential , and the energy dependence is subsequently traded for momentum dependence . this procedure is similar to using the equations of motion in many - body physics to trade the energy dependence in favor of a more complicated quasiparticle dispersion relation . in the previous section , we obtained the rg equation for @xmath22 , eq . ( [ gflow ] ) , from @xmath1 matrix equivalence alone . we now derive the same equation using the kuo - lee - ratcliff ( klr ) folded diagram technique , originally designed for constructing energy independent model space interactions in shell model applications @xcite . first , we define an energy dependent vertex function called the @xmath23 box , that is irreducible with respect to cutting intermediate low momentum propagators . the @xmath23 box resums the effects of the high momentum modes we wish to integrate out of the theory , @xmath24 it is the same energy dependent effective potential studied by weinberg in the context of chiral lagrangians @xcite and by birse _ et al . _ for the purpose of an rg treatment of two - body scattering @xcite . in terms of the @xmath23 box the hos @xmath1 matrix reads @xmath25 the original literature develops klr folded diagrams using time - ordered perturbation theory for systems with degenerate unperturbed model space spectra . in this framework , the folded diagrams can be viewed as correction terms one must add , if one factorizes the nested time - integrals one finds in time - ordered perturbation theory for dyson equations . for the present work , it is much easier to understand the meaning of folded diagrams in the time - independent framework . iterating eq . ( [ tqeq ] ) , it is clear that all @xmath23 box vertices are fully off - shell with the exception of the final @xmath23 box in each term . the folded diagrams are correction terms one must add , if all @xmath23 boxes are evaluated right - side on - shell in the dyson equation . in order to see how this works , the second order @xmath23 box contribution in eq . ( [ tqeq ] ) can be written as @xmath26 where @xmath27 denotes the `` one - fold correction '' and @xmath28 is shorthand for the model space principle value measure . with the expression for @xmath27 , we write the klr folded diagram series for @xmath0 to one - fold , or equivalently to second order in the @xmath23 box as @xmath29 analogously , the two - fold correction to the klr @xmath0 is obtained by considering the @xmath30 contribution of eq . ( [ tqeq ] ) and correcting for putting all @xmath23 boxes right - side on - shell . the resulting two - fold correction is given by @xmath31 - \iint\limits_{p , p ' } \frac{\widehat{q}(k',p';p^{2})-\widehat{q}(k',p';p^{\prime2 } ) } { ( p^{2}-p^{\prime2})(k^{2}-p^{2 } ) } \ , \widehat{q}(p',p;p^{2 } ) \ , \widehat{q}(p , k;k^{2}).\end{gathered}\ ] ] proceeding in a similar manner for the higher order terms in the dyson equation , one can write the klr expression for @xmath0 as @xmath32 having motivated the klr folded diagram series , we now derive a flow equation that tells us how the klr @xmath0 changes as one changes the scale @xmath4 . the first step is to notice that the @xmath23 box , and hence the klr series , obeys the semi - group composition law . we can therefore perform the decimation to @xmath4 in one step using the bare @xmath5 as input , or equivalently we can integrate out recursively in small steps ; each subsequent step in the decimation uses the @xmath0 of the previous step as input . let us therefore start with the klr @xmath0 at a cutoff @xmath4 , and ask how @xmath0 changes as we integrate out the states lying in the momentum shell @xmath33 . referring to the definition of the @xmath23 box , eq . ( [ qvbare ] ) , it is clear that each intermediate loop integration is @xmath34 . therefore , we have for @xmath35 @xmath36 where the superscript denotes at which step the low momentum potential and the @xmath23 box are evaluated . although we have given explicit formulas only up to two folds , this will be sufficient to derive the flow equation . the generalization to all orders in @xmath0 can be found in the appendix , where energy independent effective interactions are related to the spectral representation of the hos @xmath1 matrix . we insert the expression for the @xmath35 box , eq . ( [ qinfdec ] ) , into the klr folded diagram series , eq . ( [ klrv ] ) , and find @xmath37 + \iint\limits_{p , p ' } \frac{v^{\lambda}_{\text{low k}}(\lambda , p ) \ , v^{\lambda}_{\text{low k}}(p , p ' ) \ , v^{\lambda}_{\text{low k}}(p',\lambda)}{(\lambda^{2}- p^{2})\ , ( \lambda^{2}-p^{\prime2 } ) } + \mathcal{o}\bigl ( ( v^{\lambda}_{\text{low k}})^4 \bigr ) \ , \biggr\ } \ , + \mathcal{o}(\delta\lambda^{2 } ) .\end{gathered}\ ] ] passing to the limit @xmath38 , we sum the flow equation and obtain @xmath39 this proves that the klr @xmath0 obeys the same scaling equation one obtains using rg arguments based on @xmath1 matrix preservation alone . it does not come as a surprise , as it has been shown diagrammatically that the klr effective theory preserves the low momentum hos @xmath1 matrix of the input bare potential @xcite . nevertheless , we have here derived the same flow equation in a manner that makes no reference to @xmath1 matrix preservation . it may be regarded as an algebraic proof that the folded diagram series preserves the hos @xmath1 matrix . this firmly establishes that the klr folded diagram theory is equivalent to a rg decimation , at least for the two - nucleon sector . it is easy to see how the folded diagram series is constructed . the n - folded diagram is the correction terms one must add , if one evaluates all soft - mode irreducible vertex functions right - side on - shell in the term of @xmath40 of any dyson equation . as we have seen above , the expression for the two - fold contribution is already rather cumbersome . the lee - suzuki scheme is an iterative method that is often used in the calculation of shell model effective interactions @xcite . it has been shown to converge to the folded diagram series @xcite . while the ls method does not lend itself to a simple diagrammatic interpretation , we show how the same flow equation evolves from the iteration scheme , as this provides a check to our rg equation . the details of the ls method may be found in @xcite . the defining equations of the ls iteration are given by @xmath41 \widehat{q}^{(n ) } & \equiv \frac{1}{n ! } \ , \frac{d^{n}}{d\omega ^{n } } \widehat{q}(\omega ) \arrowvert_{\omega=0 } .\end{aligned}\ ] ] as for the klr folded diagram theory , we derive the flow equation for @xmath0 directly from the ls equations by invoking the semi - group composition law and considering an infinitesimal decimation from @xmath4 to @xmath42 . suppose we have solved the ls equations for @xmath43 at some value of @xmath4 , where @xmath44 denotes the projection operator on the model space . keeping terms to first order in @xmath45 , we find @xmath46 \widehat{q}^{(n ) \lambda-\delta\lambda}(p',p ) & = - \delta\lambda \ , \frac{2}{\pi } \ , \frac{v^{\lambda}_{\text{low k}}(p',\lambda ) \ , v^{\lambda}_{\text{low k}}(\lambda , p)}{\lambda^{2n } } .\end{aligned}\ ] ] substituting the above expressions into the defining ls equations and keeping terms only to @xmath34 , one obtains @xmath47 where @xmath48 the expression for @xmath49 can be summed to give the interacting green s function @xmath50 of the effective theory . in the limit @xmath38 , we thus obtain the following rg equation directly from the lee - suzuki equations @xmath51 & = \frac{2}{\pi } \frac{v_{\text{low k}}(p',\lambda ) \ : t(\lambda , p;\lambda ^{2})}{1-(p / \lambda)^{2 } } .\end{aligned}\ ] ] thus , we obtain the same scaling equation from the ls and klr effective interaction methods as we obtain from rg arguments based on @xmath1 matrix equivalence . hence , it is clear that the traditional model space methods of nuclear structure can be employed to perform rg decimations in the two nucleon sector . we have provided four different derivations of the same rg equation , which describes the scaling of the effective potential @xmath0 with @xmath4 so that the low energy observables are preserved by the effective theory . for completeness , we provide numerical confirmation of the rg equation in this section . starting from a model of the nuclear force such as the paris , bonn - a , argonne v-18 , or idaho - a potential , we have integrated out the model dependent high momentum modes of the theory using the klr / ls methods to obtain an effective potential @xmath0 @xcite . the results of bogner _ _ are striking @xcite : the high momentum modes of all studied potential models renormalize the low momentum interaction to a unique @xmath0 for cutoffs @xmath52 . the differences resulting from a chiral treatment of the 2@xmath53 exchange can be seen in the offdiagonal matrix elements of @xmath0 , and flow studies of the crucial s - wave matrix elements @xmath54 exhibit the relevant scales ( pions , deuteron , and higher order tensor interactions ) and for small model spaces are in excellent agreement with the power divergence subtraction scheme of kaplan , savage , and wise @xcite . for a discussion of all results see @xcite . performing the calculation for a range of @xmath4 values , we calculate the numerical derivative @xmath55 and compare it with the theoretical expression , eq . ( [ tflow ] ) . as in @xcite , the rg decimation is carried out with the iterative implementation of the ls method of andreozzi @xcite . referring to fig . ( [ 1s0flowdiag ] ) , it is evident that the numerical derivative ( numerical flow ) is in excellent agreement with the exact `` beta '' function ( theoretical flow ) . our units are such that @xmath56 . ] , @xmath57 . similar results are found for offdiagonal elements as well as other partial waves . we have shown that model space effective interaction methods , such as the kuo - lee - ratcliff folded diagram series and the lee - suzuki similarity transformation , scale with the model space boundary exactly as one would expect from rg arguments . we have derived the exact beta function for the energy independent , low momentum effective theory of two - body scattering . our theoretical beta function is in perfect agreement with the numerical scaling behaviour of @xmath0 . this is a genuinely new result , as previous rg studies of two - body scattering consider energy dependent effective interactions @xcite . in the rg formulation for effective theories , one can evolve the effective interaction according to the beta function given suitable initial conditions ; in our case , the bare interaction for very large model spaces . _ have performed such an rg decimation in the two - nucleon sector starting from realistic nucleon - nucleon potentials such as the paris , bonn - a , argonne v-18 , and idaho - a interaction @xcite . they found that the initially very different bare potentials renormalize to a unique @xmath0 starting at a scale of @xmath58 . moreover , the authors have used the unique @xmath0 directly in shell model calculations without first calculating the brueckner @xmath59 matrix . they found very promising results using the same input @xmath0 for two valence nuclei in different mass regions , and the results were nearly independent of the cutoff in the neighborhood of @xmath60 @xcite . referring to fig . ( [ 1s0flowdiag ] ) , these results are consistent with the observation that the beta function is nearly zero in this region . for energy dependent effective interactions , birse _ et al . _ have studied the power counting imposed by means of a wilson - kadanoff rescaling . their results have been very powerful in predicting only two well defined power - counting schemes for s - wave scattering @xcite with short ranged forces . their arguments can be extended to include long - range forces in a distorted wave basis @xcite , with qualitatively similar conclusions . we hope to generalize and apply the methods of birse _ et al . _ to obtain a well defined power counting scheme for the energy independent @xmath0 . this is complicated since the `` beta '' function includes @xmath0 to all orders . in conclusion , the equivalence of model space effective interaction and rg methods in the two - body sector is a first step towards reformulating traditional effective interaction methods in a manner that fully exploits their similarities to rg - eft methods , particularily their power to reliably estimate errors . the ultimate goal is to extend the rg arguments for @xmath0 to the shell model effective interaction . at this point , it is restricted to two - particle correlations in the nucleus . this work is supported by the us - doe grant de - fg02 - 88er40388 . in this appendix , we provide the fourth and final derivation of the rg equation , eq . ( [ tflow ] ) , based on the spectral representation of the hos @xmath1 matrix . moreover , we show that hos @xmath1 matrix preservation necessarily implies a non - hermitian energy independent @xmath0 . without loss of generality , we employ a notation which suggests there are no bound states in the channel under consideration ; bound states are easily handled by extending the completeness relation to include the relevant bound states in the summation . we first prove that the energy independent @xmath0 is necessarily non - hermitian , if it preserves the hos @xmath1 matrix . we begin by assuming the contrary , i.e. we assume the energy independent @xmath0 is hermitian and preserves the hos @xmath1 matrix . we have @xmath61 multiplying from the right with @xmath62 and integrating over @xmath63 , the completeness of the model space eigenfunctions yields @xmath64 expanding the @xmath65 dependence of the @xmath23 box , one can use the equations of motion @xmath66 to convert the @xmath65 dependence to @xmath67 dependence . for example , the @xmath68 dependence is eliminated by letting @xmath69 act on @xmath70 and using eq . ( [ nec ] ) to trade the extra @xmath23 for @xmath22 . it is clear from this simple example that the energy dependence is asymmetrically converted to @xmath71 dependence . therefore , we conclude that the requirement of hos @xmath1 matrix preservation necessarily implies a non hermitian @xmath0 . technically speaking , this means that the completeness relation for the model space scattering states should be modified by replacing @xmath72 with the bi - orthogonal complement , @xmath73 . having shown that @xmath0 is non - hermitian , we now derive the rg equation from the spectral representation of the hos @xmath1 matrix . inserting a complete set of low momentum plane waves between the @xmath23 box and the low energy scattering wave function in eq . ( [ nec1 ] ) , we obtain @xmath74 in terms of @xmath1 matrices , this can be written as @xmath75 this is just a rearrangement of the spectral representation for the @xmath1 matrix . differentiating eq . ( [ eavtt ] ) with respect to the cutoff we obtain the rg flow equation @xmath76 now we show that the latter two terms in eq . ( [ apprg1 ] ) cancel against each other . to this end , we write the left - side on - shell @xmath1 matrix in terms of the right - side on - shell one : @xmath77 by differentiating with respect to the cutoff , eq . ( [ leftright ] ) yields @xmath78 we insert the integral equation for the right - side on - shell @xmath1 matrix , eq . ( [ lowtmat ] ) , into the last term of flow equation , eq . ( [ apprg1 ] ) . it results in the two terms @xmath79 & \hspace{1 cm } + \iint\limits_{p , q } \ : v_{\text{low k}}(k^\prime , q ) \ : \frac{1}{p^2 - q^2 } \ : t(q , p;p^2 ) \ : \frac{1}{p^2-k^2 } \ : \frac{d t(p , k;p^2)}{d \lambda } . \label{lastterm}\end{aligned}\ ] ] finally , we combine eq . ( [ lastterm ] ) with the integral equation for @xmath80 , eq . ( [ rshosrg ] ) , inserted into the @xmath81 term . now the last two terms of the flow equation , eq . ( [ apprg1 ] ) , read @xmath82 & \ ; + \ , \frac{2}{\pi } \ , \int_p \ : v_{\text{low k}}(k^\prime , p ) \ : \frac{1}{p^2-k^2 } \ : t(p,\lambda;\lambda^2 ) \ : \biggl\ { \frac{\lambda^2}{p^2 - \lambda^2 } - \frac{\lambda^2}{k^2 - \lambda^2 } \biggr\ } \ : t(\lambda , k;\lambda^2 ) \nonumber \\[1 mm ] & \;\;\ ; + \iint\limits_{p , q } \ : v_{\text{low k}}(k^\prime , p ) \ : \frac{1}{p^2-k^2 } \ : t(p , q;q^2 ) \ : \biggl\ { \frac{1}{p^2 - q^2 } - \frac{1}{k^2 - q^2 } \biggr\ } \ : \frac{d t(q , k;q^2)}{d \lambda } \nonumber \\[1 mm ] & \;\;\;\ ; + \iint\limits_{p , q } \ : v_{\text{low k}}(k^\prime , p ) \ : \frac{1}{q^2 - p^2 } \ : t(p , q;q^2 ) \ : \frac{1}{q^2-k^2 } \ : \frac{d t(q , k;q^2)}{d \lambda } , \end{aligned}\ ] ] and we find the result that these terms cancel pairwise upon combining the energy denominators , resulting in the familiar rg equation . therefore , we have provided the final derivation of the `` beta '' function for the two - body scattering rg , eq . ( [ tflow ] ) , based on the spectral representation of the @xmath1 matrix . moreover , we have shown that @xmath0 is necessarily non - hermitian . beane , p.f . bedaque , w.c . haxton , d.r . phillips , and m.j . savage , nucl - th/0008064 . lepage , how to renormalize the schrdinger equation , nucl - th/9706029 . s.k . bogner , t.t.s . kuo , a. schwenk , d.r . entem , and r. machleidt , nucl - th/0108041 . s. k. bogner and t.t.s . kuo , phys . * b500 * ( 2001 ) 279 . haxton and c.l . song , phys . * 84 * ( 2000 ) 5454 . haxton and t. luu , nucl . phys . * a690 * ( 2001 ) 15 . kuo and e. osnes , springer lecture notes of physics , vol . 364 ( 1990 ) 1 . k. suzuki and s.y . lee , prog . * 64 * ( 1980 ) 2091 . birse , j.a . mcgovern , and k.g . richardson , phys . b464 * ( 1999 ) 169 . brown and a.d . jackson , the nucleon - nucleon interaction , north - holland amsterdam , 1976 . s. weinberg , nucl . b363 * ( 1991 ) 3 . bogner , t.t.s . kuo , and l. coraggio , nucl - th/9912056 . kaplan , m.j . savage , and m.b . wise , nucl . * b534 * ( 1998 ) 329 . f. andreozzi , phys . rev * c54 * ( 1996 ) 684 . bogner , t.t.s . kuo , l. corragio , a. covello , and n. itaco , nucl - th/0108040 . t. barford and m.c . birse , talk presented at mesons and light nuclei , prague , 2001 , nucl - th/0108024 .	we consider two nonperturbative methods originally used to derive shell model effective interactions in nuclei . these methods have been applied to the two nucleon sector to obtain an energy independent effective interaction @xmath0 , which preserves the low momentum half - on - shell @xmath1 matrix and the deuteron pole , with a sharp cutoff imposed on all intermediate state momenta . we show that @xmath0 scales with the cutoff precisely as one expects from renormalization group arguments . this result is a step towards reformulating traditional model space many - body calculations in the language of effective field theories and the renormalization group . the numerical scaling properties of @xmath0 are observed to be in excellent agreement with our exact renormalization group equation . _ pacs : _ 21.30.fe ; 13.75.cs ; 11.10.hi ; 03.65.nk + _ keywords : _ nucleon - nucleon interactions ; effective interactions ; renormalization group ; scattering theory
we are sketching a `` simple and natural '' explanation of the probabilistic nature of quantum mechanics . this is achieved by utilizing the mathematical formalism of nonstandard analysis and feynman s interpretation of the heisenberg uncertainty principle , i.e. , of the quantum fluctuations . * non - technical presentation of the main ideas . * this summary is intended for readers who are not familiar with non - standard analysis . as often in nonstandard analysis ( see , e.g. , @xcite ) , we replace a continuous time interval by an infinite `` discrete '' set of infinitely closed time instants . substituting @xmath0 to @xmath1 in the well known expression of the uncertainty principle , where @xmath2 is the position , @xmath3 the mass , @xmath4 the momentum , @xmath5 the planck constant , yields equation ( [ rhei ] ) . we rewrite it by postulating that the quantity ( [ fond ] ) , where @xmath6 is a given infinitesimal , is limited and appreciable , i.e. , it is neither infinitely large nor infinitely small . those computations are stemming from feynman s interpretation @xcite of the uncertainty principle ( see , also , @xcite ) : the `` quantum trajectories '' are fractal curves , of hausdorff dimension @xmath7 . `` weak '' mathematical assumptions permit to derive the infinitesimal difference equation ( [ ma ] ) . the lack of any further physical assumption yields the equiprobability of @xmath8 and @xmath9 . if @xmath2 is _ markov _ , i.e. , if @xmath10 and @xmath11 are are functions of @xmath12 and @xmath13 , and not of @xmath14 , the corresponding infinitesimal random walk is equivalent to a stochastic differential equation in the usual sense ( see , e.g. , @xcite ) . more or less analogous random walks have already been introduced in the literature ( see , e.g. , @xcite ) , but in another context . * non - standard analysis . * replace the interval @xmath15 $ ] by the set @xmath16 , @xmath17 , where @xmath18 is an unlimited integer . a function @xmath19 is said to verify the _ heisenberg condition _ if , and only if , for any @xmath20 , equation ( [ heis ] ) is satisfied , where @xmath21 , @xmath22 , is limited and appreciable . the lack of any further physical assumption leads us to postulate the following properties for @xmath23 in equation ( [ heis - prob ] ) : the random variables @xmath23 , @xmath24 , are independent ; @xmath25 . the next properties are `` natural '' for the stochastic processes @xmath26 and @xmath2 ( see @xcite and @xcite ) : the random variables @xmath27 and @xmath23 , @xmath28 , are independent ; @xmath13 , @xmath29 , is function of @xmath30 . set @xmath31 . equation ( [ qf ] ) yields the decomposition ( [ decomp ] ) @xcite , where @xmath32 is a stochastic process , such that @xmath33 , @xmath34 . then , for all @xmath24 , @xmath35 , @xmath36 . we will not make any distinction between two equations of type ( [ decomp ] ) if the coefficients are limited and infinitely closed . the process @xmath2 is said to be _ markov _ , or to satisfy the _ markov condition _ , if , and only if , there exist @xmath37 , @xmath38 , such that they are functions of @xmath12 et @xmath13 , and not of @xmath39 . assume that @xmath2 is markov . consider the infinitesimal random walk ( [ mad ] ) . according to @xcite ( see , also , @xcite ) it defines a diffusion if , and only if , the following conditions are satisfied ( see @xcite for another approach ) : @xmath10 and @xmath11 are of class @xmath40 ; the shadows @xmath41 and @xmath42 of @xmath10 et @xmath11 are of class @xmath43 ; the function @xmath42 is always strictly positive ; the random variable @xmath27 is almost surely limited . le caractre probabiliste du quantique , confirm par tant dexpriences , procde , depuis born , bohr et lcole dite de copenhague - gttingen , dune axiomatique de la fonction donde de schrdinger , qui , comme on le sait , a fait natre , cause de son tranget et de ses apparents paradoxes , dinnombrables commentaires ( cf . @xcite ) . cette note ( voir @xcite pour une version prliminaire ) vise dduire cette nature alatoire de considrations mathmatiques simples et naturelles . elle repose sur * linterprtation par feynman @xcite du principe dincertitude de heisenberg , cest - - dire des fluctuations quantiques ; * l_analyse non standard _ de robinson @xcite , dj employe en quantique ( voir , par exemple , @xcite , sa bibliographie , et @xcite ) , et , plus prcisment , * * une discrtisation infinitsimale du temps , * * le calcul non standard des probabilits de nelson @xcite et son extension aux quations diffrentielles stochastiques @xcite . on aboutit des quations diffrentielles stochastiques grce de marches alatoires infinitsimales . nous empruntons ainsi des degrs divers trois alternatives aux fondements du quantique , parfois trs controverses , do nt les liens avaient dj t constats dans la littrature ( cf . @xcite ) : lapproche fractale , la mcanique stochastique et les marches alatoires . 1 . nous suivons la premire , porte par nottale @xcite ( voir aussi @xcite ) , dans sa mise en valeur de linterprtation due feynman du principe dincertitude , qui semblait plus ou moins oublie . 2 . les quations diffrentielles stochastiques ordinaires sont un pivot de la seconde , invente par fnyes @xcite , dveloppe et popularise par nelson @xcite . cette mcanique continue susciter une littrature abondante ( cf . @xcite ) , en dpit de critiques virulentes ( cf . voir , par exemple , @xcite pour une comparaison avec la mcanique quantique _ euclidienne _ , que beaucoup , nelson y compris , jugent plus convaincante . renvoyons pour la troisime aux fameux chiqier ( _ chessboard _ ) de feynman @xcite , aux travaux dord ( cf . @xcite ) et @xcite . on se restreint un espace de dimension @xmath44 afin dallger lcriture . alors que certains travaux autour de la mcanique stochastique se rattachent lautomatique et , plus prcisment , la commande optimale stochastique ( cf . @xcite ) , notre point de vue fait suite une approche nouvelle @xcite de lestimation et de lidentification , en automatique et signal . rappelons les liens entre le quantique actuel et certains aspects du traitement de signal ( cf . @xcite ) . on sefforcera de dgager les rapports avec des travaux actuels sur le plongement stochastique @xcite . voici une synthse des principales ides directrices , sans prtention de rigueur , mais , esprons - le , claire et transparente , destine aux lecteurs peu au fait de lanalyse non standard : 1 . on remplace , comme souvent en non - standard ( cf . @xcite ) , le temps continu par un ensemble infini discret dinstants infiniment proches . les calculs suivants manent dune reformulation du principe dincertitude , due feynman @xcite ( voir , aussi , @xcite ) : la trajectoire quantique est une courbe continue , non drivable , cest - - dire fractale , de dimension de hausdorff @xmath7 , traduisant la , cest - - dire les fluctuations quantiques . soient @xmath2 la position , @xmath3 la masse , @xmath4 la quantit de mouvement , @xmath5 la constante de planck . le symbole @xmath45 dsigne un petit accroissement ; ainsi @xmath46 et , donc , @xmath47 . lexpression familire @xmath48 du principe dincertitude devient ( cf . @xcite , p. 85 ) : @xmath49 soit @xmath50 infinitsimal donn . on rcrit ( [ rhei ] ) en postulant que @xmath51 est limit et apprciable , cest - - dire ni infiniment grand ni infiniment petit . + voir , propos de lirrversibilit thermodynamique , @xcite pour une condition voisine de ( [ rhei])-([fond ] ) dans un espace - temps discret . 3 . on en dduit , moyennant des hypothses mathmatiques faibles , lquation aux diffrences infinitsimales @xmath52 4 . labsence de toute hypothse physique supplmentaire conduit postuler lquiprobabilit de @xmath53 . si @xmath2 est _ markovien _ , cest - - dire si @xmath10 et @xmath11 sont fonctions de @xmath12 et @xmath13 , et non de @xmath54 , la marche alatoire infinitsimale dfinie par ( [ ma ] ) quivaut une quation diffrentielle stochastique au sens usuel ( cf . @xcite ) , comme en mcanique stochastique . on rdige selon la terminologie _ ist _ de @xcite . remplaons lintervalle @xmath15 $ ] par @xmath55 , @xmath17 , o @xmath18 est un entier illimit . appelons @xmath56 et @xmath57 , respectivement , l_chelle _ et le _ pas de temps quantiques_. une fonction @xmath19 est dite vrifier la _ condition de heisenberg _ si , et seulement si , pour tout @xmath24 , notons @xmath64 lesprance conditionnelle du processus @xmath65 en linstant @xmath28 , sachant le pass du processus @xmath26 . posons @xmath66 supposons @xmath67 limit . on obtient la dcomposition @xcite ( voir , aussi , @xcite ) : @xmath68 o @xmath32 est un processus stochastique , tel que @xmath33 , @xmath34 . la proposition suivante relie les quantits @xmath69 et @xmath23 des paragraphes prcdents aux nouvelles : nous ne distinguerons pas deux quations de type ( [ decomp ] ) avec coefficients limits et infiniment proches . on dit que @xmath2 est _ markovien _ , ou satisfait la _ condition de markov _ , si , et seulement si , il existe des quantits @xmath78 , @xmath38 fonctions de @xmath12 et @xmath13 , et non de @xmath79 . supposons , dornavant , @xmath2 markovien et considrons la marche alatoire infinitsimale @xmath80 renvoyons @xcite ( voir , aussi , @xcite ) pour la notion d_quivalence _ de processus stochastiques . le reprsentant ( [ mad ] ) dfinit , selon @xcite ( voir , aussi , @xcite ) , une diffusion si , et seulement si , les conditions suivantes sont satisfaites : * @xmath10 et @xmath11 sont de classe @xmath40 , * les ombres @xmath41 et @xmath42 de @xmath10 et @xmath11 sont de classe @xmath43 , * la fonction @xmath42 est toujours strictement positive , * la variable alatoire @xmath27 est presque srement limite . * remerciements*. lauteur exprime sa reconnaissace e. benot ( la rochelle ) pour lui avoir communiqu la rfrence @xcite , j .- lvy - leblond ( nice ) pour des remises en question salutaires , p. rouchon ( paris ) pour des discussions prcieuses , et t. sari ( mulhouse ) pour des conseils utiles . e. nelson , dynamical theories of brownian motion , princeton university press , princeton , nj , 1967 ( @xmath83 d . , datant de 2001 , accessible sur http@xmath81//www.math.princeton.edu/%7enelson / books / bmotion.pdf ) .	cette note esquisse une construction mathmatique simple et naturelle du caractre probabiliste de la mcanique quantique . elle utilise lanalyse non standard et repose sur linterprtation due feynman , mise en avant dans certaines approches fractales , du principe dincertitude de heisenberg , cest - - dire des fluctuations quantiques . on aboutit ainsi des quations diffrentielles stochastiques , comme dans la mcanique stochastique de nelson , dcoulant de marches alatoires infinitsimales . _ pour citer cet article : m. fliess , c. r. acad . sci . paris , ser . i 344 ( 2007 ) . _ 0.5 * abstract * 0.5probabilities and quantum fluctuations . this note is sketching a simple and natural mathematical construction for explaining the probabilistic nature of quantum mechanics . it employs nonstandard analysis and is based on feynman s interpretation of the heisenberg uncertainty principle , i.e. , of the quantum fluctuations , which was brought to the forefront in some fractal approaches . it results , as in nelson s stochastic mechanics , in stochastic differential equations which are deduced from infinitesimal random walks . _ to cite this article : m. fliess , c. r. acad . sci . paris , ser . i 344 ( 2007 ) . _ physique mathmatique , reu le * * * * * ; accept aprs rvision le + + + + + + prsent par
in recent years , indoor visible light communication by light emitting diodes ( leds ) has attracted extensive academic attention @xcite ( and references therein ) , driven by advancements in designing and manufacturing of leds @xcite . adoption of leds as lighting source can significantly reduce energy consumption and at the same time offering high speed wireless communication , which is the primary focus of visible light communication ( vlc ) research @xcite . most of the existing schemes employ blue leds with a yellow phosphor coating , while with red / green / blue ( rgb ) leds higher data rate is possible because of wavelength division multiplexing . with rgb leds , color - shift keying ( csk ) was recommended by the ieee 802.15.7 visible light communication task group @xcite . a few authors have promoted this idea by designing constellations using signal processing tools . drost et al . proposed an efficient constellation designed for csk based on billiard algorithm @xcite . monteiro et al . designed the csk constellation using an interior point method , operating with peak and color cross - talk constraints @xcite . bai et al . considered the constellation design for csk to minimize the bit error rate ( ber ) subject to some lighting constraints @xcite . despite the fact that the three - dimensional constellation design problems have been formulated in @xcite , a few important questions have not been addressed . they include how to compare a system with csk employed and a conventional decoupled system , the constellation design , and the peak - to - average power ratio ( papr ) reduction @xcite . in this paper , we propose a novel constellation design scheme in high dimensional space , termed csk - advanced . in our design , arbitrary number of red , blue , and green leds can be selected . with any average optical intensity and average color selected , we formulate an optimization problem to minimize the system symbol error rate ( ser ) by maximizing the minimum euclidean distance ( med ) among designed symbol vectors . further , other important lighting factors such as color rendering index ( cri ) and luminous efficacy rate ( ler ) are also considered . further , optical papr is included as an additional constraint . the remainder of this paper is organized as follows . in section ii , we consider the constellation design problem assuming ideal channel . in section iii , we consider the constellation design for channel with cross - talks ( cwc ) . an svd - based pre - equalizer is applied and the constellations are redesigned subject to a transformed set of constraints . in section iv , we discuss the optimization of constellations under arbitrary color illuminations . in section v , we compare our scheme with a decoupled scheme and provide performance evaluation . finally , section vi provides conclusions . the system diagram is shown in fig . 1 , with @xmath0 red leds , @xmath1 green leds , and @xmath2 blue leds . in one symbol interval of length @xmath3 , a random bit sequence @xmath4 of size @xmath5 is first mapped by a bsa mapper @xmath6 to a symbol vector @xmath7 of size @xmath8 , where @xmath9 . the symbol @xmath7 is chosen from a constellation @xmath10 where @xmath11 denotes the constellation size . each component @xmath12 is applied to the corresponding led as intensity to transmit , such that @xmath13 . the intensity vector @xmath14 is then multiplied with the optical channel @xmath15 of size @xmath16 . the output of the color filters can be written as follows , @xmath17 where @xmath18 is the electro - optical conversion factor , @xmath19 is the photodetector responsivity . without loss of generality ( w.l.o.g . ) , assume @xmath20 . the noise @xmath21 is the combination of shot noise and thermal noise @xcite , assuming @xmath22 . it should be noted that the imaging detector is followed by imperfect color filters such that cross - talks may exist . the received intensity vector @xmath23 is passed through a symbol detector to obtain an estimate of the transmitter symbol , which is then de - mapped by @xmath24 to recover the bit sequence . we assume line - of - sight ( los ) links without inter - symbol interference . we first consider ideal channel , i.e. @xmath25 . define a joint constellation vector @xmath26^t$ ] , and the @xmath27-th symbol is written as @xmath28^t=\mathbf{j}_i\boldsymbol{\mathbf{c_t}},\end{aligned}\ ] ] where @xmath29 $ ] is a selection matrix with all zeros except for an identity matrix at the @xmath27-th block . our objective is to minimize the system ser subject to several visible lighting constraints . we aim to max the minimum med @xmath30 , i.e. , maximize @xmath31 such that the following holds for all @xmath32 @xcite @xmath33 where the parameter @xmath31 will be optimized and we obtain @xmath30 through this optimization . @xmath34 , @xmath35 ( kronecker product ) , @xmath36 of size @xmath8 has all zeros except the @xmath37-th element being one , @xmath38 , and @xmath39 the distance constraints are nonconvex in @xmath40 . we approximate by a first order taylor series approximation around @xmath41 , i.e. @xmath42 where @xmath41 is either a random initialization point or a previously attained estimate . a designer may wish to constrain the average color , as non - white illumination could be useful in many places . the average of all leds intensities can be written as the following @xmath8 vector @xmath43^t.\end{aligned}\ ] ] we consider the average power of each color , i.e. , a @xmath44 vector @xmath45 given as follows , @xmath46^t=\mathbf{k}\bar{\mathbf{c}}=\mathbf{k}\bar{\mathbf{j}}\boldsymbol{\mathbf{c_t}},\ ] ] where @xmath47 is a selection matrix summing up r / g / b intensities accordingly , @xmath48 is the average optical power , and @xmath49 @xmath50 where @xmath51 . by properly selecting @xmath45 , the cri and ler constraints can be met @xcite . for each led , the optical papr is defined as the ratio of the highest power over the average power . mathematically , the papr of the @xmath52-th led can be written as follows , @xmath53,\ ] ] where @xmath54 denotes the summation of all elements of vector , @xmath55 is a selection matrix of size @xmath56 , @xmath57 denotes the largest element of vector @xmath58 . the papr of an individual led can be constrained as follows @xmath59.\ ] ] cri stands for a quantitative measure of ability of light sources to reproduce the colors of objects faithfully , comparing with an ideal lighting source @xcite . ler measures how well light sources creates visible light . it is the ratio of luminous flux to power . depending on context , the power can be either the radiant flux of the source s output , or it can be the total power ( electric power , chemical energy , or others ) consumed by the source @xcite . the cri and ler are important practical lighting constraints . by properly selecting @xmath45 , specific cri and ler constraints can be met . when @xmath25 the problem can be formulated as follows , @xmath60 which can be straightforwardly proven as a convex optimization problem . with the first three constraints , it is termed as a regular optimization problem and with all constraints a papr - constrained problem . by iteratively solving , a local optimal constellation @xmath61 can be obtained . with multiple runs starting from different initial point @xmath41 , the best of solutions , @xmath62 is selected . the channel cross - talks exist when the transmitting led s emission spectral does not match the receiver filter s transmission spectral . it can be described by the following structure assuming single rgb led is employed based on @xcite and experiments , @xmath63 where the parameter @xmath64 characterizes both attenuation and interference effects . by singular value decomposition ( svd ) , @xmath65 , where @xmath66 and @xmath67 are unitary matrices of size @xmath68 , @xmath69 is a diagonal matrix of size @xmath70 . in this case , @xmath71 is the dimension of space for constellation design instead of @xmath72 . we apply a pre - equalizer @xmath73 at the transmitter - side and a post - equalizer @xmath74 at the receiver - side to equalize the channel and @xmath75 have the same distribution , since @xmath74 is unitary . ] . define @xmath76 , and the optimization in can be transformed as @xmath77 it should be noted that @xmath40 now is of dimension @xmath78 , i.e. , the constellation is designed in a @xmath71-dimensional space . to further minimize the system ber with a fixed ser , a good bit - to - symbol mapping function @xmath6 as shown in fig . 1 need to be designed . in this paper , we apply the binary switching ( bsa ) algorithm to optimize the mapping . since it is not the main focus of this paper , the details of bsa are omitted ( we refer the readers interested to @xcite ) . we provide numerical illustration of advantages of the csk - advanced with one rgb led , i.e. @xmath79 . both the csk - advanced and the conventional decoupled scheme can work with arbitrary color illumination . with one rgb led , @xmath80)$ ] for the decoupled scheme takes value from ook ( 2-pam ) constellations . to make a fair comparison , an 8-csk - advanced constellation is designed , with equal spectrum efficiency ( bits / sec / hz ) , equal average optical powers , and equal average color . if the average intensity of each color is similar , we call the corresponding system `` balanced lighting system '' . for example , we choose average color as @xmath81^t$ ] and the average power @xmath82 . for the conventional scheme , each led can simply take value independently from the following binary constellations @xmath83.\ ] ] the med for each branch is @xmath84 . for our scheme , the optimized constellation is as follows ( column 1 to 4 and column 5 to 8 are separated due to space limit . ) @xmath85 @xmath86 the med equals @xmath87 , such that we could expect a lower ser with sufficient snr . the asymptotic power gain is approximately @xmath88db ( = @xmath89 ) . we choose the average color as @xmath90^t$ ] . with the conventional scheme , the leds take value from constellations @xmath91~~\mathcal{c}_{u , g}=[0,6.66]~~\mathcal{c}_{u , b}=[0,4.44].\ ] ] with our scheme , the optimized constellation is as follows @xmath92 @xmath93 the med is approximately @xmath94 , which is smaller than med of one branch but larger than meds of two branches of the conventional scheme . we choose the average color as @xmath95^t$ ] . with the conventional scheme , the leds take value from constellations @xmath96~~\mathcal{c}_{e , g}=[0,3]~~\mathcal{c}_{e , b}=[0,3].\ ] ] with our scheme , the optimized constellation is @xmath97 @xmath98 the med is approximately @xmath99 , which is smaller than med of one branch but larger than meds of two branches of the conventional scheme . .med with varying papr and average color . [ cols="^,^,^,^,^",options="header " , ] instead of redesign the constellations subject to a transformed set of constraints due to employment of a pre - equalizer @xmath100 , zero - forcing ( zf ) @xmath101 or linear minimum - mean - squared - error ( lmmse ) based post - equalizer @xmath102 can be employed at the receiver @xcite to mitigate the cross - talks . 5 shows the corresponding bers against increased crosstalks for a balanced system employing different schemes when osnr is fixed to 5db . it is seen that our svd - based scheme significantly outperforms systems employing either zf or lmmse post - equalizers . 6 shows the bers against osnr for a balanced system when @xmath103 is fixed to 0.1 . with this particular parameters chosen , there is no significant difference between zf and lmmse based system performance and therefore we only included the lmmse based results . with osnr=5db for a balanced system . ] for a balanced system . ] a novel constellation design scheme , named csk - advanced , for vlc with arbitrary number of rgb leds , is proposed in this paper . with both optimized constellation and bits - to - symbols mapping , significant power gains are observed compared with conventional decoupled systems . for more unbalanced color illumination , the larger power gains can be expected . to avoid excessive nonlinear distortion , optical papr constraints is included into the optimization . furthermore , to deal with cwc , an svd - based pre - equalizer is introduced . it is shown by simulations that the proposed scheme significantly outperforms various benchmarks employing zf or lmmse - based post - equalizers . 9 s. watson , m. tan , s.p . najda , p. perlin , m. leszczynski , g. targowski , s. grzanka , and a. e. kelly , `` visible light communications using a directly modulated 422 nm gan laser diode , '' _ opt . 3792 - 3794 , 2013 . c. chen , p. wu , h lu , y. lin , j. wen , and f. hu , `` bidirectional phase - modulated hybrid cable television / radio - over - fiber lightwave transport systems , '' _ opt . 404 - 406 , 2013 . j. k. kim and e. f. schubert , `` transcending the replacement paradigm of solid - state lighting , '' _ opt . express _ , 21835 - 21842 , 2008 . h. elgala and t.d.c . little , `` reverse polarity optical - ofdm ( rpo - ofdm ) : dimming compatible ofdm for gigabit vlc links , '' _ opt . 24288 - 24299 , 2013 . j. vucic and k.d . langer , `` high - speed visible light communications : state - of - the - art , '' _ ofc / nfoec _ , pp . 1 - 3 , mar . y. wang , y. wang , n. chi , j. yu , and h. shang , `` demonstration of 575-mb / s downlink and 225-mb / s uplink bi - directional scm - wdm visible light communication using rgb led and phosphor - based led , '' _ opt . 21 , no . 1 , pp . 1203 - 1208 , 2013 . ieee 802.15.7 visible light communication task group , https://mentor.ieee.org/802.15/documents?is group=0007 . r.j . drost and b.m . sadler , `` constellation design for color - shift keying using billiards algorithms , '' _ ieee globecom workshop _ 980 - 984 , dec . 2010 . e. monteiro and s. hranilovic , `` constellation design for color - shift keying using interior point methods , '' _ ieee owc - ws _ , pp . 1224 - 1228 , dec . 2012 . b. bai , q. he , z. xu , and y. fan , `` the color shift key modulation with non - uniform signaling for visible light communication , '' _ ieee iccc - ws - owcc _ , pp . 37 - 42 , aug . 2012 . z. yu , r.j . baxley , and g.t . zhou , `` peak - to - average power ratio and illumination - to - communication efficiency considerations in visible light ofdm systems , '' _ ieee icassp _ , pp . 5397 - 5401 , may . 2013 . l. zeng , d. obrien , h. minh , g. faulkner , k. lee , d. jung , y. oh , and e. won , `` high data rate multiple input multiple output ( mimo ) optical wireless communications using white led lighting , '' jsac , vol . 27 , no.9 , pp . 1654 - 1662 , dec . 2009 . broadbent , `` a critical review of the development of the cie1931 rgb color - matching functions , '' _ color research and applications _ 267 - 272 , aug . 2004 . k. zeger and a. gersho , `` pseudo - gray coding , '' _ ieee trans . 2147 - 2158 , dec . 1990 . m. beko and r. dinis , `` designing good multi - dimensional constellations , '' _ ieee wireless commun . _ , vol . 1 , no . 3 , pp . 221 - 224 , 2012 . f. schreckenbach , n. gortz , j. hagenauer , and g. bauch , `` optimization of symbol mappings for bit - interleaved coded modulation with iterative decoding , '' _ ieee commun . _ , vol . 7 , no . 12 , pp . 593 - 595 , dec j. karout , e. agrell , k. szczerba , and m. karlsson , `` optimizing constellations for single - subcarrier intensity - modulated optical systems , '' _ ieee trans . inf . theory 58 , no . 7 , pp . 4645 - 4659 , apr . cie ( 1999 ) , `` colour rendering ( tc 1 - 33 closing remarks ) , '' _ publication 135/2 , vienna : cie central bureau , isbn 3 - 900734 - 97 - 6_. a. stimson , `` photometry and radiometry for engineers , '' _ new york : wiley and son_. t.p . crauss , m.d . zoltowski , and g. leus , `` simple mmse equalizers for cdma downlink to restore chip sequence : comparison to zero - forcing and rake , '' _ ieee icassp _ , vol . 5 , pp . 2865 - 2868 2000 .	in this paper , we propose a novel high dimensional constellation design scheme for visible light communication ( vlc ) systems employing red / green / blue light emitting diodes ( rgb leds ) . it is in fact a generalized color shift keying ( csk ) scheme which does not suffer efficiency loss due to a constrained sum intensity for all constellation symbols . crucial lighting requirements are included as optimization constraints . to control non - linear distortion , the optical peak - to - average - power ratio ( papr ) of leds is individually constrained . fixing the average optical power , our scheme is able to achieve much lower bit - error rate ( ber ) than conventional schems especially when illumination color is more `` unbalanced '' . when cross - talks exist among the multiple optical channels , we apply a singular value decomposition ( svd)-based pre - equalizer and redesign the constellations , and such scheme is shown to outperform post - equalized schemes based on zero - forcing or linear minimum - mean - squared - error ( lmmse ) principles . to further reduce system ber , a binary switching algorithm ( bsa ) is employed the first time for labeling high dimensional constellation . we thus obtains the optimal bits - to - symbols mapping . * optical wireless communication , constellation design , constellation labeling , multi - color optical , csk , im / dd . *
in many practical scenarios of network design , input graphs have a natural drawing on the sphere or equivalently the plane . in most cases , these embeddings have few crossings , either to avoid digging multiple levels of tunnels for fiber or cable or to avoid building overpasses in road networks . but a few crossings are common , and can easily come in bunches where one tunnel or overpass might carry several links or roads . thus we naturally arrive at graphs of small ( bounded ) genus , which is the topic of this paper . we develop a ptas framework for subset - connectivity problems on edge - weighted graphs of bounded genus . in general , we are given a subset of the nodes , called _ terminals _ , and the goal is to connect the terminals together with some substructure of the graph by using cost within @xmath1 of the minimum possible cost . our framework applies to three well - studied problems in this framework . in _ steiner tree _ , the substructure must be connected , and thus forms a tree . in _ subset tsp ( traveling salesman problem ) _ , the substructure must be a cycle ; to guarantee existence , the cycle may traverse vertices and edges multiple times , but pays for each traversal . in _ @xmath2-edge - connectivity survivable network design _ , the substructure must have @xmath3 edge - disjoint paths connecting terminals @xmath4 and @xmath5 , where each @xmath6 ; we allow the substructure to include multiple copies of an edge in the graph , but pay for each copy . in particular , if @xmath7 for all terminals @xmath4 and @xmath5 , then we obtain the steiner tree problem ; if @xmath8 for all terminals @xmath4 and @xmath5 , then we obtain the _ minimum - cost @xmath9-edge - connected submultigraph _ problem . our framework yields the first ptas for all of these problems in bounded - genus graphs . these ptass are efficient , running in @xmath10 time for graphs embedded on orientable surfaces and non - orientable surfaces . ( we usually omit the mention of @xmath11 and @xmath12 by assuming @xmath13 and @xmath14 are constant , but we later bound @xmath11 as singly exponential in a polynomial in @xmath15 and @xmath14 and @xmath12 as singly exponential in @xmath14 . ) in contrast , the problems we consider are apx - complete ( and @xmath16-approximable ) for general graphs . we build upon recent ptas framework of borradaile , klein , and mathieu @xcite for subset - connectivity problems on planar graphs . in fact , our result is strictly more general : any problem to which the previous planar - graph framework applies automatically works in our framework as well , resulting in a ptas for bounded - genus graphs . for example , borradaile , klein and pritchard @xcite have recently claimed a ptas for the _ @xmath17-edge - connectivity survivable network design _ problem using the planar framework . this will imply a similar result in bounded genus graphs . in contrast to the planar - graph framework , our ptases have the attractive feature that they run correctly on all graphs with the performance degrading with genus . our techniques for attacking bounded - genus graphs include two recent results from soda 2007 : decompositions into bounded - treewidth graphs via contractions @xcite and fast algorithms for finding the shortest noncontractible cycle @xcite . we also use a simplified version of an algorithm for finding a short sequence of loops on a topological surface from soda 2005 @xcite , and sophisticated dynamic programming . all graphs @xmath18 have @xmath19 vertices , @xmath20 edges and are undirected with edge lengths ( weights ) . the length of an edge @xmath21 , subgraph @xmath22 , and set of subgraphs @xmath23 are denoted @xmath24 , @xmath25 and @xmath26 , respectively . the shortest distance between vertices @xmath4 and @xmath5 in graph @xmath27 is denoted @xmath28 . the boundary of a graph @xmath27 embedded in the plane is denoted by @xmath29 . for an edge @xmath30 , we define the operation of _ contracting _ @xmath21 as identifying @xmath31 and @xmath32 and removing all loops and duplicate edges . we use the basic terminology for embeddings as outlined in @xcite . in this paper , an embedding refers to a _ 2-cell embedding _ , i.e. a drawing of the vertices and faces of the graph as points and arcs on a surface such that every face is homeomorphic to an open disc . such an embedding can be described purely combinatorially by specifying a _ rotation system _ , for the cyclic ordering of edges around vertices of the graph , and a _ signature _ for each edge of the graph ; we use this notion of a _ combinatorial embedding_. a combinatorial embedding of a graph @xmath27 naturally induces such a 2-cell embedding on each subgraph of @xmath27 . we only consider compact surfaces without boundary . when we refer to a planar embedding , we actually mean an embedding in the 2-sphere . if a surface contains a subset homeomorphic to a mbius strip , it is _ non - orientable _ ; otherwise it is _ orientable_. for a 2-cell embedded graph @xmath27 with @xmath33 facial walks , the number @xmath34 is called the euler genus of the surface . the euler genus is equal to twice the usual genus for orientable surfaces and equals the usual genus for non - orientable surfaces . the _ dual _ of an embedded graph @xmath27 is defined as having the set of faces of @xmath27 as its vertex set and having an edge between two vertices if the corresponding faces of @xmath27 are adjacent . we denote the dual graph by @xmath35 and identify each edge of @xmath27 with its corresponding edge in @xmath35 . a cycle of an embedded graph is _ contractible _ if it can be continuously deformed to a point ; otherwise it is _ non - contractible_. the operation of _ cutting along a 2-sided cycle _ @xmath36 is essentially : partition the edges adjacent to @xmath36 into left and right edges and replace @xmath36 with two copies @xmath37 and @xmath38 , adjacent to the left or right edges , accordingly . the inside of these new cycles is `` patched '' with two new faces . if the resulting graph is disconnected , the cycle is called _ separating _ , otherwise _ non - separating_. cutting along a 1-sided cycle @xmath36 on non - orientable surfaces is defined similarly , only that @xmath36 is replaced by one bigger cycle @xmath39 that contains every edge of @xmath36 exactly twice . next we define the notions related to treewidth as introduced by robertson and seymour @xcite . a _ tree decomposition _ of a graph @xmath27 is a pair @xmath40 , where @xmath41 is a tree and @xmath42 is family of subsets of @xmath43 , called _ bags _ , such that 1 . every vertex of @xmath27 appears in some bag of @xmath44 ; 2 . for every edge @xmath30 of @xmath27 , there exists a bag that contains both @xmath31 and @xmath32 ; 3 . for every vertex @xmath32 of @xmath27 , the set of bags that contain @xmath32 form a connected subtree @xmath45 of @xmath46 . the _ width _ of a tree decomposition is the maximum size of a bag in @xmath44 minus @xmath47 . the _ treewidth _ of a graph @xmath27 , denoted by @xmath48 , is the minimum width over all possible tree decompositions of @xmath27 . the input graph is @xmath49 and has genus @xmath50 ; the terminal set is @xmath51 . we assume @xmath52 is equipped with a combinatorial embedding ; such an embedding can be found in linear time , if the genus is known to be fixed , see @xcite . let @xmath53 be the considered subset - connectivity problem . in section [ sect : preprocess ] , we show how to find a subgraph @xmath54 of @xmath52 , so that for @xmath55 any @xmath56-approximate solution of @xmath53 in @xmath52 also exists in @xmath27 . hence , we may use @xmath27 instead of @xmath52 in the rest of the paper . note that as a subgraph of @xmath52 , @xmath27 is automatically equipped with a combinatorial embedding . let @xmath57 denote the length of an optimal steiner tree spanning terminals @xmath51 . we define @xmath58 to be the length of an optimal solution to problem @xmath53 . for the problems that we solve , we require that @xmath59 and in particular that @xmath60 . the constant @xmath61 will be used in section [ sec : mg - proof ] and is equal to @xmath9 for both the subset tsp and @xmath2-edge - connectivity problems . this requirement is also needed for the planar case ; see @xcite . because @xmath62 , upper bounds in terms of @xmath57 hold for all the problems herein . as a result , we can safely drop the @xmath53 subscript throughout the paper . we show how to obtain a @xmath63 solution for a fixed constant @xmath64 . to obtain a @xmath65 solution , we can simply use @xmath66 as input to the algorithm . in @xcite and @xcite , borradaile , klein and mathieu developed a ptas for the steiner tree problem in planar graphs . the method involves finding a grid - like subgraph called the _ mortar graph _ that spans the input terminals and has length @xmath67 . the set of feasible steiner trees is restricted to those that cross between adjacent faces of the mortar graph only at a small number ( per face of the mortar graph ) of pre - designated vertices called _ portals_. a structure theorem guarantees the existence of a nearly optimal solution ( one that has length at most @xmath68 ) in this set . we review the details that are relevant to this work and generalize to genus-@xmath14 graphs . here we define the mortar graph in such a way that generalizes to higher genus graphs . a path @xmath69 in a graph @xmath27 is _ @xmath70-short in @xmath27 _ if for every pair of vertices @xmath4 and @xmath5 on @xmath69 , the distance from @xmath4 to @xmath5 along @xmath69 is at most @xmath71 times the distance from @xmath4 to @xmath5 in @xmath27 : @xmath72 . given a graph @xmath27 embedded on a surface and a set of terminals @xmath51 , a mortar graph is a subgraph of @xmath27 with the following properties ( where @xmath73 , to be defined later , will depend polynomially on @xmath74 and @xmath14 ) : given an embedded graph @xmath27 and a set of terminals @xmath51 , a mortar graph is a subgraph @xmath75 of @xmath27 spanning @xmath51 such that each facial walk of @xmath75 encloses an area homeomorphic to an open disk . for each face of @xmath75 , we construct a _ brick _ @xmath76 of @xmath27 by cutting @xmath27 along the facial walk . @xmath76 is the subgraph of @xmath27 embedded inside the face , including the facial walk . each brick satisfies the following properties : 1 . @xmath76 is planar . the boundary of @xmath76 is the union of four paths in the clockwise order @xmath77 , @xmath78 , @xmath79 , @xmath80 . 3 . every terminal of @xmath51 that is in @xmath76 is on @xmath78 or on @xmath80 . @xmath78 is 0-short in @xmath76 , and every proper subpath of @xmath80 is @xmath13-short in @xmath76 . there exists a number @xmath81 and vertices @xmath82 ordered from left to right along @xmath80 such that , for any vertex @xmath4 of @xmath83 , the distance from @xmath4 to @xmath84 along @xmath80 is less than @xmath13 times the distance from @xmath4 to @xmath78 in @xmath76 : @xmath85 . the mortar graph and the set of bricks are illustrated in figures [ fig : mg ] ( a ) , ( b ) and ( c ) . constructing the mortar graph for planar graphs first involves finding a 2-approximate steiner tree @xmath46 and cutting open the graph along @xmath46 creating a new face with boundary @xmath22 and then : 1 . finding shortest paths between certain vertices of @xmath22 . these paths result in the @xmath78 and @xmath80 boundaries of the bricks.[mg : step-3 ] 2 . finding shortest paths between vertices of the paths found in step [ mg : step-3 ] . these paths are called _ columns _ , do not cross each other , and have a natural order . [ mg : step-4 ] 3 . taking every @xmath73th path found in step [ mg : step-4 ] . these paths are called _ supercolumns _ and form the @xmath79 and @xmath77 boundaries of the bricks . [ mg : step-5 ] the mortar graph is composed of the edges of @xmath46 ( equivalently , @xmath22 ) and the edges found in steps [ mg : step-3 ] and [ mg : step-5 ] . in @xcite , it is shown that the total length of the mortar graph edges is @xmath86 . the correction is forthcoming in a journal version . ] . for the purposes of this paper , we bound the length of the mortar graph in terms of @xmath25 . the following theorem can be easily deduced from @xcite and @xcite : @xcite [ thm : planar - mg ] given a planar graph with boundary @xmath22 containing the terminals and a number @xmath73 , there is a mortar graph containing @xmath22 whose length is at most @xmath87 and whose supercolumns have length at most @xmath88 . the mortar graph can be found in @xmath89 time . we will use this theorem to prove the existence of a mortar graph for genus-@xmath14 embedded graphs . section [ sec : mg - proof ] is devoted to proving the following theorem : [ thm : genus - mortar - graph ] let @xmath27 be an embedded graph with fixed genus @xmath14 . let @xmath51 be a subset of vertices . for @xmath90 and @xmath91 , there is a mortar graph @xmath75 of @xmath27 such that the length of @xmath75 is @xmath92 and the supercolumns of @xmath75 have length @xmath93 . the mortar graph can be found in @xmath89 time . along with the mortar graph , borradaile et al . @xcite define an operation @xmath94 called _ brick - copy _ that allows a succinct statement of the structure theorem . for each brick @xmath76 , a subset of @xmath95 vertices are selected as _ portals _ such that the distance along @xmath96 between any vertex and the closest portal is at most @xmath97 . for every brick @xmath76 , embed @xmath76 in the corresponding face of @xmath75 and connect every portal of @xmath76 to the corresponding vertex of @xmath75 with a zero - length _ portal edge _ : this defines @xmath98 . @xmath98 is illustrated in figure [ fig : mg ] ( d ) . we denote the set of all portal edges by @xmath99 . the following simple lemma , proved in @xcite holds also for bounded - genus graphs : @xcite [ lem : soln ] if @xmath100 is a connected subgraph of @xmath101 , then @xmath102 is a connected subgraph of @xmath27 spanning the same nodes of @xmath27 . the following structure theorem is the heart of the correctness of the ptass . let @xmath27 be an embedded graph of genus @xmath14 . there exists a constant @xmath103 depending polynomially on @xmath104 and @xmath105 such that @xmath106 here @xmath107 for steiner tree and @xmath2-edge connectivity and @xmath108 for subset tsp . recall that @xmath73 and @xmath104 depend polynomially on @xmath74 and @xmath14 ( theorem [ thm : genus - mortar - graph ] ) . since the bricks are always planar , the proof for bounded - genus graphs follows as for the planar cases : see @xcite for the steiner tree problem and @xcite for the @xmath2-edge - connectivity problem . in the remainder of this section , we sketch the proof for the subset - tsp problem . we will use the following lemma due to arora : [ lem : patching ] let @xmath80 be any line segment of length @xmath109 and @xmath110 be a closed path that crosses @xmath80 at least thrice . then we can break the path in all but two of these places and add to it line segments lying on @xmath80 of total length at most @xmath111 such that @xmath110 changes into a closed path @xmath112 that crosses @xmath80 at most twice . this lemma applies to embedded graphs as well . note : the patching process connects paths in the tour that end on a common side of @xmath80 by a subpath of @xmath80 . let @xmath22 be a subgraph of a graph @xmath27 and let @xmath69 be a path in @xmath22 . a _ joining vertex _ of @xmath22 with @xmath69 is a vertex of @xmath69 that is the endpoint of an edge of @xmath113 . the proof of theorem [ sec : structure - thm ] for subset - tsp follows from the following tsp property of bricks using the same technique as the planar case for , e.g. , steiner tree . see @xcite for details . the condition that @xmath79 and @xmath77 are crossed at most twice each is achieved by including two copies of each supercolumn in the solution and rerouting the tour around these copies . [ thm : tsp - prop ] let @xmath76 be a brick of graph @xmath27 with boundary @xmath114 ( where @xmath79 and @xmath77 are supercolumns ) . let @xmath46 be a tour in @xmath27 such that @xmath46 crosses @xmath79 and @xmath77 at most @xmath9 times each . let @xmath22 be the intersection of @xmath46 with @xmath76 . then there is another subgraph of @xmath76 , @xmath115 , such that : 1 . @xmath115 has at most @xmath116 joining vertices with @xmath96 . 2 . @xmath117 . 3 . there is a tour in the edge set @xmath118 that spans the vertices in @xmath119 . in the above , @xmath120 . let @xmath22 be the subgraph of @xmath46 that is strictly enclosed by @xmath76 ( i.e. , @xmath22 contains no edges of @xmath96 ) . we can write @xmath22 as the union of 4 sets of minimal @xmath96-to-@xmath96 paths @xmath121 where paths in : @xmath122 start and end on @xmath80 ; @xmath123 start and end on @xmath78 ; @xmath124 start on @xmath79 or @xmath77 and end on @xmath96 ; @xmath125 start on @xmath80 and end on @xmath78 . because @xmath46 crosses @xmath79 and @xmath77 at most 4 times , @xmath126 and @xmath124 results in at most 8 joining vertices with @xmath96 . for each path @xmath127 , let @xmath128 be the minimal subpath of @xmath78 that spans @xmath69 s endpoints and let @xmath129 be the resulting set . similarly define @xmath130 . because @xmath78 is 0-short and @xmath80 is @xmath13-short , @xmath131 because @xmath129 and @xmath130 are subpaths of @xmath96 , they result in no joining vertices with @xmath96 . it remains to deal with the paths in @xmath125 . let @xmath132 ( where @xmath81 ) be the vertices of @xmath80 guaranteed by the properties of the bricks ( see section [ sec : mg ] ) . let @xmath133 be the subset of paths of @xmath125 that start on @xmath80 strictly between @xmath84 and @xmath134 . let @xmath135 be the remaining paths ( note : @xmath136 ) . if @xmath137 , we do the following : let @xmath138 be the path in @xmath133 whose endpoint @xmath4 on @xmath80 is closest to @xmath134 . let @xmath139 be the subpath of @xmath80 from @xmath84 to @xmath4 . by the properties of the bricks , @xmath140 . apply the patching lemma to the tour @xmath46 and path @xmath139 : as a result , at most two paths of @xmath133 occur right before ( or after ) crossings of @xmath51 in the new tour @xmath141 . let @xmath142 whose endpoints are not crossings in @xmath141 : @xmath143 . let @xmath144 be the subpaths of @xmath139 that are added to the tour . by the patching lemma , @xmath145 while @xmath146 , we do the following : let @xmath69 be any path in @xmath142 . as a result of the patching process @xmath69 is connected to another path @xmath147 in @xmath133 via a path @xmath148 : that is , @xmath149 is a subpath of the tour . let @xmath150 be the minimal subpath of @xmath78 connecting the endpoints of @xmath69 and @xmath147 . because @xmath78 is @xmath151-short , @xmath152 . by the patching process , the endpoints of @xmath69 and @xmath147 on @xmath80 are spanned by a part of the tour `` on the other side '' of @xmath139 . remove @xmath139 from @xmath144 and add @xmath150 . remove @xmath69 and @xmath147 from @xmath133 . when we are done , @xmath153 is empty . let @xmath154 be the set resulting from all such replacements . let @xmath155 be the union of @xmath135 , @xmath156 and @xmath154 over all @xmath157 . because @xmath154 is a set of subpaths of @xmath78 and @xmath80 , these paths result in no joining vertices with @xmath158 . because @xmath159 , these paths result in at most 4 joining vertices with @xmath96 . because @xmath160 , these paths have at most @xmath161 joining vertices with @xmath96 . therefore @xmath162 has at most @xmath163 joining vertices with @xmath158 . because the only added length is introduced via the patching process , by equation ( [ eq : q ] ) , @xmath164 let @xmath115 be the union of the paths in @xmath124 , @xmath165 , @xmath129 , and @xmath155 . combining equations ( [ eq : n ] ) and ( [ eq : ns ] ) , we find that @xmath166 . by construction , the edges in @xmath118 contains a tour . @xmath115 has @xmath167 joining vertices with @xmath96 . in @xcite , klein gave a framework for designing ptases in planar graphs that is based on finding a _ spanner _ for a problem , a subgraph containing a nearly optimal solution having length @xmath67 . it is possible , using the techniques in this paper and in @xcite , to find such a spanner for bounded - genus graphs . we omit the details in favour of generalizing the framework of borradaile et al . @xcite . while both methods result in @xmath89 algorithms , the first method is doubly exponential in a polynomial in @xmath14 and @xmath74 and the second is singly exponential . ) . our presentation here is chosen to unify the methods for all problems studied . ] the idea in @xcite is to perform dynamic programming over the bricks of the brick decomposition after performing a _ thinning _ step which groups the bricks into manageable groups . to this end , the operation _ brick - contraction _ @xmath168 is defined to be the application of the operation @xmath94 followed by contracting each brick to become a single vertex of degree at most @xmath95 ( see figure [ fig : mg](e ) ) . the thinning algorithm decomposes the mortar graph @xmath75 into parts , called _ parcels _ , of _ bounded dual radius _ ( implying bounded treewidth ) . applying @xmath168 to each parcel maintains bounded dual radius . the dynamic program computes optimal steiner trees inside the bricks using the method of @xcite only at the leaves of the dynamic programming tree , thus eliminating the need of an a - priori constructed spanner . the interaction between subproblems of the dynamic programming is restricted to the portals , of which there are a few . for embedded graphs with genus @xmath169 , the concept of bounded dual radius does not apply in the same way . we deal with treewidth directly by applying the following contraction decomposition theorem due to demaine et al . @xcite : ( * ? ? ? * theorem 1.1 ) [ thm : genus - baker ] for a fixed genus @xmath14 , and any integer @xmath170 and for every graph @xmath27 of euler genus at most @xmath14 , the edges of @xmath27 can be partitioned into @xmath171 sets such that contracting any one of the sets results in a graph of treewidth at most @xmath172 . furthermore , such a partition can be found in @xmath173 time . ( recent techniques @xcite have improved the above running time to @xmath89 . ) we apply the above theorem directly to @xmath174 and contract a set of edges @xmath175 in @xmath174 . in @xmath174 , we modify the definition of contraction : after contracting an edge , we do not delete parallel portal edges . because portal edges connect the mortar graph to the bricks , they are not parallel in the graph in which we find a solution via dynamic programming . with this modified definition , we have the following algorithm : [ lem : thinning ] the algorithm thinning(@xmath176 ) returns a set of edges @xmath175 and a tree decomposition @xmath40 of @xmath174 , so that a. the treewidth of @xmath40 is at most @xmath177 where @xmath178 ; b. every brick is either marked as ignored or none of its portal edges are in @xmath175 ; c. @xmath179 ; we first verify that @xmath40 is indeed a tree decomposition . for a vertex @xmath32 and a tree decomposition @xmath180 , let @xmath181 denote the subtree of @xmath141 that contains @xmath32 in all of its bags . let us denote the tree decomposition of step ( 4 ) by @xmath182 . for each brick vertex @xmath183 and each of its portals @xmath184 , we know that @xmath185 is connected and @xmath186 is connected and that these two subtrees intersect ; it follows that after the replacement in step ( 5.2 ) , we have that @xmath187 is a connected subtree of @xmath46 and hence , @xmath40 is a correct tree decomposition . note that theorem [ thm : genus - baker ] guarantees a tree decomposition of width @xmath172 if any of @xmath188 are contracted ; and in step ( 3 ) , we only add to the set of edges to be contracted . hence , the treewidth of @xmath182 is indeed @xmath172 and with the construction in line ( 5.1 ) , the size of each bag will be multiplied by a factor of at most @xmath95 . this shows the correctness of claim ( i ) . the correctness of claim ( ii ) is immediate from the construction in line ( 3 ) . it remains to verify claim ( iii ) . let @xmath150 denote the weight of @xmath174 after setting the weights of the portal edges according to step ( 1 ) of the algorithm . we have that @xmath189 hence , the weight of @xmath175 , as selected in step ( 2 ) , is at most @xmath190 . the operation in step ( 3 ) does not add to the weight of @xmath175 : when the boundary of a face @xmath191 is added to @xmath175 , its weight is subtracted again when resetting the weights of the portal edges back to @xmath151 in step ( 6 ) . if a brick is `` ignored '' by thinning , the boundary of its enclosing mortar graph face is completely added to @xmath175 . because @xmath175 can be added to the final solution , every potential connection through that brick can be rerouted through @xmath175 around the boundary of the brick . the interior of the brick is not needed . an almost standard dynamic programming algorithm for bounded - treewidth graphs ( cf . @xcite ) can be applied to @xmath192 and @xmath40 . however , for the leaves of the tree decomposition that are added in step ( 5.2 ) of the thinning procedure , the cost of a subset of portal edges is calculated as , e.g. , the cost of the minimum steiner tree interconnecting these portals in the corresponding brick . because the bricks are planar , this cost can be calculated by the algorithm of erickson et al . @xcite for steiner tree or @xcite for @xmath9-edge connectivity . because all the portal edges of this brick are present in this bag ( recall that we do not delete parallel portal edges after contractions ) , all possible solutions restricted to the corresponding brick will be considered . because the contracted brick vertices only appear in leaves of the dynamic programming tree , the rest of the dynamic programming algorithm can be carried out as in the standard case , considering the portal edges , whenever they occur again , as having zero length . [ [ analysis - of - the - running - time . ] ] * analysis of the running time . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as was shown in @xcite , the total time spent in the leaves of the dynamic programming is @xmath193 . the rest of the dynamic programming takes time @xmath194 . the running time of the thinning algorithm is dominated by the contraction decomposition of theorem [ thm : genus - baker ] which is @xmath195 @xcite . hence , the total time is @xmath196 for the general case ; particularly , this is singly exponential in @xmath74 and @xmath14 , as desired . let @xmath49 be the input graph of genus @xmath50 and @xmath51 be the terminal set . in a first preprocessing step , we delete a number of unnecessary vertices and edges of @xmath52 to obtain a graph @xmath54 of genus @xmath197 that still contains every @xmath56-approximate solution for terminal set @xmath51 for all @xmath198 while fulfilling certain bounds on the length of shortest paths . in the next step , we find a _ cut graph _ @xmath199 of @xmath27 that contains all terminals and whose length is bounded by a constant times @xmath57 . we cut the graph open along @xmath199 , so that it becomes a planar graph with a simple cycle @xmath200 as boundary , where the length of @xmath200 is twice that of @xmath199 . see figure [ fig : cutgraph ] for an illustration . afterwards , the remaining steps of building the mortar graph can be the same as in the planar case , by way of theorem [ thm : planar - mg ] . for an edge @xmath201 in @xmath52 , we let @xmath202 and say that @xmath21 is _ at distance _ @xmath203 from @xmath204 . if the root vertex represents a contracted graph @xmath22 , we use the same terminology with respect to @xmath22 . any deleted vertex or edge is at distance @xmath205 from any terminal and hence , can not be part of a @xmath56-approximation for any @xmath206 . we call the resulting graph @xmath54 and henceforth use @xmath27 instead of @xmath52 in our algorithm . the preprocessing step can be accomplished in linear time : @xmath207 can be calculated with the recent improvement on mehlhorn s algorithm @xcite by tazari and mller - hannemann @xcite , and the shortest path tree with henzinger et al.s algorithm @xcite . trivially , we have start again with @xmath207 contracted to a vertex @xmath204 . we construct our cut graph as a _ system of loops _ rooted at @xmath204 : iteratively find short non - separating cycles through @xmath204 and cut the graph open along each cycle . erickson and whittlesey @xcite showed that , for orientable surfaces , taking the _ shortest _ applicable cycle at each step results in the shortest system of loops through @xmath204 . they suggest an implementation of their algorithm using the _ tree - cotree decomposition _ , introduced by eppstein @xcite , that runs in linear time on bounded - genus graphs . a tree - cotree decomposition of an embedded graph @xmath27 is a triple @xmath209 , so that @xmath46 is a spanning tree of @xmath27 rooted at @xmath204 , @xmath210 is a spanning tree of the dual of @xmath211 , and @xmath212 is the remaining set of edges ( recall that we identify the edges of @xmath27 and @xmath35 ) . eppstein showed that the set of elementary cycles @xmath213 is a cut graph of @xmath27 where @xmath214 is the cycle formed by the paths in @xmath46 from @xmath204 to the endpoints of @xmath21 plus the edge @xmath21 . eppstein s decomposition also works for non - orientable embeddings . as we only need to bound the length of our cut graph , we present a simpler algorithm : [ lem : cut - graph ] the algorithm planarize returns a cut graph @xmath199 , so that cutting @xmath27 open along @xmath199 results in a planar graph @xmath215 with a distinguished face whose facial walk @xmath200 clearly , @xmath217 is tree - cotree decomposition of @xmath27 and so , by eppstein s lemma @xcite , @xmath199 is a cut graph . by euler s formula , we get that @xmath218 , the euler genus of @xmath27 . each edge @xmath219 completes a ( non - surface - separating , not - necessarily simple ) cycle as follows : a shortest path @xmath220 from @xmath207 to @xmath32 , the edge @xmath21 , a shortest path @xmath221 from @xmath222 to @xmath207 and possibly a path @xmath223 in @xmath207 . by proposition [ prop : preprocess ] , we know that @xmath21 is at distance at most @xmath208 from @xmath207 and so , @xmath220 , @xmath221 , and at least one of @xmath224 and @xmath225 have length at most @xmath226 . hence , we have that @xmath227 . because there are ( exactly ) @xmath14 such cycles in @xmath199 , we get that @xmath228 because @xmath199 is a cut graph , it follows that it consists of a single facial walk @xmath229 ; this follows easily from euler s formula , because @xmath199 has euler genus @xmath14 ( because @xmath230 is planar ) , with some @xmath171 vertices and @xmath231 edges . so , @xmath229 contains every edge of @xmath199 exactly twice ( cf . @xcite ) , i.e. @xmath232 . cutting the graph open along @xmath229 results in a planar graph with a simple cycle @xmath233 as its boundary , as desired ( see fig . [ fig : cutgraph ] ) . as mentioned in the previous section , @xmath207 and @xmath234 can be computed in linear time on bounded - genus graphs @xcite . @xmath235 can be obtained , for example , by a simple breadth - first - search in the dual . the remaining steps can also easily be implemented in linear time . let @xmath215 be the result of planarizing @xmath27 as guaranteed by lemma [ lem : cut - graph ] . @xmath215 is a planar graph with boundary @xmath200 spanning @xmath51 and with length @xmath236 . let @xmath75 be the mortar graph guaranteed by theorem [ thm : planar - mg ] . every edge of @xmath75 corresponds to an edge of @xmath27 . let @xmath237 be the subgraph of @xmath27 composed of edges corresponding to @xmath75 . every face @xmath33 of @xmath75 ( other than @xmath200 ) corresponds to a face @xmath238 of @xmath237 and the interior of @xmath238 is homeomorphic to a disk on the surface in which @xmath27 is embedded . it is easy to verify that @xmath237 is a mortar graph of @xmath27 satisfying the length bounds of theorem [ thm : genus - mortar - graph ] . g. borradaile and p. n. klein . the two - edge connectivity survivable network problem in planar graphs . in _ icalp 08 : proceedings of the 35th international colloquium on automata , languages and programming _ , volume 5125 of _ lecture notes in computer science _ , pages 485501 . springer , 2008 . g. borradaile , c. mathieu , and p. n. klein . a polynomial - time approximation scheme for steiner tree in planar graphs . in _ soda 07 : proceedings of the 18th annual acm - siam symposium on discrete algorithms _ , pages 12851294 , 2007 . g. borradaile , c. mathieu , and p. n. klein . steiner tree in planar graphs : an @xmath89 approximation scheme with singly exponential dependence on epsilon . in _ wads 07 : proceedings of the 10th workshop on algorithms and data structures _ , volume 4619 of _ lecture notes in computer science _ , pages 275286 . springer , 2007 . s. cabello and e. w. chambers . multiple source shortest paths in a genus g graph . in _ soda 07 : proceedings of the 18th annual acm - siam symposium on discrete algorithms _ , pages 8997 , philadelphia , pa , usa , 2007 . e. d. demaine , m. hajiaghayi , and b. mohar . approximation algorithms via contraction decomposition . in _ soda 07 : proceedings of the 18th annual acm - siam symposium on discrete algorithms _ , pages 278287 , philadelphia , pa , usa , 2007 . j. erickson and k. whittlesey . greedy optimal homotopy and homology generators . in _ soda 05 : proceedings of the 16th annual acm - siam symposium on discrete algorithms _ , pages 10381046 , philadelphia , pa , usa , 2005 . s. tazari and m. mller - hannemann . shortest paths in linear time on minor - closed graph classes , with an application to steiner tree approximation . , in press , 2008 . an extended abstract appeared in wg 08 : proceedings of the 34th workshop on graph theoretic concepts in computer science , lncs 5344 , pp . 360371 , springer , 2008 .	we present the first polynomial - time approximation schemes ( ptases ) for the following subset - connectivity problems in edge - weighted graphs of bounded genus : steiner tree , low - connectivity survivable - network design , and subset tsp . the schemes run in @xmath0 time for graphs embedded on both orientable and non - orientable surfaces . this work generalizes the ptas frameworks of borradaile , klein , and mathieu @xcite from planar graphs to bounded - genus graphs : any future problems shown to admit the required structure theorem for planar graphs will similarly extend to bounded - genus graphs . glencora borradaile erik d. demaine siamak tazari
in the present paper we further study random two - dimensional sets that satisfy the conformal invariance property combined with the restriction property , following the work of lawler , schramm and werner in @xcite and the paper of wu in @xcite . measures that satisfy conformal restriction property were introduced and first studied by lawler , schramm and werner in @xcite : for a simply connected domain @xmath2 with two marked boundary points @xmath3 and @xmath4 ( we will say `` boundary points '' instead of prime ends in the present introduction ) , they studied a class of random simply connected and relatively closed sets @xmath5 such that @xmath6 intersects @xmath7 only at @xmath3 and @xmath4 . such a set ( or rather , its distribution ) is said to satisfy _ chordal conformal restriction property _ if the following two conditions hold : * ( conformal invariance ) the law of @xmath6 is invariant under any conformal map from @xmath8 onto itself that leave the boundary points @xmath3 and @xmath4 invariant . * ( restriction ) for any simply connected subset @xmath9 of @xmath8 such that @xmath10 , the conditional distribution of @xmath6 given @xmath11 is equal to the image of the law of @xmath6 under @xmath12 , where @xmath12 is any conformal map from @xmath8 onto @xmath9 that leaves the points @xmath13 invariant ( property ( i ) actually ensures that if this holds for one such map @xmath12 , then it holds also for any other such map ) . see figure [ fig : two - point - restriction ] . and @xmath14 is a conformal map from @xmath15 onto @xmath8 that leaves @xmath13 invariant . the conditional law of @xmath16 given @xmath17 is equal to the ( unconditional ) law of @xmath6 . , scaledwidth=78.0% ] it is straightforward to check that if a random set @xmath6 satisfies this chordal conformal restriction property in one simply connected domain @xmath8 with boundary points @xmath3 and @xmath4 , then if we map @xmath8 conformally to another simply connected domain @xmath18 via some fixed deterministic map @xmath19 , then @xmath20 satisfies chordal conformal restriction in @xmath21 with boundary points @xmath22 and @xmath23 ( and property ( i ) ensures that the image law depends only on the triplet @xmath24 , and not on the particular instance of @xmath19 ) . hence , it is sufficient to study this property in one particular given domain @xmath8 , such as the unit disc or the upper half - plane . recall that conformal invariance is believed to hold in the scaling limit for a large class of two - dimensional models from statistical physics . a chordal restriction property can be interpreted as follows : on a lattice , one can associate to each set @xmath6 an energy , and the restriction property means that this energy of @xmath6 is `` intrinsic '' in the sense that it does not depend on the domain in which it lives but only on @xmath6 ( and the extremal points @xmath3 and @xmath4 ) itself . for example , for a simple random walk on a square lattice in a discretized domain @xmath8 , which is conditioned to go from one boundary point @xmath3 to another boundary point @xmath4 , the probability of a given path @xmath25 will be proportional to @xmath26 where @xmath27 is the number of steps of @xmath25 . this weight @xmath26 is then intrinsic because it only depends on the path @xmath25 itself but not on the domain @xmath8 . as a consequence , if we condition such a random walk excursion from @xmath3 to @xmath4 in @xmath8 to stay in a subdomain @xmath9 of @xmath8 which still has @xmath3 and @xmath4 on its boundary , then one gets exactly a random walk excursion from @xmath3 to @xmath4 in this smaller domain @xmath9 . in the continuous limit , the brownian excursion from @xmath3 to @xmath4 in @xmath8 ( or rather its `` filling '' in order to get a simply connected set ) does indeed satisfy chordal conformal restriction . chordal restriction measures therefore describe the natural conformally invariant and intrinsic ways to join two boundary points in a simply connected domain . lawler , schrammm and werner proved in @xcite that such measures are fully characterized by one real parameter @xmath28 ( here and in the sequel , this means that there is an injection from the set of conformal restriction measures into @xmath29 ) that can be described as follows . for each chordal conformal restriction measure , there exists a positive @xmath28 such that for all @xmath30 such that @xmath15 is again a simply connected domain and @xmath31 , one has @xmath32 ( it is straightforward to see that the product @xmath33 does not depend on the choice of @xmath14 among the one - dimensional family of conformal maps from @xmath34 onto @xmath8 that leave @xmath3 and @xmath4 invariant ) . conversely , it is easy to see that for each given @xmath28 , there exists at most one law of @xmath6 satisfying this relation for all @xmath35 . the more challenging part is then to investigate for which values of @xmath28 such a random set @xmath6 does indeed exist . lawler , schramm and werner showed in the same paper @xcite that such a probability measure exists if and only if @xmath36 , and they provided a detailed description of these measures : when the smallest value @xmath37 , it is exactly the law of chordal sle@xmath0 from @xmath3 to @xmath4 in @xmath8 , so that @xmath6 is almost surely a simple curve from @xmath3 to @xmath4 in @xmath8 . when @xmath38 , they showed that @xmath6 is almost surely not a simple curve anymore ( the case @xmath39 is the aforementioned law of the filling of a brownian excursion in @xmath8 from @xmath3 to @xmath4 ) . in fact , they also described the law of the right boundary of @xmath6 for all @xmath40 in terms of a variant of sle@xmath0 ( the sle@xmath41 processes ) , which showed in particular that the boundary of all these random sets @xmath6 look locally like an sle@xmath0 or equivalently like the boundary of a two - dimensional brownian motion . intuitively , the larger @xmath28 is , the `` fatter '' the random set @xmath6 is . for instance , one can prove that @xmath6 will contain cut - points if and only if @xmath42 , see @xcite ( where the law of the left boundary given the right one is also described ) . these chordal restriction measures can in fact be viewed as special limiting cases of a larger class of restriction measures defined similarly , but with three marked boundary points instead of two ; this class will be the topic of the present paper : for a simply connected domain @xmath2 with three ( different ) marked boundary points @xmath13 and @xmath43 , we consider probability measures supported on simply connected and relatively closed @xmath5 such that @xmath44 . such a measure is said to satisfy the _ trichordal conformal restriction property _ if for @xmath45 such that @xmath9 is simply connected and @xmath46 , the law of @xmath6 conditioned to stay in @xmath9 is identical to the law of @xmath47 where @xmath12 is the unique conformal map from @xmath8 onto @xmath9 that leaves the points @xmath48 invariant , see figure [ fig : trichordal - restriction ] . one expects intuitively this family of measures to be larger ( ie . parametrized by more than one real parameter ) because condition ( i ) of the chordal case is no longer required , so that the measures are invariant under a smaller semi - group of transformations . is the conformal map from @xmath15 onto @xmath8 that leaves @xmath13 and @xmath43 invariant . the law of @xmath49 given @xmath17 is equal to the ( unconditioned ) law of @xmath6 . , scaledwidth=78.0% ] before going into the details of this trichordal case , let us say a few words about the radial case studied by wu in @xcite . given that the group of conformal automorphism of a domain is three - dimensional , this is the other natural variant to consider : for two simply connected domains , if we fix one interior point and one boundary point for each domain , then there is a unique conformal map from one domain onto the other that sends both of the prefixed interior point and boundary point of one domain to the corresponding points of the other . for a simply connected domain @xmath2 with an interior point @xmath3 and a boundary point @xmath4 , one looks at probability measures supported on simply connected and relatively closed sets @xmath5 such that @xmath50 and @xmath51 . such a measure is said to satisfy _ radial conformal restriction property _ if for @xmath45 such that @xmath9 is simply connected and @xmath52 , @xmath6 conditioned to stay in @xmath9 has the same law as @xmath47 where @xmath12 is the unique conformal map from @xmath8 onto @xmath9 that leaves the points @xmath13 invariant . wu showed that this family was characterized by two real parameter : for such measures , there exists @xmath28 and @xmath53 such that for all @xmath30 such that @xmath15 is again a simply connected domain that contains @xmath3 in its interior and @xmath4 on its boundary , one has @xmath54 she also showed that the range of admissible values of @xmath28 and @xmath53 is given by @xmath55 and @xmath56 where @xmath57 is the so - called disconnection exponent of @xmath28 . among the natural conformal restriction measures , the trichordal case was in some sense , the only one that was left to be studied . if ones adds more marked points , then there will typically be no conformal map @xmath14 that fix simultaneously all these points . even if it is possible to define a similar notion of @xmath58-point conformal restriction , we will not be able to speak of one measure that satisfies conformal restriction , but one has to consider instead a family of measures indexed by conformally equivalent configurations of the @xmath58 marked points . for @xmath59 , the family of all families of @xmath58-point chordal restriction measures will be much bigger than that of chordal , trichordal or radial cases , and it will in fact be infinite dimensional . intuitively speaking , one can associate to a four - point restriction sample @xmath6 , an intrinsic conformally invariant quantity , such as the modulus @xmath60 of the quadrilateral @xmath61 in @xmath6 ( where @xmath3 , @xmath4 , @xmath43 and @xmath62 are the four marked boundary points ) . one can then weight the law of @xmath6 by any function @xmath63 with mean @xmath64 , one thus still gets a four - point restriction measure . indeed , dubdat computed in @xcite formulae about @xmath58-point chordal restriction measures which involve such functions @xmath65 . this will be further explained in the preliminary section [ sec : dubedat ] . as in the chordal and radial cases , the goal in the trichordal case is to characterize all random sets satisfying the trichordal conformal restriction property . recall that one heuristic motivation is that one would like to describe all possible ways to connect three boundary points of a domain via random sets with `` intrinsic energy '' , and one major question is to see what the `` thinest '' sets @xmath6 look like . let us first make the following observation : if we consider @xmath66 three independent chordal restriction measures in @xmath8 that intersect @xmath7 respectively at pairs of points @xmath67 , then the filling @xmath6 of @xmath68 obviously satisfies trichordal restriction . moreover , if @xmath69 respectively have exponents @xmath70 then @xmath71 this suggests that the family of trichordal restriction measures might depend on three parameters . this indeed turns out to be the case . the following characterization is of the same type as the characterization of the chordal and radial cases : [ prop : charac][characterization ] for each trichordal restriction measure , one can find three real numbers @xmath72 so that for all @xmath30 such that @xmath15 is simply connected and @xmath73 , one has @xmath74 conversely , for each @xmath75 , there exists at most one probability measure so that this holds . we then denote the law of @xmath6 by @xmath76 . let us insist on the fact that this result does not tell us for which values of @xmath28 , @xmath53 and @xmath25 such a @xmath76 exists . the main contribution of the present paper will be to give the exact range of @xmath72 for which @xmath76 exist and to construct the corresponding restriction sets . the exponents @xmath72 of the trichordal case play symmetric roles and are interrelated , as opposed to the radial case . let us first make a few comments : * first note that the points @xmath48 cut @xmath7 into three arcs , namely @xmath77 and @xmath78 . if @xmath79 is a subset of the arc @xmath80 ( which is the case depicted in figure [ fig : trichordal - restriction ] ) , then it is easy to see that @xmath81 and @xmath82 . in order for the right - hand side of ( [ thm1 ] ) to be a probability , @xmath28 can therefore not be much bigger than both @xmath53 and @xmath25 . by symmetry , this indicates that each of the three exponents needs to be bounded from above by a certain function of the other two . on the other hand , it is easy to see that @xmath83 is always smaller than @xmath64 . hence , one can still say that in the symmetric family @xmath84 , the smaller @xmath28 is , the skinnier @xmath6 is ( because for each @xmath35 , the probability to intersect @xmath35 is an increasing function of @xmath28 ) . * another type of condition that one should expect is that @xmath28 , @xmath85 and @xmath25 should all be greater or equal to @xmath86 . intuitively , a random set @xmath6 with law @xmath76 should look like , in a infinitesimal neighborhood of one marked point ( say @xmath3 ) , a chordal restriction measure of the corresponding exponent ( which is @xmath28 in the case of @xmath3 ) . more precisely , if we let the point @xmath4 tend to the point @xmath43 , the limiting measure can be proved to be equal to the chordal restriction measure of exponent @xmath28 . since chordal restriction measures of exponent @xmath28 as characterized by ( [ two - point - formula ] ) exist only for @xmath36 , the same condition should hold here . a natural question is what is smallest @xmath28 for which the measure @xmath87 exists . for the chordal case , the measure of exponent @xmath37 corresponds to a simple curve , which is the thinest that one can obtain . if the measure @xmath88 exists , then the corresponding random set would a.s . be a simple curve near each of the boundary points @xmath48 . would the three legs of such an sle@xmath0 spider then merge at one single point in the middle or would there be a fat `` body '' set near the center ? the previously described method of taking the filled union of chordal restriction measures only constructs such measure with @xmath89 , which is unlikely to be the thinest choice . the answer to all these questions turns out to be somewhat surprising : a particular case of the following theorem is that that the smallest @xmath28 for which @xmath90 exists is actually @xmath91 . in this special case , the corresponding random set @xmath6 consists of three parts , that have only one common point which is the unique _ triple disconnecting point _ : when one removes this point , one disconnects @xmath6 into three disjoint connected components , that respectively contain @xmath3 , @xmath4 and @xmath43 . however , since @xmath92 , each of these three `` legs '' of this symmetric restriction spider are a bit `` fatter '' than simple curves ( even though they each of them will have quite a lot of cut - points ) . before stating this main result , let us first recall the formulas for the half - plane and whole - plane _ brownian intersection exponents _ that have been determined by lawler , schramm and werner in @xcite : @xmath93 we will use here only the formulas and not the interpretation in terms of non - intersection probabilities for brownian paths ( but the latter are of course related to restriction properties ) . [ thm : existence][existence ] the measure @xmath76 exists if and only if @xmath72 satisfy all the following conditions : * @xmath94 * @xmath95 * @xmath96 furthermore , if @xmath6 is a random set with law @xmath76 , then * @xmath6 has a triple disconnecting point if and only if @xmath97 . moreover in this case , each of the three legs away from this unique triple disconnecting point have almost surely infinitely many cut - points ( this can be viewed as a consequence of the that fact that necessarily , @xmath98 ) . see figure [ fg : tripple - cut - point ] . * the boundary point @xmath43 is a cut - point of @xmath6 if and only if @xmath99 . see figure [ fig : two - conditioned ] . + 0.4 + 0.4 note that if we define @xmath100 , @xmath101 , @xmath102 to be the respective image of @xmath28 , @xmath53 and @xmath25 under the monotone bijection from @xmath103 onto itself defined by @xmath104 , then one can easily express the conditions ( i ) , ( ii ) and ( iii ) without reference to the brownian intersection exponents : the condition ( i ) becomes @xmath105 , condition ( iii ) becomes @xmath106 ( while ( i ) implied only that this sum can not be smaller than 9 ) , and condition ( ii ) becomes three inequalities of the type @xmath107 . in other words , all these conditions can be summed up as @xmath108 we have already highlighted to special role of @xmath109 . another interesting case worth mentioning is the measure @xmath110 ( that also exhibits a triple disconnecting point ) . our construction will in fact implicitly describe this measure as being constructed by an sle@xmath0 from @xmath3 to @xmath4 , weighted according to its ( renormalized ) harmonic measure seen from @xmath43 ( this will make it more likely for the path to wander closer to @xmath43 ) , and to which one attaches a brownian excursion from @xmath43 to a point ( chosen according to this renormalized harmonic measure ) on this sle . in general , when @xmath111 , it is natural to guess that conditionally on the triple disconnecting point , @xmath6 consists of three radial restriction samples ( from the center point to the three points @xmath3 , @xmath4 and @xmath43 respectively ) conditioned not to intersect each other except at this center point . however , in practice , it would be a rather tedious and intricate venture to define this conditioning rigorously , and we will not follow this route . for the @xmath99 case , one can make a similar remark , and this time , it can be made rigorous more easily . it will indeed follow from our construction that conditioned on one branch of @xmath6 , the other branch is a chordal restriction sample in one of the connected components of the complement of the first branch . in this sense , one can view @xmath6 as the union of two chordal restriction samples conditioned not to intersect each other except at @xmath43 . an interesting example is of course @xmath112 , which corresponds to two chordal sle@xmath0 from @xmath3 to @xmath43 and from @xmath4 to @xmath43 , conditioned not to intersect . this ( and the three symmetric images when interchanging @xmath3 , @xmath4 and @xmath43 ) is the only trichordal restriction measure that consists of a simple path in @xmath113 . the family of measure @xmath114 exists exactly in the range @xmath115 $ ] , and the two cases @xmath116 and @xmath112 that we have just briefly described are the two extremal ones . the case @xmath117 corresponds to ( the filling of ) the union of two independent sle@xmath0 curves from @xmath3 to @xmath43 and from @xmath4 to @xmath43 . as explained above , when @xmath25 increases , as opposed to the symmetric case , one can not really argue anymore that the sets @xmath6 become larger ( and indeed , the @xmath118 case corresponds to a simple curve , while this is not the case for @xmath119 ) . instead , when @xmath25 grows , the sets are in some sense `` pushed towards @xmath43 '' . the proof of the characterization result will follow somewhat similar ideas than in the chordal and than in the radial cases . the proof of the existence part will be divided in two parts : first , we will construct trichordal restriction measures for all admissible values of the parameters @xmath28 , @xmath53 and @xmath25 , and then we will see that if a measure @xmath120 exists for the non - admissible part , one gets a contradiction . our construction of trichordal restriction measures will rely on a special family of sle processes with driving functions involving hypergeometric functions , that we call hypergeometric sles and denote by hsle . zhan first defined in @xcite intermediate sles to describe the inverse of sle@xmath121 processes . the intermediate sles are hypergeometric sles with special parameters . in our case , for @xmath122 , the hsle curves correspond to the boundaries of trichordal restriction samples . the construction goes in fact through several steps . first of all , we define the notion of one - sided restriction ( in this trichordal case ) which is reminiscent and analogous to the one - sided chordal restriction measures that were studied by lawler , schramm and werner . we consider measures supported on simply connected and relatively closed @xmath123 such that @xmath124 is the union of the arcs @xmath125 and @xmath80 . for all @xmath126 such that @xmath15 is again a simply connected domain and @xmath127 is a subset of the arc @xmath78 . such a measure is said to satisfy _ one - sided ( trichordal ) conformal restriction property _ if @xmath6 conditioned on @xmath17 has the same law as @xmath16 where @xmath12 is the unique conformal map from @xmath15 onto @xmath8 that preserves the points @xmath48 . see figure [ fig : one - sided ] . is the conformal map from @xmath15 onto @xmath8 that preserves the points @xmath48 . the law of @xmath16 conditioned on @xmath17 is equal to the ( unconditioned ) law of @xmath6.,scaledwidth=76.0% ] one - sided restriction measures are also characterized by ( [ thm1 ] ) with three parameters @xmath72 and we denote the measure by @xmath128 . the range of @xmath72 for which the measure @xmath128 exists is larger than the range for which the three - sided restriction measure @xmath76 exists . in particular , since we only consider @xmath35 that intersect @xmath7 at the arc @xmath129 , there is nothing that prevents @xmath28 or @xmath25 to be much bigger than the other two exponents . moreover , some negative values of @xmath53 will be allowed . we will show that the range is the following : a one - sided measure @xmath130 exists if and only if @xmath75 satisfy the following conditions : @xmath131 moreover , if @xmath6 has law @xmath130 where @xmath132 , then the point @xmath4 is on the right boundary of @xmath6 . these one - sided restriction measures @xmath128 will be constructed , in most of the cases , by a hsle@xmath0 curve . however for a certain range of parameters , this construction does not work , and we will use poisson point process of brownian excursions instead . next we will study two - sided restriction measures . their definition is similar to that of one - sided restriction , except that two - sided restriction measures are supported on @xmath6 such that @xmath133 is the union of @xmath134 and the arc @xmath80 , and the sets @xmath30 are such that @xmath127 is a subset of the union of the arcs @xmath78 and @xmath125 , see figure [ fig : two - sided ] . is the conformal map from @xmath15 onto @xmath8 that preserves the points @xmath48 . the law of @xmath16 conditioned on @xmath17 is equal to the ( unconditioned ) law of @xmath6.,scaledwidth=78.0% ] two sided restriction measures are also characterized by ( [ thm1 ] ) with three parameters @xmath72 and we denote the measure by @xmath135 . the range of @xmath72 for which the measure @xmath135 exists is given in the following theorem . the two - sided measure @xmath136 exists if and only if @xmath75 satisfy the following conditions : @xmath137 moreover , if @xmath6 has the law @xmath136 where @xmath138 , then there exist a point on the arc @xmath80 which is both on the left and the right boundary of @xmath6 . the two - sided restriction measures will be constructed , in most of the cases , as follows : construct first its boundary at one side ( say the right side ) which is a one - sided restriction measure , by a hsle , then conditionally on this side , sample an independent one - sided restriction measure in the connected component which is to the other side ( say the left side ) of the first boundary . however we will again use other methods to construct some limiting cases . in order to prove that the hsle define such trichordal restriction measures , we will use a martingale - type argument , in the spirit of the proof of the chordal restriction measure construction . however , the technical implementation of this strategy can appear somewhat daunting , but things are not as bad as they look : one derives that a certain process is a local martingale by longish but straightforward it formula calculations ( having the a priori information that the computation should work out nicely ) and in order to be able to apply the optional stopping theorem , we have to prove that the semi - martingale is anyway bounded ( this is not obvious from the expression of the process as product of terms that are not bounded , but we have again the a priori knowledge that in the end , this quantity will be a conditional probability , so that it should be bounded ) , which one can then show using some a priori knowledge on the hypergeometric functions that we use . having a full description of all possible one - sided and two - sided restriction measures , we will be able to construct the family of three - sided restriction measures by first constructing the boundary on one side ( say the right side ) as a one - sided measure and then conditionally on this side , sampling a two - sided restriction measure in the connected component which is to the other side ( say the left side ) of this first boundary . the determination of the exact range of admissible parameters in one - sided and two - sided cases , will be obtained by investigating some geometric properties of the corresponding random sets that we construct in the limiting cases of the parameter - range . in the three - sided case , we will need to do a kind of reverse procedure of the previous construction ( showing that a three - sided restriction measure can be decomposed as described above ) . this will be explained in [ sec : three - sided ] . first we would fix some notations and give a more complete definition of the trichordal restriction property . after that , we will briefly recall some relevant background material on sles and brownian excursions . we will also review in more detail the chordal and the radial conformal restriction properties . let @xmath139 be a configuration where @xmath140 is a simply connected domain and @xmath141 denote three different prime ends of @xmath8 . when the boundary of @xmath8 is a jordan curve , then the set of prime ends is exactly in bijection with @xmath7 . in the sequel , by slight abuse of notation , we will often just refer to the set of prime ends as @xmath7 . let @xmath142 be the collection of simply connected and closed @xmath143 such that @xmath144 . let @xmath145 be the collection of @xmath126 such that @xmath146 , @xmath15 is simply connected and @xmath147 . we endow @xmath142 with the @xmath148algebra generated by events @xmath149 for @xmath150 . in the sequel , @xmath151 will always implicitly denote a probability measure on the space @xmath152 . a family of probability measures @xmath151 is said to verify _ trichordal conformal restriction property _ if it satisfies the following properties . * ( conformal invariance ) if @xmath153 is a bijective conformal map , then @xmath154 * ( restriction ) for all @xmath150 , let @xmath155 be the unique bijective conformal map such that @xmath156 for @xmath157 . let @xmath6 be a sample with law @xmath151 , then the law of @xmath158 conditioned on @xmath159 is equal to the ( unconditioned ) law of @xmath6 . by conformal invariance , one can restrict the study of such measures for the @xmath160 . in this upper - half plane setting , we let @xmath161 be the set of bounded closed @xmath162 such that @xmath163 is simply connected . we will also frequently use the following subsets of @xmath161 : * for @xmath164 , let @xmath165 . we will mainly consider @xmath166 and @xmath167 . * let @xmath168 . let @xmath169 . for each @xmath170 , there is a unique conformal map @xmath171 with the normalization @xmath172 where @xmath173 is called the _ half - plane capacity _ of @xmath35 . in the @xmath174 setting , the formula ( [ thm1 ] ) in proposition [ prop : charac ] becomes @xmath175=g_a'(-1)^{\beta}g_a'(0)^{\gamma}(g_a(0)-g_a(-1))^{\alpha-\beta-\gamma}.\end{aligned}\ ] ] note that ( [ c ] ) can be alternatively written as @xmath176=\varphi_a'(-1)^\beta\varphi_a'(0)^\gamma\varphi_a'(\infty)^\alpha,\end{aligned}\ ] ] where @xmath177 is the conformal map from @xmath163 to @xmath178 that preserves the three points @xmath179 , and @xmath180 is the limit as @xmath181 of @xmath182 . suppose that @xmath183 is a real - valued continuous function . for each @xmath184 , let @xmath185 be the solution of the initial value problem @xmath186 for each @xmath187 there is a time @xmath188 $ ] such that the solution @xmath185 exists for @xmath189 $ ] and @xmath190 if @xmath191 . the set @xmath192 is then the chordal loewner hull generated by the driving function @xmath193 at time @xmath194 . it is easy to check that @xmath195 is the unique conformal map from @xmath196 onto @xmath178 such that @xmath197 as @xmath181 and that @xmath198 the _ schramm - loewner evolution _ ( denoted by sle ) was introduced by schramm in @xcite as candidate for scaling limits of discrete models : chordal sle@xmath199 in @xmath178 from @xmath200 to @xmath201 is the loewner evolution that one obtains when the driving function is @xmath202 and @xmath203 is a standard one - dimensional brownian motion . a useful variant of sle are the sle@xmath204 processes ( see @xcite ) is the following : let @xmath205 . let @xmath206 be the solution to the system of stochastic differential equations @xmath207 the corresponding random loewner chain generated by the driving function @xmath208 is called a chordal sle@xmath209 having @xmath210 as marked points . it is possible to choose one of the marked points @xmath211 to be @xmath212 or @xmath213 . in the upper - half plane , for any point @xmath214 on the real line , let @xmath215 be a probability measure which is the rescaled limit as @xmath216 of a brownian motion started at @xmath217 and killed at the exit of @xmath178 . it is possible to decompose the measure according to the location of the exit point @xmath218 . then , @xmath219 where @xmath220 is the probability measure of a brownian bridge from @xmath214 to @xmath218 that stays in @xmath178 . it is also possible to integrate the starting point @xmath214 according to the lebesgue measure : @xmath221 note that @xmath222 is a measure with infinite total mass . it is then well - known ( @xcite ) that * @xmath222 is conformal invariant under mobius transformation from @xmath178 to itself . * @xmath222 satisfies the following restriction property : for all @xmath170 , @xmath223 . for @xmath224 , consider a poisson point process @xmath225 of brownian excursions with intensity @xmath226 for any set @xmath227 , let @xmath228 be the function that sends @xmath229 to the filling of @xmath229 , i.e. @xmath230 is the closure of the complement in @xmath178 of the unbounded connected component of @xmath231 . then , as explained in @xcite , it turns out that @xmath232 satisfies one - sided chordal conformal restriction with an exponent @xmath233 for certain given positive proportionality factor @xmath43 . here we make a brief summary of the main results on the chordal case studied by lawler , schramm and werner @xcite . let @xmath234 be the collection of simply connected and relatively closed @xmath235 such that @xmath236 . recall that a probability measure @xmath237 on @xmath234 ( or the random set @xmath6 under the law @xmath237 ) is said to verify the ( two - sided ) _ chordal conformal restriction property _ if * ( conformal invariance ) the law of @xmath6 is invariant under any mobius transform from @xmath178 to itself . * ( restriction ) for all @xmath238 , @xmath16 conditioned on @xmath159 has the same law as @xmath6 . for all @xmath238 , let @xmath239 . proposition 3.3 in @xcite states that if @xmath6 is a chordal restriction measure , then there exist a constant @xmath28 such that for all @xmath238 , @xmath240 and that this formula characterizes uniquely the restriction measure @xmath237 . in order to show that the range of admissible values of @xmath28 is @xmath241 , the argument in @xcite goes as follows : * the measures are constructed explicitly for all @xmath242 . * sle@xmath0 satisfies two - sided conformal restriction with exponent @xmath86 , and it is a simple curve . * sets that satisfy a one - sided conformal restriction properties are all constructed ( this corresponds to all positive values of @xmath28 ) . if the two - sided restriction measure exists , the right - boundary of its samples does defines a one - sided conformal restriction measure . by symmetry , the probability that the point @xmath243 lies to the right of the two - sided restriction measure can not be larger than @xmath244 ( because it is also equal to the probability that it lies to the left ) . the fact that sle@xmath0 is a simple curve and that this probability is decreasing with @xmath28 then implies that when @xmath245 , this probability would be larger than @xmath244 , and therefore that two - sided restriction measures can not exist . let us mention some ways to construct chordal restriction measures ( other constructions , for instance via sle@xmath246 processes are also possible ) : 1 . first construct in @xmath178 a sle@xmath199 from @xmath200 to @xmath201 with @xmath247 . then consider independently a brownian loop - soup in @xmath178 with intensity @xmath248 . finally , define the filling @xmath6 of the union of the sle curve and the loops that it intersects . 2 . in section 5.2 of @xcite , two - sided restriction samples are constructed by first constructing their right boundary @xmath25 ( that is a one - sided restriction sample ) , then putting another one - sided restriction sample in the domain between @xmath249 and @xmath25 ( and we will be using similar ideas in the trichordal setting ) . it is possible to construct one - sided restriction samples of any exponent @xmath250 using sle@xmath251 processes for @xmath252 , or by brownian excursions as recalled in [ sec : brownian - excursions ] . note that once restriction measures are described in @xmath178 , one gets for free the description in any simply connected domain @xmath8 other than @xmath253 by conformal invariance . the radial case of conformal restriction was studied by wu @xcite . here we make a short review of the results and of some of the arguments . let @xmath234 be the collection of simply connected and relatively closed subsets @xmath6 of the closed unit disc @xmath254 such that @xmath255 and @xmath256 . a probability measure @xmath237 on @xmath234 ( or the random set @xmath6 under the law @xmath237 ) is said to verify _ radial conformal restriction property _ if for any closed subset @xmath35 of the closed unit disc such that @xmath257 , and such that @xmath258 is simply connected , the law of @xmath16 conditioned on @xmath17 is equal to the law of @xmath6 where @xmath14 is the conformal map from @xmath258 onto @xmath259 that fixes both the boundary point @xmath64 and the origin . as already mentioned in the introduction , the main theorem of @xcite states on the one hand that a radial restriction measure is fully characterized by a pair of real numbers @xmath260 such that @xmath261 and on the other hand that the range of admissible values of @xmath28 and @xmath53 is exactly @xmath262 the proof uses the following steps : * first , the characterization part is established directly by rather soft arguments reminiscent of the chordal case . * the measures @xmath263 for @xmath55 and @xmath264 are constructed explicitly using variants of the radial sle@xmath0 processes . it turns out that they all have the property that a sample of this measure always has the origin on its boundary . * when a measure @xmath263 exists , it is always possible to construct @xmath265 for all @xmath266 by adding to the former a poisson point process of brownian loops that surround the origin . in particular , a sample of the latter then do not have the origin on its boundary anymore . this shows in particular that @xmath263 can not exist for @xmath55 and @xmath267 . * finally , by comparison with the chordal case , one can see that @xmath53 can not be smaller than @xmath86 . there is a natural extension of the conformal restriction property to the @xmath58-point case . let @xmath268 be a configuration where @xmath269 is a simply connected domain and @xmath270 . let @xmath271 be the collection of simply connected and relatively closed @xmath143 such that @xmath272 . let @xmath273 be a probability measure on @xmath274 . a family of probability measures @xmath273 is said to verify _ @xmath58-point chordal conformal restriction property _ if it satisfies the following properties . * ( conformal invariance ) if @xmath153 is a bijective conformal map , then @xmath275 * ( restriction ) for all @xmath30 such that @xmath15 is simply connected and @xmath276 is positive . let @xmath6 be a sample with law @xmath273 . then @xmath273 conditionally on @xmath17 is equal to @xmath277 . in @xcite ( see also @xcite ) , dubdat , based on formal algebraic arguments ( in the spirit of the formal section 8.5 in @xcite ) argues that if @xmath278 is a family of @xmath58-point restriction measures , then , under some regularity assumptions , there should exist * real numbers @xmath279 * a continuous function @xmath280 which is invariant under @xmath281 mobius transformations such that for all @xmath282 such that @xmath283 and @xmath163 is simply connected , @xmath284 for @xmath285 and @xmath286 , the function @xmath65 can only be constant hence families of chordal and trichordal restriction measures are characterized respectively by @xmath64 and @xmath286 parameters . for @xmath287 , the function @xmath65 comes up and the families will be infinite - dimensional . this formal argument is indeed convincing on heuristic level , but the actual formal regularity issues are not addressed in @xcite . in the next section of the present paper , we will provide a complete self - contained characterization of the special case of trichordal restriction measures . let us stress also that the formula ( [ dubedat ] ) does not indicate for which values of @xmath288 the measure exists , which will be one main focus of the present paper . as pointed out in @xcite , the structure of brownian half - plane intersection exponents @xmath289 is of the form @xmath290 for some function @xmath291 ( this relation was then interpreted and exploited by duplantier @xcite in the light of a quantum gravity approach to these exponents , see also @xcite for further references ) . it was then shown in @xcite that the explicit expression @xmath289 corresponds to the function @xmath292 ( note that the relation between @xmath291 and @xmath289 determines @xmath291 up to a multiple constant ; for future notational convenience , we choose to divide the function @xmath293 that appeared in the introduction by this factor @xmath294 ) . in particular , one gets that @xmath295 is smaller than @xmath64 if and only if @xmath296 is smaller than @xmath297 . it was further shown in @xcite , that the full - plane disconnection exponent @xmath298 was in fact a function of the half - plane exponent @xmath299 , and the explicit expression for this function was determined in @xcite . in particular , it turns out that the former exponent @xmath300 is smaller than @xmath301 if and only the latter exponent @xmath289 is smaller than @xmath302 ie . if and only if @xmath303 . in the present section , we will adapt ideas used in @xcite and @xcite in order to derive proposition [ prop : charac ] : let us first define a family of deterministic loewner curves in @xmath178 that is stable under the mapping down procedure with respect to conformal maps that preserve the three marked points @xmath179 : for all @xmath304 and all @xmath187 , let @xmath305 be the solution to the initial value problem @xmath306 where @xmath307 the evolving hull generated by ( [ perfect equation ] ) is a continuous curve that we will denote by @xmath308 and @xmath309 is the conformal map from @xmath310)$ ] onto @xmath178 such that @xmath311 . here @xmath312 is not constantly equal to @xmath301 as in ( [ loewner ] ) , which corresponds to a continuous and monotone reparametrization in time of the standard loewner evolution . let @xmath313 be the unique conformal map from @xmath310)$ ] onto @xmath178 that sends @xmath179 to themselves . the following lemma shows a nice semi - group property of the curves @xmath314 . for all @xmath315 , @xmath316 it is not difficult to verify that @xmath317 are solutions of equation ( [ perfect equation ] ) . then we differentiate @xmath318 using chain rules to get @xmath319 let @xmath320 for all @xmath187 . then we have @xmath321 . the fact that @xmath65 does not depend on @xmath194 readily implies ( [ semi - group ] ) . let us immediately compute @xmath322 and @xmath323 that we will need later . differentiating ( [ perfect equation ] ) w.r.t . @xmath324 at @xmath325 leads to @xmath326 solving the equation with initial condition @xmath327 leads to @xmath328 in the same way , we have @xmath329 we will also need @xmath330 there exist @xmath331 such that , @xmath332)=\emptyset \right)=\exp(-\lambda(x ) t).\end{aligned}\ ] ] the semigroup property ( [ semi - group ] ) of the curve @xmath314 and the conformal restriction property of @xmath237 implies that @xmath333)=\emptyset \right)={\mathbf{p}}\left ( k\cap \eta_x([0,t])=\emptyset \right){\mathbf{p}}\left ( k\cap \eta_x([0,s])=\emptyset \right).\end{aligned}\ ] ] the existence of @xmath334 follows . we will endow @xmath335 with the same topology @xmath336 as in @xcite ( which is also closely related to carathodory topology , see the comments in @xcite ) : given @xmath337 and a sequence @xmath338 , we say that @xmath339 converges to @xmath35 if @xmath340 converges to @xmath341 uniformly on compact subsets of @xmath342 and @xmath343 is bounded away from @xmath344 and @xmath201 . then we have the following lemma which is an analogue of ( * ? ? ? * lemma 3.5 ) . [ topology ] on @xmath167 ( resp . on @xmath345 , on @xmath346 ) with the topology @xmath336 , the semigroup generated by @xmath347 ) , t\ge 0 , x\in(0,\infty)\}$ ] ( resp . by @xmath347 ) , t\ge 0 , x\in(-1,0)\}$ ] , by @xmath347 ) , t\ge 0 , x\in(-\infty , -1)\}$ ] ) is dense , the application @xmath348 is continuous , and @xmath349 is continuous . we refer the readers to ( * ? ? ? * lemma 3.5 ) for a complete proof . the idea is that any simple loewner curve can be approximated by a curve which has a driving function that takes piecewise the form of the driving function of @xmath314 . any hull @xmath35 can be approximated by a fattened and smoothed hull @xmath350 whose boundary is a simple curve . [ lem : continuous ] the function @xmath351 is continuous . for a fixed @xmath352 , as @xmath353 , the hausdorff distance between @xmath354)$ ] and @xmath355)$ ] goes to @xmath200 , which implies the convergence of @xmath354)$ ] to @xmath355)$ ] under the topology @xmath336 . according to lemma [ topology ] , we have @xmath356)=\emptyset)-{\mathbf{p}}(k\cap \eta_y([0,t])=\emptyset ) \right\|\underset{x\to y}{\longrightarrow } 0.\ ] ] therefore @xmath357 as @xmath353 . [ lem : differentiable ] the function @xmath351 is differentiable . first note that if @xmath358 , then for all @xmath359 such that @xmath360 , we have @xmath361 where @xmath362 is a constant depending on @xmath291 . this is a consequence of the two facts : for small @xmath194 , @xmath363)\not=\emptyset)=\lambda(x)t+o(t^2)$ ] where @xmath194 is the capacity of @xmath354)$ ] ; and @xmath35 can be approximated by a curve which has a driving function piece - wisely taking the form of the driving function of @xmath314 for @xmath364 . secondly , note that by conformal restriction property of @xmath6 , we have @xmath365)=\emptyset , k\cap \eta_y([0,1])\not=\emptyset)={\mathbf{p}}(k\cap \varphi_{x,1}(\eta_y([0,1]))\not=\emptyset){\mathbf{p}}(k\cap \eta_x([0,1])=\emptyset).\end{aligned}\ ] ] the set @xmath366))$ ] gets smaller as @xmath353 in the following sense ( for example see ( * ? ? ? * proposition 3.82 ) ) : @xmath367 hence @xmath368 . then by ( [ estimation ] ) , we have @xmath369 therefore @xmath370)=\emptyset , k\cap \eta_y([0,1])\not=\emptyset)\le c_0 ' \|x - y\| $ ] and consequently @xmath371)=\emptyset)-{\mathbf{p}}(k\cap \eta_y([0,t])=\emptyset ) \right\| \\ \le & { \mathbf{p}}(k\cap \eta_x([0,1])=\emptyset , k\cap \eta_y([0,1])\not=\emptyset)+ { \mathbf{p}}(k\cap \eta_x([0,1])\not=\emptyset , k\cap \eta_y([0,1])=\emptyset)\\ \le & c_0''\|x - y\|.\end{aligned}\ ] ] therefore @xmath372 is differentiable hence @xmath351 is differentiable . in fact , it is possible to iterate the procedure to see that @xmath351 is @xmath373 , but differentiability of @xmath351 will be enough for our purposes . from ( [ varphi ] ) we know that @xmath374 therefore as @xmath375 , we have @xmath376 by differentiating ( [ varphi ] ) , we get @xmath377 ( here and in the following lines @xmath378 will mean the derivative with respect to the second variable ) . therefore as @xmath375 , we have @xmath379 hence we have @xmath380 now compute @xmath381)\not=\emptyset , k\cap \eta_y([0,\delta t])\not=\emptyset \right)\\ = & { \mathbf{p}}(k\cap \eta_{x}([0,\delta s])\not=\emptyset)-{\mathbf{p}}(k\cap \eta_{x}([0,\delta s])\not=\emptyset , k\cap \eta_y([0,\delta t])=\emptyset)\\ = & 1-\exp(-\delta s\lambda(x))-\exp(-\delta t\lambda(y ) ) { \mathbf{p}}(k\cap \eta_x([0,\delta s])\not=\emptyset \| k\cap \eta_y([0,\delta t])=\emptyset)\\ = & 1-\exp(-\delta s\lambda(x))-\exp(-\delta t\lambda(y ) ) { \mathbf{p}}(k\cap \varphi_{y,\delta t}(\eta_x([0,\delta s]))\not=\emptyset)\\\end{aligned}\ ] ] note that @xmath382)$ ] is approximately @xmath383)$ ] where @xmath384 and @xmath385 . their distance is of order @xmath386 . for this , the readers can find estimates in @xcite and argue similarly as in lemma [ lem : differentiable ] . the previous expression is then equal to @xmath387+o(\delta^2)\\ = & s\lambda(x)\delta-\frac{\lambda(x)^2}{2}s^2\delta^2-(1-\delta t\lambda(y))\left(\varphi_{y,\delta t}'(x)^2\delta s\lambda(\varphi_{y,\delta t}(x))-\frac{\lambda(x)^2}{2}s^2\delta^2\right)+o(\delta^2)\\ = & s\lambda(x)\delta-\frac{\lambda(x)^2}{2}s^2\delta^2-\varphi_{y,\delta t}'(x)^2\delta s\lambda(\varphi_{y,\delta t}(x))+\frac{\lambda(x)^2}{2}s^2\delta^2-\delta t\lambda(y)\varphi_{y,\delta t}'(x)^2\delta s\lambda(\varphi_{y,\delta t}(x))+o(\delta^2)\\ = & s\lambda(x)\delta-\frac{\lambda(x)^2}{2}s^2\delta^2-\delta s\lambda(x ) -\delta^2 st f(y , x)\lambda'(x ) -\delta^2 st \lambda(x ) 2\partial_2 f(y , x)\\ & + \frac{\lambda(x)^2}{2}s^2\delta^2-\delta^2 st\lambda(y)\lambda(x)+o(\delta^2)\\ = & -\delta^2 st \left(f(y , x)\lambda'(x)+2\lambda(x)\partial_2 f(y , x)+\lambda(x)\lambda(y)\right ) + o(\delta^2).\end{aligned}\ ] ] note that we can do the same computation if we exchange the roles of @xmath388 and @xmath389 , therefore we must have the following _ commutation relation _ : @xmath390 we now fix @xmath391 and let @xmath392 in ( [ x - y ] ) . comparison of the coefficients of taylor development of the @xmath393 term on both sides yields @xmath394 this is equivalent to @xmath395 therefore , there exist three constants @xmath396 such that for all @xmath391 , @xmath397 hence for @xmath398 we have that for all @xmath391 @xmath399 putting ( [ aaa ] ) back into ( [ exp - lambda ] ) and knowing ( [ -1 ] ) , ( [ 0 ] ) , ( [ 0 - 1 ] ) , we get @xmath400)=g_t'(-1)^\beta g_t'(0)^\gamma ( g_t(0)-g_t(-1))^{\alpha-\beta-\gamma}.\end{aligned}\ ] ] in the same way as ( [ aaa ] ) we can get that for @xmath401 , there exist @xmath402 such that @xmath403 and for @xmath404 , there exist @xmath405 such that @xmath406 [ lem:123 ] for a same @xmath6 , we have @xmath407 , @xmath408 , @xmath409 . using the commutation relation ( [ x - y ] ) for @xmath410 and @xmath411 yields @xmath412 \frac{y(1-y)}{x(y - x)(1-x)}\\ & + 2\left [ \frac{\alpha_2}{y^2}+\frac{\beta_2}{(y-1)^2}+\frac{\gamma_2}{y(y-1 ) } \right ] \frac{-y^2 + 2xy - x}{x(y - x)^2(1-x)}\\ = & \left [ -\frac{2\alpha_1}{x^3}-\frac{2\beta_1}{(x-1)^3}-\frac{\gamma_1(2x-1)}{x^2(x-1)^2 } \right ] \frac{x(1-x)}{y(x - y)(1-y)}\\ & + 2\left [ \frac{\alpha_1}{x^2}+\frac{\beta_1}{(x-1)^2}+\frac{\gamma_1}{x(x-1 ) } \right ] \frac{-x^2 + 2xy - y}{y(x - y)^2(1-y)}\end{aligned}\ ] ] by simplifying it , we get for all for @xmath410 and @xmath411 @xmath413 hence @xmath414 . in the same way we can identify @xmath72 with @xmath415 . therefore we get the formula ( [ kkk ] ) for all @xmath416 . this , together with lemma [ topology ] and the restriction property of @xmath6 , proves proposition [ prop : charac ] . in this section , we will define the variants of sle@xmath0 that will turn out to describe boundaries of trichordal restriction samples . the driving functions of these sle are brownian motions with drift terms that involve hypergeometric functions ( the fact that this type of sle drift terms arise for multiple sles was observed in the context of commutation relations and/or cft partition function considerations , see @xcite ) . in the present paper , they will form a family of curves that depend on three parameters @xmath417 and we will call them hsle@xmath418 ( this paramater @xmath351 is unrelated to the function @xmath351 used in the previous function we will similarly use @xmath419 , @xmath420 , @xmath3 etc . for different purposes , but it will be each time clear from the context , which one we will be talking about ) . let us give a brief heuristic . let @xmath25 be the right boundary of a restriction sample in @xmath178 which intersects the points @xmath421 . parametrize @xmath25 with loewner s differential equation , i.e. map @xmath422 onto @xmath178 with the normalized maps @xmath195 . it is possible to prove ( although we do not rely on such proof ) , in a similar way as we shall prove lemma [ lem : j ] later on , that conditioned on @xmath423 , the curve @xmath424 is independent of @xmath425 and moreover it satisfies a kind of four - point restriction with marked points @xmath426 and @xmath201 . this domain - markov property suggests that there should exist a function @xmath427 such that the driving function of @xmath25 is given by @xmath428 in this section , we will introduce some preliminaries about hypergeometric functions , and then give the expression of @xmath427 that we use to define our hypergeometric sles . in [ s5 ] , we will then use these processes and show that they satisfy restriction properties . even though we will define @xmath427 without further explanations , the form of @xmath427 can be heuristically understood and guessed by the fact that , in the case @xmath122 , the hsle@xmath0 defined by ( [ 1st ] ) and @xmath427 should eventually be viewed as an ordinary sle@xmath0 conditioned to avoid a trichordal restriction sample @xmath6 to its left ( such that @xmath6 intersects @xmath179 note that at this point , the trichordal restriction measures have not yet been constructed ) . this is analogous to the interpretation of sle@xmath251 curves as ordinary sle@xmath0 conditioned to avoid a chordal restriction sample , see @xcite . first of all we introduce some parameters that are to be used until the end of this paper , even if we will not recall their definition each time . as mentioned earlier , the family of hypergeometric sles will depend on three parameters @xmath417 . we also introduce three triplets of parameters @xmath429 and @xmath430 , each depending on @xmath417 . the first triplet of parameters is given by @xmath431 they will be used to define the hypergeometric functions . recall the definition of the function @xmath291 mentioned in [ sec : brownian - exponents ] , which is related to the brownian disconnection exponent @xmath289 : @xmath432 it can be viewed as a bijection from @xmath433 into @xmath434 ( or as a bijection from @xmath103 onto itself ) , and its inverse function is defined on @xmath434 by @xmath435 we will also sometimes use @xmath436 as a shorthand notation for this polynomial when @xmath437 . the second triplet of parameters is defined to be @xmath438 the third triplet of parameters is defined to be @xmath439 which will turn out to be the restriction exponents associated to the corresponding hsle@xmath418 . from now on until the end of the paper , we also restrict @xmath417 to stay in the range @xmath440 note that ( [ align : cond ] ) makes @xmath441 in ( [ alphabetagamma ] ) a bijection , which allows us to write @xmath442 the condition @xmath443 in ( [ align : cond ] ) is therefore equivalent to @xmath444 , hence is also equivalent to @xmath445 . note that the @xmath441 in ( [ align : lmn ] ) is not necessarily a bijection . now let us summarize all the conditions that we imposed . the range ( [ align : cond ] ) can be easily translated to equivalent ranges on @xmath446 or @xmath430 . the equivalence with the @xmath430 range is to be understood in the sense that we assume above all @xmath447 ( in this case ( [ align : bijection ] ) is true ) . @xmath448 \displaystyle{\mu\ge-\frac14 } & \iff & a+b\ge c & \iff & \displaystyle{\beta\ge-\frac1{24}}\\[2.8 mm ] \displaystyle{\nu>-\frac34 } & \iff & c>0 & \iff & \gamma>0\\[2 mm ] \displaystyle{\mu<\lambda+\nu+\frac32 } & \iff & b < c & \iff & \beta<\tilde\xi(\alpha,\gamma ) . \end{array } \label{first - condition}\tag{range 1 } \right.\ ] ] the ( [ first - condition ] ) is chosen to make the process and other related functions well defined . lemma [ g ] , lemma [ lem : nu ] and lemma [ lem : lambda ] will explain why all these conditions are indeed needed . hypergeometric functions are defined for @xmath449 and for @xmath450 by @xmath451 where @xmath452 is the pochhammer symbol defined by @xmath453 and @xmath454 if @xmath455 . the hypergeometric function @xmath456 is a particular solution of euler s hypergeometric differential equation @xmath457 it is known that @xmath456 can be analytically extended to a function @xmath458 defined on the larger domain @xmath459 , which is a maximal solution of ( [ euler ] ) ( see @xcite ) . this function @xmath460 restricted to the domain of definition @xmath461 $ ] that will play in important role in the be sequel . note that @xmath462 is real on @xmath463 , hence @xmath460 is real on @xmath461 $ ] . from now on , except when stated otherwise , @xmath464 will stand for @xmath465 . for @xmath466 in ( [ first - condition ] ) , let @xmath467 . then @xmath468 . there exist some constant @xmath469 such that as @xmath470 @xmath471 therefore we also have as @xmath470 @xmath472 it can be found in @xcite that when @xmath473 are not negative integers and when @xmath474 is not an integer , we have @xmath475 note that @xmath476 and ( [ first - condition ] ) implies @xmath477 . the lemma is obviously true in this case . the ( [ first - condition ] ) ensures that @xmath478 and @xmath479 are not negative integers . if @xmath4 is a negative integer , then @xmath460 becomes elementary : a polynomial with leading term ( note that @xmath480 ) @xmath481 the lemma is also true in this case . if @xmath482 is a negative integers , then we can apply the identity ( see @xcite ) @xmath483 in this case @xmath484 is a polynomial in @xmath324 of order @xmath485 , hence @xmath464 is a polynomial in @xmath324 of order @xmath486 . thus the lemma is true in this case . if @xmath487 is an integer , explicit formula of the expansion of @xmath464 at @xmath201 can be found in @xcite such that ( [ lim2 ] ) is still true . [ g ] for @xmath48 in ( [ first - condition ] ) , @xmath488 for all @xmath489 . we have the identity ( see @xcite ) @xmath490 therefore it is equivalent to show that @xmath491 on @xmath492 . since @xmath493 , it is sufficient to show that @xmath494 is increasing on @xmath492 . note that @xmath495 because ( [ first - condition ] ) implies @xmath496 . if @xmath497 is not increasing on @xmath492 , then there must be a @xmath498 which corresponds to the first point at which @xmath494 stops being increasing . consequently , @xmath499 and @xmath497 is at its local maximum . note that @xmath494 satisfies the euler s differential equation @xmath500 if we evaluate ( [ w(a , c - b , c ) ] ) at @xmath501 , knowing that @xmath502 , we have @xmath503 the left - hand side is negative but the right - hand side is strictly positive , which leads to a contradiction . hence @xmath494 is increasing on @xmath492 and the lemma follows . now we define the related functions @xmath505 and make some preparatory estimations . later on , the function @xmath427 will be used to define the hsle via the formula ( [ 1st ] ) . the function @xmath506 will be used in the definition of the martingale that will be related to the hsle . the function @xmath507 has an auxiliary role and will be referred to in some computations . for @xmath508 within ( [ first - condition ] ) , define for all @xmath509 @xmath510 it is worth noting that @xmath506 is the solution of the following differential equation @xmath511 it is immediate that @xmath512 by ( [ expansion ] ) , we know that there exist a constant @xmath513 such that @xmath514 lemma [ g ] ensures that @xmath506 is non zero on @xmath515 . hence we can define @xmath516 according to earlier estimates , @xmath517 has a limit as @xmath324 goes to @xmath200 or @xmath201 . @xmath518 always for @xmath48 under condition ( [ first - condition ] ) , for @xmath519 such that @xmath520 and @xmath521 , define @xmath522 in ( [ first - condition ] ) , if we allow @xmath523 , then we have equivalently @xmath524 . then @xmath525 and ( [ lim2 ] ) no longer holds . instead we have @xmath526 the functions @xmath505 become @xmath527 in the present section , we are first going to define the hypergeometric sle processes by giving the explicit form of their driving functions . we will then interpret these processes as weighted sle@xmath121 processes up to certain ( boundary touching ) stopping times . this shows that the hsles are indeed well defined and for @xmath528 they are continuous simple curves . in order to see whether the hsle touches certain parts of the boundary , we will do a coordinate change of the type schramm and wilson did for sle@xmath121 in @xcite . we define the hypergeometric sle process to be the loewner chain generated by the driving function @xmath529 , such that @xmath530 is the solution of the following system of stochastic differential equations @xmath531 where @xmath532 is a standard brownian motion . we denote such a process hsle@xmath533 where @xmath417 are the parameters associated to the function @xmath427 . the solution to ( [ hsle ] ) is clearly well defined at times when @xmath534 and @xmath535 are not @xmath200 . in the next two sections , we will show that for @xmath417 in ( [ first - condition ] ) , ( [ hsle ] ) can be well defined even when @xmath534 or @xmath535 hits @xmath200 , by showing that the law of the hsle is locally absolutely continuous w.r.t . that of an sle@xmath121 process . let @xmath536 be a standard brownian motion for the probability measure @xmath237 and the filtration @xmath537 . let @xmath538 be the solution of the initial value problem @xmath539 the loewner curve @xmath540 generated by the equation @xmath541 is a sle@xmath121 process . let @xmath542 . let @xmath543 define the following @xmath544 adapted local martingale @xmath545 note that the function @xmath427 has a pole when @xmath546 . however when @xmath547 is chosen to be @xmath548 , @xmath549 remains bounded , because : * by definition @xmath550 * by ( [ q1 ] ) , @xmath551 as @xmath552 . * for any @xmath553 , we have @xmath554 . in this case , we can then define the measure @xmath555 on the @xmath556-field @xmath557 by the following girsanov transformation @xmath558 the measure @xmath555 is absolutely continuous w.r.t . moreover , under @xmath555 , the process @xmath560 is a standard brownian motion up to time @xmath561 . therefore @xmath562 satisfies the differential equations @xmath563 hence , under @xmath555 , @xmath25 is a hsle@xmath533 up to time @xmath564 . this proves that the law of a hsle@xmath533 is absolutely continuous w.r.t . the law of a sle@xmath565 process up to the stopping time @xmath564 . for @xmath566 , a hsle@xmath533 is therefore almost surely a simple curve up to @xmath567 . note that the measures @xmath555 are compatible for all @xmath568 hence we can let @xmath569 and say that the hsle@xmath533 is a simple curve up to the time @xmath570 however the absolute continuity of @xmath555 w.r.t . @xmath237 is not preserved as @xmath571 since @xmath572 might be no longer bounded away from @xmath200 or @xmath201 . let us first mention that whether a hsle curve touches the boundary in a neighborhood of the starting point ( i.e. for @xmath573 ) can be deduced from the knowledge on sle@xmath121 processes , due to absolute continuity . we state this property only in the @xmath574 case because it is all we need . [ lem : nu ] for @xmath417 in ( [ first - condition ] ) , let @xmath25 be a hsle@xmath418 curve as defined in ( [ hsle ] ) which starts at @xmath200 and has @xmath575 as boundary marked points . let @xmath576 be a stopping time defined as ( [ ttt ] ) . * if @xmath577 , then @xmath578 ) \cap ( -1,0]=\emptyset$ ] a.s . * if @xmath579 , then @xmath578 ) \cap ( -1,0]\not=\emptyset$ ] a.s . the condition @xmath580 is due to the @xmath581 threshold for sle@xmath121 . now we need to determine whether @xmath582 . let us observe that @xmath583 in fact corresponds to the first time @xmath25 hits @xmath584 $ ] on the real line ; and @xmath582 if and only if @xmath25 does not touch @xmath584 $ ] . more precisely , by knowledge on deterministic conformal maps , it is not difficult to see that as @xmath569 , * if @xmath585 , then @xmath582 and @xmath586 as @xmath587 ; * if @xmath588 converges to a point on @xmath589 , then @xmath590 and @xmath586 as @xmath587 ; * if @xmath588 converges to the point @xmath591 , then @xmath590 and @xmath592 oscillates between @xmath200 and @xmath201 as @xmath587 . the oscillation in case ( iii ) is due to the fact that @xmath25 approaches @xmath591 in a fractal way . in order to determine in which case we are in , we will zoom into the neighborhood of the point @xmath591 , by doing a coordinate ( and time ) change of the type that schramm and wilson did for sle@xmath121s in @xcite . let @xmath593 be a hsle@xmath533 as previously defined . let @xmath594 be a mbius transformation from @xmath178 to itself , that sends the points @xmath179 respectively to @xmath595 . let @xmath596 be a mbius transformation from @xmath178 to itself such that @xmath597 is normalized at infinity in a way that there exist @xmath598 , @xmath599 then the image of @xmath593 by @xmath65 reparametrized by @xmath600 is a loewner chain with driving function @xmath601 : for all @xmath602 , @xmath603 the above - mentioned conformal maps are illustrated in figure [ fig : coordinate change ] . since @xmath596 sends @xmath604 to @xmath201 , there exist @xmath605 such that @xmath606 we expand @xmath607 at @xmath608 , @xmath609 in order that @xmath610 be normalized as ( [ normalized gs ] ) , we should have @xmath611 moreover , the time change should be @xmath612 note that @xmath613 in order to get @xmath614 , let us calculate @xmath615 applying the it calculus rule , we have @xmath616 let @xmath617 , then we have @xmath618 thus we have proved the following lemma . the image of @xmath619 by @xmath65 under a proper time change @xmath620 ( see ( [ align : time - change ] ) ) is a loewner curve @xmath621 generated by the loewner equation @xmath622 where @xmath623 is the solution of the differential system @xmath624 and where @xmath625 note that the curve @xmath626 is well defined when @xmath627 ( or equivalently when @xmath626 hits @xmath461 $ ] ) , because the curve @xmath25 is well defined when it hits @xmath628 $ ] ( or we can argue similarly as in [ sec : girsanov ] by doing girsanov transformation for @xmath629 directly ) . to see whether @xmath630 ( whether @xmath631 ) , we need to do again a girsanov transformation as in [ sec : girsanov ] . note that when @xmath632 is small , by the estimate ( [ q2 ] ) , the quantity @xmath633 is approximately @xmath634 hence when @xmath632 is small , @xmath635 is absolutely continuous w.r.t . a bessel process of dimension @xmath636 . in the case when @xmath637 , we have @xmath638 . hence @xmath631 a.s . note that the constraint for @xmath632 to be small will be satisfied for some @xmath639 because the process @xmath640 oscillates between @xmath641 $ ] and has positive probability to get to any neighborhood of @xmath200 . note that @xmath631 corresponds to the case ( i ) and ( ii ) at the end of [ sec : girsanov ] for the original curve @xmath25 and the case @xmath642 corresponds to the case ( iii ) in which @xmath25 hits the point @xmath591 . hence for @xmath417 in ( [ first - condition ] ) , we are in case ( i ) or ( ii ) and the curve hsle@xmath418 does not hit the point @xmath591 a.s . let us now make a remark on the degenerate case . if we allow @xmath524 in ( [ first - condition ] ) , then this corresponds to the degenerate case in [ degenerated case ] , and we have @xmath643 in this case @xmath626 is a sle@xmath644 with @xmath645 . since @xmath637 , we have @xmath646 . this means @xmath642 . we are in case ( iii ) mentioned at the end of [ sec : girsanov ] . the original curve @xmath25 hits @xmath591 a.s . under condition ( [ first - condition ] ) , we are either in case ( i ) or ( ii ) mentioned after lemma [ lem : nu ] . therefore @xmath647 by the expansion ( [ q2 ] ) , @xmath648 behaves approximately like @xmath649 . by redoing the girsanov transformation as in [ sec : girsanov ] , we have that as @xmath587 , the hsle curve @xmath25 is locally absolutely continuous w.r.t . a sle@xmath650 process . similarly to lemma [ lem : nu ] , we have the following . [ lem : lambda ] in the @xmath122 case , under condition ( [ first - condition ] ) , * @xmath582 and @xmath651 if and only if @xmath652 . * @xmath653 and @xmath654 if and only if @xmath655 . in the @xmath653 case , for @xmath656 , @xmath657 will stick to each other and the function @xmath427 degenerates to a rational function . the rest of the curve is just a @xmath658 with marked point @xmath659 . now we can summarize the above - discussed properties of hsle processes . [ prop : hsle ] the hsle@xmath418 process is well defined for @xmath417 inside ( [ first - condition ] ) . it is almost surely a simple curve and moreover * it does not hit the point @xmath591 ; * it hits @xmath660 if and only if @xmath661 ; * it hits @xmath662 if and only if @xmath663 . when @xmath417 are such that @xmath664 , the hsle@xmath418 defined by ( [ hsle ] ) where @xmath427 is given by ( [ j ] ) is actually an sle@xmath644 process that hits the point @xmath591 . it is unable to continue thereafter . as in the case of the chordal conformal restriction , it will be useful in the trichordal case to first study the following larger family of random sets , that satisfy conformal restriction on just one side : in the @xmath665 setting , let @xmath666 be the collection of simply connected and closed @xmath667 such that @xmath668 $ ] and @xmath6 is unbounded . a probability measure @xmath669 on @xmath666 is said to verify _ one - sided trichordal conformal restriction for the side @xmath515 _ if for all @xmath670 , the law of @xmath158 conditioned on @xmath17 is the same as the law of @xmath6 . clearly , if a random set satisfies ( three - sided ) trichordal conformal restriction , it also defines a set that satisfies trichordal one - sided conformal restriction ( just add the connected components of the complement of the restriction sample , that have part of the negative half - line on their boundary ) . the argument given in [ characterization ] shows also that there exist three constants @xmath671 such that for all @xmath670 , @xmath176=\phi_a'(\infty)^\alpha\phi_a'(-1)^{\beta}\phi_a'(0)^\gamma\end{aligned}\ ] ] and conversely , it is easy to see there exists at most one one - sided restriction measure , that we denote by @xmath672 , that satisfies this identity . our goal is now to determine the range of accessible values for these three parameters . in the following subsections , we are first going to construct the right boundaries of the @xmath128 samples using hsle curves , for a certain range of @xmath72 . then we will construct some limiting cases that are not covered by this first construction , using sle@xmath673 curves . we will then use poisson point process of brownian excursions to construct yet another sub - family of @xmath128 distributions . in the end , we will prove that these three constructions combined together will give the full range of possible values for @xmath72 . in this section , we are going to construct one - sided restriction measures by constructing their right boundaries that are actually the hsles defined in the previous section . to this purpose , we will adopt the same strategy as the one used by lawler , schramm and werner in the chordal case @xcite : we will first find out an explicit martingale @xmath674 , which we will then show to correspond to the quantity @xmath675)$ ] by inspecting its @xmath676 convergence . let @xmath25 be a hsle@xmath418 such that @xmath677 are in ( [ first - condition ] ) . let @xmath670 . let @xmath678 and @xmath679 . for all @xmath352 , when neither @xmath534 nor @xmath680 is @xmath200 , let us define @xmath681 knowing that as @xmath375 @xmath682 and knowing the expansion ( [ z - to-0 ] ) , we define by continuity @xmath683 if @xmath25 hits @xmath628 $ ] at a time @xmath684 , then @xmath685 . define by continuity @xmath686 if @xmath25 hits @xmath589 at a time @xmath687 , then @xmath688 . moreover , @xmath689 let us define by continuity @xmath690 after the first time @xmath691 that @xmath25 hits @xmath589 , the points @xmath604 and @xmath692 will stick together . thereafter let @xmath693 note that @xmath694 is well defined only for @xmath695 . note that this @xmath696 should not be confused with the @xmath576 of the previous section ; we will keep this definition of @xmath696 until the end of the paper . as mentioned at the beginning of [ sec : hsle ] , conditionally on @xmath697 $ ] , the remainder of the curve should satisfy a kind of four - point restriction property , and the martingale @xmath694 indeed take the form of four - point restriction probabilities as suggested by ( [ dubedat ] ) . [ martingale-1-side ] the process @xmath674 is a local martingale . moreover for all @xmath670 , there exist a constant @xmath229 , possibly dependent on @xmath417 and @xmath35 , such that @xmath698 for all @xmath699 a.s . the fact that @xmath700 is a local martingale can be shown by straightforward but longish it computation that we postpone to the [ appendix ] . without loss of generality , we can assume that @xmath701 is a simple curve , since the semigroup generated by such @xmath35 will be dense in @xmath167 . for a fixed time @xmath194 , let @xmath702 . let @xmath703\to\overline{\mathbb{h}}$ ] be the simple curve @xmath704 starting from @xmath705 and parametrized by half plane capacity . note that @xmath706 for all @xmath707 because that @xmath708 by the well - known subadditivity property of the half - plane capacity . set @xmath709}$ ] and @xmath710 . by chordal version of loewner s theorem , @xmath711 . we have @xmath712 since both are the normalized map from @xmath713 onto @xmath178 . since @xmath714 , it follows that @xmath715 note that @xmath716 because @xmath717 . let @xmath718 . then @xmath719 . we have @xmath720 let @xmath721 then @xmath722 therefore , for @xmath723 we have @xmath724 ds . \end{split}\ ] ] since @xmath725 is uniformly bounded , it is enough to prove that the quantity @xmath726 is uniformly bounded from below for all @xmath727 . it is true because @xmath65 is a continuous function and @xmath727 is a compact set . [ prop : one - sided ] for @xmath417 in ( [ first - condition ] ) , let @xmath25 be a hsle@xmath418 from @xmath200 to @xmath201 , having @xmath575 as marked points . the closure of the connected component of @xmath728 at the left of @xmath25 has the law @xmath128 . recall that @xmath72 are given by ( [ alphabetagamma ] ) hence @xmath729 by similar arguments as in ( * ? ? ? * lemma 6.2 , lemma 6.3 ) , we can show that * if @xmath730 , then @xmath731 . as @xmath732 , @xmath733 and @xmath734 by the expansion ( [ z - to - infty ] ) , we have @xmath735 . * if @xmath736 , then @xmath737 . as @xmath732 , @xmath738 and the quantities @xmath739 all go to finite limits . hence @xmath740 . therefore @xmath741 as @xmath732 . since lemma [ martingale-1-side ] says that @xmath674 is a bounded martingale , we get by martingale convergence theorem and ( [ m0 ] ) that @xmath742 for @xmath417 in ( [ first - condition ] ) , using the family of hsle@xmath743 curves , we can construct @xmath128 for all @xmath744 such that @xmath745 in this section , we want to construct the case @xmath747 . this corresponds to the degenerate case in [ degenerated case ] , in which we have @xmath748 by proposition [ prop : hsle ] , an hsle with such limiting parameters is actually an sle@xmath749 . we will construct a family of random sets @xmath6 using sle@xmath750 processes and later prove that they satisfy one - sided trichordal restriction . the construction goes as the following . * for @xmath751 , let @xmath752 be a sle@xmath750 in @xmath753 from @xmath200 to @xmath201 having @xmath754 as marked points . let @xmath755 be the closure of the connected component of @xmath756 which is to the left of @xmath752 . * let @xmath757 be a one - sided chordal restriction measure of exponent @xmath758 in configuration @xmath759 for the side @xmath760 , independent of @xmath752 . let @xmath12 be the conformal map from @xmath178 to @xmath761 that sends the boundary @xmath762 $ ] to @xmath763 $ ] . then @xmath764 is a one - sided chordal restriction measure of exponent @xmath758 in the configuration @xmath765 for the side @xmath766 . * let @xmath767 where @xmath768 is the mbius transformation from @xmath178 to itself that sends @xmath595 to @xmath179 . .,scaledwidth=86.0% ] we will again use a proper martingale in order to prove the restriction property of @xmath6 . for all @xmath769 , let @xmath770 be associated to the loewner curve @xmath752 in a way that @xmath771 is the driving function that generates @xmath25 and @xmath772 . define @xmath773 in the same way as in [ sec : construction - by - hsle ] . the following is a bounded martingale : @xmath774 where @xmath775 we can first prove by it computation that @xmath776 is a local martingale ( for example see @xcite ) . then we should write @xmath776 in the form of ( [ exp - m ] ) . using the same notations as in the proof of lemma [ martingale-1-side ] , we have @xmath777 where @xmath778 note that in our case @xmath779 . hence it is enough to prove that @xmath780 is bounded from below for @xmath781 . this is true because as @xmath782 , the leading term of @xmath780 is @xmath783 where @xmath784 . [ prop : limit ] the random set @xmath6 has law @xmath785 . let @xmath786 ) \cap { a}\not=\emptyset\}$ ] . similarly to the proof of proposition [ prop : one - sided ] , note that * if @xmath730 , then @xmath582 . as @xmath787 , the terms @xmath788 all tend to @xmath64 almost surely however , the term @xmath789 does not tend to @xmath200 . in the same way as explained in @xcite , noting that @xmath790 , we have @xmath791 * if @xmath736 , then @xmath653 . as @xmath587 , @xmath792 and all the other terms in @xmath793 tend to a finite limit . hence @xmath794 . we have thus proved that @xmath795 by the martingale convergence theorem , we have @xmath796 then it is easy to see that @xmath797 the desired proposition follows . proposition [ prop : limit ] constructs @xmath128 for all @xmath744 such that @xmath798 the following lemmas are obvious from the construction . conditionally on @xmath799 , @xmath800 is a one - sided chordal restriction sample in @xmath801 . [ lem:-1 ] the right boundary of the random set @xmath6 with law @xmath802 intersects @xmath591 almost surely . similarly to what we mentioned in [ sec : brownian - excursions ] ( or see @xcite ) , we can also construct one - sided trichordal measures by taking the filling of a poisson point process of brownian excursions . more precisely , let @xmath222 be the half - plane brownian excursion measure as in ( [ brownian - excursion - measure ] ) . for @xmath803 , let @xmath804 be a poisson point process of brownian excursions of intensity @xmath805 but restricted to the set of excursions that start in @xmath589 and end in @xmath660 . [ sub - family-1 ] there exist @xmath806 such that @xmath807 hence @xmath808 has the law @xmath809 . the conformal invariance and restriction property of @xmath804 follows from the conformal invariance and restriction property of @xmath222 . hence there exist @xmath75 such that @xmath810 let @xmath811 be the vertical slit from @xmath812 to @xmath813 , then @xmath814 because there will be only finitely many excursions that are near the point @xmath200 . however @xmath815 as @xmath216 , so we must have @xmath816 . by symmetry , we also have @xmath817 . hence @xmath818 for all @xmath819 and all @xmath670 , we have @xmath820 therefore @xmath53 must be linear with respect to @xmath821 . note that @xmath822 so we must have @xmath823 so that @xmath824 can be a probability . hence there exist @xmath806 such that @xmath825 . since the filling of the union of independent restriction samples still satisfy restriction , we have many ways to play with it . for example , let @xmath826 be a poisson point process of brownian excursions that start and end on @xmath589 which corresponds to a chordal one - sided restriction measure of exponent @xmath827 . let @xmath828 be an poisson point process of brownian excursions that start and end on @xmath660 which corresponds to a chordal one - sided restriction measure of exponent @xmath829 . let @xmath830 as in lemma [ sub - family-1 ] which has law @xmath809 for @xmath831 . let @xmath832 be independent . it is easy to see that @xmath833 has the law @xmath834 . thus we have constructed using poisson point process of brownian excursions all measures @xmath835 for @xmath744 such that @xmath836 this construction requires no sle knowledge and can be seen as one of the first evidences why trichordal restriction measures form a three - parameter family . however this construction does not give the entire family of desired restriction measures . we can also take the union of restriction samples with law @xmath837 as constructed in [ sec : limit ] and restriction samples with law @xmath838 as in lemma [ sub - family-1 ] . this constructs all measures @xmath128 for @xmath744 such that @xmath839 which is a larger range than that of ( [ range:1 ] ) . note that this construction does not use hsle processes . however these processes will turn out to be necessary in later constructions of two - sided and three - sided restriction measures . now we can make a remark about the existence of cut points for a sample @xmath6 with law @xmath128 , which is an immediate consequence of the construction above and of cut - point properties for chordal restriction measures in @xcite . [ prop : geom - one - sided ] let @xmath6 be a sample which has law @xmath128 where @xmath72 are in range ( [ one - sided - final - range ] ) . * if @xmath840 , then @xmath6 has infinitely many cut points in any neighborhood of @xmath200 . * if @xmath841 , then @xmath6 has no cut point on the line segment @xmath628 $ ] . * if @xmath816 , then @xmath842}$ ] and @xmath843}$ ] has no cut point on @xmath628 $ ] . * the point @xmath591 is a cut point of @xmath6 if and only if @xmath132 . due to symmetry , the situation in the neighborhood of @xmath201 is the same as in the neighborhood of @xmath200 if we consider @xmath28 in stead of @xmath25 . in figure [ fig : one - sided ] we make some illustrations of the above - mentioned geometric properties for some one - sided restriction measures in the unit disk ( which is a more symmetric domain ) . the points @xmath48 have respectively restriction exponents @xmath72 . 0.3 0.3 0.3 we will now show that ( [ one - sided - final - range ] ) is the full range of values for which @xmath128 exist . @xmath128 does not exist for @xmath844 where @xmath845 let @xmath670 . for @xmath846 , the scaled hull @xmath847 . note that @xmath848 if @xmath6 has the law of @xmath128 for some @xmath849 , then @xmath850 since @xmath851 . this is impossible . the same argument applied to the sequence of hulls @xmath852 shows that @xmath128 does not exist either for @xmath853 . now it is left to show that @xmath128 does not exist for @xmath854 . suppose that @xmath128 exists for some @xmath855 and let @xmath856 be a sample . let @xmath755 be an independent restriction sample with law @xmath857 which exists according to lemma [ sub - family-1 ] . then the fill - in of @xmath858 has the law @xmath859 . the right boundary of @xmath755 does not intersect @xmath860 almost surely , however according to lemma [ lem:-1 ] , the right boundary of @xmath858 intersects @xmath860 almost surely since it has law @xmath859 . this is a contradiction . now we are able to state a corollary that justifies the fact that we did impose the condition that @xmath6 does hit all three points @xmath3 , @xmath4 and @xmath43 in our definition of the trichordal restriction property . [ corollary ] let @xmath6 be a random set that satisfies all axioms of the three - sided trichordal restriction sample except that we replace the condition @xmath861 by the condition @xmath862 . then , * almost surely , @xmath124 has at least two points . * if @xmath124 contains exactly two points with positive probability , then the law of the set @xmath6 is exactly a chordal restriction measure . the statement ( i ) is due to the fact that there is no one - sided trichordal restriction sample @xmath6 such that the closure of the interior of @xmath6 intersects @xmath7 at only one point . for the statement ( ii ) , note that one - sided trichordal restriction sample @xmath427 such that the closure of the interior of @xmath427 intersects only one arc of @xmath7 are just one - sided chordal restriction samples . if @xmath863 , then the right boundary of @xmath6 is a one - sided chordal restriction sample . hence for all @xmath30 such that @xmath15 is simply connected and @xmath864 , there is a @xmath865 such that @xmath866 by lemma [ lem:123 ] , we know that ( [ corollary_align ] ) is true for all @xmath30 such that @xmath15 is simply connected and @xmath867 . we now turn our attention to the two - sided case : in the @xmath665 setting , let @xmath868 be the collection of simply connected and closed @xmath667 such that @xmath869 $ ] and @xmath6 is unbounded . a probability measure @xmath870 on @xmath868 is said to satisfy _ two - sided trichordal conformal restriction for the sides @xmath589 and @xmath515 _ if for all @xmath871 , the law of @xmath158 conditioned on @xmath17 is the same as the law of @xmath6 . again , the same argument as before shows that there exist three constants @xmath671 such that for all @xmath871 , @xmath176=\phi_a'(\infty)^\alpha\phi_a'(-1)^{\beta}\phi_a'(0)^\gamma\end{aligned}\ ] ] and that conversely there exists at most one two - sided restriction measure , that we denote by @xmath872 that satisfies this . our goal is now to determine the range of accessible values for these three parameters . in the following subsections , we are first going to construct a big class of @xmath135 samples using hsle curves and the one - sided restrictions samples that have been constructed in the previous section . then we will construct a class of limiting cases using a mixture of chordal restriction samples and poisson point process of brownian motions . in the end we will prove that these two constructions together will give the full range of accessible values for @xmath72 . in the present section , our goal is to construct a two - sided restriction sample by first constructing its right boundary by a hsle curve and then putting a one - sided restriction sample to the left of the first curve . to this end , we are going to use the same local martingale @xmath674 as before , but for @xmath871 . in order for @xmath674 to be a bounded martingale , we need to restrict @xmath417 to a range which is smaller than ( [ first - condition ] ) . this is because two - sided restriction implies one - sided restriction but is harder to achieve . let us give the range right now . @xmath873 \displaystyle{\mu\ge 0 } & \iff & \displaystyle{a+b\ge c+\frac12 } & \iff & \displaystyle{\beta\ge0}&\iff & m\ge 0\\[2 mm ] \displaystyle{\nu>-\frac34 } & \iff & c>0 & \iff & \gamma>0 & \iff & \displaystyle{n\ge-\frac1{24 } } \\[2 mm ] \displaystyle{\mu<\lambda+\nu+\frac32 } & \iff & b < c & \iff & \beta<\tilde\xi(\alpha,\gamma)&\iff & ( 1 ) \\[2 mm ] \displaystyle{\nu\le \lambda+\mu}&\iff & \displaystyle{{{c}}\le { { a}}-\frac12 } & \iff & \gamma\le\tilde\xi(\alpha,\beta)&\iff & ( 2)\\[3 mm ] \displaystyle{\lambda+\mu+\nu \ge -\frac12 } & \iff & \displaystyle{{{a}}\ge \frac32 } & \iff & \tilde\xi(\alpha,\beta,\gamma)\ge1&\iff & ( 3 ) . \end{array } \label{second - condition}\tag{range 2 } \right.\ ] ] the equivalence here is to be understood in the sense that we assume above all @xmath874 . in this case , ( [ align : bijection ] ) is true and moreover we have @xmath875 . however , note that we have @xmath876 only if @xmath877 , otherwise @xmath878 . therefore if @xmath877 , then we have @xmath879 if @xmath661 , then we have @xmath880 note that @xmath881 is always true and hence we can say @xmath882 before proving the boundedness of the local martingale @xmath700 , let us first prove a lemma which gives some estimation about the function @xmath460 . [ auxiliary lemma ] for @xmath48 in ( [ second - condition ] ) that satisfy in addition @xmath883 and @xmath884 , we have @xmath885 note that @xmath886 and @xmath887 . hence @xmath888 and there is a @xmath568 such that @xmath889 for @xmath890 . we can therefore look at the function @xmath891 on @xmath892 and compute @xmath893 note that @xmath460 is a solution of ( [ euler ] ) , hence @xmath894 observe that @xmath895 therefore we must have @xmath896 for all @xmath897 . now we are ready to prove our important lemma . [ martingale2 ] let @xmath417 be in ( [ second - condition ] ) . for all @xmath898 , there exist a constant @xmath229 , possibly dependent on @xmath417 and @xmath35 , such that @xmath698 for all @xmath707 a.s . since ( [ second - condition ] ) is included in ( [ first - condition ] ) , for @xmath670 the lemma follows from lemma [ martingale-1-side ] . hence it is enough to prove the lemma for @xmath899 . let us keep the notations from lemma [ martingale-1-side ] . recall that @xmath900 and @xmath901 were defined by ( [ p_lmn ] ) and ( [ qw ] ) as @xmath902 here @xmath214 and @xmath218 actually represent the quantities @xmath903 and @xmath904 . the only modification from the proof of lemma [ martingale-1-side ] is : for @xmath899 , we have @xmath905 . we need to prove that the quantity @xmath906 is bounded from below for @xmath907 . since @xmath65 is continuous , it is enough to look at @xmath65 when either @xmath214 or @xmath218 goes to @xmath201 . in this case we also have @xmath908 and @xmath909 the condition @xmath910 and ( [ second - condition ] ) also imply @xmath911 note that in this case we have @xmath912 and @xmath884 . we can apply lemma [ auxiliary lemma ] : @xmath913 since @xmath914 , we have @xmath915 in fact @xmath916 . if @xmath214 stays bounded and @xmath218 goes to @xmath201 , then @xmath65 behaves asymptotically as a polynomial in @xmath218 with the leading term @xmath917 where @xmath918 . it is obviously bounded from below . if @xmath214 goes to @xmath201 then @xmath218 must also go to @xmath201 . by ( [ lim2 ] ) and knowing that @xmath919 , we have as @xmath181 , @xmath920 hence as @xmath921 , we have @xmath922 it follows that @xmath923 let us make a change of variable @xmath924 for @xmath925 and get @xmath926 note that @xmath921 is equivalent to @xmath927 . so the @xmath928 in the equation above is to be understood in the @xmath927 limit . * if @xmath929 then @xmath930 and ( [ inequality - f ] ) becomes @xmath931 since in this case we also have @xmath932 and @xmath916 , it follows that @xmath65 is bounded from below as @xmath927 . * if @xmath933 ( we also have @xmath916 ) , then if ever @xmath65 is not bounded from below then we must have @xmath934 and the discriminant @xmath935 . note that in this case on the one hand @xmath936 is increasing in @xmath58 , on the other hand by ( [ 2 + 3 ] ) we have @xmath937 hence @xmath938 which leads to a contradiction . therefore @xmath65 is bounded from below . for @xmath417 in ( [ second - condition ] ) , we are going to construct two - sided restriction samples @xmath6 by the following steps , see figure [ fig : construction - two - sided ] . [ cons : two - sided ] * let @xmath752 be a hsle@xmath939 from @xmath200 to @xmath201 . let @xmath940)\cap { \mathbb{r}}$ ] . according to proposition [ prop : hsle ] , we have that @xmath941 when @xmath877 and @xmath942 when @xmath579 . let @xmath943 be the connected component of @xmath756 which has @xmath944 as a part of its boundary . * let @xmath757 be an independent sample @xmath945 . note that ( [ second - condition ] ) together with ( [ 2 + 3 ] ) ensures that @xmath946 and @xmath945 exists . let @xmath12 be the conformal map from @xmath178 to @xmath943 that sends the boundary points @xmath179 to @xmath947 . * let @xmath948 . [ prop : two - sided ] for @xmath417 in the range of lemma [ martingale2 ] , the law of the set @xmath6 constructed above is @xmath135 . recall again that @xmath949 for @xmath670 , this proposition follows from proposition [ prop : one - sided ] . it is thus enough to consider the case @xmath899 . recall @xmath950 ) \cap { a}\not=\emptyset\}$ ] . similarly to the proof of proposition [ prop : one - sided ] and proposition [ prop : limit ] , note that * if @xmath951 , then @xmath731 . as @xmath952 , we have @xmath953 by the expansion ( [ z - to - infty ] ) , we have @xmath954 * if @xmath955 , then @xmath737 . as @xmath732 , @xmath792 and other terms in @xmath694 tend to a finite limit . hence @xmath956 . we have thus proved that @xmath957 by the martingale convergence theorem , we have @xmath958 according to ( [ second - condition ] ) , we have constructed the measures @xmath135 for @xmath744 such that @xmath959 the construction [ cons : two - sided ] together with the geometric properties of one - sided restriction measures mentioned in proposition [ prop : geom - one - sided ] immediately imply some geometric properties of two - sided restriction samples . we list the two following properties without mentioning other obvious ones . [ lem : goemetry2 ] let @xmath6 be a sample which has law @xmath135 where @xmath72 are within the range ( [ range : two - sided ] ) , see figure [ fig : construction - two - sided ] . then : * the point @xmath200 is a cut - point of @xmath6 if and only if @xmath99 . * there exist a point @xmath942 , which is the unique triple disconnecting point of @xmath6 , if and only if @xmath960 . when @xmath99 , we have @xmath961 and @xmath752 intersects @xmath29 only at @xmath200 . moreover , @xmath99 implies @xmath962 , hence @xmath591 is on the right boundary of @xmath757 , consequently @xmath200 is on the left boundary of @xmath963 , hence must be a cut point . when @xmath960 , since @xmath36 we have @xmath964 hence @xmath752 is boundary touching and @xmath965 . @xmath966 is obviously on the right boundary of @xmath6 . when @xmath960 , we actually have @xmath962 . hence @xmath591 is on the right boundary of @xmath757 . hence @xmath966 is on the left boundary of @xmath963 hence it is on both the left and right boundaries of @xmath6 , hence is a triple disconnecting point . it is obviously unique . in figure [ fig : two - sided - geom ] we illustrate some of the above - mentioned geometric properties for some two - sided restriction measures in the unit disk ( which is a more symmetric domain ) . the points @xmath48 have respectively restriction exponents @xmath72 . 0.3 0.3 0.3 we can see that ( [ range : two - sided ] ) does not seem to be the full range , because some limiting cases are missing , such as @xmath816 and @xmath132 . when @xmath816 , the right - boundary of the restriction sample does not start at @xmath200 and thus can not be given by a hsle curve which is defined to start at @xmath200 . however since the case @xmath967 is included in ( [ range : two - sided ] ) , by symmetry between @xmath53 and @xmath25 , the case @xmath968 should also exist and can be constructed in a symmetric way , i.e. by first constructing its left boundary which is a hsle curve starting at @xmath591 . for the @xmath132 case , we can either use again the sle@xmath750 curves as in [ sec : limit ] , or say it is symmetric to the @xmath99 case . the @xmath969 case is still not constructed . however this case should exist . consider a brownian excursion in @xmath178 starting from @xmath201 and exiting at a uniformly chosen point of the segment @xmath970 $ ] , then its outer boundary should satisfy two - sided restriction with exponents @xmath971 ( see figure [ fig : two - sided3 ] ) . the union of @xmath58 independent such brownian excursions should satisfy restriction with exponents @xmath972 . now let us construct @xmath973 samples for all @xmath974 . as illustrated in figure [ fig : alpha00 ] : for @xmath975 , let @xmath976 be a chordal restriction sample of exponent @xmath977 in @xmath978 . let @xmath979 be a poisson point process of brownian excursions of intensity @xmath980 but restricted to the excursions which start in @xmath981 and end in @xmath982 , so that it satisfies restriction in @xmath983 for exponent @xmath984 . this is possible because @xmath985 and we have lemma [ sub - family-1 ] . let @xmath986 be the filling of the union of @xmath976 and @xmath987 . let @xmath547 be the probability density function supported on @xmath970 $ ] defined by @xmath988 now let @xmath6 be the random set constructed by first choosing @xmath576 according to the density @xmath547 and then a ( conditionally ) independent @xmath989 for @xmath990 . .,scaledwidth=39.0% ] the law of this set @xmath6 is @xmath973 ( see figure [ fig : alpha00 ] ) . note that the map @xmath991 is a conformal map from @xmath163 to @xmath178 that fixes @xmath992 and @xmath201 . hence @xmath993 note that the conformal map from @xmath163 to @xmath35 that fixes @xmath994 is given by @xmath995 it yields that @xmath996 for all @xmath871 , we have @xmath997 since @xmath987 and @xmath976 are independent , we further have @xmath998 hence @xmath999 since @xmath341 sends @xmath344 to themselves , we can view @xmath1000 as a change of variable on @xmath970 $ ] . hence @xmath1001 by adding these limiting cases to what have been constructed in the previous section , we have constructed @xmath135 for all @xmath744 in the following domain ( which is an extension of ( [ range : two - sided ] ) ) @xmath1002 in this section we are going to prove that ( [ two - sided - final - range ] ) is actually the largest range for which @xmath135 can exist . the conditions @xmath1003 are inherited from the one - sided range ( [ one - sided - final - range ] ) , because the left and right boundaries of a two - sided restriction measure should both satisfy one - sided restriction . the condition @xmath1004 is inherited from the chordal two - sided restriction measures . if @xmath6 is a sample with law @xmath128 , then the sequence @xmath1005 ( @xmath6 rescaled by the factor @xmath1006 ) admits a subsequence that converges in law to a two - sided chordal restriction measure @xmath1007 , due to some standard facts about tightness and weak convergence of measures . moreover , for @xmath670 , by approaching the event @xmath1008 by open and closed sets ( for example see 4.2 @xcite ) , we have @xmath1009 hence we must have @xmath36 . now it is only remains to prove the following . the measure @xmath135 does not exist if @xmath1010 . if there exist a sample @xmath856 with law @xmath135 for some @xmath430 such that @xmath1011 but @xmath1010 , then since @xmath1012 is continuous and strictly increasing in its arguments , there exist @xmath1013 such that @xmath1014 . let @xmath755 be an independent one - sided chordal restriction sample of exponent @xmath1015 in @xmath1016 . then @xmath1017 has law @xmath1018 . this construction shows that the left and right boundaries of @xmath6 have no common point on @xmath970 $ ] . however @xmath1019 is in the range ( [ range : two - sided ] ) , and according to proposition [ lem : goemetry2 ] @xmath6 should have a triple disconnecting point on @xmath970 $ ] . this is a contradiction . we are now finally going to be able to determine which of the three - sided restriction measures exist described via proposition [ prop : charac ] do indeed exist . in order to distinguish them from the two - sided and one - sided ones , we will refer to them as @xmath1020 instead of @xmath1021 in the present section . we still consider the same martingale @xmath674 as before , but this time for all @xmath359 . in order for @xmath674 to be a bounded martingale , we need to restrict @xmath417 to a range which is even smaller than ( [ second - condition ] ) . let us give the range right now . @xmath1022 \displaystyle{\mu\ge \frac34 } & \iff & { { a}}+{{b}}-{{c}}\ge 2 & \iff & \displaystyle{\beta\ge\frac58 } & \iff&\displaystyle { m\ge \frac58}\\[2.5 mm ] \displaystyle{\nu\ge 0 } & \iff & \displaystyle{{{c}}\ge \frac32 } & \iff&\displaystyle { \gamma\ge \frac58 } & \iff & n\ge 0\\[2.5 mm ] \displaystyle{\mu<\lambda+\nu+\frac32 } & \iff & { { b}}<{{c}}&\iff & \beta<\tilde\xi(\alpha,\gamma ) & & \\[2.5 mm ] \nu\le \lambda+\mu & \iff&\displaystyle { c\le a-\frac12 } & \iff & \gamma\le\tilde\xi(\alpha,\beta)&\iff & n\le\tilde\xi(l , m)\\[2.5 mm ] \lambda\le\mu+\nu & \iff & b\ge 0 & \iff & \alpha\le\tilde\xi(\beta,\gamma ) & \iff & l\le\tilde\xi(m , n)\\[2.5 mm ] \lambda+\mu+\nu \ge 1 & \iff & { { a}}\ge 3 & \iff & \xi(\alpha,\beta,\gamma)\ge 2 & \iff&\tilde\xi(l , m , n)\ge1 . \end{array } \label{third - condition } \tag{range 3 } \right.\ ] ] the equivalence here is to be understood in the sense that we assume above all @xmath1023 . in this case , ( [ align : bijection ] ) is true and moreover @xmath1024 . [ lem7.1 ] for @xmath48 in ( [ third - condition ] ) , we have @xmath1025 for all @xmath489 . note that @xmath1026 . if @xmath460 is not always strictly increasing on @xmath760 , then there exist @xmath1027 such that @xmath1028 and @xmath1029 . putting this back to the equation ( [ euler ] ) yields @xmath1030 the two sides of the equation above should have the same sign , hence we have @xmath1031 . this is in contradiction with lemma [ g ] . [ martingale3 ] let @xmath417 be in ( [ third - condition ] ) . for @xmath359 , there exist a constant @xmath229 , possibly dependent on @xmath417 and @xmath35 , such that @xmath698 for all @xmath707 a.s . since ( [ third - condition ] ) is included in ( [ second - condition ] ) , it is enough to prove the lemma for @xmath1032 . let us keep the notations from the proofs of lemma [ martingale-1-side ] and lemma [ martingale2 ] . recall that @xmath900 and @xmath901 were defined by ( [ p_lmn ] ) and ( [ qw ] ) as @xmath902 here @xmath214 and @xmath218 actually represent the quantities @xmath903 and @xmath904 . the difference here is that @xmath1033 for @xmath1032 . hence we need to prove that the quantity @xmath906 is bounded from below for @xmath1034 . since @xmath65 is continuous , it is enough to look at @xmath65 when either @xmath214 or @xmath218 goes to @xmath201 . , then : * if @xmath1035 , then @xmath65 is asymptotically equivalent to the leading term @xmath1036 . hence @xmath65 is bounded from below . * if @xmath1037 , then we also have @xmath1038 and @xmath1039 as @xmath1040 , it is asymptotically equivalent to the leading term @xmath1041 x.\end{aligned}\ ] ] first we can exclude the trivial case where @xmath1042 for which @xmath1043 . then note that @xmath1044 and @xmath1045 . also note that @xmath1046 for all @xmath1047 because of lemma [ lem7.1 ] , lemma [ g ] , and @xmath1048 . hence @xmath1049 hence @xmath65 is bounded from below . , then @xmath65 is asymptotically a polynomial in @xmath218 with leading term @xmath917 where @xmath918 hence @xmath65 is bounded from below . , then @xmath1050 . in the same way as ( [ estim6 ] ) , we have @xmath1051 * if @xmath1037 , then we also have @xmath1038 and @xmath1052 note that in this case we always have @xmath1053 . if @xmath1054 , then @xmath65 is obviously bounded from below . if @xmath1055 , then we also have @xmath1056 , hence @xmath1057 , hence @xmath460 is constantly equal to @xmath64 . hence @xmath901 is constantly equal to @xmath200 and @xmath1058 this is bounded from below because @xmath916 . * if @xmath1035 , then note that we also have @xmath916 . assume @xmath65 is not bounded from below , we must have @xmath1059 in this case @xmath1060 is increasing in @xmath1061 hence in @xmath351 . since @xmath1062 , we have @xmath1063 this implies that @xmath1060 gets to its maximum when @xmath1064 . hence @xmath1065 which leads to a contradiction . hence @xmath65 is bounded from below . now we can construct a three - sided restriction sample @xmath6 by the following steps , see figure [ fig : construction - three - sided ] . [ construction3 ] let @xmath417 be inside ( [ third - condition ] ) . * let @xmath752 be a hsle@xmath939 from @xmath200 to @xmath201 . note that since @xmath1066 , according to proposition [ prop : hsle ] @xmath752 does not hit the real axis except at @xmath200 . let @xmath943 be the connected component of @xmath756 which is to the left of @xmath752 . * let @xmath757 be an independent sample @xmath1067 . note that ( [ third - condition ] ) implies that @xmath757 exists . let @xmath12 be the conformal map from @xmath178 to @xmath943 that sends the boundary points @xmath421 to @xmath1068 . let @xmath1069 . the above - constructed @xmath6 has law @xmath1070 , where @xmath949 the proof is identical to that of proposition [ prop : two - sided ] , except that we also need to check for @xmath1032 . we constructed all three - sided restriction samples of law @xmath1070 for @xmath72 in ( [ third - condition ] ) . note that the limiting case @xmath132 can be either constructed using sle@xmath750 curves as in [ sec : limit ] , or alternatively in a symmetric way as the cases @xmath99 and @xmath1071 . thus , we have now constructed ( and proved the existence of ) the measures @xmath1070 for @xmath744 such that @xmath1072 the construction [ construction3 ] together with the geometric properties of two - sided restriction measures mentioned in proposition [ lem : goemetry2 ] immediately imply some geometric properties of three - sided restriction samples . the following proposition is also a part of theorem [ thm : existence ] . let @xmath6 be a sample which has law @xmath1070 where @xmath72 are within range ( [ three - sided - final - range ] ) . * @xmath6 has a triple disconnecting point which is unique if and only if @xmath111 . * the boundary point @xmath200 is a cut point of @xmath6 if and only if @xmath99 . we illustrate these geometric properties in figure [ fig : existence ] for samples in the unit disk . according to construction [ construction3 ] , @xmath111 if and only if @xmath1073 . by proposition [ lem : goemetry2 ] , the set @xmath757 has a unique triple disconnecting point @xmath966 on the boundary @xmath660 . this point will be mapped by @xmath12 to a point on @xmath752 , which will be the unique triple disconnecting point of @xmath6 . when @xmath99 , then @xmath1074 hence @xmath591 is a cut point of @xmath757 , hence @xmath200 will be a cut point of @xmath6 . the range ( [ three - sided - final - range ] ) is the same as in theorem [ thm : existence ] . in this section , we will prove the following proposition . [ prop : range-3 ] three - sided trichordal restriction measures @xmath1070 do not exist for @xmath430 outside of range ( [ three - sided - final - range ] ) . it is easy to see that , in range ( [ three - sided - final - range ] ) , the conditions @xmath1075 are inherited from chordal two - sided restriction measures and the conditions @xmath1076 are inherited from trichordal one - sided restriction measures . hence it is enough to justify the condition @xmath1077 . as we are now going to explain , this condition is actually inherited from the two - sided condition @xmath1078 in ( [ two - sided - final - range ] ) . for this purpose , we are going to explore the relation between two - sided and three - sided restriction measures . we can already see that if we take @xmath6 a three - sided restriction measure as in construction [ construction3 ] and send the right boundary of @xmath6 to the half - line @xmath515 by the conformal map @xmath1079 , then the image @xmath1080 is a two - sided restriction measure . we would like to prove this transitivity of restriction property in a way that does not rely on the construction [ construction3 ] . this is because we want to know whether there exists any other @xmath1070 than those constructed in construction [ construction3 ] . let us reformulate the desired property and fix notations for the rest of this section . let @xmath1081 be the bottom part of @xmath6 which has @xmath591 and @xmath200 as end points . let @xmath1082 be the conformal map from the unbounded connected component of @xmath1083 to @xmath178 which sends @xmath179 to themselves . let @xmath1084 . ( see fig [ relation - multi ] . ) we will prove in lemma [ lem : j ] that @xmath427 satisfies two - sided restriction . [ indep ] @xmath427 and @xmath1081 are independent . for all @xmath1032 , because of the restriction property of @xmath6 , we have the following identity in law : @xmath1085 equivalently , ( [ lem : indep ] ) says that @xmath1086 has the same law as @xmath427 . since the events @xmath1087 for all @xmath1032 generate the @xmath556-algebra for the law of @xmath1081 , we conclude that @xmath427 is independent of @xmath1081 . lemma [ indep ] says that in order to sample @xmath6 , we can first sample @xmath1081 and then sample the pre - image by @xmath1082 of an independent sample @xmath427 . this gives us an alternative way to describe the law of @xmath427 : it is the law of @xmath6 conditioned on @xmath1088 $ ] . of course this conditioning must be taken as a limit . let @xmath970^{\varepsilon}:=\left\{z\in\overline{\mathbb{h } } , { \mathrm{dist}}(z,[-1,0])\le{\varepsilon}\right\}.$ ] let @xmath1089 be @xmath6 conditioned on @xmath1090^{\varepsilon}$ ] . the sequence @xmath1091 converges in law to @xmath427 as @xmath216 . let @xmath1092 be the space of continuous curves from @xmath591 to @xmath200 that stay in @xmath178 except for the two extremities . let @xmath868 be as in the definition of the two - sided measures . let @xmath1093 be as in the definition of our three - sided measures . let @xmath1094 be a function from @xmath1095 to @xmath1093 that sends @xmath1096 to the pre - image by @xmath1097 of @xmath1098 . let @xmath65 be a mbius transformation from the unit disk @xmath259 to @xmath178 . endow @xmath1092 , @xmath868 and @xmath1093 with the metric which is the image by @xmath65 of the hausdorff metric on @xmath259 , then @xmath1099 is continuous in both of its arguments . by definition @xmath1089 has the law of @xmath1100 where @xmath1101 is @xmath1081 conditioned on @xmath1090^{\varepsilon}$ ] and @xmath427 is independent of @xmath1102 . since @xmath1101 converges to the interval @xmath970 $ ] almost surely within hausdorff distance , we have @xmath1089 converges in law to @xmath1103 , j)=j$ ] . [ lem : j ] @xmath427 satisfies two - sided restriction . let @xmath1104 be a measurable continuous and bounded function on @xmath1105 . let @xmath871 . we have @xmath1106^{\varepsilon}\right)}{{\mathbf{p}}(k\cap a=\emptyset \| \gamma_b\subset[-1,0]^{\varepsilon } ) } = \frac{{\mathbf{e}}\left(f(\varphi_a(k))\mathbf{1}_{k\cap a=\emptyset } \mathbf{1 } _ { \gamma_b\subset[-1,0]^{\varepsilon}}\right)}{{\mathbf{p}}\left(k\cap a=\emptyset , \gamma_b\subset[-1,0]^{\varepsilon}\right ) } \\ = & \frac{{\mathbf{e}}\left(f(\varphi_a(k ) ) \mathbf{1 } _ { \gamma_b\subset[-1,0]^{\varepsilon } } \| { k\cap a=\emptyset}\right)}{{\mathbf{p}}(\gamma_b\subset[-1,0]^{\varepsilon}\| k\cap a=\emptyset ) } = \frac{{\mathbf{e}}\left(f(k ) \mathbf{1 } _ { \varphi_a^{-1}(\gamma_b)\subset[-1,0]^{\varepsilon } } \right)}{{\mathbf{p}}\left(\varphi_a^{-1}(\gamma_b)\subset[-1,0]^{\varepsilon}\right ) } \\ = & { \mathbf{e}}\left(f(k ) \| { \varphi_a^{-1}(\gamma_b)\subset[-1,0]^{\varepsilon } } \right).\end{aligned}\ ] ] on the one hand , by the fact that @xmath1089 converges to @xmath427 and by some open and closed approximations as needed for ( [ kacvg ] ) , we have @xmath1107 on the other hand , @xmath1108^{\varepsilon } } \right)\underset{{\varepsilon}\to 0}{\longrightarrow } { \mathbf{e}}\left(f(j ) \right)\ ] ] for the same reason why @xmath1089 converges in law to @xmath427 . hence we have @xmath1109 and @xmath427 satisfies two - sided restriction . note that for two - sided ( trichordal ) restriction measures , we can do a similar conformal transformation to obtain a one - sided ( trichordal ) restriction measure ; this follows either via the same proof , or directly from our construction [ cons : two - sided ] . moreover construction [ cons : two - sided ] gives us a precise relation between the exponents . [ lem : aaa ] let @xmath6 be a sample of law @xmath135 where @xmath1110 and let @xmath1111 be its right boundary . then @xmath1112 has law @xmath1113 where @xmath1114 recall that @xmath291 , as defined in [ sec : brownian - exponents ] , can be viewed as a bijection from @xmath1115 onto itself , and that for @xmath1116 , we have @xmath1117 so that there are no definition problems here . furthermore , as @xmath1118 , it follows that @xmath1119 can be equivalently described as the only non - negative real such that @xmath1120 . loosely speaking , when we unfold a corner by the conformal maps as we defined , the restriction exponent @xmath36 goes to @xmath1121 for that corner . there is also a relation for the case @xmath1122 but we will not need it . the same holds for chordal two - sided measures . it was proved in @xcite that we can construct all chordal restriction measure of exponent @xmath28 by first constructing its right boundary @xmath25 then putting an independent one - sided chordal restriction sample of exponent @xmath1121 in the connected component of @xmath728 which contains @xmath249 . [ proof of proposition [ prop : range-3 ] ] let @xmath744 be such that @xmath1123 assume that the law @xmath76 exists and let @xmath6 be a sample , then by lemma [ lem : j ] , @xmath1124 is a two - sided restriction measure . we can in particular view @xmath6 as a two - sided restriction measure and apply lemma [ lem : aaa ] , so @xmath427 has law @xmath1125 . therefore @xmath1126 must fall into the range ( [ two - sided - final - range ] ) . hence @xmath1127 which means that @xmath1128 which can be rephrased as @xmath1129 which is equivalent to @xmath1130 . we need to show that the process @xmath1131 is a local martingale , where @xmath530 satisfy @xmath1132 this looks like a daunting task , but it is a rather straightforward direct it formula computation : to simplify the notations , set @xmath1133 first calculate the following it derivatives : @xmath1134 dt + \sqrt{\frac83 } h_t'(w_t ) d b_t\\ & d\ , h_t'(w_t ) = \left [ \frac{h_t''(w_t)^2}{2h_t'(w_t ) } + h_t''(w_t ) j_t \right ] dt + \sqrt{\frac83 } h_t''(w_t ) d b_t\\ & d\ , h_t(o_t)=\frac{2h_t'(w_t)^2}{h_t(o_t)-h_t(w_t ) } dt , \quad h_t(v_t)=\frac{2h_t'(w_t)^2}{h_t(v_t)-h_t(w_t ) } dt\\ & d\ , h_t'(o_t)=\left[-\frac{2h_t'(w_t)^2 h_t'(o_t)}{(h_t(o_t)-h_t(w_t))^2 } + \frac{2h_t'(o_t)}{(o_t - w_t)^2 } \right ] dt\\ & d\ , h_t'(v_t)=\left[-\frac{2h_t'(w_t)^2 h_t'(v_t)}{(h_t(v_t)-h_t(w_t))^2 } + \frac{2h_t'(v_t)}{(v_t - w_t)^2 } \right ] dt.\end{aligned}\ ] ] then we have @xmath1135 from the formula above , we see that the local martingale term of @xmath1136 is @xmath1137 ( and therefore @xmath1138 ) , where @xmath1139 the drift term of @xmath1136 can be written as @xmath1140 where @xmath1141-\frac{5h_t''(w_t)^2}{6 h_t'(w_t)^2 } \\ & + n \left[-\frac{2h_t'(w_t)^2 } { ( h_t(o_t)-h_t(w_t))^2 } + \frac{2}{(o_t - w_t)^2 } \right]+ m \left[-\frac{2h_t'(w_t)^2 } { ( h_t(v_t)-h_t(w_t))^2 } + \frac{2}{(v_t - w_t)^2 } \right]\\ & + ( l - m - n+\lambda ) \frac{-2 h_t'(w_t)^2 } { \left ( h_t(o_t)-h_t(w_t ) \right ) \left ( h_t(v_t)-h_t(w_t ) \right)}-(l - m - n+\lambda ) \frac{-2}{(o_t - w_t)(v_t - w_t)}\\ & + \frac{g'(\tilde x_t)}{g(\tilde x_t)}\cdot\frac{1}{h_t(o_t)-h_t(v_t ) } \left [ \left(-\frac53 h_t''(w_t)+h_t'(w_t ) j_t \right ) - \frac{2h_t'(w_t)^2}{h_t(o_t)-h_t(w_t ) } \right]\\ & -\frac{g'(\tilde x_t)}{g(\tilde x_t)}\cdot\frac{h_t(w_t)-h_t(o_t)}{\left(h_t(o_t)-h_t(v_t)\right)^2 } 2 h_t'(w_t)^2 \frac{h_t(v_t)-h_t(o_t)}{\left ( h_t(o_t)-h_t(w_t ) \right ) \left ( h_t(v_t)-h_t(w_t ) \right)}\\ & + \frac{g''(\tilde x_t)}{2g(\tilde x_t)}\cdot \frac83 \frac{h_t'(w_t)^2}{\left ( h_t(o_t)- h_t(v_t ) \right)^2 } -\frac{4g'(\tilde x_t)^2}{3g(\tilde x_t)^2 } \left ( \frac{h_t'(w_t)}{h_t(o_t)-h_t(v_t ) } \right)^2\\ & -\frac{g'(x_t)}{g(x_t)}\cdot \left ( \frac{j_t}{o_t - v_t}-\frac{2}{(o_t - w_t)(o_t - v_t ) } \right ) + \frac{g'(x_t)}{g(x_t)}\cdot \frac{w_t - o_t}{(o_t - v_t)^2 } \cdot\frac{2(v_t - o_t)}{(o_t - w_t)(v_t - w_t)}\\ & -\frac{g''(x_t)}{2g(x_t ) } \cdot \frac83 \frac{1}{(o_t - v_t)^2 } + \frac{4g'(x_t)^2}{3g(x_t)^2}\frac{1}{\left ( o_t - v_t \right)^2 } + \frac{1}{2 } l_t^2.\end{aligned}\ ] ] recall that by ( [ def : j ] ) we have @xmath1142 in @xmath1143 , we now interpret the terms that only involve @xmath770 as constant and we regroup the coefficients in front of the terms which involve @xmath773 . the coefficients for each of these terms are as follows : the constant term is @xmath1144 and the remaining ones can be listed as follows : @xmath1145=0\end{aligned}\ ] ] adding up the terms in ( [ a.3]-[a.8 ] ) , we get @xmath1146 where @xmath1147 as @xmath506 satisfies the differential equation ( [ gdiff ] ) , we can conclude that @xmath1148 . now , it remains only to deal with the constant term . it is equal to @xmath1149 , with @xmath1150 which is also equal to @xmath200 because of ( [ gdiff ] ) . this finally proves that @xmath1151 hence @xmath694 is a local martingale . the author thanks wendelin werner for many discussions and suggestions , and for his help throughout the preparation of this paper . bertrand duplantier . conformal fractal geometry & boundary quantum gravity . in _ fractal geometry and applications : a jubilee of benot mandelbrot , part 2 _ , volume 72 of _ proc . pure math . _ , pages 365482 . soc . , providence , ri , 2004 .	the study of conformal restriction properties in two - dimensions has been initiated by lawler , schramm and werner in @xcite who focused on the natural and important chordal case : they characterized and constructed all random subsets of a given simply connected domain that join two marked boundary points and that satisfy the additional restriction property . the radial case ( sets joining an inside point to a boundary point ) has then been investigated by wu in @xcite . in the present paper , we study the third natural instance of such restriction properties , namely the `` trichordal case '' , where one looks at random sets that join three marked boundary points . this case involves somewhat more technicalities than the other two , as the construction of this family of random sets relies on special variants of sle@xmath0 processes with a drift term in the driving function that involves hypergeometric functions . it turns out that such a random set can not be a simple curve simultaneously in the neighborhood of all three marked points , and that the exponent @xmath1 shows up in the description of the law of the skinniest possible symmetric random set with this trichordal restriction property .
young massive star clusters , which often surpass the globular clusters in the galaxy in terms of brightness , mass , and density , are seen to result from intense episodes of star - formation in galaxies . however , star clusters are also found in relatively quiescent , low star - formation rate ( sfr ) galaxies , albeit at much lower masses ( e.g. larsen & richtler 1999 , 2000 ) . this difference in the types ( mass ) of clusters produced in various galactic environments has been suggested to be caused by size - of - sample effects , in which galaxies with high sfrs form proportionally more clusters , and hence are able to sample the cluster mass function out to higher masses ( larsen 2002 ) . this effect has been quantitatively observed through the use of the relation ( larsen 2002 ) , where m@xmath0 is the brightest cluster in v - band absolute magnitude , in the sense that the most luminous clusters in galaxies with high sfrs are brighter . this trend , along with the similar log n vs. @xmath4 relation ( where n is the number of clusters brighter than a certain magnitude limit ; whitmore 2003 ) , have been used to argue for a universality of cluster formation , i.e. stochastic sampling from a universal underlying mass function . size - of - sample effects , together with cluster population synthesis models ( e.g. gieles et al . 2005 ) have become a common means to investigate the properties of clusters and cluster systems . for example , hunter et al . ( 2003 ) used the relation of the most massive cluster per logarithmic age bin in the lmc and smc in order to estimate the exponent of the cluster initial mass function ( @xmath5 ) . this proceedure was recently revisited by gieles & bastian ( 2008 ) who used the same relation to rule out mass independent , long duration ( @xmath6 myr ) cluster disruption models . gieles et al . ( 2006a ) used the log n vs. @xmath4 relation to constrain @xmath5 , and found a value of @xmath7 , which is similar ( @xmath8 ) to that derived by weidner , kroupa , & larsen ( 2004 ) using the relation . this is significantly steeper than that derived from direct measurements of the mass / luminosity function of galaxies , namely @xmath9 ( e.g. de grijs et al . this discrepancy will be addressed in [ sec : discussion ] . wilson et al . ( 2006 ) have tested whether the above relations still hold in the extreme environment of galaxy merger starbursts . they studied the ultra - luminous infrared galaxy ( ulirg ) , arp 220 , and found that despite its high sfr ( @xmath10 /yr ) , being an order of magnitude higher than any of the galaxies in the previous samples , falls nicely on the extrapolated fit to the more quiescent star - forming galaxies . weidner et al . ( 2004 ) used the relation to constrain cluster formation scenarios , namely the timescale over which clusters form , which they estimate to be on the order of a crossing time . they further suggest that a cluster population formation epoch ( i.e. the timescale where a statistically full population of clusters is formed ) is on the order of 10 myr . however , their analysis was based on the assumption that within a `` cluster population formation epoch '' the brightest cluster of a galaxy is also the most massive , hence that the trend is simply reflecting a relation between the sfr of a galaxy and the most massive cluster within it . observationally , it appears that this assumption is not valid , as the brightest cluster in a galaxy tends to be young , and more massive , older clusters may appear less luminous due to stellar evolution ( gieles et al . 2006a ) . in this work our goals are threefold . the first is to test the claim by weidner et al . ( 2004 ) that the brightest cluster within a galaxy is also the most massive . this naturally leads to a discussion as to why the observed relation holds . the second is to investigate the implications of the observed relation , paying particular attention to the cluster initial mass function , and the implied connection between the cluster and star formation rates within a galaxy . thirdly , using the observed trend , combined with a correction for stellar evolutionary fading , to estimate the sfr in a sample of post - starburst merger galaxies . this , in turn , allows us to place limits on the duration of the starburst phase of ulirgs as well as trace their subsequent evolution . in [ sec : data ] we present archival observations of two ongoing galaxy mergers and a collection of data taken from the recent literature . [ sec : why ] presents a series of monte carlo simulations of cluster populations in order to investigate why the observed relation holds . in [ sec : discussion ] we investigate the implications for the underlying cluster initial mass function , the relation between star and cluster formation , and use the observed relation to derive the peak sfr of post - starburst galaxies . our conclusions are presented in [ sec : conclusions ] . ngc 2623 is a luminous infrared galaxy which shows clear evidence of an ongoing merger , namely two long tidal tails and a large amount of ongoing star - formation . it was observed with the advanced camera for surveys ( acs ) wide - field camera ( wfc ) onboard hst on june 2nd , 2004 ( f555w ; prop . i d 9735 ) and november 11th , 2005 ( f435w , f814w ; prop . i d 10592 ) . we obtained the reduced and calibrated drizzled images through the eso / hst archive . we adopt a distance to ngc 2623 of 77.1 mpc ( assuming @xmath11 km / s / mpc ) . aperture photometry was carried out ( using a 10 pixel aperture and a background annulus from 12 to 14 pixels ) on the brightest source in the f555w image and zeropoints from the acs website were applied . the brightest v - band cluster has b ( f435w ) , v ( f555w ) , and i ( f814w ) apparent magnitudes of 20.7 , 20.3 , and 19.6 , respectively ( vegamag system ) . a correction for galactic extinction was then applied ( a@xmath12 - schlegel et al . 1998 ) . in order to estimate the extinction of this cluster we employed a @xmath13 vs. @xmath14 colour - colour plot . adopting the galactic extinction law of savage & mathis ( 1979 ) we found an extinction , a@xmath15 , of 0.3 mag was necessary in order to bring the cluster colours into agreement with the ssp models . applying this , along with the adopted distance modulus , results in an absolute v - band magnitude of -14.5 for this cluster . ngc 3256 is part of the toomre sequence of merging galaxies and has the highest current sfr and x - ray luminosity of any galaxy in the sequence . it s known to harbor an extensive cluster population which has been studied photometrically ( zepf et al . 1999 ) as well as spectroscopically ( trancho et al . 2007a , b ) . we adopt a distance of 37 mpc ( zepf et al . 1999 ) and a galactic foreground extinction of a@xmath16 mag ( schegel et al . ngc 3256 was observed with the acs - wfc onboard hst on november 18th , 2003 ( f555w - prop . i d 9735 ) and november 6th , 2005 ( f435w , f814w - prop . i d 10592 ) . we obtained the images in the same manner as for ngc 2623 . we performed aperture photometry of the brightest cluster visable in the f555w image using the same techniques as above . we find that the brightest cluster has b ( f435w ) , v ( f555w ) , and i ( f814w ) magnitudes of 16.9 , 17.0 and 16.36 , respectively . we have not applied any correction for intrinsic extinction . applying the galactic extinction and distance modulus we find m@xmath17 for this cluster . in table [ table : literature ] we list the @xmath18 of the brightest cluster and the estimated sfr for a sample of galaxies taken from the literature . we have focussed our study on moderate to high sfr galaxies ( @xmath19 ) in order to strengthen the observed trend for extrapolation to higher cluster luminosities and sfrs . references for the brightest cluster and sfr are given , where s03 refers to a sfr estimated from the infrared luminosity ( taken from sanders et al . 2003 ) and sfr / l@xmath20 relation of kennicutt ( 1998 ) . in the case of iras 19115 - 2124 ( vaisanen et al . 2008 ) , we used the brightest i - band cluster ( @xmath21 ; vaisanen priv . comm . ) and applied a v i colour of 0.7 ( typical of young clusters ) , in order to estimate the v - band magnitude . the galaxy with the highest sfr galaxy in the sample is arp 220 , which was studied in detail by wilson et al . they showed that this cluster fits the observed relation quite well , even though the galaxy has a sfr that is an order of magnitude higher than any other galaxies previously used in the relation . here we fill in that gap and show that it is indeed a continuous relation ( see fig . [ fig : relation ] ) . in addition , we include the low - luminosity hii regions in the extreme outskirts of ngc 1533 which have recently been studied by werk et al . these low sfr regions ( @xmath22 ) are a welcome opportunity to test the low sfr regime and also test at what physical scale ( i.e. galactic , hii region , etc ) does the relation break down . the points lie at the lower - left of fig . [ fig : relation ] and can be seen to follow the extrapolated relation reasonably well . one additional caveat is that all points in fig . [ fig : relation ] are in fact lower limits in the y - direction . the reason for this is that most of the studies used in the construction of fig . [ fig : relation ] were based on optical studies , and hence possibly effected by extinction . thus , it is impossible to rule out the possibility that a brighter cluster in the v - band was missed due to extinction effects . however , the tight observed correlation suggests that this does not significantly bias the results . 10 myr ) . the diamond represents the third brightest cluster in ngc 34 as the first two have ages of @xmath23 myr . the best fit to the data from weidner et al . ( 2004 ) is shown as a dashed line . regions occupied by ( ultra / hyper ) luminous infrared galaxies are also shown , assuming the relation between infrared luminosity and star - formation rate of kennicutt ( 1998 ) . the dotted line shows the expected relation for a pure - power law ( @xmath2 ) case if all stars formed in bound clusters , see [ sec : discussion ] for details . expected error bars from stochastic sampling are shown in the upper left of the panel , along with assumed errors in the sfr.,width=321 ] the relation between the sfr of a galaxy and the brightest cluster within it is empirically based , and was originally given a statistical ( i.e. size - of - sample ) explanation ( larsen 2002 ) . however , weidner et al . ( 2004 ) suggested that the underlying cause was that a full , statistically complete cluster population was formed every @xmath24 myr , and that this rapid formation timescale was at the basis of the observed sfr / cluster relation . for this the authors assumed that the most massive cluster in a population is normally also the brightest . in order to test this assertion we performed a simple series of monte carlo simulations of cluster populations . we create 100 realisations of a cluster population in a galaxy . the clusters are drawn stochastically from a mass function ( a power - law with an index of @xmath25 ) with a lower limit of 100 and an upper limit @xmath26 ( effectively no upper limit ) , and originally we create 5000 clusters . the clusters are then assigned ages randomly between 0 and 100 myr ( in order to simulate a constant cluster formation rate ) . we then calculate the absolute magnitude of each cluster , using the cluster s age and mass , from simple stellar population models ( bruzual & charlot 2003 ; solar metallicity , salpter stellar imf ) . once this is carried out for a single realisation , we search the population for the brightest cluster and the most massive ( which are not necessarily the same cluster ) along with the corresponding clusters ages . the average cluster formation rate ( cfr ) is found by dividing the total mass formed in clusters by the duration of the experiment ( i.e. 100 myr ) . [ fig : mc ] shows the cumulative distribution of the fraction of realizations whose brightest cluster is younger than a given age ( solid black line ) . we also show the age distribution of the most massive cluster in each sample . the 50-percentile mark is also shown , which we see for this simulation happens at @xmath27 myr for the brightest cluster ( i.e. in 50% of the realizations the brightest cluster had an age of 15 myr or less ) and @xmath28 myr for the most massive cluster . gieles ( 2008 ) has shown that if these simulations were to be carried out over a much longer timescale ( e.g. a few gyr ) the brightest cluster would usually be old , as the most massive cluster increases faster ( due to the size - of - sample effect ) than the fading of clusters past a few 10s of myr . however , in reality a cluster is expected to lose a large fraction of its mass over time due to tidal and internal effects , which is why we limit our simulation to 100 myr . we note that if we use the starburst99 models ( leitherer et al . 1999 ) the results remain unchanged . from this we conclude that for a power - law mass distribution ( with an index of @xmath25 ) at least half of the time the brightest cluster will be young and not necessarily the most massive . note that we have not included the effects of infant - weight loss ( goodwin & bastian 2006 ) nor cluster mass loss due to tidal effects ( e.g. lamers et al . 2005 , gieles et al . 2006c ) , both of which would make older clusters less massive ( on average ) and hence decrease their luminosity faster than would be predicted based solely on stellar evolution . thus , the shown distribution is in fact a lower - limit . we also conclude from fig . [ fig : mc ] that the most massive cluster in a galaxy is likely to be old , hence the brightest cluster is often not the most massive cluster . these results differ from those of weidner et al ( 2004 ) due to the duration of the experiment . if we would use a duration of 500 myr , instead of 100 myr , then we reproduce their results , namely that in @xmath29% of the cases the most massive cluster is also the brightest . this is because over long time spans , the mass of the most massive cluster increases due to size - of - sample effects faster than stellar evolutionary fading reduces its brightness ( gieles 2008 ) . observationally , however , the most massive cluster in a galaxy does not continue to increase at the rate expected for the size - of - sample effect , but grows more slowly after @xmath30 myr ( gieles & bastian 2008 ) , presumably due to disruption effects . additionally , we have tested the expected distribution ( fraction of cases where the brightest cluster is young ) for an underlying schechter mass function . this type of distribution behaves as a power - law in the low - mass regime ( which we assume to have an index of @xmath25 ) , but decreases more rapidly at high - masses a certain characteristic mass , . we carried out the same set of simulations as above , but using a schechter mass function with @xmath31 . this particular limit is similar to that found by gieles et al . ( 2006a , b ) and used by jordn et al . ( 2005 ) , and will be justified further in [ sec : cimf ] . we have run a series of simulations with different cluster - formation rates ( cfr ) , which is necessary as the higher the cfr the more clusters are sampled near / above . the results are shown in fig . [ fig : mc ] where the number next to each curve is the cfr ( in /yr ) . the dash - dot - dot - dot ( purple ) line represents a low cfr ( 0.05 /yr ) which , as expected , is very similar to the pure power - law case , since it does not sample near . not surprisingly , as the cfr increases and one samples closer to , and above , the maximum cluster mass does not increase rapidly . hence the population will be dominated by youth , rather than mass . for high cfr ( @xmath32 /yr ) the chance of finding an older cluster which is the brightest in a galaxy essentially drops to zero . _ thus we are left to conclude that the tight relation between the observed sfr of a galaxy and the brightest cluster in the population is due to the fact that the youngest clusters are generally the brightest , and hence accurately reflect the current ( recent ) sfr . _ the above experiments assumed that the cfr of a galaxy was approximately constant during the duration of the simulations ( 100 myr ) . however , this assumption clearly does not hold for starbursts , in which a galaxy can undergo a significant enhancement of its sfr for extended periods . to demonstrate this effect we use the merger remnant ngc 7252 as an example . this galaxy is likely to have been produced by an equal mass gas - rich spiral / spiral galaxy merger ( hibbard & mihos 1995 ) approximately 500 - 1000 myr ago ( schweizer & seitzer 1998 ) . ngc 7252 is an outlying point in the relation , being significantly above the fit to the other galaxies ( @xmath33 is calculated from the difference between the observed @xmath34 and the best fit relation and error by weidner et al . ] , i.e. its brightest cluster is too bright for its current sfr ) , and as such , was not included in the fit by weidner et al . ( 2004 ) . during this merger , many extremely massive star clusters were formed , including a handful of clusters with masses exceeding @xmath35 ( schweizer & seitzer 1998 , maraston et al . 2004 , bastian et al . after this strong starburst , the sfr has presumably been decreasing to its present rate of 5.4 ( kneirman et al . 2003 ) , which is localized mainly to a relatively small inner star - forming disk ( schweizer 1982 ) . the current sfr is @xmath36 orders of magnitude less than the implied sfr ( see [ sec : sfr ] ) , hence the currently forming clusters are expected to be much less massive than those formed during the main starburst event . thus , even though these clusters are much younger than the clusters formed in the burst , they are fainter , given the extremely high masses of the clusters formed in the merger . if one were to use the brightest _ young _ cluster in ngc 7252 ( which is found in the star - forming disk ) , which has @xmath37 ( miller et al . 1997 ) then this galaxy sits well on the sfr - brightest cluster diagram , shown in fig . [ fig : relation ] ( the filled star towards the upper - right of the figure ) . the conclusion is then that the _ current _ sfr of a galaxy can be inferred from the absolute magnitude of the brightest _ young _ cluster of the population . in [ sec : discussion ] we will expand this idea and infer the peak sfr of a galaxy merger by the ( evolutionarily corrected ) luminosity of its brightest cluster . it has been noted that dwarf / irregular starburst galaxies often lie significantly above the relation shown in fig . [ fig : relation ] ( e.g. larsen 2002 , billett , hunter & elmegreen 2002 ) , a fact which led weidner et al . ( 2004 ) not to include them in their fit to the data ( shown as a dashed line in fig . [ fig : relation ] ) . a number of theories have been put forward to explain this offset . billett et al . ( 2002 ) suggest that this offset is due to different conditions governing cluster formation in small galaxies . weidner et al . ( 2004 ) , on the other hand , attribute this offset to star - formation proceeding in short bursts in small galaxies ( e.g. searle & sargent 1972 ) . here , we use the dwarf irregular ( post ) starburst galaxy , ngc 1569 , to argue for the latter model , which we summarize below . one clear difference between spirals and smaller dwarfs is the amount of energy and the timescale required to disturb the galaxy as a whole . in the case of small galaxies , a single burst of star - formation may be enough to completely rid the galaxy of any gas reservoir . studies of ngc 1569 have shown that the inner region of the galaxy hosts two massive ( @xmath38 ) clusters , with ages between 15 - 30 myr ( anders et al . 2004 , larsen et al . the inner region is currently largely devoid of dense star - forming gas and a powerful supergalactic wind is seen centred on the central region ( e.g. westmoquette et al . 2007 ) which was presumably powered by the energy input into the ism from these two massive clusters . thus , as the galaxy began to form stars / clusters more vehemently , it eventually reached the point where a single ( or in this case two ) massive clusters formed , which caused a blowout of the ism within the central regions of the galaxy . the sfr then dropped due to the lack of dense gas , and henceforth only smaller clusters could be formed . a schematic diagram of this process is shown in fig . [ fig : dwarf - evo ] . in such a model , galaxies which fall to the left of the observed trend in fig . [ fig : relation ] should have older brightest clusters than the average cluster used in the diagram , i.e. have ages larger than @xmath39 myr , so that enough time has passed for the clusters to have had a large influence on the surrounding ism . additional evidence can be found in the young stellar clusters in this galaxy . anders et al . ( 2004 ) have catalogued and age - dated approximately 160 clusters in ngc 1569 . using their dataset , and looking at only those clusters with ages less than 10 myr , we find that the brightest young cluster in the galaxy has m@xmath40 . it is this cluster which should be used in comparison with the current sfr . in fig . [ fig : relation ] we show the position of this cluster in the diagram as the lower filled star . this cluster appears to fit the relation splendidly , hence arguing that the cluster formation process is not intrinsically different in dwarfs and spirals . ngc 1705 , the other dwarf post - starburst galaxy which lies significantly to the left of the observed relation , also has had an explosive outflow attributed to the formation of its most massive cluster which occured @xmath41 myr ago ( vzquez et al . this galaxy as well seems to have drastically reduced its sfr through the feedback associated with forming a massive clusters . however , we note that ngc 1140 appears to be 1.5 - 2@xmath42 off the expected relation . this galaxy has been classified as an amorphous galaxy ( hunter et al . 1994 ) , having recently undergone a strong interaction . however , it may have originally been a low - luminosity late type spiral ( hunter et al . this galaxy has a current sfr of about 0.8 and a brightest cluster of @xmath43 which has an age of 47 myr ( moll et al . 2007 ) . due to the youth of the cluster , we would expect it to accurately reflect the current sfr in the galaxy . if this cluster / galaxy represents a statistical fluctuation or a real physical difference can not be concluded at this time . this conceptual model makes two predictions . the first , is that there should be few / no galaxies which lie significantly below the observed relation ( i.e. more than random sampling would predict ) , only above . the second is that galaxies which lie significantly above the relation should have brightest clusters which are older than the average age of clusters whose galaxies fit the relation . plane . , width=321 ] this same effect , i.e. feeback from massive stars / clusters causing a drop in the sfr , is likely to happen within individual star - forming regions within spiral galaxies ( e.g. shetty & ostriker 2008 ) , like those found in m51 or the antennae galaxies ( bastian et al . 2005 , 2006a ) . here the important length scale is that of the gmc . once the gmc begins to form clusters , the clusters , through radiative feedback , destroy the host gmc , and eventually halt star - formation within the region / complex . thus we would predict older complexes to lie to the left of the observed relation while young ones ( i.e. currently forming ) to lie on the relation , where the sfr should be estimated over the region of interest . the or equivocally n@xmath44 vs. m@xmath0 relations have been used previously in order to constrain the initial mass function of clusters . multiple studies have reported that the observed correlation is consistent with a power - law mass function with index , @xmath45 ( whitmore 2003 ; larsen 2002 ; weidner et al . 2004 ; gieles et al . this , however , is inconsistent with direct measurements of the mass function which generally find @xmath46 ( e.g. zhang & fall 1999 ; de grijs et al . 2003 , mccrady & graham 2007 ) . gieles et al . ( 2006a ) have suggested that instead of a power - law mass function , the underlying mass distribution of clusters is better described by a schechter function . this gives the standard @xmath2 form on the low mass side , but is truncated on the high mass side above some critical value . this type of function is physically motivated , as gmcs ( i.e. the material from which clusters form ) appear to have a power - law mass function which is truncated at the high mass end ( e.g. rosolowsky et al . 2007 ) . in order to test this assertion we develop the following model . we generate 50 cluster populations for eleven cluster formation rates ( cfr ) and stochastically sample an underlying schechter mass function with a lower cluster mass limit of 100 , a truncation mass , and an upper mass limit @xmath47 . each population is generated with @xmath48 clusters which are assigned ages randomly between 0 and 100 myr . we then calculate the absolute magnitude of each cluster , in the same way as was done in [ sec : why ] . finally , we calculate the average _ cluster formation rate _ ( cfr ) by dividing the total mass in clusters formed in each population by the age range sampled ( i.e. 100 myr ) . the top panel of fig . [ fig : relation - sim ] shows the results of the simulations for various values of , where the points and lines represent the mean in the logarithm of the sfr and magnitude of the 550 realizations for each value of . the values of and the resulting slopes from linear least - square fits to the unbinned data points ( i.e. in the same way that weidner et al . ( 2004 ) have fit the data ) , are also shown in the figure . the bottom panel of fig . [ fig : relation - sim ] shows the same simulations but with the cfr translated into a sfr using @xmath49 ( the fraction of star - formation which happens in bound clusters ) , which will be discussed in detail in [ sec : cfr - sfr ] . the dashed line in both panels is the observed relation , taken from weidner et al . ( 2004 ) , which has a slope of @xmath50 . as can be seen , low values of ( @xmath51 ) do not fit the data well in the high sfr regime , and high values ( i.e. pure power - laws with index of @xmath52 ) over - predict the magnitude of the brightest cluster for a given sfr . however , values of of @xmath53 appear to fit the observed trend quite nicely . weidner et al . ( 2004 ) found that the observations were best fit by m@xmath54 . this is @xmath55 off that expected from the pure power - law , @xmath2 case ( seen as the green line with triangles in fig . [ fig : relation - sim ] ) , hence we conclude that this is ruled out by the current data . the rather high values of which do fit the data ( @xmath56 ) make a direct comparison ( i.e. a binned mass distribution ) between a power - law and a schechter mass function extremely difficult , as the two populations only differ by a handful of the most massive clusters . an example of this is found in whitmore , chandar , & fall . ( 2007 ) who found a lack of massive clusters at the high mass end of the mass distribution in the antennae galaxies compared to a pure power - law with an index of @xmath25 , and they note that an index of @xmath57 to @xmath58 is required . this difference was near the poissonian limit so no strong conclusion could be reached . however , size - of - sample effects , like the relation , which rely on only the brightest cluster in a sample , are quite sensitive to the difference between the two functions in the high mass regime , allowing for a clear separation . it should be noted that it is unlikely that the value of is universal and that it probably depends on the ambient conditions of the host galaxy , i.e. to what mass can the galaxy form the molecular concentrations required to form star clusters . for example , the truncation in the gmc mass function in the inner regions of m33 happens at @xmath59 ( rosolowsky et al . 2007 ) , which would lead to a cluster mass truncation below this value . in contrast , major galaxy mergers such as ngc 7252 and arp 220 may be able to form extremely dense and high mass gas concentrations . a schechter function with of @xmath60 can be ruled out in the case of ngc 7252 due to the large number of clusters with masses well in excess of this value ( bastian et al . additionally , a schechter function results in a slightly curved relation where at low sfr ( when a population does not sample near ) the slope is the same as for the power - law exponent in the schechter function . if , however , weakly correlates with the sfr ( as might be expected if is dependent on galaxy type ) then the slope would be preserved beyond the plotted bounds in fig . [ fig : relation - sim ] . applied , i.e. converting the cfr to sfr . the dashed lines shows the observed relation as fit by weidner et al . ( 2004 ) , which has a slope of @xmath50 . the filled triangles represent a case where is sufficiently high as to not affect the mass distribution ( i.e. a representation of the pure - power law mass function case ) . as can be seen , a schechter function with @xmath61 provides an excellent fit to the data , while the pure power - law case over - predicts the luminosity of the brightest cluster for a given sfr.,title="fig:",width=321 ] applied , i.e. converting the cfr to sfr . the dashed lines shows the observed relation as fit by weidner et al . ( 2004 ) , which has a slope of @xmath50 . the filled triangles represent a case where is sufficiently high as to not affect the mass distribution ( i.e. a representation of the pure - power law mass function case ) . as can be seen , a schechter function with @xmath61 provides an excellent fit to the data , while the pure power - law case over - predicts the luminosity of the brightest cluster for a given sfr.,title="fig:",width=321 ] in order to compare our monte carlo cluster population simulations to the observed data , we need to apply a correction factor between the cluster - formation rate ( cfr ) and the star - formation rate ( sfr ) , which we define as @xmath62 . this factor enters as a horizontal scaling in fig . [ fig : relation - sim ] ( i.e. the horizontal shift between the top and bottom panels ) . in order to equate our best fit simulations ( in terms of the slope of the relation , i.e. a schechter function with @xmath63 ) to the observations we needed to use @xmath64 , meaning that optically selected `` bound '' clusters represent only 8% of the total star - formation of a galaxy . if we instead adopt a pure - power law with index @xmath65 then @xmath66 . by `` bound '' we refer to clusters which have survived the transition from being embedded in molecular gas to being exposed , and hence possible to be included optically selected cluster samples ( see goodwin 2008 for a recent review of this transition period ) . additionally , the cluster must be compact enough so as to make it into cluster samples , for which it is often assumed that the cluster must then be gravitationally bound . note , however , that we can not constrain the fraction of stars which are _ formed _ in dense clusters relative to the field or loose association . we can only constrain the fraction of stars , relative to the total , which end up in bound clusters after the transition for embedded to exposed . it is also possible to shift the other simulations onto the observations using a conversion factor which is dependent on the sfr . however , for the pure power - law case with an index of @xmath25 , this implies that cluster formation becomes _ less _ efficient for high sfr . this issue is discussed in detail in bastian & gieles ( in prep . ) . , hence shifting the curves along their slopes . ] this value of @xmath67 is in good agreement with that found for the solar neighbourhood by lada & lada ( @xmath68 ; 2003 ) and lamers & gieles ( @xmath69 ; 2007 ) . it is also in relatively good agreement with that found by gieles & bastian ( 2008 ) for the smc ( @xmath70 ) . the interpretation and implications of such a low value of @xmath67 are also treated in detail in bastian & gieles ( in prep . ) . interestingly , the same efficiency factor was needed , for all values of the cfr , in order to fit the data . this means that the same fraction of stars which are forming are going into clusters , independent of the sfr . along with the smooth trend in fig . [ fig : relation ] , _ this implies that there do not exist multiple modes of cluster formation_. starburst galaxies do not preferentially form high - mass clusters , they are simply able to sample further out into the cluster mass function . since the cfr appears to make up a more or less constant fraction of the sfr ( for @xmath71 sfr ( /yr ) @xmath72 ) , it naturally follows that the star and cluster formation histories of a galaxy should coincide . exceptions to this ansatz may be due to the combined uncertainties of the cluster disruption correction and the sampling of small fields for the star - formation history which can suffer from severe variance ( e.g. maschberger et al . , in prep ) . combining equation 2 in weidner et al . ( 2004 ) , which is shown as a dashed line in fig . 1 , and equation 4 in kennicutt ( 1998 ) , one can find the relation between the brightest cluster , the sfr , and the infrared luminosity of a galaxy . however , as noted above , the brightest star cluster _ associated with that epoch of star - formation _ must be used . by using the brightest cluster in a galaxy , and accounting for cluster fading due to stellar evolution using simple stellar population models , one can obtain an estimate of the sfr during the epoch of the formation of that cluster . this is similar to what has been done by maschberger & kroupa ( 2007 ) , however , instead of deriving the full star - formation history of a galaxy , we are interested in what was the peak sfr in that galaxy s ( recent ) history . for such a study , post - galaxy mergers are an ideal place to look , as they are known to form copious amounts of massive clusters ( e.g. schweizer 1987 , ashman & zepf 1992 ) . since major mergers of galaxies are relatively rare in the local universe , one can look at the fossil remnants which are much more prevalent , in order to infer the conditions during the merger . many nearby ( @xmath73 mpc ) merger remnants have had their star - cluster populations catalogued and here we will estimate the sfr and peak log ( l@xmath74 ) in order to see the prevailing conditions at the time of the cluster formation and whether these galaxies passed through a ulirg / hlirg phase . combining the observed relation and that between the sfr and far - infrared luminosity ( kennicutt 1998 ) results in the equations : @xmath75 @xmath76 @xmath77 the results for seven post - merger galaxies are given in table [ table : mergers ] . we have used the published values of @xmath18 and cluster age , and used the bruzual & charlot ( 2003 ) ssp models ( solar metallicity and salpeter stellar imf ) in order to estimate the absolute magnitude of the brightest cluster at an age of 10 myr , @xmath78 . the estimated values of are lower limits since we have not attempted to correct for mass loss due to tidal / dynamical effects , which are expected to be relatively small since the mass - loss is inversely related to cluster mass ( e.g. baumgardt & makino 2003 , lamers et al . 2005 ) . additionally , the peak values quoted here are also lower limits since they do not consider any agn component which may have been present . it has been noted ( e.g. veilleux , sanders , & kim 1999 ) that the fraction of galaxies with a clear agn component increases with increasing luminosity . however , verma et al . ( 2002 ) note that in their sample of four hlirgs , three needed substantial fraction of the to come from an obscured starburst ( from 3075% of the total luminosity ) . ngc 7252 stands out among the others in its estimated star formation rate ( @xmath79 ) and its peak luminosity would clearly place it in the range of hyper - luminous infrared galaxies ( hlirgs ) . verma et al . ( 2002 ) have decomposed the infrared spectra of two hlirgs into their starburst and agn components , and have estimated star - formation rates above 3000 . if , instead of w3 , we used the second brightest cluster in the ngc 7252 , w30 , the implied sfr and peak would be @xmath80 and @xmath81 , respectively . while still well within the ulirg category , this shows the sensitivity and limitations of the method used here . one possibility to decrease this sensitivity would be to use the say three or five brightest clusters , which would reduce the sampling scatter . young clusters do not form in isolation but are often , instead , found in cluster complexes ( e.g. bastian et al . 2005,2006 ) . within these complexes , cluster merging may be an important mechanism in building massive clusters ( e.g. fellhauer & kroupa 2002 ; kissler - patig , jordan , & bastian 2006 ) . through merging , it is possible to raise a galaxy in the relation , as the brightness of the most luminous cluster would increase through cluster mergers , but the sfr would remain unchanged . we do not expect this to bias our results substantially , as the relation is empirically derived , and this effect may already be included . however , if a significant amount of merging happens after the first 1520 myr of a cluster s life , the current ( stellar evolutionary corrected ) magnitudes would be over - estimates leading to high sfr and derivations . one galaxy , ngc 34 , stands out as unique in the sample assembled here . it is currently a lirg with a far - ir luminosity of 10@xmath82 , which translates to a sfr of 75 - 90 ( s03 ; prouton et al . . however , it appears to have had a higher sfr in the past . schweizer & seitzer ( 2007 ) have obtained spectroscopic ages of the two brightest clusters ( @xmath83 ) in the galaxy , and have derived ages of @xmath84 myr ( although they note that cluster 2 may be @xmath85 times older ) . correcting for evolutionary fading and applying eqns . @xmath86 gives an absolute v - band magnitude at 10 myr , peak sfr , and peak of -17.3 mag , @xmath80 , and @xmath87 , respectively . therefore , we can infer that the sfr and have both been declining for the past 150 myr to the current levels , and that this timescale is a lower limit to the duration of the lirg / ulirg phase . if cluster 2 is indeed four times older , then the ulirg duration for this galaxy is at least @xmath80 myr , or if the sfr has not been continuous , then ngc 34 has gone through multiple ulirg phases . using the empirically derived relation between the star - formation rate ( sfr ) and the luminosity of the brightest star cluster within the galaxy ( larsen 2002 ) we have investigated the conditions ( sfr and infrared luminosity ) that were present during the peak star - forming epoch of galactic mergers . we began with a series of simple monte carlo simulations of cluster populations with different assumptions on the underlying cluster mass function , and showed that , independent of the model construction , the youngest clusters ( @xmath88 myr ) were predominantly the brightest . this , along with size - of - sample effects , can explain the observed relation , as the young clusters should be a good representation of the current sfr . using data from the literature as well as archival hst imaging , we have further tested the high sfr regime of the relation . we found , in agreement with the results of wilson et al . ( 2006 ) , that the relation continues to hold up to sfrs of a few hundred solar masses per year . we have also presented a schematic model ( following on weidner et al . 2004 ) to explain why some dwarf ( post ) starburst galaxies lie on the left of the observed relation . the basic model is that the formation of a massive cluster is enough to disturb a substantial part of the galaxy , and cause an ism blowout ( as seen in ngc 1569 ) which effectively terminates further star - formation . this same process is expected to take place within star - forming complexes in spiral galaxies . by generating a large sample of monte carlo simulations of cluster populations with varying underlying initial mass functions and cluster - formation rates , we have shown that the observed relation is best fit by a schechter mass function with @xmath89 . this rather high value of makes it extremely difficult to detect a difference between a pure power - law or a schechter type mass function directly ( i.e. through binned cluster mass distributions ) . however , size - of - sample effects which rely heavily on the most massive or brightest object in a sample are much more sensitive to small differences between these two functions at the high mass end . we find that the data are inconsistent with the pure power - law , @xmath2 case at the @xmath55 level . using this set of simulations , we also showed that a constant fraction of star - formation goes into clusters . this fraction , which we term @xmath67 , is @xmath90 , meaning that optically selected clusters represent only 8% of the total star - formation in a galaxy . this is similar to that found for the solar neighborhood by lada & lada ( 2003 ) and lamers & gieles ( 2007 ) , and it is also in fair agreement with that derived for the smc , namely @xmath91% ( gieles & bastian 2008 ) . @xmath67 appears to be independent of the star - formation rate over 6 orders of magnitude . from this we conclude that cluster formation does not have multiple modes and that the cluster formation history of a galaxy should accurately reflect its star - formation history ( once a proper accounting of the cluster formation rate and variance in the stellar fields is taken into account ) . using the fit to the relation by weidner et al . ( 2004 ) , and extrapolating to higher sfrs , we have estimated the peak sfr and infrared luminosity of seven ( post - starburst ) galactic mergers by estimating the absolute magnitude of the brightest star cluster when corrected for stellar evolutionary fading . we estimate sfrs between 50 and @xmath79 and infrared luminosities between @xmath92 . these are lower limits as we have not corrected for cluster mass - loss ( fading ) due to tidal fields and internal dynamics . additionally , the infrared luminosities are also lower limits as they do not take into consideration of any additional contribution to the by an agn component . these results further demonstrate that many mergers pass through a lirg / ulirg / hlirg phase during their evolution . ngc 34 , which currently has a high sfr ( @xmath93 ) , also hosts a population of bright massive clusters with ages of @xmath94 myr ( schweizer & seitzer 2007 ) . using these clusters we have estimated a peak sfr of @xmath80 which corresponds to an infrared luminosity of @xmath81 , well within the ulirg regime . thus we have placed a lower - limit to the duration of the lirg / ulirg phase of 150 myr for this galaxy . nb thanks mark gieles for helpful discussions and for providing the stochastic power - law and schechter function sampling programmes , used throughout this work . franois schweizer , cathie clarke , thomas maschberger , & john hibbard , are gratefully acknowledged for interesting and helpful discussions . the anonymous referee is acknowledged for helpful suggestions . this paper is based in part on observations with the nasa / esa _ hubble space telescope _ which is operated by the association of universities for research in astronomy , inc . under nasa contract nas5 - 26555 . lamers , h. j. g. l. m. , & gieles , m. 2007 , in de koter a. , waters r. , smith l. j. , eds , asp conf . ser . , mass loss from stars and the evolution of stellar clusters . , san francisco , in press ( astro - ph/0702166 )	we further the recent discussion on the relation between the star - formation rate ( sfr ) of a galaxy and the luminosity of its brightest star - cluster ( sfr vs. m@xmath0 ) . we first show that the observed trend between sfr vs. m@xmath0 is due to the brightest cluster in a galaxy being preferentially young ( @xmath1 myr - for a constant sfr ) and hence a good tracer of the current sfr , although we give notable exceptions to this rule . archival hst imaging of high - sfr galaxies , as well as additional galaxies / clusters from the literature , are used to further confirm the observed trend . using a series of monte carlo simulations we show that a pure power - law mass function with index , @xmath2 , is ruled out by the current data . instead we find that a schechter function ( i.e. a power - law with an exponential truncation at the high mass end ) provides an excellent fit to the data . additionally , these simulations show that bound cluster formation ( in /yr ) represents only @xmath3% of the total star - formation within a galaxy , independent of the star - formation rate . from this we conclude that there is only a single mode of cluster formation which operates over at least six orders of magnitude in the sfr . we provide a simple model of star / cluster formation feedback within dwarf galaxies ( and star - forming complexes within spirals ) which highlights the strong impact that a massive cluster can have on its surroundings . using this relation , we can extrapolate backwards in time in order to estimate the peak sfr of major merger galaxies , such as ngc 7252 , ngc 1316 , and ngc 3610 . the derived sfrs for these galaxies are between a few hundred and a few thousand solar masses per year . the inferred far infrared luminosity of the galaxies , from the extrapolated sfr , places them well within the range of ultra - luminous galaxies ( ulirgs ) and for ngc 7252 within the hyper - luminous infrared galaxy regime . thus , we provide evidence that these post merger galaxies passed through a ulirg / hlirg phase and are now evolving passively . using the current and extrapolated past sfr of ngc 34 , we infer that the ulirg phase of this galaxy has lasted for at least 150 myr . [ firstpage ] galaxies : star clusters galaxies : starburst
the experimental achievement of bose - einstein condensation ( bec ) in dilute alkali atoms @xcite , spin - polarized hydrogen @xcite and recently in metastable helium @xcite has enormously stimulated the theoretical research @xcite on the ultracold bosons . among the several intriguing questions on the statistical properties of trapped interacting bose gases , the problem of condensate fluctuations @xmath2 of the mean ground state occupation number @xmath3 is of central importance . apart from the intrinsic theoretical interest , it is foreseeable that such fluctuations will become experimentally testable in the near future @xcite . on the other hand , the calculations of @xmath4 are crucial to investigate the phase collapse time of the condensate @xcite . it is well known that within a grand canonical ensemble the fluctuations of the condensate are given by @xmath5 , implying that @xmath6 becomes of order @xmath7 when the temperature approaches zero . to avoid this sort of unphysically large condensate fluctuations , a canonical ( or a microcanonical ) ensemble has to be used to investigate the fluctuations of the condensate . on the other hand , because in the experiment the trapped atoms are cooled continuously from the surrounding , the system can be taken as being in contact with a heat bath but the total number of particles in the system is conserved . thus it is necessary to use the canonical ensemble to investigate the statistical properties of the trapped weakly interacting bose gas . within the canonical as well as the microcanonical ensembles , the condensate fluctuations have been studied systematically in the case of an ideal bose gas in a box @xcite , and in the presence of a harmonic trap @xcite . recently , the question of how interatomic interactions affect the condensate fluctuations has been an object of several theoretical investigations @xcite . @xcite investigated the condensate fluctuations of interacting bose gases using the lowest - order perturbation theory and a two - gas model , while giorgini _ @xcite addressed this problem within a traditional particle - number - nonconserving bogoliubov approach . recently , kocharovsky _ et al . _ @xcite supported and extended the results of the work of giorgini _ et al . _ @xcite using a particle - number - conserving operator formalism . although the condensate fluctuations are thoroughly investigated in ref.@xcite , to best our knowledge up to now an analytical description of the probability distribution function for the interacting bose gas directly from the microscopic statistics of the system has not been given . note that as soon as the probability distribution function of the system is obtained , it is straightforward to get the mean ground state occupation number and the condensate fluctuations . the purpose of the present work is an attempt to provide such an analytical description of the probability distribution function of interacting and non - interacting bose gases based on the analysis of the partition function of the system . we shall investigate in this paper the condensate fluctuations of interacting and non - interacting bose gases confined in a harmonic trap . the analytical probability distribution function of the condensate will be given directly from the partition function of the system using a canonical ensemble approach . for an ideal bose gas , we find that the probability distribution of the condensate is a gaussian function . in particular , our method can be easily extended to discuss the probability distribution function for a weakly interacting bose gas . a unified way is given to calculate the condensate fluctuations from the lowest - order perturbation theory and from bogoliubov theory . we find that different methods of approximation for the interacting bose gas give quite different predictions concerning the condensate fluctuations . we show that the fluctuations based on the lowest - order perturbation theory follow the law @xmath8 , while the fluctuations based on the bogoliubov theory behave as @xmath9 . the paper is organized as follows . ii is devoted to outline the canonical ensemble , which is developed to discuss the probability distribution function of bose gases . in sec . iii we investigate the condensate fluctuations of the ideal bose gas confined in a harmonic trap . in sec . iv the condensate fluctuations of the weakly interacting bose gas are calculated based on the lowest order perturbation theory . in sec . v the condensate fluctuations due to collective excitations are obtained based on bogoliubov theory . finally , sec . vi contains a discussion and summary of our results . according to the canonical ensemble , the partition function of the system with @xmath7 trapped interacting bosons is given by @xmath10 = \sum_{\sigma _ { { \bf { n}}}n_{{\bf { n}}}=n}\exp \left [ -\beta \left ( \sigma _ { { \bf { n}}}n_{{\bf { n}}}\varepsilon _ { { \bf { n}}% } + e_{int}\right ) \right ] } , \label{par1}\ ] ] where @xmath11 and @xmath12 are occupation number and energy level of the state @xmath13 , respectively . @xmath14 and @xmath15 are non - negative integers . @xmath16 is the interaction energy of the system . for convenience , by separating out the ground state @xmath17 from the state @xmath18 , we have @xmath10 = \sum_{n_{{\bf 0}}=0}^{n}\left\ { \exp \left [ -\beta \left ( e_{{\bf 0}}+e_{int}\right ) \right ] z_{0}\left ( n , n_{{\bf 0}}\right ) \right\}}% , \label{par2}\ ] ] where @xmath19 stands for the partition function of a fictitious system comprising @xmath20 trapped ideal non - condensed bosons : @xmath21 . } \label{ii - function-1}\ ] ] assuming @xmath22 is the free energy of the fictitious system , we have @xmath23 the calculation of the free energy @xmath24 is nontrivial because there is a requirement that the number of non - condensed bosons is @xmath25 in the summation of the partition function @xmath26 . using the saddle - point method developed by darwin and fowler @xcite , it is straightforward to obtain a useful relation between the free energy @xmath27 and the fugacity @xmath28 of the fictitious @xmath29 non - interacting bosons @xmath30 where the fugacity @xmath28 is determined by @xmath31 z_{0}^{-1}-1}. } \label{relation2}\ ] ] we have given a simple derivation of eqs . ( [ relation1 ] ) and ( [ relation2 ] ) in the appendix . using the free energy @xmath32 , the partition function of the system becomes @xmath10 = \sum_{n_{{\bf 0}}=0}^{n}\exp \left [ q\left ( n , n_{{\bf 0}% } \right ) \right ] } , \label{par3}\ ] ] where @xmath33 it is obvious that @xmath34)\exp \left [ q\left ( n , n_{{\bf { 0}}}\right ) \right]$ ] represents the probability finding @xmath35 atoms in the condensate . to obtain the probability distribution function of the system , let us first investigate the largest term in the sum of the partition function @xmath36 $ ] . assume the number of the condensed atoms is @xmath37 in the largest term of the partition function . the largest term @xmath38 $ ] is determined by requiring that @xmath39 , _ i.e. _ , @xmath40 using eq . ( [ relation1 ] ) we obtain @xmath41 in addition , from eq . ( [ relation2 ] ) , the most probable value @xmath42 is determined by @xmath43 \left ( z_{0}^{p}\right ) ^{-1}-1}. } \label{non1}\ ] ] in the case of an ideal bose gas , from eq . ( [ z0p ] ) one obtains @xmath44 . thus @xmath45 is the same as the mean ground state occupation number obtained by using a grand canonical ensemble approach . for sufficiently large @xmath7 , the sum @xmath46 in ( [ par3 ] ) may be replaced by the largest term , since the error omitted in doing so is statistically negligible . in this situation , ( [ non1 ] ) shows the equivalence between the canonical ensemble and the grand canonical ensemble for large @xmath7 . the other terms in the partition function ( [ par3 ] ) will contribute to the fluctuations of the condensate , and lead to the deviation of @xmath47 from the most probable value @xmath48 . if @xmath49 , we have @xmath50 . assuming @xmath51 from eqs . ( [ relation1 ] ) and ( [ qqq ] ) , we obtain @xmath52 by eqs . ( [ relation2 ] ) and ( [ z00 ] ) , we have @xmath53 @xmath54 \exp \lbrack -\beta { \frac{\partial } { \partial n_{{\bf { 0}}}}% \left ( e_{{\bf { 0}}}+e_{int}\right ) -\alpha \left ( n , n_{{\bf { 0}}}\right ) } % \rbrack -1}. } \label{main3}\ ] ] combining eqs . ( [ non1 ] ) and ( [ main3 ] ) , we get the following equation for determining @xmath55 @xmath56 \exp \lbrack -\beta { % \frac{\partial } { \partial n_{{\bf { 0}}}^{p}}\left ( e_{{\bf { 0}}% } + e_{int}\right ) } \rbrack -1}}\ ] ] @xmath57 \exp \lbrack -\beta { \frac{\partial } { \partial n_{{\bf { 0}}}}\left ( e_{{\bf { 0}}}+e_{int}\right ) -\alpha \left ( n , n_{{\bf { 0}}}\right ) } \rbrack -1}. } \label{alpha}\ ] ] once we know @xmath58 and @xmath16 of the system , it is straightforward to obtain @xmath59 from eq . ( [ alpha ] ) . using @xmath59 , one can obtain the probability distribution function of the system . from eq . ( [ q - alpha ] ) , we obtain the following result for @xmath60 @xmath61 thus the partition function of the system becomes @xmath10 = \sum_{n_{{\bf { 0}}}=0}^{n}\left\ { \exp \left [ q\left ( n , n_{{\bf { 0}}}^{p}\right ) \right ] g\left ( n , n_{{\bf { 0}}}\right ) \right\ } , } \label{par - alpha}\ ] ] where @xmath62 . } \label{ideal - dis}\ ] ] assuming @xmath63 is the probability to find @xmath64 atoms in the condensate , @xmath65 represents the ratio @xmath66 , _ i.e. _ , the relative probability to find @xmath35 atoms in the condensate . from eq . ( [ ideal - dis ] ) , the normalized probability distribution function is given by @xmath67 , } \label{norm - di}\ ] ] where @xmath68 is a normalization constant and is given by the condition @xmath69 . as soon as we know @xmath70 , the statistical properties of the system can be clearly described . from eqs . ( [ par - alpha ] ) and ( [ ideal - dis ] ) one obtains the mean ground state occupation number @xmath3 and fluctuations @xmath71 in the canonical ensemble : @xmath72 } { \sum_{n_{{\bf { 0}}}=0}^{n}\exp \left [ q\left ( n , n_{{\bf { 0}}}\right ) \right ] } = \frac{% \sum_{n_{{\bf { 0}}}=0}^{n}n_{{\bf { 0}}}g\left ( n , n_{{\bf { 0}}}\right ) } { % \sum_{n_{{\bf { 0}}}=0}^{n}g\left ( n , n_{{\bf { 0}}}\right ) } } \label{mean - ideal}\ ] ] @xmath73 @xmath74 ^{2}. } \label{fluc - ideal}\ ] ] starting from eqs . ( [ mean - ideal ] ) and ( [ fluc - ideal ] ) , one can calculate the mean ground state occupation number and fluctuations for ideal and interacting bose gases . we now study the condensate fluctuations of the system with @xmath7 non - interacting bosons trapped in an external potential . the potential is a harmonic one with the form @xmath75 where @xmath76 is the mass of atoms , @xmath77 , @xmath78 , and @xmath79 are frequencies of the trap along three coordinate - axis directions . the single - particle energy level has the form @xmath80 from eq . ( [ non1 ] ) one can get easily the most probable value @xmath81 , which reads @xmath82 ^{2/3}}\left ( \frac{t}{t_{c}^{0}}\right ) ^{2}n^{2/3 } , } \label{ideal - most}\ ] ] where @xmath83 is the critical temperature of the ideal bose gas in the thermodynamic limit . @xmath84 and @xmath85 are arithmetic and geometric averages of the oscillator frequencies , respectively . when obtaining ( [ ideal - most ] ) we have used the following expression of the density of states @xcite @xmath86 on the basis of the same density of states a detailed study of the critical temperature and the ground state occupation number was given recently in @xcite . in a thermodynamic equilibrium the deviation from the most probable value @xmath81 is small , therefore we can use the approximation @xmath87 \approx 1-\alpha \left ( n , n_{{\bf { 0}}}\right ) $ ] . from eq . ( [ alpha ] ) and the single - particle energy level ( [ energy - trap ] ) we find the result for @xmath88 @xmath89 when obtaining @xmath59 we have used the expansion @xmath90 @xcite , where @xmath91 belongs to the class of functions @xmath92 and @xmath93 is riemann @xmath94 function . from ( [ ideal - dis ] ) and ( [ alpha - har ] ) we obtain the normalized probability distribution function of the harmonically trapped ideal bose gas @xmath95 , } \label{dis - har}\ ] ] where @xmath96 is a normalization constant . it is interesting to note that the expression ( [ dis - har ] ) is a gaussian distribution function . from the formulas ( [ mean - ideal ] ) , ( [ fluc - ideal ] ) , ( [ ideal - most ] ) , and ( [ dis - har ] ) we can obtain @xmath97 and @xmath98 for the ideal bose gas . in fig . 1(a ) and fig . 1(b ) we plot @xmath99 as a function of temperature for @xmath100 and @xmath101 ideal bosons confined in an isotropic harmonic trap . the dashed line displays @xmath102 in the thermodynamic limit , while the solid line displays @xmath103 within the grand canonical ensemble ( or @xmath81 within the canonical ensemble ) . the dotted line displays @xmath104 within the canonical ensemble . when @xmath105 , @xmath106 from the canonical ensemble agrees well with that from the grand canonical ensemble . obviously , in the case of @xmath107 , @xmath102 obtained from the canonical ensemble coincides with that from the grand canonical ensemble . from the formulas ( [ mean - ideal ] ) and ( [ fluc - ideal ] ) and the results ( [ ideal - most ] ) and ( [ dis - har ] ) we can obtain the condensate fluctuations of the ideal bose gas . in fig . 2 we plot the numerical result of @xmath108 ( solid line ) for @xmath109 ideal bosons confined in an isotropic harmonic potential . the dashed line displays the result of holthaus _ et al . _ @xcite , where the saddle - point method is developed to avoid the failure of the standard saddle - point approximation below the onset of bec . in fig . 2 the dotted line displays the result given in refs.@xcite . our results coincide with those of refs.@xcite when @xmath110 is smaller than @xmath111 , which corresponds to the maximum fluctuations @xmath112 . in fact , when @xmath113 , from ( [ mean - ideal ] ) , ( [ fluc - ideal ] ) and ( [ dis - har ] ) , we obtain the analytical result for the condensate fluctuations : @xmath114 which recovers the result given in refs.@xcite . this shows the validity of the probability distribution function ( [ dis - har ] ) for studying the statistical properties of the system . at the critical temperature , however , our results give @xmath115 which is much smaller than the result of ref.@xcite . this difference is apprehensible because the analysis of giorgini _ et al . _ @xcite holds in the canonical ensemble except near and above @xmath116 , while our result holds also for the temperature near @xmath117 . near the critical temperature , our result ( solid line ) agrees with that of holthaus _ et al . the results given by ( [ analytical ] ) and ( [ critical - fluc ] ) show a normal behavior of the condensate fluctuations for the harmonically trapped ideal bose gas . the fluctuations of the condensate can also be evaluated at @xmath118 . in the case of @xmath119 , from ( [ dis - har ] ) we get @xmath120 if @xmath121 , while @xmath122 when @xmath123 . therefore , we obtain @xmath124 and @xmath125 when @xmath119 . note that our results are reliable although the disputable saddle - point method is used to investigate the fluctuations of the condensate . it is well known that the applicability of the saddle - point approximation for the condensed bose gas has been the subject of a long debate @xcite . recently , the analysis given in ref.@xcite showed that the fluctuations are overestimated , and do not appear to vanish properly with temperature using the conventional saddle - point method . our discussions on the condensate fluctuations are reasonable because of two reasons : ( i ) as proved given in ref.@xcite , the most probable value eq . ( [ non1 ] ) for the non - interacting bose gas is still correct , even when carefully dealing with the failure of the standard saddle - point method below the critical temperature . ( ii ) in the usual statistical method @xmath126 and @xmath127 are obtained through the first and second partial derivatives of partition function , respectively . when the saddle - point approximation is used to calculate the partition function of the system , the error will be overestimated in the second partial derivative of the partition function . thus one can not obtain correct condensate fluctuations using usual method . however , in our approach here what we used is the reliable result given by eqs . ( [ non1 ] ) and ( [ main3 ] ) . the probability distribution function of the ground state occupation number can be obtained directly from eqs . ( [ non1 ] ) and ( [ main3 ] ) , without resorting to the second partial derivative of the partition function . @xmath126 and @xmath127 are obtained from the probability distribution function in our approach . the correct description of @xmath6 near zero temperature and critical temperature also shows the validity of our method . thus our approach has provided in some sense a simple method recovering the applicability of the saddle - point method through the calculations of the probability distribution function of the system . below the critical temperature , bose - einstein condensation results in a sharp enhancement of the density in the central region of the trap . this makes the interacting effect between atoms be much more important than above @xmath128 . the correction to the condensate fraction and critical temperature due to the interatomic interaction has been discussed within grand canonical ensemble @xcite and canonical ensemble@xcite . in this section we investigate the role of interaction on the condensate fluctuations of a weakly interacting bose gas . using the lowest - order perturbation theory the interaction energy of the system takes the form @xmath129 where @xmath130 is the coupling constant fixed by the s - wave scattering length @xmath131 . @xmath132 and @xmath133 are the density distributions of the condensate and normal gas , respectively . below the critical temperature , by thomas - fermi approximation the density distribution of the condensate reads @xmath134 where @xmath135 is the chemical potential of the system . the temperature dependence of the chemical potential is then fixed by the number of atoms in the condensate @xmath136 where @xmath137 is the harmonic oscillator length . moreover , since @xmath138 , the energy per particle in the condensate turns out to be @xmath139 as a first - order approximation , omitting the interaction between condensed and non - condensed atoms , the partition function of the system is given by @xmath140 = \sum_{n_{{\bf { 0}}}=0}^{n } \left\ { \exp \left [ -\beta n_{{\bf { 0}}}\varepsilon _ { { \bf { 0}}}^{tf}\right ] z_{0}\left ( n , n_{% { \bf { 0}}}\right ) \right\}. } \label{par - hf}\ ] ] from eq . ( [ non1 ] ) the most probable value reads @xmath141 -1}. } \label{most - hf-1}\ ] ] using the density of states , _ i.e. _ , ( [ density - state ] ) , one obtains the result for the most probable value @xmath81 @xmath142 @xmath143 ^{2/3}}t^{2}n^{2/3 } , } \label{most - hf}\ ] ] where @xmath144 is the reduced temperature . introducing the scaling parameter @xmath145 @xcite @xmath146 ( [ most - hf ] ) becomes @xmath142 @xmath147^{2/3}}t^{2}n^{2/3}. } \label{most - hf - tra}\ ] ] note that the corrections due to the interatomic interaction and finite number of particle of the system can be obtained simultaneously when ( [ most - hf - tra ] ) is used to calculate @xmath148 and @xmath149 of the system . the second term on the right hand side of ( [ most - hf - tra ] ) accounts for the correction of the interaction effect . the correction due to the interatomic interaction coincides with the lowest - order thermal depletion obtained in the grand canonical ensemble approach @xcite . for other @xmath64 , assuming @xmath150 , we get @xmath151 \exp \left [ -\alpha \left ( n , n_{{\bf { 0}}}\right ) \right ] -1}. } \label{alpha - hf-1}\ ] ] combining eqs . ( [ most - hf-1 ] ) and ( [ alpha - hf-1 ] ) , one obtains the result for @xmath152 : @xmath153 the probability distribution function of the interacting bose gas is then @xmath154 = } } \ ] ] @xmath155 where @xmath156 is a normalization constant . @xmath157 is the probability distribution function given by the formula ( [ dis - har ] ) for the ideal harmonically trapped bose gas . the correction @xmath158 originating from the interatomic interaction takes the form @xmath159 @xmath160 \right\}. } \label{r - hf}\ ] ] note that @xmath161 is not a gaussian distribution function because of the existence of the non - gaussian factor @xmath162 . from ( [ fluc - ideal ] ) and ( [ dis - hf ] ) we can obtain the numerical result of @xmath163 . in fig . 3 we have provided the numerical result of @xmath164 for @xmath165 interacting bosons confined in an isotropic harmonic trap with @xmath166 and @xmath167 , respectively . the crossover from the interacting to the non - interacting bose gases is clearly shown . from fig . 3 we find that the repulsive interaction between atoms results in a decrease of the condensate fluctuations . for an attractive interaction , we anticipate that the corrections between atoms result in an increase for the condensate fluctuations . the interaction between condensed and non - condensed atoms gives high - order correction to the thermodynamic properties of the system . near the critical temperature , _ i.e. _ , when @xmath168 @xcite , we have @xmath169 . in addition , we can adopt the semiclassical approximation for the normal gas @xcite , _ i.e. _ , @xmath170 with @xmath171 ^{1/2}$ ] being the thermal wavelength . from eqs . ( [ non1 ] ) and ( [ inter - energy ] ) , it is straightforward to obtain the most probable value @xmath172 near the critical temperature : @xmath173 \theta t^{7/2 } -\frac{3\zeta \left ( 2\right ) } { 2 \left [ \zeta \left ( 3\right ) \right ] ^{2/3 } } \frac{\overline{\omega } } { \omega _ { ho}}t^{2}n^{-1/3}% } { 1+\frac{\zeta \left ( 2\right ) \theta n^{1/2}t^{2 } } { \left [ \zeta \left ( 3\right ) \right ] ^{1/2}\zeta \left ( 3/2\right ) } } . } \label{near}\ ] ] where @xmath174 ^{3/2}}$ ] . when obtaining ( [ near ] ) we have introduced a scaling parameter @xmath175 . @xmath176 can also be written in the form of @xmath177 . by setting @xmath178 , from ( [ near ] ) we obtain the shift of the critical temperature : @xmath179^{2/3}}\frac{% \overline{\omega } } { \omega _ { ho}}n^{-1/3}. \label{shift}\ ] ] the first term on the right hand side of ( [ shift ] ) is the shift due to the interatomic interaction . it agrees with the results based on the local density approximation @xcite obtained by using the grand - canonical ensemble approach . the second term in ( [ shift ] ) gives exactly the usual results due to effects of the finite number of particles @xcite . thus in our approach , within the canonical ensemble the corrections due to the effects of the finite particle number and the interatomic interactions can be obtained simultaneously . below the critical temperature , the most probable value is given by @xmath180^{2/3}}t^{2}n^{-1/3 } } \ ] ] @xmath181 , } \label{below1}\ ] ] where @xmath182 . @xmath183 is defined by @xmath184 omitting the high - order terms of the parameter @xmath185 , the expression ( [ below1 ] ) gives exactly the lowest - order correction of ( [ most - hf - tra ] ) . from ( [ inter - energy ] ) we can obtain the probability distribution function of the condensate when the interaction between condensed and non - condensed atoms is considered . combining with the most probable value , one obtains @xmath126 and @xmath186 of the interacting bose gases . in fig . 4 the experimental parameter by ensher _ et al . _ @xcite is used to plot @xmath99 within the canonical ensemble . our results ( solid line ) agree well with the conclusion of ref.@xcite ( circles ) where semiclassical approximation is used in the frame of the grand - canonical ensemble . we can also obtain the numerical results for @xmath164 in the presence of the interaction between condensed and non - condensed atoms . the numerical result of @xmath164 is displayed in fig . 3 with @xmath187 ( circles ) and @xmath167 ( squares ) , respectively . our calculations show that the repulsive interaction between the condensed and non - condensed atoms lowers the condensate fluctuations further . condensate fluctuations due to collective excitations have been recently investigated by giorgini _ et al . _ @xcite within the traditional particle - number - nonconserving bogoliubov approach . in ref.@xcite the fluctuations from collective excitations are shown to follow the law @xmath188 . in this section , the bogoliubov theory will be developed based on our canonical statistics to discuss the condensate fluctuations originating from collective excitations . according to the bogoliubov theory @xcite , the total number of particles out of the condensate is given by @xmath189 the real quantities @xmath190 and @xmath191 satisfy the relations @xmath192 ^{1/2}}{2\varepsilon _ { nl}^{b } } } , \label{uv-1}\ ] ] @xmath193 where @xmath194 is the number of the collective excitations excited in the system at the thermal equilibrium @xmath195 -1}}. \label{bdistribution}\ ] ] in addition , the energy of the collective excitations entering eqs . ( [ uv-1 ] ) and ( [ uv-2 ] ) is given by the dispersion law @xcite @xmath196 these phonon - like collective excitations are in excellent agreement with the measurement of experiments . the dispersion law ( [ benergy ] ) is valid if the conditions @xmath197 and @xmath198 are satisfied . the contribution to the condensate fluctuations due to these discrete low energy modes are important because @xmath199 at low excitation energies . in eq . ( [ bthermal ] ) @xmath200 can be regarded as the effective occupation number of non - condensed atoms , while @xmath201 is the occupation number of the collective excitations . from the form of @xmath194 one can construct the partition function of the collective excitations in the frame of canonical ensemble @xmath202 . } \label{bpar1}\ ] ] from eq . ( [ bogo ] ) @xmath203 becomes @xmath204 , } \label{bpar2}\ ] ] where @xmath205 can be taken as an effective energy level of the thermal atoms . in this case @xmath203 is the partition function of a fictitious boson system , which is composed of @xmath7 non - interacting bosons whose energy level is determined by @xmath206 . from ( [ bpar2 ] ) the most probable value @xmath81 is given by @xmath207 -1}. } \label{bmost}\ ] ] it is obvious that the occupation number of low @xmath208 in eq . ( [ bmost ] ) coincides with that of eq . ( [ bthermal ] ) . other @xmath64 is determined by @xmath209 \exp \left [ -\alpha \left ( n , n_{{\bf { 0}}}\right ) \right ] -1}. } \label{bother}\ ] ] from eqs . ( [ bmost ] ) and ( [ bother ] ) we obtain @xmath210 when getting ( [ balpha ] ) we have used the approximation @xmath211 for low energy collective excitations . thus the probability distribution function of the condensate is given by @xmath212 , } \label{bgauss - dis}\ ] ] where @xmath213 is a normalization constant . therefore , the condensate fluctuations due to the collective excitations reads @xmath214 @xmath215 ^{2}. } \label{f - b}\ ] ] eqs . ( [ bgauss - dis ] ) and ( [ f - b ] ) provide the formulas for calculating the condensate fluctuations originating from the collective excitations . below the temperature @xmath216 which corresponds to the maximum fluctuations , we obtain the analytical result for the condensate fluctuations @xmath217 @xmath218 where @xmath219 is a dimensionless parameter , which is the same as that obtained in ref.@xcite . note that compared with the result obtained by ref.@xcite , the coefficient in ( [ b - below ] ) differs by a factor @xmath220 . the expression ( [ b - below ] ) shows clearly that the condensate fluctuations due to the collective excitations are anomalous , _ i.e. _ , proportional to @xmath1 . note that @xmath221 is a gaussian distribution function , the anomalous behavior of the condensate fluctuations comes from the factor @xmath222 , which is proportional to @xmath223 . at the critical temperature the probability distribution is given by @xmath224 $ ] , where @xmath225 . in this case , we obtains the analytical result of the condensate fluctuations @xmath226 from ( [ b - below ] ) and ( [ bnear - trap ] ) we find that the behavior of the condensate fluctuations based on the bogoliubov theory is rather different from that of the lowest - order perturbation theory . in this paper , a canonical ensemble approach has been developed to investigate the mean ground state occupation number and condensate fluctuations for interacting and non - interacting bose gases . different from the conventional methods , the analytical probability distribution function of the condensate has been obtained directly from the partition function of the system . based on the probability distribution function , we have calculated the thermodynamic properties of the bose gas , such as the condensate fraction and the fluctuations . through the calculations of the probability distribution function , we have provided a simple method to recover the applicability of the saddle - point method for studying the condensate fluctuations . in fact , the theory of the improved saddle - point method developed in this work can be applied straightforwardly to consider the condensate fluctuations in other physical systems , such as the interacting bose gas confined in a box @xcite , the interacting bose gas in low - dimensions , etc .. the probability distribution function can also be used to discuss other interesting problems , such as the phase diffusion of the condensate . for the harmonically trapped interacting bose gas , we found that different approximations for weakly interacting bose gases give quite different theoretical predictions concerning the condensate fluctuations . in our opinion the lowest - order perturbation theory gives in some sense the condensate fluctuations due to normal thermal atoms , while the bogoliubov theory gives the condensate fluctuations originating from the collective excitations . the contributions to the condensate fluctuations due to the collective excitations mainly come from the low energy modes , and it is obvious that the condensate fluctuations based on the lowest - order perturbation theory miss the contributions coming from the collective excitations . considering the fact that the contributions due to low energy thermal atoms in the lowest order perturbation theory is relatively small , the overall condensate fluctuations may be written as @xmath227 where @xmath163 and @xmath228 are condensate fluctuations due to the normal thermal atoms and the collective excitations , respectively . this work was supported by the science foundation of zhijiang college , zhejiang university of technology and natural science foundation of zhejiang province . g. x. huang was supported by the national natural science foundation of china , the trans - century training programme foundation for the talents and the university key teacher foundation of chinese ministry of education . s. j. liu and h. w. xiong thank professors g. s. jia and j. f. shen for their enormous encouragement . in this appendix , the method of saddle - point integration described by darwin and fowler @xcite is used to investigate the partition function of the fictitious @xmath25 non - interacting bosons . the partition function of the fictitious system is given by because of the restriction @xmath231 in the summation of eq . ( [ a - function-1 ] ) , @xmath232 can not be explicitly evaluated . to proceed we define a generating function for @xmath233 in the following manner . for any complex number @xmath234 , we take noting that eq . ( [ a - thermal ] ) is exactly the equation to determine the number of condensed atoms within the grand canonical ensemble , the saddle point @xmath28 can be also regarded as the fugacity of the fictitious @xmath25 non - interacting bosons . anderson , j.r . ensher , m.r . matthews , c.e . wieman and e.a . cornell , science * 269 * , 198 ( 1995 ) ; k.b . davis , m .- o . mewes , m.r . andrews , n.j . van druten , d.s . durfee , d.m . kurn and w. ketterle , phys . lett . * 75 * , 3969 ( 1995 ) ; c.c . bradley , c.a . sackett , j.j . tollett and r.g . hulet , phys . * 75 * , 1687 ( 1995 ) . killian , d.g . fried , l. willmann , d. landhuis , s.c . moss , t.j . greytak and d. kleppner , phys . * 81 * , 3807 ( 1998 ) ; d.g . fried , t.c . killian , l. willmann , d. landhuis , s.c . moss , d. kleppner and t.j . greytak , phys . lett . * 81 * , 3811 ( 1988 ) . a. robert , o. sirjean , a. browaeys , j. poupard , s. nowak , d. boiron , c.i . westbrook , a. aspect , science * 292 * , 461 ( 2001 ) ; f. pereira dos santos , j. leonard , j. wang , c.j . barrelet , f. perales , e. rasel , c.s . unnikrishnan , m. leduc , c. cohen - tannoudji , phy . lett . * 86 * , 3459 ( 2001 ) . s. grossmann and m. holthaus , phys . . a * 208 * , 188 ( 1995 ) ; w. ketterle and n.j . van druten , phys . a * 54 * , 656 ( 1996 ) ; k. kirsten and d.j . toms , phys . rev . a * 54 * , 4188 ( 1996 ) ; t. haugset , h. haugerud and j.o . andersen , phys . a * 55 * , 2922 ( 1997 ) .	the mean ground state occupation number and condensate fluctuations of interacting and non - interacting bose gases confined in a harmonic trap are considered by using a canonical ensemble approach . to obtain the mean ground state occupation number and the condensate fluctuations , an analytical description for the probability distribution function of the condensate is provided directly starting from the analysis of the partition function of the system . for the ideal bose gas , the probability distribution function is found to be a gaussian one for the case of the harmonic trap . for the interacting bose gas , using a unified approach the condensate fluctuations are calculated based on the lowest - order perturbation method and on bogoliubov theory . it is found that the condensate fluctuations based on the lowest - order perturbation theory follow the law @xmath0 , while the fluctuations based on bogoliubov theory behave as @xmath1 .
topological methods are increasingly used to describe observed states in condensed matter systems . prominent examples are the description of vortex textures in superfluid helium;@xcite band theory for topological insulators;@xcite topological superconductivity in a helical dirac gas@xcite and in dirac semimetals;@xcite and topological defects in liquid crystals,@xcite ferromagnets,@xciteand antiferromagnets.@xcite in this context , thin curvilinear films of ordered matter are in the focus of strongly growing interest , because in these systems a nontrivial geometry can induce topological defects in the order parameter field@xcite and can result in new effective interactions.@xcite among curvilinear films the most promising candidates for new physical effects are closed surfaces due to the natural appearance of topological invariants in the system . in this case the normalized vector field @xmath0 defined on the surface realizes a map of the surface into a sphere @xmath1 . the degree @xmath2 of this map is an integer topological invariant,@xcite i.e. each given distribution of the vector field @xmath0 on a closed oriented surface is characterized by an integer number @xmath3 which is conserved for any continuous deformation ( homotopy ) of the field @xmath0 . moreover , any two distributions of the field @xmath0 are topologically equivalent ( homotopic ) , i.e. they can be matched by means of a continuous deformation provided they have the same @xmath3.@xcite since a discontinuity in the physical field @xmath0 is usually energetically non - favorable , two solutions with different @xmath3 are separated by a high energy barrier . this causes topological stability . for example , an isolated magnetic skyrmion@xcite in a planar film with dzyaloshinskii - moriya interaction ( dmi ) is an excited state of the system ( for the case of low temperature and absence of external magnetic fields ) . however , this excitation is topologically stable , because the invariant is @xmath4 for the skyrmion,@xcite while @xmath5 for the ground state . topological stability occurs for a variety of defects in ordered matter , such as disclination loops , hedgehog point defects and knots in nematic liquid crystals;@xcite and vortices@xcite and bloch points@xcite in ferromagnets . conservation of the topological index @xmath3 for a closed surface raises two fundamental questions : ( i ) what is the lowest energy equilibrium solution @xmath6 for a given @xmath3 , and ( ii ) which @xmath3 corresponds to the ground state for a given surface ? the answers can lead to new phenomena , specific to the physical system under consideration . in this paper we answer these questions for the case of thin ferromagnetic spherical shells . even such a relatively simple model brings a number of surprising results . we show that for a spherical shell a skyrmion solution exists as a topologically stable excitation above the hedgehog ground state . an important feature is that the skyrmion may be stabilized by curvature effects only , specifically by the curvature - induced , exchange - driven _ effective _ dmi.@xcite this is in contrast to the planar case , where the intrinsic dmi is required for the skyrmion stabilization.@xcite the case of the spherical shell is topologically opposite to that of the planar film : the skyrmion has the index @xmath5 , in other words it is topologically trivial , while the ground state is characterized by @xmath4 . this is due to a shift of the topological index of the vector field , caused by topology of the surface itself . since the skyrmion solution on a spherical shell is homotopic to a uniform state , it can be induced by means of a uniform external magnetic field , similarly to the excitation of onion magnetic states in nanorings.@xcite in a continuous medium the switching between states with different @xmath3 is topologically forbidden . however , in discrete spin lattices such a transition is possible , though it requires a strong external influence . we first present a set of general results valid for an arbitrary thin curvilinear shell . in the following we apply these results to calculate the magnetic energy and topological properties of magnetization states of spherical shells . the degree @xmath3 of a map , realized by a normalized three - dimensional vector field @xmath0 defined on a two - dimensional closed surface @xmath7 , reads@xcite @xmath8 . in this particular case the mapping jacobian @xmath9 can be presented in the form of the triple product@xcite @xmath10/2 $ ] , where the minus sign is introduced solely to conform with the traditional notation used in ferromagnetic research . here and everywhere below the greek indices @xmath11 numerate the curvilinear coordinates @xmath12 , introduced on the surface , and the vector components defined in the corresponding curvilinear local basis @xmath13 ; while the latin indices @xmath14 numerate coordinates and vector components in the cartesian basis @xmath15 . the summation over repeated dummy indices is implied , unless stated otherwise . the local basis @xmath13 is assumed to be orthonormal @xmath16 , therefore the metric tensor @xmath17 is diagonal . details on the definition of the orthonormal basis for a given surface are presented in the appendix [ app : basis ] . the operator @xmath18 , where the summation over @xmath19 is not implied and @xmath20 , denotes the corresponding component of the surface del operator @xmath21 . the surface element reads @xmath22 , where @xmath23 . since using cartesian components of the vector field @xmath0 is not convenient for curvilinear systems , we will switch to curvilinear coordinates @xmath24 , where @xmath25 is the surface normal . moreover , it is useful to incorporate the constraint @xmath26 by means of the angular parameterization @xmath27 , @xmath28 , where @xmath29 and @xmath30 represent colatitude and longitude , the spherical angles of the local curvilinear basis , respectively . in this case one can show ( see appendix [ app : gyrovector ] ) that @xmath31 . \end{split}\ ] ] here , @xmath32 , where @xmath33 is the normalized projection of the vector @xmath0 on the tangential plane and @xmath34 is a tensor known as the weingarten map or modified second fundamental form.@xcite vector @xmath35 denotes the spin connection and @xmath36 is the gau curvature . the corresponding definitions are presented in the appendix [ app : basis ] . one can easily check that for the case of a plane with a cartesian frame of reference the expression results in the well known@xcite formula @xmath37 . remarkably , for a strictly normal distribution of the vector field @xmath38 ( normal gau map ) one obtains the well known@xcite result @xmath39 . applying the gau - bonnet theorem we obtain the famous relation @xmath40 between degree of the normal gau map @xmath41 and genus @xmath42 of the surface . thus , @xmath43 for a normally magnetized sphere ( hedgehog ) , @xmath44 for a normally magnetized torus , etc . the vector @xmath45 is the limit for the two - dimensional case for the gyrocoupling vector@xcite(topological density , topological current , vorticity ) @xmath46 , whose cartesian components read @xmath47/2 $ ] . the gyrocoupling vector is widely used for the topological description of a unit vector field @xmath0 defined in a three - dimensional domain . if the shell thickness @xmath48 is small enough to ensure the uniformity of @xmath0 along the normal direction : @xmath49 , then @xmath50 , see appendix [ app : gyrovector ] . in magnetism , the gyrocoupling vector @xmath46 is the key quantity to describe the dynamics of topologically nontrivial solutions , such as domain walls,@xcite vortices,@xcite skyrmions,@xcite skyrmion lines@xcite and bloch points.@xcite it determines important integrals of motion in the dynamics of ferromagnetic media.@xcite recently it was shown@xcite that @xmath46 is proportional to the emergent magnetic field , which appears due to the hund s coupling between spins of the conducting electrons and localized magnetic moments . this gives rise to the topological hall effect.@xcite let us provide physically illustrative explanations why the topological charge or index @xmath3 is an integer number and a conserved quantity . a direct consequence of the definition of @xmath46 with the constraint @xmath26 is @xmath51 , where @xmath52 is the dirac delta - function and the vector @xmath53 determines the position of a bloch point ( monopole ) , @xmath54 , whose infinitesimal neighborhood of the center has the structure@xcite @xmath55 . here , @xmath56 is an arbitrary matrix of three - dimensional rotations and @xmath57 is the monopole charge . thus , the monopoles are sources and sinks of the gyrovector field.@xcite likewise , electrical charges are sources and sinks of the electrical field . for any closed surface @xmath7 enclosing the volume @xmath58 , the integral @xmath59 yields an integer number @xmath2 equal to the difference of negatively and positively charged monopoles inside @xmath7 . since two monopoles with opposite charges are connected by the dirac string , which may be considered a skyrmion line,@xcite one can also say that @xmath3 is the difference of outgoing and incoming skyrmion lines.@xcite thus the only way to change @xmath3 for a given closed @xmath7 is to replace a monopole across @xmath7 . when the monopole center crosses the surface , i.e. it is located exactly on @xmath7 , the vector field @xmath60 is discontinuous at the point @xmath53 . thus , one can conclude that _ a continuous deformation of the continuous field @xmath61 can not change the mapping degree @xmath3 of @xmath7_. the rigorous proof of the latter statement can be found elsewhere , for instance see ref . . the expression for the mapping jacobian is general : it is the key formula for topological analysis of a normalized vector field of an arbitrary physical nature on an arbitrary curvilinear surface . the topological analysis is independent of the physical nature of the vector field @xmath0 . in the following , we focus on possible equilibrium magnetization states of thin ferromagnetic curvilinear shells . to this end , we introduce the energy functional @xmath62\mathrm{d}\mathcal{s}$ ] . here , we take into account three magnetic interactions . the first term of the integrand represents the exchange energy with the energy density @xmath63 and the exchange constant , @xmath64 . the second term is a uniaxial anisotropy : easy - normal for @xmath65 or easy - surface for @xmath66 . the last term in @xmath67 is the dmi with the energy density@xcite @xmath68 and the dmi constant , @xmath69 . this kind of dmi originates from the spin - orbit coupling and is related to the inversion symmetry breaking on the film interface ; it is typical for ultrathin films@xcite or bilayers.@xcite in the curvilinear basis one can represent the dmi density as follows @xmath70 where @xmath71 is the mean curvature , see appendix [ app : interactions ] for details . it is clear from , that an effective uniaxial anisotropy along @xmath72 appears with a coefficient equal to @xmath73 . as the simplest example we consider a thin spherical shell with radius @xmath74 . for the case of easy - normal anisotropy ( @xmath65 ) there exists a class of azimuthally symmetric solutions @xmath75 , see appendix [ app : spherical_shell ] . the basis vector @xmath76 points , tangential to the surface , towards the direction of increasing polar angle @xmath77 and @xmath72 is the outward normal . the function @xmath78 satisfies the following equation @xmath79\\ + & 2\cot\vartheta\sin^2\theta\left(1+\frac{d}{d_c}\right)=0 . \end{split}\ ] ] here , @xmath80 is the characteristic magnetic length and @xmath81 is the strength of the curvature - induced effective dmi that is solely exchange - driven.@xcite this geometrical dmi contribution competes with the intrinsic spin - orbit - driven dmi . full compensation takes place when @xmath82 . there are two kinds of boundary conditions ( bc ) possible for eq . , namely ( i ) @xmath83 , @xmath84 , and ( ii ) @xmath85 , @xmath86 . here , @xmath87 , the helicity number , is formally a winding number of the magnetization along a circle loop passing through both pole points @xmath88 and @xmath89 . using the helicity number one can introduce the chirality of the structure : @xmath90 . thus , the skyrmions shown in fig . [ fig : q0]a and a@xmath91 have the chiralities @xmath92 and @xmath93 , respectively . from the general expression for the gyrocoupling vector , it follows that the mapping index for an azimuthally symmetrical solution @xmath94 is @xmath95\ ] ] which implies that @xmath96 for the mentioned class of solutions . it is interesting to note that a one - dimensional magnetization in the planar case , @xmath97 , results in @xmath5 . however , in the case of a spherical shell a solution with @xmath98 is possible even if @xmath0 depends on one coordinate only . according to an even @xmath99 results in @xmath5 for both kinds of bc and an odd @xmath99 results in @xmath100 and @xmath101 for bc of type ( i ) and ( ii ) , respectively . note that , in contrast to the mapping degree @xmath3 , _ the helicity number @xmath99 is not a topological invariant _ : we merely use it for the classification of solutions . any two solutions with different @xmath99 but with the same @xmath3 belong to the same homotopy class and they can be transformed into each other by means of a continuous deformation of the vector field @xmath0 . any two solutions of , obtained under different kinds of bc but for the same @xmath99 , differ by sign only : @xmath102 . the latter transformation does not change the energy of the system , as the energy functional @xmath67 , in the absence of external fields , is quadratic with respect to components of vector @xmath0 . however , it changes the sign of @xmath3 because the mapping jacobian @xmath103 is cubic in the magnetization . thus a state with given @xmath99 is doubly degenerate with respect to replacements @xmath102 and @xmath104 . equation can have skyrmion solutions for the cases @xmath105 and @xmath106 , see fig . [ fig : q0]a and fig . [ fig : q0]a@xmath91 , respectively . in contrast to the planar case , where the skyrmion solution has @xmath98 , on a spherical shell the skyrmion is topologically trivial ( @xmath5 ) . let us define the skyrmion radius as @xmath107 , where @xmath108 . for the planar case @xmath109 with @xmath110 being distance to the skyrmion center and @xmath111 can be defined analogously : @xmath112 . in planar films skyrmions are widely studied ; it is well known@xcite that the skyrmion radius strongly depends on the dmi constant @xmath69 : the skyrmion collapses , @xmath113 , when @xmath114 ; and @xmath115 when @xmath116 , see dashed line in fig . [ fig : q0 ] . for this type of dmi , the so called hedgehog ( nel ) skyrmions appear with zero azimuthal magnetization component , see fig . [ fig : q0]b , b@xmath91 . in planar films such type of skyrmions have been predicted theoretically@xcite and were observed experimentally.@xcite the same type of skyrmions appear on a spherical shell with an analogous dependence @xmath117 , see the solid line in fig . [ fig : q0 ] . there are , however , a number of new , important features : \(i ) skyrmions collapse for a finite value of the dmi constant , @xmath82 , and as a consequence a skyrmion of finite radius exists for the case @xmath118 , see point a in fig . [ fig : q0 ] and the corresponding inset . the shift along the @xmath69 axis is due to the additional curvature - induced dmi , which appears as an effective term in the exchange interaction.@xcite \(ii ) for a given radius @xmath74 of the spherical shell the skyrmion exists for a certain range of the dmi constant , @xmath119 , see fig . [ fig : q0 ] and fig . [ fig : gr - states ] . beyond this range , at @xmath120 , see fig . [ fig : q0 ] , the skyrmion transforms into the 3d - onion state . \(iii ) in contrast to the planar case where the function @xmath121 is even , the corresponding curve for spherical shells is highly asymmetrical . [ fig : q0 ] shows possible equilibrium states for the case @xmath5 , answering our introductory question ( i ) about the physically stable magnetization structures in topologically different sectors . though a continuous transition between solutions with @xmath5 and @xmath98 is not possible ( topological stability ) , a solution with @xmath98 can have lower energy than the corresponding solution with @xmath5 for some range of parameters . in order to clarify this picture and answer question ( ii ) about the globally stable magnetization configurations , we build the diagram of the ground states for the class of azimuthally symmetrical solutions determined by eq . , see fig . [ fig : gr - states ] . one can distinguish two main states : the hedgehog states with @xmath122 ( @xmath98 ) and 3d - onion state with @xmath105 ( @xmath5 ) . however , for large enough sphere radii and magnitudes of the dmi constant a variety of states with higher helicity numbers appear . these states can be interpreted as helical structures on a spherical shell . similar skyrmionic structures were recently observed in disk - shaped chiral nanomagnets.@xcite like the solitary skyrmion on a planar film , the skyrmion on a spherical shell does not form the magnetic ground state , yet skyrmions with @xmath105 and @xmath106 can exist as topologically stable excitations , see the dashed areas @xmath123 and @xmath124 , respectively . . inset ( a ) shows the dependence of the energy difference @xmath125 of the solution @xmath126 for the case @xmath105 and the hedgehog solution @xmath127 , when @xmath122 . the critical radius @xmath128 separates two phases of the ground state : 3d - onion state for @xmath129 and hedgehog state for @xmath130 , see fig . [ fig : gr - states ] . for @xmath131 the 3d - onion state experiences instability , which results in the skyrmion formation . skyrmion exists as a topologically stable excitation of the hedgehog state for the case @xmath132 , see also region @xmath123 in fig . [ fig : gr - states ] . inset ( b ) demonstrates the dependence of the skyrmion radius ( angular @xmath133 as well as lateral @xmath111 ) on the sphere radius @xmath74 . symbols correspond to the results of micromagnetic simulations : disks and squares the magnetostatic interaction is reduced to the easy - surface anisotropy , triangles the full scale simulations with magnetostatics is included . ] in order to describe a spherical shell of radius @xmath74 we use the parameterization @xmath192 . here , @xmath193 $ ] and @xmath194 are polar and azimuthal spherical angles , respectively . basis vectors are @xmath195 , @xmath196 and the normal vector @xmath197 is directed outward of the sphere . in this case the weingarten map is the diagonal matrix with components @xmath198 , and consequently @xmath199 , @xmath200 and @xmath201 . the spin - connection vector has only one ( azimuthal ) component : @xmath202 . taking into account and one can show that in the case of easy - normal anisotropy ( @xmath65 ) the functions @xmath203 and @xmath204 , which minimize the energy functional @xmath67 , are a solution of the following euler - lagrange equations [ eq : theta - phi - sphere ] @xmath205\\ \nonumber & + \frac{2}{r}\sin^2\theta\left(1+\frac{d}{d_c}\right)({\bm{\nabla}}\phi-{\bm{\omega}})\cdot\partial_\phi{\bm{\varepsilon}}=0 , \\ \label{eq : phi - sphere}&{\bm{\nabla}}\cdot\left[\sin^2\theta({\bm{\nabla}}\phi-{\bm{\omega}})\right]-\frac{2}{r}\sin^2\theta\left(1+\frac{d}{d_c}\right){\bm{\nabla}}\theta\cdot\partial_\phi{\bm{\varepsilon}}=0 . \end{aligned}\ ] ] when deriving we used that for a spherical shell @xmath199 and @xmath200 . in addition , we use the following general properties of vector @xmath206 @xmath207 taking into account that the vector of spin connection is @xmath208 one can see that eqs . have a solution @xmath78 , @xmath209 . in this case . turns to identity and the function @xmath126 can be determined as a solution of eq . . since @xmath210 for the considered class of solutions , one can consider the colatitude angle @xmath78 as the only parameter : @xmath211 . in this case the function @xmath126 is determined by equation . for the axially symmetric solutions @xmath78 and @xmath212 the gyrocoupling vector for the spherical shell can be written as follows @xmath213 the integration over the sphere @xmath1 results in the topological index . component of @xmath134 in an external magnetic field along the @xmath135 axis . the arrows represent the increase ( red / dashed ) and subsequent decrease ( blue / solid ) of the magnetic field . the field is normalized by the saturation field @xmath136 . insets , which show the vertical cross - section of the spherical shell , trace the formation of a skyrmion . ] the results on static skyrmion state configurations on spherical shells immediately pose the problem whether and how these states can be realized . in the following , we discuss in more detail an intriguing case of the skyrmion formation , when @xmath118 . for this purpose , we will move along the vertical axis of the ground states diagram ( fig . [ fig : gr - states ] ) starting from small sphere radii @xmath74 . when the sphere s radius is small enough ( @xmath137 ) , the ground state of the system is topologically trivial ( @xmath5 ) and close to the uniform state , we call it a 3d - onion state , see fig . [ fig : gr - states ] and fig . [ fig : bifurcation]a . the ground state of the spheres with @xmath130 is one of the hedgehog states with @xmath4 . due to the topological stability the 3d - onion state survives when the sphere radii @xmath138 . at the point @xmath131 the 3d - onion state becomes unstable resulting in a fork - like bifurcation , see fig . [ fig : bifurcation]b , where the lines are obtained by solving eq . [ eq : theta ] with the boundary conditions @xmath83 , @xmath139 and dots correspond to the micromagnetic simulations . as a result of the bifurcation a skyrmion is formed either on the north or on the south pole of the sphere , see fig . [ fig : bifurcation ] . the skyrmion exists as a topologically stable excitation of a hedgehog state for the case @xmath132 , see region @xmath123 in figs . [ fig : gr - states ] , [ fig : bifurcation]a . however the radius of the skyrmion @xmath111 decreases rapidly when the radius of the sphere further increases , see fig . [ fig : bifurcation]b . the observed behavior of this system can be explained as follows . the 3d - onion state can be interpreted as a skyrmion solution with the angular radius @xmath140 , or with lateral radius @xmath120 . however , it is well known@xcite that skyrmions on a planar film collapse when the intrinsic spin - orbit - driven dmi vanishes , i.e. , in the limit @xmath118 . thus , it is natural to expect the skyrmion to collapse for the case @xmath141 when the curvature effect vanishes . that is why the instability of the 3d - onion state appears for a certain value of @xmath131 and with a further increase of the radius of the sphere the skyrmion collapses to either the north or the south pole , see fig . [ fig : bifurcation]b . however , for a sphere radius @xmath142 skyrmions with finite radii exist , see fig . [ fig : bifurcation]b . this can be interpreted as a skyrmion stabilization due to the curvature - induced , exchange - driven effective dmi.@xcite the obtained results agree very well with micromagnetic simulations based on a model where the magnetostatic interaction was reduced to the easy - surface anisotropy , see disk- and square - shaped symbols in fig . [ fig : bifurcation]b . it is important that taking into account the magnetostatic interaction does not change the physical picture , but results in an increase of the skyrmion radius , see triangle - shaped markers in fig . [ fig : bifurcation]b . the increase of the skyrmion radius appears due to the volume magnetostatic charges . thus , the full micromagnetic simulation results vindicate our analytical approach that neglect the dipolar stray - fields to construct the diagrams of equilibrium states . it is noteworthy that the energy difference @xmath125 of the skyrmion ( 3d - onion ) and the hedgehog solutions varies in the region @xmath143 , where @xmath144 is the energy of the hedgehog state sphere with @xmath145 , and @xmath146 is the energy of the belavin - polyakov soliton,@xcite which is equal to the energy of a skyrmion with infinitesimal radius , see fig . [ fig : bifurcation]a . it is also remarkable that @xmath147 , which matches the energy of the vortex - antivortex pair with the opposite polarities just before their annihilation.@xcite finally , we demonstrate how the skyrmion configuration can be created by means of a uniform magnetic field . we consider a spherical shell , whose radius @xmath142 corresponds to the hedgehog ground state . by means of micromagnetic simulations ( see appendix [ app : simuls ] ) we find that the adiabatically slow increase of the external uniform magnetic field results in the transition from the hedgehog state to the 3d - onion state , see fig . [ fig : hyst ] . a subsequent decrease of the field leads to a skyrmion state formation . this transition from @xmath148 to @xmath5 is topologically forbidden for a continuous system and appears here merely due to the discretization . however , since real magnetic crystals have a discrete structure , one can expect this behavior in strong enough external magnetic fields . this mechanism of the skyrmion formation is similar to the formation of onion magnetic states in nanorings.@xcite in conclusion , we demonstrate that different types of axially symmetrical solutions of the magnetization @xmath94 exist for a thin ferromagnetic spherical shell . these solutions can be divided into three homotopic classes with topological index @xmath149 . to calculate @xmath3 we developed the general expression for the mapping jacobian valid for an arbitrary curvilinear shell . skyrmion solutions are found in the topologically trivial class with @xmath5 . remarkably , a skyrmion solution on a spherical shell can be stabilized by curvature effects only , namely by the curvature - induced , exchange - driven effective dmi.@xcite this is in contrast to the planar case , where the spin - orbit - driven intrinsic dmi is required for the skyrmion stabilization.@xcite since a skyrmion on a spherical shell is homotopic to a uniformly magnetized sphere , it can be induced by a strong uniform external magnetic field . v.p.k . acknowledges the alexander von humboldt foundation for the support and ifw dresden for kind hospitality . this work was funded in part by the european research council under the european union s seventh framework programme ( fp7/2007 - 2013)/erc grant agreement no . 306277 and the european union future and emerging technologies programme ( fet - open grant no . d.d.s . acknowledges prof . avadh saxena for fruitful discussions . in order to formalize the geometry of the shell we use the parametric representation @xmath150 . here , @xmath151 is the 3d position vector , which determines a 2d surface @xmath7 embedded in @xmath152 with @xmath12 being local curvilinear coordinates on @xmath7 . the unit vector @xmath72 denotes the surface normal and the parameter @xmath153 $ ] is the corresponding curvilinear coordinate along the normal direction . we restrict ourselves to the limiting case @xmath48 . specifically , we assume that the thickness @xmath154 is much smaller than the curvature radius as well as the characteristic magnetic length @xmath155 . as a consequence , we assume that the magnetization is uniform along the normal direction : @xmath49 . the parameterization @xmath156 induces the natural tangential basis @xmath157 with the corresponding metric tensor @xmath158 . assuming that vectors @xmath159 are orthogonal , one can introduce the orthonormal basis @xmath160 , where @xmath161 and @xmath25 . using the gau - codazzi formula and weingarten s equations@xcite one can obtain the following differential properties of the basis vectors @xmath162 here , matrix @xmath163 is a tensor , known as the weingarten map or modified second fundamental form.@xcite the formula @xmath164 with @xmath165 being components of the second fundamental form , is practically useful . the weingarten map determines the gau curvature @xmath36 and the mean curvature @xmath71 . components of the spin connection vector @xmath35 are determined by the relation @xmath166 . in general , the basis @xmath159 need not be orthogonal . however , if the vectors @xmath167 and @xmath168 are not collinear then one can always introduce the orthonormal basis @xmath169 in the following way : @xmath170 , @xmath171 , and @xmath172 , where @xmath173 is dual tangential basis with @xmath174 . however , in this case the relations should be revised . we start from the cartesian representation @xmath47/2 $ ] . taking into account that the magnetization is uniform along the shell thickness , i.e. the normal coordinate , one can replace the del operator by its surface analogue @xmath175 and represent the cartesian derivatives via its curvilinear counterparts @xmath176 . simple calculations result in @xmath177 , where @xmath9 is the mapping jacobian defined in the main text . introducing the curvilinear magnetization components @xmath24 and taking into account differential properties of the curvilinear basis one obtains @xmath178\\ & + \epsilon_{\alpha\beta}\omega_\beta m_\gamma\left[m_n\nabla_{\!\!\alpha } m_\gamma - m_\gamma\nabla_{\!\!\alpha } m_n - h_{\alpha\gamma}\right]-\mathcal{k}m_n . \end{split}\ ] ] when deriving the constraint @xmath26 is obeyed . substituting the angular representation for the magnetization components into , one obtains formula . notice the difference between volume @xmath46 and surface @xmath45 gyrocoupling vectors . @xmath46 can be introduced for a vector field @xmath0 defined in a 3d region @xmath179 . one can cut out from @xmath179 a curvilinear shell with small but finite thickness @xmath154 . if @xmath0 is _ uniform along the shell thickness _ ( otherwise @xmath45 can not be introduced ) , then @xmath180 . however , if we build a 2d surface @xmath7 ( @xmath181 ) in @xmath179 , then @xmath182 on the surface . let us first consider the exchange interaction . under the main assumption of magnetization uniformity along the shell thickness , one can represent the exchange energy density in its curvilinear form , [ eq : eex - detail ] @xmath183 here , @xmath184 is the `` common '' isotropic exchange , @xmath185 and @xmath186 can be treated as an effective curvature - induced dmi and anisotropy , respectively . in the angular representation the exchange energy reads@xcite @xmath187 ^ 2+\left[\sin\theta({\bm{\nabla}}\phi-{\bm{\omega}})-\cos\theta\,\partial_\phi{\bm{\gamma}}\right]^2.\ ] ] applying the same procedure for the dmi energy density @xmath188 one obtains [ eq : ed - detail ] @xmath189 where an additional term @xmath190 can be interpreted as an effective curvature - induced anisotropy . substituting the angular parameterization into results in the expression . in the angular representation the density of the anisotropy energy looks particularly simple @xmath191 , because the anisotropy has the symmetry of the surface . in order to verify our analytical results we performed micromagnetic simulations with the ` finmag ` code , which is the successor to the ` nmag ` tool.@xcite we used the material parameters of cobalt : @xmath214 j / m , @xmath215 a / m and @xmath216 j / m@xmath217 , which are typical for pt / co / alo@xmath218 layer structures.@xcite for comparison , we also performed a simulation neglecting the magnetostatic interaction , assuming that it can be reduced to an effective easy - surface anisotropy , we used @xmath219 j / m@xmath217 . these parameters correspond to @xmath220 nm . in all simulations the ratio @xmath221 is kept constant . the size of the discretization mesh is @xmath222 . to simulate the formation of skyrmions by means of a uniform field , see fig . [ fig : hyst ] , we consider a spherical shell with radius @xmath223 nm and thickness @xmath224 nm . the applied magnetic field is increased from zero up to the value 1.5 t with a rate of 330 mt / ns . the field is then decreased back to zero with rate 82 mt / ns . in order to break the symmetry and avoid the unstable equilibrium state we introduce a radial @xmath225 field with a small amplitude @xmath226 mt . fields in fig . [ fig : hyst ] are normalized to the value @xmath227 t. in this numerical experiment we reduce magnetostatics to the effective easy - surface anisotropy . 93 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1 '' '' @noop [ 0]secondoftwo sanitize@url [ 0 ] + 12$12 & 12#1212_12%12 @startlink[1 ] @endlink[0 ] @bib@innerbibempty link:\doibase 10.1103/physrevlett.38.508 [ * * , ( ) ] http://www.zentralblatt-math.org/zmath/search/?an=01866310[__ ] ( , , ) link:\doibase 10.1103/revmodphys.82.3045 [ * * , ( ) ] link:\doibase 10.1038/nature08916 [ * * , ( ) ] link:\doibase 10.1038/nature06843 [ * * , ( ) ] link:\doibase 10.1038/nphys3139 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.187001 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.84.497 [ * * , ( ) ] http://www.informaworld.com/10.1080/14786430600593016 [ * * , ( ) ] http://link.aps.org/abstract/prl/v30/p230 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) http://www.sciencedirect.com/science/article/b6tvc-470f3hy-36/2/69a5e1a128fb5b7ef2ee0c512c3d78fc [ * * , ( ) ] http://www.sciencedirect.com/science/article/b6tvk-3spcrw6-6/2/f95e4629d5db97bbd9d8bdcbdbd476f3 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.116.147203 [ * * , ( ) ] link:\doibase 10.1080/00018730903043166 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.93.215301 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.82.1301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.207803 [ * * , ( ) ] link:\doibase 10.1039/c3sm50605c [ * * , ( ) ] link:\doibase 10.1103/physrevlett.112.257203 [ * * , ( ) ] \doibase http://dx.doi.org/10.1088/1751-8113/48/12/125202 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.51.591 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ , gtm093 ( , ) http://www.sciencedirect.com/science/article/b6tvp-46sxp0p-19/2/d0a99bbbccf078602123db3e5d601202 [ * * , ( ) ] @noop _ _ , cambridge monographs on mathematical physics ( , , ) @noop * * , ( ) link:\doibase 10.1016/0304 - 8853(94)90046 - 9 [ * * , ( ) ] link:\doibase 10.1016/s0304 - 8853(98)01038 - 5 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.87.037203 [ * * , ( ) ] link:\doibase 10.1126/science.1240573 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.114.177203 [ * * , ( ) ] link:\doibase 10.1038/nphys3234 [ * * , ( ) ] link:\doibase 10.1088/1367 - 2630/18/6/065003 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.064412 [ * * ( ) ] link:\doibase 10.1073/pnas.1308225110 [ * * , ( ) ] link:\doibase 10.1103/physreve.84.031703 [ * * , ( ) ] link:\doibase 10.1038/nature11710 [ * * , ( ) ] in @noop _ _ , ( , , ) pp . @noop * * , ( ) link:\doibase 10.1038/nnano.2013.210 [ * * , ( ) ] link:\doibase 10.1103/physrevb.88.184422 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.1098 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.74.953 [ * * , ( ) ] link:\doibase 10.1063/1.3466747 [ * * , ( ) ] link:\doibase 10.1063/1.3643037 [ * * , ( ) ] link:\doibase 10.1103/physrevb.52.6712 [ * * , ( ) ] http://www.sciencedirect.com/science/article/b6tvm-4ndmn8t-6/1/5884027a6e920c09249e2e818d0d3e71 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.144433 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.99.117202 [ * * , ( ) ] link:\doibase 10.1063/1.330661 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.207202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.214419 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.180401 [ * * , ( ) ] link:\doibase 10.1126/science.1234657 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.060401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.224401 [ * * , ( ) ] link:\doibase 10.1038/nphys2231 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.186602 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.117202 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.106.156603 [ * * ( ) ] link:\doibase 10.1016/j.physleta.2012.10.027 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0375960114010573 [ ( ) ] link:\doibase 10.1016/j.jmmm.2015.04.078 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.134450 [ * * , ( ) ] link:\doibase 10.1016/j.physleta.2013.03.028 [ * * , ( ) ] link:\doibase 10.1016/s0304 - 8853(97)01044 - 5 [ * * , ( ) ] link:\doibase 10.1209/0295 - 5075/100/57002 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.267210 [ * * , ( ) ] http://www.journals.royalsoc.ac.uk/openurl.asp?genre=article&id=doi:10.1098/rspa.1997.0013 [ * * , ( ) ] http://dx.doi.org/10.1098/rspa.2004.1342 [ * * , ( ) ] http://dx.doi.org/10.1007/s00205-005-0372-7 [ * * , ( ) ] link:\doibase 10.1142/s021820250500087x [ * * , ( ) ] http://stacks.iop.org/0022-3727/44/i=39/a=392001 [ * * , ( ) ] link:\doibase 10.1038/srep08787 [ * * , ( ) ] link:\doibase 10.1038/srep17137 [ * * , ( ) ] link:\doibase 10.1002/pssb.2221860223 [ * * , ( ) ] http://link.aps.org/abstract/prb/v75/e012408 [ * * , ( ) ] http://link.aps.org/abstract/prl/v97/e177202 [ * * , ( ) ] link:\doibase 10.1088/0022 - 3727/49/36/363001 [ * * , ( ) ] link:\doibase 10.1021/cm902114d [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.094419 [ * * ( ) , 10.1103/physrevb.79.094419]link:\doibase 10.1002/smll.201401049 [ * * , ( ) ] link:\doibase 10.1038/nmat1324 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.96.077202 [ * * , ( ) ] link:\doibase 10.1063/1.2993334 [ * * , ( ) ] link:\doibase 10.1063/1.3260243 [ * * , ( ) ] @noop _ _ , ed . , , vol . ( , ) link:\doibase 10.1109/tmag.2007.893843 [ * * , ( ) ]	topologically stable structures include vortices in a wide variety of matter , such as skyrmions in ferro- and antiferromagnets , and hedgehog point defects in liquid crystals and ferromagnets . these are characterized by integer - valued topological quantum numbers . in this context , closed surfaces are a prominent subject of study as they form a link between fundamental mathematical theorems and real physical systems . here we perform an analysis on the topology and stability of equilibrium magnetization states for a thin spherical shell with easy - axis anisotropy in normal directions . skyrmion solutions are found for a range of parameters . these magnetic skyrmions on a spherical shell have two distinct differences compared to their planar counterpart : ( i ) they are topologically trivial , and ( ii ) can be stabilized by curvature effects , even when dzyaloshinskii - moriya interactions are absent . due to its specific topological nature a skyrmion on a spherical shell can be simply induced by a uniform external magnetic field .
the calculation of accurate collision strengths for atomic transitions has been a long standing problem in the field of quantitative spectroscopy . any calculation involving atoms in non - lte conditions requires the knowledge of vast numbers of collision strengths in order to make these calculations realistic and accurate . until recently the computing power was simply not available to calculate collision strengths in a systematic way . one either had to resort to simpler and more approximate methods or one had to limit the calculations to only the most important transitions . this situation has now changed with the start of the iron project ( hummer et al . 1993 ) , which aims to produce a large database of accurately calculated collision strengths . the collision strength for an atomic transition depends + strongly on the energy of the colliding electron and shows many resonances ( e.g. , aggarwal 1984 ) . such resonances occur when the total energy of the target ion and the colliding electron correspond to an auto - ionizing state . in order to calculate these resonances accurately , a fine grid of energy points is necessary . this is a type of problem for which r - matrix methods are very well suited . however , a source of uncertainty in these calculations is that the energies of most auto - ionizing states have not been measured in the laboratory and therefore need to be derived from calculations . it is a well known fact that the resulting energies are not very accurate and hence the positions of the resonances are also uncertain . since the collision strengths are usually folded with a maxwellian energy distribution , this is not a major problem for high temperature ( i.e. , x - ray ) plasmas where the distribution is much broader than the uncertainty in the position of the resonances . however , for low temperature ( e.g. , photo - ionized ) plasmas this can lead to problems if a resonance is present near the threshold energy for the transition . if only the high - energy tail of the maxwellian distribution is capable of inducing a collisional transition , then a small shift in the position of a near - threshold resonance can have a severe impact on the effective collision strength . this effect would be even more pronounced if the resonance shifts below the transition threshold and disappears completely . .comparison of various determinations of the effective collision strengths for transitions within the ground term of . the values are for @xmath1 = 10000 k. [ cols="<,>,>,>",options="header " , ] this is the nebula studied by o96 . the [ nev ] temperature and density for this nebula were already derived from the sws06 spectrum by pottasch & beintema ( 1999 ) . in our analysis we will include the sws014 spectrum as well . the expected values for the electron temperature and density are those of o96 . our analysis gives a result which deviates slightly more than 1 @xmath2 from these values . the preferred temperature of o96 is based on various determinations using ions with lower ionization potentials than . in view of the discussion in the previous section concerning temperature stratification , this estimate is probably too low . especially in view of the temperatures derived by o96 from rather low excitation line ratios like [ siii ] , [ ariii ] and [ oiii ] which range between 18100 k and 19400 k , the [ nev ] temperature may be expected to be considerably higher than 19000 k. a temperature of 22000 k is more realistic ( see the discussion in pottasch & beintema 1999 ) . this value for the temperature is indicated by an arrow in fig . [ nev : fig ] . after this correction the discrepancy is slightly less than 1 @xmath2 . this is the nebula studied by c87 . the intensity for the [ nev ] 342.6 nm line is in doubt . c87 quote @xmath3 , but it is not clear how this value was obtained . we decided to use the value quoted in aller & faulkner ( @xcite ) instead . from the discussion in that article it is not clear whether the data were corrected for interstellar extinction . the intensities they quote for other strong blue emission lines compare well with the dereddened intensities given by c87 and we therefore assume that the aller & faulkner ( @xcite ) data are corrected for interstellar extinction . they give @xmath4 . the correction for the blend with the oiii 342.9 nm and 344.4 nm bowen resonance - fluorescence lines is easy , since the oiii 313.3 nm line has been measured by c87 . the oiii 313.3 nm , 342.9 nm and 344.4 nm lines all originate from the same upper level ( @xmath5 @xmath6 ) and the intensity ratio of the lines is simply given by the ratio of the transition probabilities times the photon energy . using opacity project data ( luo et al.1989 ) one finds @xmath7 : @xmath8 : @xmath9 = 10.94 : 1.00 : 2.94 . c87 gives @xmath10 . hence @xmath11 and @xmath12 . this result is substantially lower than the value used by c87 . we were not able to correct the _ sws _ spectrum accurately for aperture effects and therefore preferred to use the _ lrs _ flux for the [ nev ] 14.32 @xmath0 m line . to complete the data set , we assumed a value for the 24.32 @xmath0 m flux such that the resulting density agreed with the expected value . none of the flux values we adopted for this nebula can be considered accurate and re - measurement is warranted . the expected values for the electron temperature and density were determined by averaging the data in table 12 of c87 . for the temperature we only used the values derived from the [ arv ] , [ neiv ] and [ nev ] line ratios , for the density we used all values except those derived from [ mgi ] , [ niv ] and [ oiv ] lines . one can see that there is a slightly more than 2 @xmath2 discrepancy for the electron temperature . again the expected value for the electron temperature may be underestimated due to temperature stratification . we think 15000 k is in all probability a more realistic , though still conservative , estimate . this value for the temperature is indicated by an arrow in fig . [ nev : fig ] . after this correction there is a 1.5 @xmath2 discrepancy . a major advance in atomic theory in the past decade has been close coupling calculations that include resonances . these new calculations ( carried out with an r - matrix code ) can raise the collision strength by an order of magnitude or more compared to older calculations . these large differences have led to a discussion of the validity of these calculations . nebulae offer powerful laboratories for verifying atomic processes , and two studies ( c87 and o96 ) used this approach to test the r - matrix calculations for . they found that spectra of planetary nebulae did not agree with the r - matrix results . this casted doubt on the validity of the close coupling calculations . in this paper we have redone the analysis carried out by c87 and o96 using newer , more accurate data . we also included the well studied nebula ngc 7027 in the analysis to obtain a larger sample . we found that the expected values of the electron temperature and density all were within 3 @xmath2 of our results . hence there is no proof for significant problems with the r - matrix calculations . on closer inspection one sees that the largest discrepancy is for ngc 7027 , the best studied nebula . also the electron temperature derived from our analysis is systematically higher than the expectation value for all nebulae . this could point to inaccuracies in the collision strengths . to check this point further , we have re - analyzed our sample using effective collision strengths which were lowered by 30 % for transitions within the @xmath13 ground term . the results are shown in the bottom row of fig . [ nev : fig ] . they are in good agreement with the expected values . hence the r - matrix calculations for could be off by 30 % , but certainly not more . we point out that this is significantly less than the factor @xmath142.7 suggested by o96 . an alternative explanation could be that the [ nev ] 342.6 nm fluxes are systematically overestimated due to a problem with the extinction curve , or a combination of the two effects . 1 . the discrepancies found by c87 can be explained by the inaccuracy of the [ nev ] 342.6 nm flux they adopted . the discrepancies found by o96 mainly stem from the inaccuracy of the _ lrs _ measurement of the [ nev ] 14.32 @xmath0 m line . 2 . based on the data presented in this paper there is _ no reason _ to assume that there are any problems with the collision strengths for calculated by lennon & burke ( 1994 ) . our analysis has shown that the data are accurate at the 30 % level or better . this confirms the validity of close coupling calculations . aggarwal k.m . , 1983 , j. phys . b 16 , 2405 aggarwal k.m . , 1984 , apjs 54 , 1 aller l.h . , faulkner d.j . , 1964 , in iau - ursi symp . 20 : ` the galaxy and the magellanic clouds ' , eds . kerr f.j . , rodgers a.w . , australian academic press , canberra , p. 45 baluja k.l . , 1985 , j. phys . b 18 , l413 beintema d.a . , pottasch s.r . , 1999 , a&a 347 , 942 beintema d.a . , van hoof p.a.m . , lahuis f. et al . , 1996 , a&a 315 , l253 blaha m. , 1969 , a&a 1 , 42 clegg r.e.s . , harrington j.p . , barlow m.j . , walsh j.r . , 1987 , apj 314 , 551 ( c87 ) de graauw th . , haser l.n . , beintema d.a . et al . , 1996 , a&a 315 , l49 harrington j.p . , lutz j.h . , seaton m.j . , stickland d.j . , 1980 , mnras 191 , 13 hummer d.g . , berrington k.a . , eissner w. , pradhan a.k . , saraph h.e . , tully j.a . , 1993 , a&a 279 , 298 lennon d.j . , burke v.m . , 1991 , mnras 251 , 628 lennon d.j . , burke v.m . , 1994 , a&as 103 , 273 luo d. , pradhan a.k . , saraph h.e . , storey p.j . , yan yu , 1989 , j. phys . b , 22 , 389 mathis j.s . , 1990 , ara&a 28 , 37 middlemass d. , 1990 , mnras 244 , 294 oliva e. , pasquali a. , reconditi m. , 1996 , a&a 305 , l21 ( o96 ) osterbrock d.e . , 1974 , in ` astrophysics of gaseous nebulae ' , freeman , san francisco pquignot d. , aldrovandi s.m.v . , stasinska g. , 1978 , a&a 63 , 313 pottasch s.r . , beintema d.a . , 1999 , a&a 347 , 975 pottasch s.r . , preite - martinez a. , olnon f.m . , jing - er m. , kingma s. , 1986 , a&a 161 , 363 rowlands n. , houck j.r . , herter t. , 1994 , apj 427 , 867	the calculation of accurate collision strengths for atomic transitions has been a long standing problem in quantitative spectroscopy . most modern calculations are based on the r - matrix method and problems pertaining to the use of this method have led to a discussion of the accuracy of these results . more in particular , based on an analysis of the spectra of ngc 3918 and ngc 6302 , clegg et al . ( 1987 ) and oliva et al . ( 1996 ) have questioned r - matrix calculations for the infrared [ nev ] fine - structure transitions . using improved flux measurements for the [ nev ] lines , we show that the conclusion that these collision strengths would be too high , is not correct . the discrepancies found by clegg et al . ( 1987 ) can be explained by the inaccuracy of the [ nev ] 342.6 nm flux they adopted . the discrepancies found by oliva et al . ( 1996 ) can be explained by the inaccuracy of the _ lrs_flux for the [ nev ] 14.32 @xmath0 m line . based on the data presented in this paper there is _ no reason _ to assume that there are any problems with the r - matrix calculations for of lennon & burke ( 1994 ) . we show that the data are accurate at the 30 % level or better . this confirms the validity of the close coupling method .
we consider a two - dimensional configuration of @xmath0 particles with @xmath105 contacts and @xmath57 polygons . for convenience of notation , only single digit particle indices are used in this example , so that the notation @xmath106 means the cartesian @xmath10 component of the unit vector from the center of particle @xmath107 to that of particle @xmath108 . + and @xmath104 matrices are shown . arrows represent the normal vectors used to construct the @xmath19 and @xmath104 matrices ( before normalization ) . different arrow colors are for visualization purposes only . ] the convention for ordering of the contacts is demonstrated in eq . [ eq : c ] ( and see also fig . [ fig : m_configuration ] ) : @xmath109 the @xmath19 matrix is used to describe the force balance condition ( eq . 1 in the main text ) and has dimension @xmath110 in the most general case when contact forces have both normal and tangential components . each row is associated with a given particle @xmath21 and each column describes one contact and has non - zero entries corresponding only to the pair of particles @xmath21 and @xmath22 forming that contact . its first @xmath0 rows store the @xmath10 components and the next @xmath0 rows store the @xmath11 components of unit normal vectors @xmath111 and unit tangential vectors @xmath112 ( counter - clockwise orthogonal to @xmath111 ) . the first @xmath105 columns of @xmath19 correspond to the normal directions and the next @xmath105 columns correspond to the tangential directions ( which can also of course be expressed using the normal directions via a simple rotation transformation ) . an example of some of the terms of the @xmath19 matrix for the configuration of fig . [ fig : m_configuration ] is given in eq . [ eq : m ] : the @xmath104 matrix is used to describe the torque balance condition ( see eq . 9 in the main text ) and is of dimensions @xmath114 . again , the row indices correspond to particles and the column indices refer to contacts . the non - zero entries in each column correspond to the radii of particles @xmath21 and @xmath22 forming that contact . an example of some of the terms of the @xmath104 matrix for the configuration of fig . [ fig : m_configuration ] is given in eq . [ eq : t ] : @xmath115 when the external torque is zero , as in our loading protocol using compression , the radii are eliminated from the torque balance equation and the @xmath104 matrix can be further simplified to the form of eq . [ eq : t_alt ] : @xmath116 the @xmath55 matrix ( cf . eq . 7 in the main text ) is used to describe the presence of closed polygons formed by particles in contact and and is of dimensions @xmath117 . here row indices correspond to polygons and column indices refer to the contacts . non - zero entries in each row describe the unit normal directions joining two particles in contact which are members of a given polygon . the first @xmath57 rows store the @xmath10 components and the next @xmath57 rows store the @xmath11 components of unit vectors @xmath111 . an example for some of the terms of the @xmath55 matrix is given in eq . [ eq : q ] ( and see fig . [ fig : q_configuration ] ) : @xmath118	the determination of the normal and transverse ( frictional ) inter - particle forces within a granular medium is a long standing , daunting , and yet unresolved problem . we present a new formalism which employs the knowledge of the external forces and the orientations of contacts between particles ( of any given sizes ) , to compute all the inter - particle forces . having solved this problem we exemplify the efficacy of the formalism showing that the force chains in such systems are determined by an expansion in the eigenfunctions of a newly defined operator . in a highly influential paper from 2005 majmudar and behringer @xcite wrote : inter - particle forces in granular media form an inhomogeneous distribution of filamentary force chains . understanding such forces and their spatial correlations , specifically in response to forces at the system boundaries , represents a fundamental goal of granular mechanics . the problem is of relevance to civil engineering , geophysics and physics , being important for the understanding of jamming , shear - induced yielding and mechanical response . " a visual example of such force chains in a system of plastic disks is provided in fig . [ mahesh ] . in this letter we present a solution of this goal . to be precise , the problem that we solve is the following : consider a granular medium with known sizes of the granules , for example the 2-dimensional systems analyzed in ref . @xcite and shown in fig . [ mahesh ] , of @xmath0 disks of known diameters @xmath1 . given the external forces , denoted below as @xmath2 and the external torques @xmath3 exerted on the granules , and given the angular orientations of the vectors connecting the center of masses of contacting granules ( but not the distance between them ! ) , determine all the inter - particle normal and tangential forces @xmath4 and @xmath5 . the method presented below applies to granular systems in mechanical equilibrium ; the issue of instabilities and abrupt changes in the force chains will be discussed elsewhere . for the sake of clarity and simplicity we will present here the two - dimensional case ; the savvy reader will recognize that the formalism and the solution presented will go smoothly also for the three - dimensional case ( as long as the system is in mechanical equilibrium ) . the full formalism will be presented in a longer publication in due course . the obvious conditions for mechanical equilibrium are that the forces and the torques on each particle have to sum up to zero @xcite . the condition of force balance is usefully presented in matrix form using the following notation . denote the ( signed ) amplitudes of the inter - particle forces @xmath6 as a vector @xmath7 , where the amplitudes @xmath8 appear first and then the amplitudes @xmath9 . the vector of @xmath10 and @xmath11 components @xmath12 and @xmath13 is denoted as @xmath14 where all the @xmath10 components are presented in @xmath14 first and then all the @xmath11 components . the vector @xmath7 has @xmath15 entries where @xmath16 is the number of contacts between particles . the vector @xmath14 has @xmath17 entries where @xmath0 is the number of particles , with zero entries for all the particles on which there is no external force . we can then write the force balance condition as @xmath18 where @xmath19 is a @xmath20 matrix . the entries in the matrix @xmath19 contain the directional information , see supplemental material at [ url will be inserted by publisher ] for an example of an @xmath19 matrix . denote the unit vector in the direction of the vector distance between the centers of mass of particles @xmath21 and @xmath22 by @xmath23 , and the tangential vector by @xmath24 orthogonal to @xmath23 . then the entries of @xmath19 display the projections @xmath25 and @xmath26 or @xmath27 and @xmath28 as appropriate . we thus guarantee that eq.([m ] ) is equivalent to the mechanical equilibrium condition @xmath29 as is well known , the friction - less granular system in the thermodynamic limit is jammed exactly at the isostatic condition @xmath30 @xcite . in the friction - less case @xmath19 is a @xmath31 matrix and as long as @xmath32 one can solve the problem by multiplying eq . ( [ m ] ) by the transpose @xmath33 , getting @xmath34 in this case the matrix @xmath35 has generically exactly three goldstone modes ( two for translation and one for rotation ) @xcite , and since the external force vector is orthogonal to the goldstone modes ( otherwise the external forces will translate or rotate the system ) , eq . ( [ mmt ] ) can be inverted with impunity by multiplying by @xmath36^{-1}$ ] . in fact even when @xmath37 but the system is small enough to be jammed , this method can be used since there are enough constraints to solve for the forces . this last comment is important for our developments below . the problem becomes under - determined above isostaticity in the frictionless case , when force chains begin to build up that span from one boundary to the other . with friction we anyway have twice as many unknowns and we need to add the constrains of torque balance . the condition of torque balance is @xmath38 on every particle , where @xmath39 is the external torque exerted on the @xmath21th disk @xcite . for disks , @xmath40 is in the normal direction , and therefore the torque balance becomes a condition that the sum of tangential forces has to balance the external tangential force . this condition can be added to eq . ( [ m ] ) using a new matrix @xmath41 in the form @xmath42 the order of the extended matrix @xmath41 is @xmath43 , see supplemental material at [ url will be inserted by publisher ] for an example of @xmath44 . above jamming when the number of contacts increases @xmath45 . the matrix @xmath46 is not square , and the matrix @xmath47 which is of size @xmath48 , has at least @xmath49 zero modes @xcite . accordingly it can not be inverted and one can conclude that * the conditions of mechanical equilibrium are not sufficient to determine all the forces . * obviously what is missing are additional constraints to remove the host of zero modes . these additional constraints are _ geometrical _ constraints @xcite which can be read from those disks which describe connected polygons . in other words , since we know the orientation @xmath23 of each contact in our system , we can determine which granules are stressed in a triangular arrangement , and which in a square or pentagonal etc . , see fig . [ geometry ] . each such arrangement is a constraint on the radius vectors adjoining the centers of mass . for example if particles @xmath50 and @xmath51 are in a triangular arrangement then @xmath52 , with the analogous constraint on squares , pentagons etc . these constraints can be written in a matrix form by denoting the _ amplitudes _ of inter - particle vector distances as @xmath53 where we again arrange the @xmath10 components first and the @xmath11 components second : @xmath54 where the matrix @xmath55 again has entries @xmath25 or @xmath26 as appropriate to represent the vectorial geometric constraints , see supplemental material at [ url will be inserted by publisher ] for an example of @xmath56 . denoting the total number of polygons by @xmath57 the dimension of the matrix @xmath55 is @xmath58 . of course @xmath53 has @xmath16 entries while @xmath7 had @xmath15 entries . note that in generic situations there can be also disks which are not stressed at all . these are referred to as rattlers " . for example in the configuration shown in fig . [ geometry ] there exist 14 rattlers . at this point we specialize the treatment to hookean normal forces with a given force constant @xmath59 @xcite . non hookean forces result in a nonlinear theory that can still be solved but much less elegantly . for the present case @xmath60 \ . \label{sig}\ ] ] denoting the amplitudes of the vectors @xmath61 as the vector @xmath62 ( again with first the @xmath10 and then the @xmath11 components ) , we can rewrite eq . ( [ q ] ) in the form @xmath63 having this result at hand we can formulate the final problem to be solved . arrange now a new matrix , say g , operating on a vector @xmath7 , with a rhs being a vector , say @xmath64 , made of a stacking of @xmath14 , @xmath65 and @xmath66 , as before with @xmath10 and then @xmath11 components : @xmath67 using these objects our problem is now @xmath68 the dimension of the matrix @xmath69 is @xmath70 and the matrix @xmath71 has the dimension @xmath48 . we can use now the euler characteristic @xcite to show that the situation has been returned here to the analog of the invertible matrix @xmath72 when @xmath73 : the euler characteristic in two dimensions requires that @xmath74 where @xmath75 is the number of rattlers " i.e. disks on which there is no force . accordingly we find that @xmath76 consequently , the matrix @xmath77 has no zero eigenmodes . thus the final solution for the forces can be obtained as @xmath78 where @xmath79 is the set of eigenfunctions of @xmath77 associated with eigenvalues @xmath80 . we compared the inter - particle forces obtained from direct numerical simulations ( see below for details ) to those computed from eq . ( [ final ] ) . both normal and tangential forces are of course identical to machine accuracy . we reiterate that we did not need to know the distances between particles . this is important in applying the formalism to experiments since it is very difficult to measure with precision the degree of compression of hard particles like , say , metal balls or sand particles . note also the remarkable fact that we never had to provide the frictional ( tangential ) force law in the formalism to obtain the correct forces ! at this point we can discuss the force chains . by definition these are the large forces in the system that provide the tenuous network that keeps the system rigid . observing eq . ( [ final ] ) we should focus on the eigenfunction @xmath79 of @xmath77 that have the smallest eigenvalues and the largest overlaps with @xmath81 . these can be found and arranged in order of the magnitude of @xmath82 independently of the calculation of @xmath7 . in fig . [ order ] we show the contribution to the total energy @xmath83 , learning that about 20% of the leading eigenfunction are responsible for 90% of the energy . we can therefore hope that the force chains will be determined by the same relatively small number of eigenfunctions . this is not guaranteed ; due to contributions to the forces that oscillate in sign the convergence can be much slower than in the case of the energy where the sum is of positive contributions . in fig . [ chains ] we show in upper left panel the force chains in the configuration of fig . [ geometry ] . in the other panels we show the prediction of the force chains using 100 , 200 and 300 of the ( energy ) leading modes . we learn that with 100 out of the 864 modes the main force chains begin to be visible . with 200 out of the 864 modes the full structure of the force chains is already apparent , although with 300 it is represented even better . since the number of geometric constraints is very large , one can ask whether all the geometric constraints are necessary , as eq . ( [ euler ] ) shows that @xmath84 . the answer is no , we could leave out constraints as long as we have enough conditions to determine the solution . there is the obvious question then why do we have a unique solution when the number of equations is larger than the number of unknowns . the answer to this question lies in the properties of the vector @xmath64 and the matrix @xmath85 which does have many zero modes . a condition for the existence of a solution is that @xmath64 is orthogonal to all the zero modes of @xmath85 , as can be easily checked . we have ascertained in our simulations that this condition is always met . in the near future we will present an extension of this formalism to three dimensions and the use of the formalism to study the instabilities of the force networks to changes in the external forces . as a final comment we should note that in fact only _ one _ external force is necessary to determine _ all _ the inter - disk forces . this single external force is necessary to remove the re - scaling freedom that this problem has by definition . * simulations * : for the numerical experiments in 2-dimensions we use disks of two diameters , a ` small ' one with diameter @xmath86 and a ` large ' one with diameter @xmath87 . such @xmath0 disks are put between virtual walls at @xmath88 and @xmath89 . these walls exert external forces on the disks . the external forces are taken as hookean for simplicity . for disks near the wall at @xmath88 we write @xmath90 here @xmath91 denotes the @xmath10 component of the position vector @xmath40 of the center of mass of the @xmath21th disk , and we have a similar equation for the @xmath11 components with @xmath92 replaced by @xmath93 . when two disks , say disk @xmath21 and disk @xmath22 are pressed against each other we define their amount of compression as @xmath94 : @xmath95 where @xmath96 is the actual distance between the centers of mass of the disks @xmath21 and @xmath22 . in our simulations the normal force between the disks acts along the radius vector connecting the centers of mass . we employ a hookean force @xmath97 . to define the tangential force between the disks we consider ( an imaginary ) tangential spring at every contact which is put at rest whenever a contact between the two disks is formed . during the simulation we implement memory such that for each contact we store the signed distance @xmath98 to the initial rest state . for small deviations we require a linear relationship between the displacement and the acting tangential force . this relationship breaks when the magnitude of the tangential force reaches @xmath99 where due to coulomb s law the tangential loading can no longer be stored and is thus dissipated . when this limit is reached the bond breaks and after a slipping event the bond is restored with a the tangential spring being loaded to its full capacity ( equal to the coulomb limit ) . the numerical results reported above were obtained by starting with @xmath100 particles on a rectangular grid ( ratio 1:2 ) with small random deviations in space and no contacts . we implement a large box that contains all the particles . the box acts on the system by exerting a restoring harmonic normal force as described in eq . ( [ external ] ) . the experiment is an iterative process in which we first shrink the containing box infinitesimally ( conserving the ratio ) . the second step is to annul all the forces and torques , to bring the system back to a state of mechanical equilibrium . we therefore annul the forces using a conjugate gradient minimizer acting to minimize the resulting forces and torque on all particles . we iterate these two steps until the system is compressed to the desired state . this work had been supported in part by an ideas " grant stanpas of the erc . we thank deepak dhar for some very useful discussions . we are grateful to edan lerner for reading an early version of the manuscript with very useful remarks . 99 t.s . majmudar and r.p . behringer , contact force measurements and stress - induced anisotropy in granular materials " , nature * 435 * , 1083 ( 2005 ) . s. alexander , amorphous solids : their structure , lattice dynamics and elasticity " , phys . rep . * 296 * 65 - 236 ( 1998 ) . m. wyart , s.r.nagel and t. a. witten geometric origin of excess low - frequency vibrational modes in weakly connected amorphous solids " europhys . lett . * 72 * 486492 ( 2005 ) . the existence of rattlers " that are not in close contact with other particles may increase the number of zero modes . to see this note that the matrices @xmath101 and @xmath102 have the same rank , but @xmath102 can not have more than @xmath103 non - zero eigenmodes . therefore @xmath101 must have at least @xmath49 zero modes . r. c. ball and r. blumenfeld , `` stress field in granular systems : loop forces and potential formulation '' , phys . rev . lett . * 88 * , 115505 ( 2002 ) . r. blumenfeld , `` stresses in isostatic granular systems and emergence of force chains '' , phys . rev . lett . * 93 * , 108301 ( 2004 ) . the relevance of the linear forces to laboratory experiments was shown in t.s . majmudar , m. sperl , s. luding and r.p . behringer , jamming transition in granular systems " , phys . rev . lett . * 98 * , 058001 ( 2007 ) and suppl . information . s.v matveev , euler characteristic " , in hazewinkel , michiel , encyclopedia of mathematics , springer , isbn 978 - 1 - 55608 - 010 - 4 , ( 2001 ) .
bayesian networks or graphical models based on directed acyclic graphs ( dags ) are widely used to represent complex causal systems in applications ranging from computational biology to epidemiology , and sociology @xcite . a dag entails a set of conditional independence relations through the markov properties . two dags are said to be _ markov equivalent _ if they entail the same conditional independence relations . in general , observational data can only identify a dag up to markov equivalence . for statistical causal inference it is therefore important to enumerate and describe the set of markov equivalence classes ( mecs ) and their sizes . if the mecs are large in size , then causal inference algorithms that operate in the space of mecs as compared to dags could significantly increase efficiency . however , gaining a full understanding of the causal relationships in a system with a large mec requires many interventional experiments that deliberately and carefully alter one or more components of the system . the purpose of this paper is to recast this important combinatorial and enumerative question from statistics in the language of combinatorial optimization . this new perspective yields complexity results on the problem in general , as well as solutions to the problem in some special cases . the problem of enumerating mecs has been studied from two fundamental perspectives : ( 1 ) enumerate all mecs on @xmath1 nodes ( as in @xcite ) , and ( 2 ) enumerate all mecs of a given size ( as in @xcite ) . at the heart of these studies is a result of verma and pearl @xcite , which states that a mec is determined by the underlying undirected graph ( or _ skeleton _ ) and the placement of immoralities , i.e. induced subgraphs of the form @xmath2 . this characterization leads to a representation of an mec by a graph with directed and undirected edges known as the _ essential graph _ @xcite ( or _ cpdag _ @xcite or _ maximally oriented graph _ @xcite ) . in @xcite , gillespie and perlman use this characterization to identify all mecs on @xmath3 nodes ; namely , they fix a skeleton on @xmath1 nodes , and then count the number of ways to compatibly place immoralities within the skeleton . the works @xcite give inclusion - exclusion formulae for mecs of a fixed size by utilizing the combinatorial structure of the essential graph described in @xcite . however , since essential graphs can be quite complicated , these formulae are only realizable for relatively constrained classes of mecs . in particular , @xcite and @xcite only consider mecs of size one , and @xcite must fix the undirected edges of the essential graphs to be enumerated . as exhibited by these results , the implementation of combinatorial enumeration techniques appears difficult from perspective ( 2 ) . on the other hand , perspective ( 1 ) has only been considered via computer - enumeration @xcite . a common approach to difficult graphical structure enumeration problems is to specify a type of graph for which to solve the problem . this approach is used in such problems as the enumeration of independent sets , matchings , and colorings @xcite . given a graph , it can be useful to consider a refined set of combinatorial statistics each of which plays a role in the enumeration question . for instance , given a graph @xmath4 researchers examine the total number of independent sets ( or the _ fibonacci number _ of @xmath4 ) @xcite , the maximum size of an independent set ( or _ independence number _ of @xmath4 ) @xcite , and/or the number of independent sets of a fixed size @xcite . these refined statistics work together to give a complete understanding of the problem of enumerating independent sets for @xmath4 . in the present paper , we initiate the combinatorial enumeration of mecs with respect to a fixed undirected graph and thereby recast this enumeration problem in the language of combinatorial optimization . for a graph @xmath4 this amounts to enumerating all possible placements of immoralities within @xmath4 @xcite . thus , we are interested in the following combinatorial statistics : 1 . @xmath5 , the total number of mecs on @xmath4 , 2 . @xmath6 , the maximum number of immoralities on @xmath4 , 3 . @xmath7 , the number of ways to place exactly @xmath8 immoralities on @xmath4 , and 4 . @xmath9 , where @xmath10 denotes the number of mecs on @xmath4 of size @xmath11 . the first three statistics fit together naturally in the polynomial presentation @xmath12 in general , computing any or all of these statistics for a given type of graph appears to be difficult . in this paper , we will prove the following theorem in support of this observation . [ thm : np - complete ] given an undirected graph @xmath4 , the problem of computing a dag @xmath13 with skeleton @xmath4 and @xmath14 immoralities is np - hard . here , we use the notion of np - hardness as defined in ( * ? ? ? * chapter 5 ) . as with most np - hard problems , restricting to special cases can make the problem tractable . in this paper , we will compute some or all of ( 1 ) , ( 2 ) , ( 3 ) , and ( 4 ) for some special types of graphs that are important in both statistical and combinatorial settings . moreover , these special cases can offer useful structural insights on the general problem . for example , it appears that the number and size of equivalence classes is guided by the number of cycles and high degree nodes in the skeleton . in order to test and verify these types of observations , we develop a computer program for the enumeration of the combinatorial statistics ( 1 ) , ( 2 ) , ( 3 ) , and ( 4 ) that expands on the original program of gillespie and perlman @xcite . using this program we can not only verify the observations that high degree nodes and cycles in the skeleton play an important role , but we are also able to make the following interesting observation , indicating the profound role played by the underlying skeleton . [ thm : frequency determinism ] for @xmath3 , every connected graph @xmath4 on @xmath1 nodes has a unique frequency vector @xmath15 . the remainder of this paper is organized as follows . in section [ sec : some first examples ] , we examine some first and fundamental examples including paths , cycles , and the complete bipartite graph @xmath16 . we compute all the desired combinatorial statistics specified by ( 1 ) , ( 2 ) , ( 3 ) , and ( 4 ) for these graphs . the first two examples exhibit an important connection to independent sets and vertex covers . in section [ sec : trees ] , we consider our enumeration question in the special setting of trees . here , we derive results for stars , bistars , complete binary trees , and caterpillar graphs . the former two examples play an important role in bounding the number and size of mecs on tree graphs , and the latter two examples are fundamental to phlyogenetic modeling @xcite . following this , we identify bounds on the number of mecs on a given tree that exactly parallel the classically known bounds for independent sets in trees . we also identify tight bounds on the size of a mec on a given tree using properties of the associated essential graphs . in section [ sec : immorality numbers and star decompositions ] , we prove theorem [ thm : np - complete ] via a reduction of the minimum vertex cover problem . to do so , we prove a correspondence between minimum vertex covers of a given triangle - free graph @xmath4 and minimum decompositions of @xmath4 into non - overlapping stars , which we call _ minimum star decompositions_. using this correspondence , we can compute the number @xmath6 for triangle - free graphs whose minimum star decompositions are isomorphic as forests . we apply this result to recover @xmath6 for the complete bipartite graph @xmath17 and some special types of circulant graphs . in section [ sec : computational analysis ] , we describe our computer program for the computation of the statistics ( 1 ) , ( 2 ) ( 3 ) , and ( 4 ) . this program collects a variety of data on markov equivalence classes and the skeleton of each class for all connected graphs on @xmath3 nodes and for triangle - free graphs on @xmath18 nodes . in particular , we compare class size and the number of mecs per skeleton to skeletal features including average degree , max degree , clustering coefficient , and the ratio of number of immoralities in the essential graph of the mec to the number of induced @xmath19-paths in the skeleton . finally , we see that this program validates theorem [ thm : frequency determinism ] , and we also use it to address the analogous result in the case of unconnected graphs . since this work draws heavily on different concepts from two different fields , statistics and combinatorics , we provide an extensive review of the required concepts and definitions from both fields in the appendix . in this section , we provide some first examples for which we can compute all of the desired combinatorial statistics ( 1 ) , ( 2 ) ( 3 ) , and ( 4 ) . the first two examples are the path and cycle on @xmath1 nodes . using some well - known results on the independent sets within these graphs , we can quickly obtain the desired numbers . the third example presented in this section is the graph @xmath16 . unlike the path and cycle , @xmath16 requires a more detailed analysis . to compute the polynomial @xmath20 and the vector @xmath15 for paths and cycles , we will use the notion of independent sets . we refer the unfamiliar reader to section [ subsec : graphs ] for all the necessary definitions . in this section , we will use two well - studied combinatorial sequences , and their associated polynomial filtrations . recall that the @xmath21 fibonacci number @xmath22 is defined by the recursion @xmath23 the _ @xmath21 fibonacci polynomial _ is defined by @xmath24 and it has the properties that @xmath25 for all @xmath26 and @xmath27 for all @xmath28 . analogously , the @xmath21 lucas number @xmath29 is given by the fibonacci - like recursion @xmath30 the @xmath21 _ lucas polynomial _ is given by @xmath31 it is a well - known result that the independence polynomial of the path of length @xmath1 , which we denote by @xmath32 , is equal to the @xmath33 fibonacci polynomial and the independence polynomial of the @xmath1-cycle @xmath34 is given by the @xmath21 lucas polynomial ; that is to say , @xmath35 with these facts in hand we prove the following theorem . [ thm : path and cycle polynomials ] for the path @xmath32 and the cycle @xmath34 on @xmath1 nodes we have that @xmath36 in particular , the number of mecs on @xmath32 and @xmath34 , respectively , is @xmath37 and the maximum number of immoralities is @xmath38 the result follows from a simple combinatorial bijection . since paths and cycles are the graphs with the property that the degree of any vertex is at most two , then the possible locations of immoralities are exactly the degree two nodes . that is , the unique head node @xmath39 in an immorality @xmath40 must be a degree two node . in the path @xmath32 , this corresponds to all @xmath41 non - leaf vertices , and for the cycle @xmath34 this is all the vertices of the graph . notice then that no two adjacent degree two nodes can simultaneously be the unique head node of an immorality , since this would require one arrow to be bidirected . thus , a viable placement of immoralities corresponds to a choice of any subset of degree two nodes that are mutually non - adjacent , i.e. that form an independent set . conversely , given any independent set in @xmath32 , a dag can be constructed by placing the head node of an immorality at each element of the set and directing all other arrows in one direction . similarly , this works for any nonempty independent set in @xmath34 . ( notice that any mec on the cycle must have at least one immorality since all dags have at least one sink node . ) the resulting formulas are then @xmath42 which completes the proof . it remains to compute the vectors @xmath43 and @xmath44 and the maximum number of immoralities @xmath45 and @xmath46 . the formulae for these combinatorial statistics follow naturally from the description of the placement of immoralities given in theorem [ thm : path and cycle polynomials ] . [ thm : path vectors ] the number @xmath47 of mecs of size @xmath48 with skeleton @xmath32 is the number of compositions @xmath49 of @xmath50 into @xmath51 parts that satisfy @xmath52 as @xmath8 varies from @xmath53 let @xmath13 be a dag with skeleton @xmath32 . we denote the markov equivalence class of @xmath13 by @xmath54 $ ] . by the proof of theorem [ thm : path and cycle polynomials ] , we know that the immorality placements in @xmath54 $ ] correspond to the nodes in an independent @xmath8-subset @xmath55 $ ] on the subpath @xmath56 of @xmath32 induced by the non - leaf nodes of @xmath32 . the induced graph of the complement of @xmath57 is a forest of @xmath51 paths . since each member of @xmath54 $ ] is a dag with skeleton @xmath32 that has no immoralities on these @xmath51 paths , then each path contains a unique sink . each independent @xmath8-subset yields a distinct forest of @xmath51 paths on @xmath58\backslash\mathcal{i}$ ] , which corresponds to a unique partition of @xmath50 into @xmath51 parts . the formula for @xmath47 is then given by considering all such possible placements of sinks on each path in the forests over all independent sets . a similar argument using integer partitions allows us to compute the number of mecs of size @xmath48 on the @xmath1-cycle . we refer the reader to subsection [ subsec : combinatorial definitions ] for the unfamiliar terminology or notation . [ thm : cycle vectors ] the number of mecs of size @xmath48 in the @xmath1-cycle is @xmath59 , \\ \ell=\prod_{i=1}^k i^{m_i}}}\frac{p}{k}{k\choose m_1,\ldots , m_{p-2k+1}},\ ] ] where @xmath60 $ ] denotes the partitions of @xmath61 with @xmath8 parts with largest part at most @xmath39 . since @xmath34 is a graph in which every node is degree 2 , then each mec of @xmath34 containing @xmath8 immoralities corresponds to an independent @xmath8-subset of @xmath58:=\{1,2 , \dots , p\}$ ] , and the subgraph of @xmath34 given by deleting this @xmath8-subset consists of @xmath8 disjoint paths . the size of this mec is then the product of the lengths of these paths . so we need only count the number of such subgraphs for which this product equals @xmath48 . to count these objects , consider that each subgraph of @xmath34 given by deleting an independent @xmath8-subset of @xmath34 forms a partition of the @xmath50 remaining vertices into @xmath8 parts with maximum possible part size being @xmath62 . such a partition is represented by @xmath63,\ ] ] where @xmath64 and @xmath65 . the partition @xmath66 corresponds to an unlabeled forest consisting of @xmath67 @xmath11-paths , and the number of subgraphs of @xmath34 isomorphic to this forest is @xmath68 the desired formula follows since the size of each corresponding mec is @xmath69 . [ rmk : lucas triangle ] it is a well - known result that the coefficient of @xmath70 in the @xmath71 fibonacci polynomial is the binomial coefficient @xmath72 , and that this is also the number of compositions of @xmath50 into @xmath51 parts . the former result says that the @xmath71 fibonacci polynomial has coefficients given by the @xmath71 diagonal of pascal s triangle , and so the latter result gives a compositional interpretation of the corresponding entry in pascal s triangle . in this subsection , we saw that this compositional interpretation of @xmath72 results in the proof of theorem [ thm : path vectors ] . diagonal of each triangle is the coefficient vector of @xmath73 and @xmath74 , respectively . , scaledwidth=75.0% ] analogously , the @xmath21 diagonal of a second triangle , called _ lucas triangle _ in @xcite , corresponds to the coefficients of the @xmath21 lucas polynomial . this triangle is depicted on the right in figure [ fig : pascal - and - lucas - triangles ] . thus , the proof of theorem [ thm : cycle vectors ] results in a combinatorial interpretation of the entries of this triangle via partitions . in particular , the entry of the lucas triangle corresponding to the @xmath75 coefficient of @xmath74 is @xmath76.l_p(x ) = \sum_{\left\langle 1^{m_1},2^{m_2},\ldots,(p-2k+1)^{m_{p-2k+1}}\right\rangle\in \mathbb{p}[p-2k+1,k , p - k]}\frac{p}{k}{k\choose m_1,\ldots , m_{p-2k+1}}.\ ] ] moreover , the binomial recursion on the triangle implies that these coefficients satisfy the identity @xmath76.l_p(x ) = [ x^{k-1}].l_{p-2}+[x^k].l_{p-1}.\ ] ] to the best of the authors knowledge , such a partition identity is new to the combinatorial literature . for convenience , we consider the partitioned vertex set of @xmath16 to be the two distinguished nodes @xmath77 . the remaining @xmath1 nodes are labeled by @xmath58 $ ] and are collectively referred to as the _ spine _ of @xmath16 . this labeling of @xmath16 is depicted on the left in figure [ fig : k - two - p ] . first , it is easy to see that the maximum number of immoralities is given by orienting the edges such that all edge heads are at the nodes @xmath78 and @xmath79 . this results in @xmath80 . next , we compute a closed form formula for the number of mecs for @xmath16 . [ thm : k_2,p number of mecs ] the number of mecs with skeleton @xmath16 is @xmath81 to arrive at the desired formula , we divide the problem into three cases : 1 . the number of immoralities at node @xmath79 is @xmath82 . the number of immoralities at node @xmath79 is strictly between @xmath83 and @xmath82 . 3 . there are no immoralities at node @xmath79 . notice that cases ( 1 ) and ( 2 ) have a natural interpretation via the indegree at node @xmath79 of the essential graph of the corresponding mecs . readers unfamiliar with the theory of essential graphs can find the basics in section [ subsec : essential graphs ] . if the indegree at @xmath79 is two or more , all edges adjacent to @xmath79 are essential , and the number of immoralities at node @xmath79 is given by its indegree . thus , we can rephrase cases ( 1 ) and ( 2 ) as follows : the indegree of node @xmath79 in the essential graph of the mec is @xmath1 . the indegree of node @xmath79 in the essential graph of the mec is @xmath84 . in case ( 1 ) , the mec is determined exactly by the mec on the star with center node @xmath78 and @xmath1 edges . one can easily check ( this is also proven as part of theorem [ thm : stars ] in the following section ) that this yields @xmath85 mecs . case ( 2 ) is more subtle . first , assume that the indegree at node @xmath79 is @xmath84 , and the arrows with head @xmath79 have the tails @xmath86 $ ] . then the remaining arrows adjacent to @xmath79 are all directed outwards with heads @xmath87 . is depicted on the left , and one of the essential graphs counted in the proof of theorem [ thm : k_2,p number of mecs ] is depicted on the right.,scaledwidth=75.0% ] notice that no immoralities can happen at nodes @xmath88 $ ] along the spine , but some may occur at the nodes @xmath58\backslash[k]$ ] . if there are no such immoralities , then node @xmath78 has indegree @xmath1 , otherwise the essential graph would contain a directed @xmath89-cycle . similarly , if , without loss of generality , we denote the nodes in @xmath58\backslash[k]$ ] that are the heads of immoralities by @xmath90 for @xmath91 , then the nodes @xmath92 are tails of the arrows adjacent to node @xmath78 . thus , if the number of immoralities with heads in @xmath58\backslash[k]$ ] is @xmath91 , then the immoralities with heads at node @xmath78 are completely determined . therefore , each @xmath93-subset of @xmath58\backslash[k]$ ] yields a single mec . figure [ fig : k - two - p ] depicts an example of one such choice of immoralities . we start by selecting the arrows to form immoralities at node @xmath79 which forces the remaining arrows at @xmath79 to point towards the spine . we then select some of these to form immoralities at the spine , and this forces the remaining arrows to be directed inwards towards @xmath78 . however , if @xmath94 , the star induced by nodes @xmath95 determines the mecs . this yields @xmath96 classes ( see again theorem [ thm : stars ] ) . in total , for case ( 2 ) the number of mecs is @xmath97 in case ( 3 ) , we consider when there are no immoralities at node @xmath79 , and we count via placement of immoralities along the spine . there are @xmath98 ways to place immoralities along the spine , one for each subset of @xmath58 $ ] . suppose the immoralities along the spine have the heads @xmath99 for @xmath100 ( the cases @xmath101 and @xmath102 are considered separately ) . then the remaining immoralities can happen at node @xmath78 . however , if there is an immorality with head at node @xmath78 then all other arrows adjacent to @xmath78 are essential , some of which may point towards the spine with heads in the set @xmath58\backslash[k]$ ] . since there are no immoralities with head in the set @xmath58\backslash[k]$ ] , then any such outward pointing arrow is part of a directed path from @xmath78 to @xmath79 . however , since there are no immoralities at node @xmath79 , there can be at most one such directed path . the presence of any such directed path forces a directed @xmath89-cycle since @xmath100 . therefore , for @xmath100 the nodes @xmath87 must be tails of arrows oriented towards node @xmath78 , thereby yielding only a single mec . since @xmath103 and @xmath104 also yield only a single mec , case ( 3 ) yields a total of @xmath98 classes . combing the total number of mecs counted for each of these cases yields the desired formula . using the case - by - case analysis from the proof of theorem [ thm : k_2,p number of mecs ] we can count the number of mecs with skeleton @xmath16 of each possible size . similarly , one can also recover the statistics @xmath105 from this proof . however , to avoid overwhelming the reader with formulae , we omit the expressions for @xmath105 . [ cor : k_2,p sizes of mecs ] the possible sizes of a mec with skeleton @xmath16 and the number of classes having each size is as follows : [ cols="^,^",options="header " , ] the theory developed in this paper and the computational verification of theorem [ thm : frequency determinism ] support the intuition that the undirected structure of the skeleton of a dag plays a fundamental role in the size of its mec as well as the number of other mecs that have the same skeleton . in the triangle - free setting , the results of section [ sec : trees ] and theorem [ thm : minimum star decompositions formula ] highlight the significance of the number and density of high degree nodes within the skeleton of the dag . in this section , we parse these observations in terms of the computational data collected via our computer program . specifically , we will compare class size and the number of mecs per skeleton to skeletal features including average degree , max degree , clustering coefficient , and the ratio of number of immoralities in the essential graph of the mec to the number of induced @xmath19-paths in the skeleton . @xmath106{clustering - coefficient - and - log - average - class - size.pdf } & \includegraphics[width=0.47\textwidth]{clustering - coefficient - and - average - number - of - mecs.pdf } \\ \end{array}$ ] recall that the _ ( global ) clustering coefficient _ of a graph @xmath4 is defined as the ratio of the number of triangles in @xmath4 to the number of connected triples of vertices in @xmath4 . the clustering coefficient serves as a measure of how much the nodes in @xmath4 cluster together . figure [ fig : clustering coefficient all graphs ] presents two plots : one compares the clustering coefficient to the log average class size and the other compares it to the average number of mecs . this data is taken over all connected graphs on @xmath3 nodes with @xmath107 edges ( to achieve a large number of mecs ) . as we can see , the average class size grows as the clustering coefficient increases . this is to be expected , since an increase in the number of triangles within the dag should correspond to an increase in the size of the chordal components of the essential graph . on the other hand , the average number of mecs decreases with respect to the clustering coefficient , which is to be expected given that the class sizes are increasing . this decrease in the average number of mecs empirically captures the intuition that having many triangles in a graph results in fewer induced @xmath19-paths , which represent the possible choices for distinct mecs with the same skeleton . @xmath108{average - degree - and - log - average - class - size.pdf } & \includegraphics[width=0.48\textwidth]{average - degree - and - average - number - of - mecs.pdf } \\ \end{array } $ ] figure [ fig : average degree vs log average class size ] presents a pair of plots , the first of which compares the average degree of the underlying skeleton of the dag to the log average class size of the associated mec . the second plot compares the average degree of the skeleton to the average number of mecs it supports . both plots present one curve for all connected graphs and a second curve for triangle - free graphs on @xmath109 nodes . for connected graphs on @xmath109 nodes the left - most plot shows a strict increase in the log average mec class size as the average degree of the nodes in the underlying skeleton increases . this is to be expected since graphs with a higher average degree are more likely to contain larger chordal components . on the other hand , the average class size for triangle - free graphs increases for average degree up until approximately @xmath110 , and then shows a steady decrease for larger average degree . since the average degree of a tree on @xmath1 nodes is @xmath111 , this suggests that the largest mecs amongst triangle - free graphs have skeleta being trees . as such , the bounds developed in section [ sec : trees ] of this paper can be , heuristically , thought to apply more generally to all triangle - free graphs . the right - most plot in figure [ fig : average degree vs log average class size ] describes the relationship between average degree and the average number of mecs for all connected graphs and triangle - free graphs on @xmath109 nodes . we see from this that in the setting of all connected graphs , the skeleta with the largest average number of mecs appear to have average degree @xmath112 , whereas in the triangle - free setting , the higher the average degree the more equivalence classes the skeleta can support . this supports the intuition that the more high degree nodes there are in a triangle - free graph the more equivalence classes the graph can support . theoretically , a portion of this intuition is captured by theorem [ thm : minimum star decompositions formula ] . @xmath106{max - degree - and - average - class - size.pdf } & \includegraphics[width=0.48\textwidth]{max - degree - and - average - number - of - mecs.pdf } \\ \end{array}$ ] the left - most plot in figure [ fig : max degree vs log average class size ] depicts the relationship between the maximum degree of a node in a skeleton and the average class size on the skeleton for all connected graphs and for triangle - free graphs on at most @xmath109 nodes . for all graphs , the relationship appears to be almost linear beginning with maximum degree @xmath113 , suggesting that average class size grows linearly with the maximum degree of the underlying skeleton . this growth in class size is due to the introduction of many triangles as the maximum degree grows . on the other hand , in the triangle - free setting we actually see a decrease in average class size as the maximum degree grows , which empirically reinforces this intuition . the right - most plot in figure [ fig : max degree vs log average class size ] records the relationship between the maximum degree of a node in a skeleton and the average number of mecs supported by that skeleton for all connected graphs and triangle - free graphs on at most @xmath109 nodes . for all graphs , we see that the average number of mecs grows with the maximum degree of the graphs , and this growth is approximately exponential . in the triangle - free setting , the average number of mecs appears to be unimodal , but would be increasing if we considered also all graphs on @xmath114 . for triangle - free graphs there is only one graph with maximum degree 9 , namely the star @xmath115 , where the number of mecs is @xmath116 . for connected graphs the average number of mecs is pushed up by those cases consisting of a complete bipartite graph where in addition one node is connected to all other nodes . -paths in the essential graph.,scaledwidth=46.0% ] the final plot of interest is in figure [ fig : class size versus immorality ratio ] , and it shows the relationship between markov equivalence class size and the ratio of the number of immoralities in the essential graph to the number of induced @xmath19-paths in the skeleton for all connected graphs and triangle - free graphs on @xmath109 nodes . that is , it shows the relationship between the class size and how many of the potential immoralities presented by the skeleton are used by the class . it is interesting to note that , in the triangle - free setting , as the class size grows , this ratio appears to approach @xmath117 , suggesting that most large mecs use about a third of the possible immoralities in triangle - free graphs . in the connected graph setting , as the class size grows , we see a steady decrease in the value of this ratio . this supports the intuition that a larger class size corresponds to an essential graph with large chordal components and few immoralities . we wish to thank brendan mckay for some helpful advice in the use of the programs nauty and traces @xcite . liam solus was partially supported by an nsf mathematical sciences postdoctoral research fellowship ( dms - 1606407 ) . caroline uhler was partially supported by darpa ( darpa - sn-16 - 37 ) and onr ( n00014 - 16-s - ba10 ) . an _ ( undirected ) graph _ @xmath4 is a pair of sets @xmath118 in which @xmath119 is some _ node _ set and @xmath120 is the _ edge _ set , where an _ edge _ is taken to be an unordered pair of nodes @xmath121 for @xmath122 . we say that two nodes @xmath123 are _ adjacent _ in @xmath124 if @xmath125 , or equivalently , we say that @xmath11 is a _ neighbor _ of @xmath39 and vice versa . the _ neighborhood _ of a node @xmath126 in a graph @xmath127 is the set of all neighbors of @xmath11 in @xmath4 including the node @xmath11 itself . a node is said to be _ incident _ to an edge @xmath128 if @xmath11 is one of the two defining nodes of @xmath129 . the _ degree _ of a node @xmath11 in a graph @xmath4 is the number of edges incident to @xmath11 in @xmath4 , and it is denoted by @xmath130 . when the graph @xmath4 is understood , we often write @xmath131 . a node @xmath11 is called a _ leaf _ of @xmath4 when @xmath132 . a _ path _ in an undirected graph @xmath4 is an alternating sequence of nodes and edges @xmath133 in which @xmath134 for all @xmath135 $ ] . cycle _ in an undirected graph @xmath4 is an alternating sequence of nodes and edges @xmath133 in which @xmath134 for all @xmath135 $ ] , and @xmath136 . a graph @xmath124 is called _ connected _ if there exists a path in @xmath4 between every pair of nodes in @xmath119 . an undirected graph is called a _ forest _ if it contains no cycles . a _ tree _ is a connected forest . a graph @xmath137 is _ directed _ if the edge set @xmath120 is a set of _ arrows _ , where an _ arrow _ is defined to be any ordered pair of nodes , denoted @xmath138 or @xmath139 for @xmath122 . in an arrow @xmath139 the node @xmath11 is referred to as the _ tail _ of the arrow and the node @xmath39 is referred to as the _ head_. a _ directed cycle _ in a directed graph @xmath13 is an alternating sequence of nodes and arrows @xmath133 in which @xmath140 for all @xmath135 $ ] , and @xmath136 . a _ directed acyclic graph _ ( or dag ) is a directed graph that does not contain any directed cycles . a node @xmath11 is called a _ parent _ of node @xmath39 in a directed graph @xmath13 if @xmath141 . in this case , the node @xmath39 is called a _ child _ of node @xmath11 . a node @xmath39 is called a _ descendant _ of a node @xmath11 , if there is a directed path in @xmath13 from @xmath11 to @xmath39 . in this case , node @xmath11 is an _ ancestor _ of @xmath39 . we denote the set of parents , descendants , and ancestors of node @xmath11 , respectively , by @xmath142 , @xmath143 , and @xmath144 . the set of _ non - descendants _ of @xmath11 is @xmath145 . a _ partially directed graph _ is a pair @xmath127 with node set @xmath119 for which the set @xmath120 consists of some edges @xmath121 and some @xmath146 for @xmath147 . both , dags and partially directed graphs have an underlying _ skeleton _ , which is the undirected graph given by replacing every arrow @xmath139 with the edge @xmath121 . the collection of all partially directed graphs include both , directed and undirected graphs , as subcollections . for this reason , we make the following definitions in the setting of partially directed graphs , and simply note that they specify to directed and undirected graphs accordingly . a _ subgraph _ of a partially directed graph @xmath148 is a partially directed graph @xmath149 for which @xmath150 and @xmath151 . an _ induced subgraph _ of @xmath152 is a subgraph @xmath149 of @xmath152 for which an edge or arrow @xmath128 is in @xmath153 if and only if its two incident nodes are in @xmath154 . we often call @xmath155 the subgraph _ induced _ by @xmath154 and denote this by writing @xmath156 . we say that two partially directed graphs @xmath148 and @xmath157 are _ isomorphic _ if there is a bijection @xmath158 such that @xmath121 is an edges of @xmath152 if and only if @xmath159 is an edge of @xmath155 and @xmath139 is an arrow of @xmath152 if and only if @xmath160 is an arrow of @xmath155 . a _ path _ in a partially directed graph @xmath4 is an alternating sequence of nodes and edges / arrows @xmath133 in which @xmath134 ( or @xmath140 ) for all @xmath135 $ ] . if the path includes no arrows , it is called _ undirected _ , and if it includes at least one arrow , it is called _ directed_. a _ cycle _ in an undirected graph @xmath4 is an alternating sequence of nodes and edges / arrows @xmath133 in which @xmath134 ( or @xmath140 ) for all @xmath135 $ ] , and @xmath136 . if the cycle contains no arrows it is called _ undirected _ and if it includes at least one arrow it is called _ directed_. a _ chain graph _ is a partially directed graph containing no directed cycles . the _ chain components _ of a chain graph are the connected components of its subgraph consisting of all its edges . we now define some special classes of graphs that we will consider in this paper , and specify their notation . @xmath1-path _ is the undirected graph @xmath161,e)$ ] for which @xmath162\}$ ] . @xmath1-cycle _ is the undirected graph @xmath163,e)$ ] for which @xmath162\}\cup\{\{1,n\}\}$ ] . a _ chordless cycle _ in a graph @xmath4 is any induced subgraph of @xmath4 that is a cycle . a graph is called _ chordal _ if every chordless cycle in @xmath4 is a @xmath19-cycle . the _ complete graph _ on @xmath1 nodes is the undirected graph @xmath164,e)$ ] for which @xmath165\}$ ] . given two disjoint node sets @xmath119 and @xmath166 for which @xmath167 and @xmath168 , the _ complete bipartite graph _ on @xmath119 and @xmath166 is the undirected graph @xmath169 . the _ @xmath1-star _ is the complete bipartite graph @xmath170 , and we often denote the @xmath1-star by @xmath171 . for @xmath172 , the _ center node _ of @xmath171 is the unique node of degree @xmath1 , and for @xmath173 the center of @xmath171 is chosen to be one of the two nodes of the graph . bistar _ can be thought of as a gluing of two stars in which a leaf node of one star is glued to the the center node of the other star . we denote the bistar given by gluing a leaf of @xmath174 to the center node of @xmath171 by @xmath175 . equivalently , the bistar @xmath175 can be defined by attaching @xmath1 leaves to one node of the @xmath176-path @xmath177 and @xmath178 leaves to the other node . more generally , we can define the graph @xmath179 to be the undirected graph given by attaching @xmath180 leaves to node @xmath11 of the @xmath1-path @xmath32 . a _ rooted _ undirected graph @xmath127 is an undirected graph with a unique distinguished node @xmath181 called the _ root _ of @xmath4 . complete binary tree _ is a rooted tree in which the root node has degree two and every other non - leaf node has degree three . in this section , we let @xmath127 be an undirected graph . independent set _ in @xmath4 is a subset of mutually non - adjacent nodes @xmath182 . vertex cover _ of @xmath4 is a subset of nodes @xmath182 for which every edge of @xmath4 is incident to some node in @xmath183 . the complement in @xmath119 of an independent set is always a vertex cover in @xmath119 . in particular , the complement of a maximum independent set is a minimum vertex cover . the _ independence number of @xmath4 _ is the size of any maximum independent subset of @xmath4 , and the _ fibonacci number of @xmath4 _ is the total number of independent sets of @xmath4 . let @xmath184 denote the number of independent subsets of size @xmath8 in @xmath4 , @xmath185 denote the independence number of @xmath4 , and @xmath186 denote the fibonacci number of @xmath4 . these numbers collect into the polynomial generating function @xmath187 which we call the _ independence polynomial of @xmath4_. in particular , @xmath188 . let @xmath189 denote a vector of random variables that are indexed by the nodes of some directed acyclic graph ( dag ) @xmath137 . the set of arrows and non - arrows in @xmath13 encode a set of conditional independence constraints on random variables @xmath190 , known as the markov properties of the dag @xmath13 . the _ directed acyclic graphical model _ associated to @xmath13 is the family of multivariate probability distributions that abide by the markov properties of @xmath13 . to specify the markov properties , we first recall the notion of connectedness or separation for directed graphs . a _ path _ in the dag @xmath13 is an alternating sequence of nodes and arrows @xmath191 in which the arrow @xmath192 is either @xmath193 or @xmath194 for all @xmath11 . a _ collider _ on a path in @xmath13 is any subsequence @xmath195 such that @xmath196 and @xmath197 . given a dag @xmath198 , we consider the graphical model associated to @xmath13 by the _ directed global markov property_. two nodes @xmath199 and @xmath200 in @xmath198 are said to be _ @xmath201-connected _ given @xmath202 if there exists a path @xmath203 from @xmath199 to @xmath200 for which if @xmath205 , @xmath206 , and @xmath152 are disjoint subsets of @xmath119 with @xmath205 and @xmath206 being nonempty , then @xmath152 is said to _ @xmath201-separate _ @xmath205 and @xmath206 if there are no @xmath207 and @xmath208 that are @xmath201-connected given @xmath152 . the _ directed global markov property _ associates the dag @xmath137 with the set of conditional independence relations @xmath209 for all triples @xmath210 , and @xmath152 for which @xmath152 @xmath201-separates @xmath205 and @xmath206 . note that multiple dags can have the same set of @xmath201-separation relations , and thereby encode the same set of conditional independence statements . if two dags encode the exact same conditional independence relations with respect to the directed global markov property , they are called _ markov equivalent_. given a dag @xmath137 , the collection of all dags that are markov equivalent to @xmath13 , including @xmath13 itself , is called the _ markov equivalence class _ of @xmath13 . in this paper we denote the markov equivalence class of @xmath13 by @xmath54 $ ] . there is a well - known characterization of when two dags are markov equivalent , and it is given in terms of their skeleta and their set of immoralities . immorality _ in a dag @xmath13 is a triple of node @xmath211 for which @xmath13 contains the arrows @xmath212 and @xmath213 but @xmath13 does not contain either of the arrows @xmath214 or @xmath215 . let @xmath13 be a dag and let @xmath54 $ ] denote the mec containing @xmath13 . an _ essential arrow _ in @xmath13 is an arrow which appears in all elements of @xmath54 $ ] . that is , @xmath139 is an essential arrow of @xmath13 if and only if @xmath139 is an arrow of @xmath216 for all @xmath217 $ ] . we also refer to such an arrow as an essential arrow of the mec @xmath54 $ ] . the _ essential graph _ of the class @xmath54 $ ] is the partially directed graph whose set of arrows is precisely the essential arrows of @xmath54 $ ] and whose set of edges is those edges @xmath121 in the skeleton of @xmath13 that support non - essential edges in @xmath13 . we denote the essential graph of @xmath54 $ ] by @xmath218 . recall that the _ chain components _ of a chain graph are the connected components of its subgraph consisting of all its undirected edges . we call the connected directed components of an essential graph its _ essential components_. an arrow @xmath139 is called _ strongly protected _ in a chain graph if it occurs in any one of the configurations depicted in figure [ fig : strongly protected edges ] . 1 . @xmath4 is a chain graph . every chain component of @xmath4 is chordal . 3 . the configuration @xmath219 does not occur as an induced subgraph of @xmath4 . every arrow @xmath139 is strongly protected in @xmath4 . a _ composition _ of a positive integer @xmath61 into @xmath8 parts is an ordered sum of @xmath8 positive integers equaling @xmath61 , i.e. @xmath220 the positive integers @xmath221 are called the _ parts _ of the composition . a partition of the positive integer @xmath61 is a sequence of integers @xmath222 satisfying @xmath223 and @xmath224 . the nonzero integers @xmath225 are called the _ parts _ of the partition , and we say that @xmath226 has @xmath8 parts if @xmath227 . we let @xmath228 $ ] denote the set of partitions of @xmath61 with @xmath8 parts , and we let @xmath60 $ ] denote the partitions of @xmath61 with @xmath8 parts with largest possible part @xmath39 . note that our notation differs slightly from the notation used in @xcite , in which @xmath229 denotes the number of partitions of @xmath61 with _ at most _ @xmath8 parts and largest part being @xmath39 . if the partition @xmath226 has @xmath67 parts equaling @xmath11 then we may write @xmath230 . when @xmath231 $ ] , then @xmath232 and so @xmath226 is uniquely identified with the integer vector @xmath233 . in this case , we simply write @xmath234 $ ] . a wealth of enumerative combinatorics , including a complete treatment of compositions and integer partitions , can be found in @xcite . m. r. garey and d. s. johnson . _ computers and intractability : a guide to the theory of np - completeness_. a series of books in the mathematical sciences . wh freeman and company , new york , ny 25.27 ( 1979 ) : 141 . s. b. gillispie and m. d. perlman . _ enumerating markov equivalence classes of acyclic digraph models . _ proceedings of the seventeenth conference on uncertainty in artificial intelligence . morgan kaufmann publishers inc . , 2001 . t. verma and j. pearl . _ an algorithm for deciding if a set of observed independencies has a causal explanation_. proceedings of the eighth international conference on uncertainty in artificial intelligence . morgan kaufmann publishers inc .	bayesian networks or directed graphical models are widely used to represent complex causal systems . from observational data alone , a bayesian network can only be identified up to _ markov equivalence _ and interventional experiments are required to distinguish between markov equivalent bayesian networks . it is therefore important to describe and understand the set of markov equivalence classes ( mecs ) and their sizes . in this paper , we initiate the combinatorial enumeration of mecs on a fixed undirected graph , and thereby recast this important statistical problem into the language of combinatorial optimization . combinatorially , two bayesian networks are markov equivalent if their underlying undirected graphs and their set of immoralities are the same . we show that the np - complete minimum vertex cover problem reduces to the computation of a bayesian network with the maximum number of immoralities . the np - hardness of this problem shows the complexity of enumerating mecs from the perspective of combinatorial optimization . we then solve this enumeration problem for classically studied families of graphs including paths , cycles , and some types of complete bipartite graphs , trees , and circulants . we also provide tight bounds on the number and size of mecs on a given tree . these bounds exactly parallel the well - known bounds for the independent set problem . finally , a computer program was written to analyze the enumeration problem on all graphs with at most 10 nodes and all triangle - free graphs with at most @xmath0 nodes . the results show a trade - off between degree and clustering coefficient with respect to the number of mecs and their sizes . in addition , this program established that the frequency vector consisting of the number of mecs of each size uniquely determines the undirected graph .
much work has been done on spin related transport properties of nanoelectronic devices resulting in interesting applications , for example the so called spin field effect transistor proposed by datta and das @xcite . there has been particular interest in using the rashba effect to manipulate the spin degree of freedom in such systems @xcite . in this paper we model a simple system exhibiting the rashba effect , viz . a ring with rashba hamiltonian attached to an arbitrary number of ` free ' wires , using so called solvable models @xcite . this means that we approximate the system by a one dimensional graph on which is defined an appropriate self adjoint schrdinger operator . the advantage of this approach is that , as the name suggests , it allows us to get explicit expressions for the scattering matrix , and hence for the transport properties of the system , in this case in terms of the greens function of the ring and the boundary conditions at the vertices . + our particular interest in considering this model is to investgate the possibility of constructing a spin filter . various approaches have been taken to filter spin : we mention @xcite in which the authors construct a spin filter using a four terminal device with the rashba effect as well as @xcite where the authors achieve spin filtering using a two terminal device and a magnetic field . a third approach , discussed in @xcite , uses a three terminal device with the rashba effect and to some extent was the motivation for this paper . + it is known that a device with two terminals and time reversal symmetry can not polarise spin currents @xcite ( the device in @xcite does not have time reversal invariance due to the magnetic field ) . nevertheless , kiselev and kim @xcite show that a three terminal device with time reversal symmetry and a particular geometric symmetry can make an effective spin filter . we consider the same geometry as considered in @xcite , viz . a ring with three wires and symmetry with respect to reflection across the line defined by the ` incoming ' wire . whereas kiselev and kim assume the rashba effect is localised at the ` incoming ' terminal in our model the rashba hamiltonian is present uniformly on the whole ring . kiselev and kim use a sophisticated numerical model of the system to calculate transport properties while our model is of course solvable . + we believe that the formalism of solvable models offers , in general , advantages over numerical studies in that it allows us to derive explicit expressions for scattering properties thereby identifying principal features of the system . ideally , these may be used to help optimise the design ( for instance for spin filtering ) . in particular , for the three terminal device described above we investigate how the polarisation is related to the resonant eigenvalues on the ring , the rashba coefficient and the angle of attachment of the wires . we observe , as did kiselev and kim , that this system may be used as an efficient spin filter . we consider a ring shaped quantum waveguide where the width of the waveguide and the incident electron energy is such that the ring may be considered one - dimensional . furthermore , we assume that there is structural inversion asymmetry @xcite so that a rashba term appears in the hamiltonian on the ring . normalising the radius to one it can be shown @xcite that the hamiltonian has the form @xmath0 where @xmath1 @xmath2 is the local coordinate on the ring ; @xmath3 , @xmath4 , @xmath5 , @xmath6 denote the pauli spin matrices and the unit matrix respectively ; and @xmath7 describes the strength of the rashba spin - orbit coupling . the solutions of the eigenequation , @xmath8 , are @xmath9 where @xmath10 and @xmath11 , @xmath12 . the eigenvalues on the ring @xmath13 correspond to the zeroes of @xmath14 . each eigenvalue @xmath15 has multiplicity two , the corresponding eigenspace is spanned by @xmath16 where @xmath17 since @xmath18 and @xmath19 we assume @xmath20 for @xmath21 and @xmath22 for @xmath23 . finally , we note that the twofold degeneracy of the eigenvalues drops to a fourfold degeneracy when @xmath24 . in this case we see that @xmath25 . + mostly we will write eigenfunctions with both spin eigenstates together in a @xmath26 matrix in order to simplify notation . in particular the solutions @xmath27 may be used to find the greens function , ie . the continuous solution of @xmath28 which is in fact @xmath29 \frac{e^{-i\sigma_{y } \varphi/2 } } { 2 i ( \kappa_{+ } + \kappa_{- } ) } e^ { i\sigma_{z } \eta/2}\ , \sigma_{z } \nonumber \\ & = & \frac { e^{-i\sigma_{z } \theta /2}\ , e^{-i\sigma_{y } \varphi/2 } } { 2 i ( \kappa_{+ } + \kappa_{- } ) } \left [ \frac { e^ { i\sigma_{z } \kappa_{+ } \left ( \theta - \eta - \pi \right ) } } { \cos ( \kappa_{+ } \pi ) } - \frac { e^ { -i\sigma_{z } \kappa_{- } \left ( \theta - \eta - \pi \right ) } } { \cos ( \kappa_{- } \pi ) } \right ] e^{-i\sigma_{y } \varphi/2 } e^ { i\sigma_{z } \eta/2}\ , \sigma_{z } \ , . \nonumber\end{aligned}\ ] ] here we take @xmath30 . + we assume that the ring is attached to @xmath31 semi - infinite wires . on each wire we have a ` free ' hamiltonian @xmath32 with generalised eigenfunctions @xmath33 where @xmath34 is the index for the wire and @xmath35 is the coordinate on the respective wire . + we write the hamiltonian on the whole system @xmath36 and consider this as an operator on the hilbert space @xmath37 of spinor valued functions on the graph @xmath38 consisting of the ring @xmath39 with @xmath31 wires @xmath40 attached . to define this as a self adjoint operator we need to correctly define the domain of @xmath41 which is related to self adjoint boundary conditions arising from the vanishing of the boundary form @xmath42 generally these boundary conditions are parameterised by a unitary matrix , for details see @xcite . here @xmath43 is the inner product on @xmath44 , @xmath45 is the inner product on spinors and @xmath46 are the points where the wires are attached to the ring . + we always assume that each vertex has three incident edges , ie . no two wires are attached at the same point on the ring . for a given vertex we denote by @xmath47 the values of the eigenspinor on edge @xmath48 in the limit as we approach the vertex and by @xmath49 the values of the outward derivative on edge @xmath48 in the limit as we approach the vertex . in this paper we assume the following boundary conditions at the vertices @xmath50 motivated by the fact that they are closely related to the ansatz for the scattering matrix of the t - junction as described in the physics literature @xcite ( we describe this relationship in the first appendix ) . here edge @xmath51 is the semi - infinite wire while edges @xmath52 are on the ring . the coefficient @xmath53 describes the strength of the coupling between the wire and the ring . + these boundary conditions ( [ bndcnd ] ) are assumed to hold with the same @xmath53 for each component of the spinor , ie . the coupling is independent of spin . however , @xmath53 may in general be different at each vertex or point where a wire is attached to the ring . it is clear that these boundary conditions are self - adjoint , ie . ( [ bform ] ) vanishes ( see @xcite for a discussion of boundary conditions in the presence of magnetic terms ) . the scattered waves @xmath54 are eigenfunctions on the quantum graph satisfying the boundary conditions at the vertices and having the following form on the wires : @xmath55 we reiterate that @xmath54 corresponds to two spinor valued waves , one with spin up and one with spin down incident waves . similarly , the components of the scattering matrix @xmath56 , @xmath57 , are @xmath26 matrix valued . due to the nature of the boundary conditions we can define the scattered wave on the ring as @xmath58 where @xmath59 , @xmath60 are the points where the wires are attached to the ring and @xmath61 is a matrix of coefficients . + the boundary conditions for the scattered waves can be neatly expressed using the greens function . defining @xmath62 we see from ( [ swwires ] ) that on the wires @xmath63 where @xmath64 denotes evaluation at the point where the @xmath65-th wire is attached to the ring and @xmath66 should be interpreted as matrix with matrix valued components . likewise from ( [ swring ] ) we have @xmath67 on the ring where we have used the property of the derivative of the greens function . it is then easy to see that the boundary conditions in ( [ bndcnd ] ) can be written @xmath68 respectively where @xmath53 is a diagonal matrix containing the coupling strengths for the @xmath31 vertices . solving for the scattering matrix we get @xmath69 this result can be generalised , at least in principle , using the techniques described in @xcite to find the scattering matrix of a quantum graph consisting of @xmath31 semi - infinite wires attached to a compact graph consisting of @xmath70 rings with rashba term connected by edges of finite length . let us consider a a three terminal device with symmetry as illustrated in figure [ sfil ] , ie . the angle @xmath71 between the first and second and first and third wires the same . to be precise we also need that the coupling constants at vertices two and three are the same in fact , for simplicity , we will set all of the coupling constants equal to one , @xmath72 . + as was shown in @xcite a three terminal device with this symmetry can potentially act as a spin filter . specifically , for unpolarised current entering the first wire the polarisation of flux measured on wires two , @xmath73 , and three , @xmath74 , along the @xmath7-axis satisfies @xmath75 a proof of these statements , following @xcite but in the context of quantum graphs , is given in appendix 2 . + appendix 3 contain an outline of the derivation of expressions for the conductance @xmath76 and the polarisation in the @xmath77-axis @xmath78 for current going from wire one to two for the device in figure [ sfil ] . in figures [ pi2][3pi4 ] we plot the conductance @xmath76 ( upper curve ) and @xmath78 ( lower curve ) against the energy @xmath79 . in these plots we assume @xmath80 ; there is no significant change in the form of the @xmath76 and @xmath78 curves with respect to @xmath7 apart from at the special values discussed below . we assume that the angle of attachment of the wires , @xmath81 , is an integer fraction of @xmath82 . and @xmath78 for @xmath80 and @xmath83.,width=529,height=566 ] and @xmath78 for @xmath80 and @xmath84.,width=529,height=566 ] and @xmath78 for @xmath80 and @xmath84.,width=529,height=566 ] and @xmath78 for @xmath80 and @xmath85.,width=529,height=566 ] and @xmath78 for @xmath80 and @xmath85.,width=529,height=566 ] the resonance eigenvalues of the ring are indicated by @xmath86 for @xmath21 and @xmath87 for @xmath23 ( the leftmost @xmath86 and @xmath87 are at @xmath88 and @xmath89 respectively with @xmath90 off the left of the plot ) . + it is clear from these plots that at some of the resonant energies , depending on the value of @xmath91 , this device would function as an efficient spin filter . our case @xmath83 , figure [ pi2 ] , most closely resembles the device considered in @xcite and we observe as did kiselev and kim that the polarisation is maximum at odd resonant eigenvalues and severely damped at even eigenvalues an explanation in terms of interference effects is given in @xcite . however , for our device at @xmath83 the polarisation changes sign at the odd resonances indicating that this may not be an ideal parameter regime for spin filtering . clearly defined and isolated peaks can be found for instance at @xmath84 and we believe would be a better parameter regime for spin filtering . + we remark that for @xmath81 , @xmath76 and @xmath78 are periodic repeating after @xmath92 resonances @xmath15 with zeroes in @xmath76 and @xmath78 at @xmath93 . it seems plausible that this behaviour can again be explained by spin interference and topological phase effects @xcite . + finally , we observe from the expression for the polarisation @xmath78 , equation ( [ fap ] ) , the following behaviour as a function of the rashba coefficient @xmath7 . as noted above , for @xmath24 the eigenvalues on the ring become fourfold degenerate . for these values of @xmath7 , @xmath94 so that see the expression for @xmath95 given after equation ( [ fap])@xmath96 and the polarisation is _ identically zero_. this is clearly a generalisation of the degenerate behaviour at @xmath97 , a possible physical explanation involves noting that there is an integral number of effective flux quanta @xcite through the ring for these values of @xmath7 and showing that for integral quanta the spin interaction is cancelled . we rather consider the gauge transformation @xmath98 on the ring where @xmath99 . here , since @xmath100 is not single valued , we take @xmath101 and assume that the point of attachment @xmath102 . we also make a change of basis on the wires , formally @xmath103 . generally this results in all spin interaction being concentrated at the boundary conditions ; however , for @xmath24 the boundary conditions remain spin independent so that , in analogy with @xmath97 , we have vanishing of the polarisation . precisely , at all vertices except @xmath104 the original boundary conditions ( [ bndcnd ] ) continue to hold while at the origin we get the new boundary conditions @xmath105 here edge @xmath51 is the semi - infinite wire , edge @xmath106 corresponds to @xmath107 , edge @xmath108 corresponds to @xmath109 on the ring and we recall that the above derivatives are in the outward direction from the vertex . these boundary conditions show that , up to an energy shift on the ring , the scattering properties of the ring fall into two classes depending on whether @xmath70 is even or odd indicating that the conductance is a periodic function of @xmath7 ( as observed in the case of a two terminal device in @xcite ) . the author has benefitted greatly from conversations with prof b.pavlov and dr u. zlicke . there is a well established description of the scattering matrix for the t - junction in the physics literature ( see @xcite , pg 173 , and @xcite ) . in this appendix we show how this anstz is related to the solvable models approach of specifying boundary conditions ( [ bndcnd ] ) at the vertex of the t - junction . + we note that ( [ bndcnd ] ) are of ` projection type ' @xcite , ie . we can express these boundary conditions in the form @xmath110 where @xmath111 , @xmath112 and @xmath113 @xmath114 are projections . now we suppose that the t - junction is at the point where three semi - infinite wires meet , instead of at the point of connection of one wire and a ring . as above we construct scattered waves for this non compact system and derive the scattering matrix which turns out to be @xcite @xmath115 this scattering matrix is the same as posited in @xcite for the t - junction ( a different ordering of the edges is used in the last two references ) motivating our choice of ( [ bndcnd ] ) for the boundary conditions . + we note that , at least for the chosen boundary conditions ( [ bndcnd ] , [ bndcnd2 ] ) , scattering at the idealised t - junction is independent of energy . it is not difficult to see that this is true for any projection type boundary condition and furthermore means that there are no discrete eigenvalues or resonances associated to the junction ( this is an equivalence @xcite , there are no discrete eigenvalues or resonances associated to the junction iff the boundary conditions are of projection type ) . we also note that projection type boundary conditions , also referred to in the degree two case as flp - tsutsui point interactions , are important in the study of ` quantum chaotic ' behavior in quantum graphs @xcite due to their scale invariance . + given a two or three dimensional quantum network , the problem of deriving appropriate boundary conditions at the vertices of an approximating one dimensional quantum graph is an active area of research ( see @xcite ) as is the ` inverse problem ' of constructing sequences of graphs with regular potentials so that in the limit we observe a chosen boundary condition from the whole @xmath116 parameter space @xcite . here we follow the argument of @xcite , which describes the symmetries of the scattering matrix , using terms appropriate for quantum graphs . + we think of @xmath38 as a graph in the plane and suppose that @xmath117 is a closed curve nowhere tangent to @xmath38 . the wronskian is defined as @xmath118 where @xmath119 and @xmath120 is the orientation of the ordered pair formed of the orientation of the local variable at @xmath121 and the orientation of @xmath117 at @xmath121 . the operator @xmath122 is one of @xmath123 or @xmath124 depending on whether @xmath121 is on the ring or the wires . + we always assume that the wronskian acts on solutions of the eigenequation , @xmath125 , @xmath126 , from which it is easy to see that @xmath127 is piecewise constant . furthermore , we assume that @xmath117 is large , in particular it encloses and has no intersections with the ring , in which case we drop the subscript and write @xmath128 from the constancy of the wronskian we see that on the ring @xmath129 while on the wires @xmath130 so that @xmath131 is actually equal to the boundary form ( [ bform ] ) . in particular , if @xmath132 and @xmath133 are eigensolutions satisfying the boundary conditions , ie . such that the boundary form ( [ bform ] ) vanishes , then @xmath134 ( in fact since our boundary conditions are ` local ' we will have @xmath135 for any @xmath117 , but we do not need this . ) + the wronskian allows us to identify symmetries of the scattering matrix . consider the wronskian of two scattered waves : @xmath136 we get immediately @xmath137 we note that in this case , since @xmath54 is matrix valued , the wronskian @xmath138 is properly thought of as a @xmath26 matrix . + further symmetries of the scattering matrix may be found from operators commuting with the hamiltonian . here we are mainly interested in the case where the graph , and boundary conditions , are invariant with respect to a reflection in one of the coordinate axes @xmath139 . it is then clear that the hamiltonian will commute with @xmath140 and we have the vanishing of the wronskian @xmath141 or @xmath142 here @xmath143 is block diagonal with @xmath4 on the diagonal . + before we apply this we need to define some notation . it is convenient for us to decompose the components of the scattering matrix in terms of spin matrices @xmath144 in terms of this decomposition we can express the conductance @xmath145 and the polarisation in the @xmath7-axis @xmath146 for waves going from wire @xmath65 to wire @xmath48 as @xmath147 now we consider the three terminal device illustrated in figure [ sfil ] which is clearly invariant with respect to @xmath148 . then ( [ rsym ] ) implies @xmath149 where @xmath150 . in particular , the polarisation satisfies @xmath151 finally we note that for the system under consideration the time reversal operator @xmath152 , where @xmath153 is complex conjugation , along with @xmath154 , where @xmath155 reverses the sign of @xmath7 or equivalently @xmath156 , both commute with the hamiltonian . proceeding as above these can be used to find yet more symmetries of the scattering matrix ( see @xcite for a further discussion ) . using equations ( [ gfn ] , [ smatrix ] ) we see that we can express the scattering matrix for the device illustrated in figure [ sfil ] as @xmath157 where @xmath158 @xmath159 , @xmath160 and we have assumed that all of the coupling constants are equal , @xmath161 . + from the form of the scattering matrix we see that @xmath162 where , using maple , @xmath163 and , due to the form of @xmath164 , the determinant @xmath165 is a complex scalar . + it is easy to show that the terms due to @xmath166 in the expression for @xmath167 make no contribution to the conductance , equation ( [ flux ] ) , @xmath168 and introduce a multiplicative factor into the polarisation in the @xmath77-axis , equation ( [ pol ] ) , @xmath169 using ( [ s1sz ] ) we write @xmath170 and @xmath171 in terms of @xmath172 and @xmath173 ( here again the form of @xmath172 is important ) which gives us @xmath174 and @xmath175 we again use maple to get explicit expressions and simplify here it is important to cancel common factors @xmath176 which appear in the numerator and denominator to aid simplification and avoid numerical instability . at this step we also put @xmath177 . the expression for the conductance and polarisation are then @xmath178 where @xmath179 \\ & & \mbox { } + 2 \kappa^4 \left [ \cos \left ( \kappa_{+ } + \kappa_{- } \right ) \left ( \xi-2\pi \right ) + \cos \left ( \kappa_{+ } + \kappa_{- } \right ) \left ( 3\xi-2\pi \right ) \right ] \\ & & \mbox { } + 4 \kappa^4 \left [ \cos \left ( \kappa_{+ } + \kappa_{- } \right ) \xi + \cos \left ( \kappa_{+ } + \kappa_{- } \right ) 2 \left ( \xi-\pi \right ) \right ] \\ q & = & \kappa^3 \left [ \cos\left(2 \kappa_{-}\pi\right ) - \cos\left(2 \kappa_{+}\pi\right ) \right . \\ & & \left . \mbox { } + \cos\left ( \kappa_{+}2\xi + \kappa_{- } 2\left(\xi - \pi \right ) \right ) - \cos\left ( \kappa_{+ } 2\left(\xi - \pi\right ) + \kappa_{-}2\xi \right ) \right ] \\ & & \mbox{}+ 2 \kappa^3 \left [ \cos\left ( \kappa_{+ } \left(\xi - 2\pi\right ) + \kappa_{-}\xi \right ) - \cos\left ( \kappa_{+}\xi + \kappa_{- } \left(\xi - 2\pi\right ) \right ) \right ] \ , .\end{aligned}\ ] ]	we consider a quantum graph consisting of a ring with rashba hamiltonian and an arbitrary number of semi - infinite wires attached . we describe the scattering matrix for this system and investigate spin filtering for a three terminal device .
helical structures in the solar features like sunspot whirls were reported long back by george e. hale in 1925 @xcite . he found that about 80% of the sunspot whirls were counterclockwise in the northern hemisphere and clockwise in the southern hemisphere . later , in 1941 the result was confirmed by richardson @xcite by extending the investigation over four solar cycles . this hemispheric pattern was found to be independent of the solar cycle . since the 90 s , the subject has been rejuvenated and this hemispheric behaviour independent of sunspot cycle is claimed to be observed for many of the solar features like active regions @xcite , filaments @xcite , coronal loops @xcite , interplanetary magnetic clouds ( imcs ) @xcite , coronal x - ray arcades @xcite and network magnetic fields @xcite etc . helicity is a physical quantity that measures the degree of linkages and twistedness in the field lines @xcite . magnetic helicity h@xmath3 is given by a volume integral over the scalar product of the magnetic field * b * and its vector potential * a * @xcite . @xmath4 with * * b*= @xmath5 a. it is well known that the vector potential a * is not unique , thereby preventing the calculation of a unique value for the magnetic helicity from the equation ( 1 ) . @xcite pointed out that the helicity of magnetic field can best be characterized by the force - free parameter @xmath0 , also known as the helicity parameter or twist parameter . the force - free condition @xcite is given as , @xmath6 alpha is a measure of degree of twist per unit axial length ( see appendix - a for details of physical meaning of alpha ) . this is a local parameter which can vary across the field but is constant along the field lines . researchers have claimed to have determined the sign of magnetic helicity on the photosphere by calculating alpha , e.g. @xmath7 @xcite , averaged alpha e.g. @xmath8 = @xmath9 @xcite with current density @xmath10 . some authors have used current helicity density @xmath11 and @xmath12 @xcite . a good correlation was found between @xmath7 and @xmath13 by @xcite and @xcite . but the sign of magnetic helicity can not be inferred from the force - free parameter @xmath0 under all conditions ( see appendix b ) . it is well known that the reliable measurements of vector magnetic fields are needed to study various important parameters like electric currents in the active regions , magnetic energy dissipation during flares , field geometry of sunspots , magnetic twist etc . the study of error propagation from polarization measurements to vector field parameters is very important @xcite . @xcite have studied the effects of realistic errors e.g. , due to random polarization noise , crosstalk between different polarization signals , systematic polarization bias and seeing induced crosstalk etc . on known magnetic fields . they derived analytical expressions for how these errors produce errors in the estimation of magnetic energy ( calculated from virial theorem ) . however , they simulated these effects for magnetographs which sample polarization at few fixed wavelength positions in line wings . it is well known that such observations lead to systematic under - estimation of field strength and also suffer from magneto - optical effects @xcite . whereas in our analysis , we simulate the effect of polarimetric noise on field parameters as deduced by full stokes inversion . the details are discussed in the section 6 . @xcite found large variations in the global @xmath0 values from repeated observations of the same active regions . it is important to model the measurement uncertainties before looking for physical explanations for such a scatter . in a study by @xcite the noise levels in the observed fields were analyzed , but a quantitative relationship between the uncertainties in fields and the uncertainties in global @xmath0 value were not established . they could only determine the extent to which the incremental introduction of noise affects the observed value of alpha . however , for the proper tracking of error propagation , we need to start with ideal data devoid of noise and with known values of @xmath0 and magnetic energy . we follow the latter approach in our present analysis . here , we estimate the accuracy in the calculation of the @xmath0 parameter and the magnetic energy due to different noise levels in the spectro - polarimetric profiles . modern instruments measure the full stokes polarization parameters within the line profile . basically there are two types of spectro - polarimeters : ( i ) spectrograph based e.g. , advanced stokes polarimeter ( asp : @xcite ) , zurich imaging polarimeter ( zimpol : @xcite ) , themis - mtr ( @xcite ) , solis - vector spectro - magnetograph ( vsm : @xcite ) , polarimetric littrow spectrograph ( polis : @xcite ) , diffraction limited spectro - polarimeter ( dlsp : @xcite ) , hinode ( sot / sp : @xcite ) , etc . and ( ii ) filter - based e.g. , imaging vector magnetograph ( ivm ) at mees solar observatory , hawaii @xcite , solar vector magnetograph at udaipur solar observatory ( svm - uso)@xcite etc . earlier magnetographs like crimea @xcite , msfc @xcite , hsp @xcite , oao @xcite , hsos @xcite , potsdam vector magnetograph @xcite , sft @xcite etc . were mostly based on polarization measurements at a few wavelength positions in the line wings and hence subject to zeeman saturation effects as well as magneto - optical effects like faraday rotation @xcite . the magnetic field vector deduced from stokes profiles by modern techniques are almost free from such effects @xcite . this paper serves three purposes . first , we estimate the error in the calculation of field strength , inclination and azimuth and thereafter in the calculation of the vector field components @xmath14 , @xmath15 and @xmath16 . second , we estimate the error in the determination of global @xmath0 due to noise in polarimetric profiles constructed from the analytical vector field data . third , we estimate the error in the calculation of magnetic energy derived using virial theorem , due to polarimetric noise . in the next section ( section 2 ) we discuss a direct method for calculation of a single global @xmath0 for an active region . in section 3 , we describe the method of simulating an analytical bipole field . section 4 contains the analysis and the results . error estimation in global @xmath0 is given in section 5 . in section 6 we discuss the process of estimating the error in the virial magnetic energy . section 7 deals with discussion and conclusions . taking the z - component of magnetic field , from the force - free field equation ( 2 ) @xmath0 can be written as , @xmath17 for a least squares minimization , we should have @xmath18 where @xmath0 is the local value at each pixel , @xmath19 is the global value of @xmath0 for the complete active region and n is total number of pixels . + since eqn.(4 ) will lead to singularities at the neutral lines where b@xmath20 approaches 0 , therefore the next moment of minimization , @xmath21 should be used . from eqn.(5 ) we have @xmath22 which leads to the following result , @xmath23 this formula gives a single global value of @xmath0 in a sunspot and is the same as @xmath24 of @xcite . we prefer this direct way of obtaining global @xmath0 which is different from the method discussed in @xcite for determining @xmath7 . the main differences are : ( 1 ) . the singularities at neutral line are automatically avoided in our method by using the second moment of minimization and ( 2 ) . the computation of constant @xmath0 force - free fields for different test values of @xmath0 is not required . @xcite used a different parameter @xmath25 to avoid the effect of faraday rotation in sunspot umbrae . however , modern inversion techniques using complete stokes profiles are free of this problem . it must be noted that one can generate different values of @xmath19 using higher moments of minimization , e.g. , by weighting @xmath26 with @xmath27 , with n=3 , 5 , 7 , ... etc . the higher moments will be more sensitive to spatial variation of @xmath28 . such large and complex variation of @xmath28 is found generally in flare productive active regions @xcite . thus we can try to use higher order @xmath19 as a global index for predicting the flare productivity in active regions . finally , to compute @xmath19 we need all the three components of magnetic field which is obtained from the measurements of vector magnetograms . however , here we use the analytically generated bipole , as discussed in the following section , with known values of all the magnetic parameters to investigate the effect of polarimetric noise . we use the analytic , non - potential force - free fields of the form derived by @xcite . these fields describe an isolated bipolar magnetic region which is obtained by introducing currents into a potential field structure . this potential field is produced by an infinite straight line current running along the intersection of the planes y = 0 and z = -a , where negative sign denotes planes below the photosphere z = 0 . at the photosphere ( z = 0 ) , the field has the following form : @xmath29 @xmath30 @xmath31 where @xmath32 is the magnitude of the field at origin and @xmath33 . the function @xmath34 is a free generating function related to the force - free parameter @xmath0 ( see eqn ( 2 ) ) by @xmath35 which determines the current structure and hence the amount and location of shear present in the region . by choosing @xmath36 we can obtain a simple potential ( current - free , @xmath37 ) field produced by the infinite line current lying outside the domain . steeper gradient of @xmath34 results in a more sheared ( non - potential ) field . in equation ( 11 ) the sign on the right hand side is taken positive in the paper by low ( 1982 ) which is a typing mistake ( confirmed by b. c. low , private communication ) . we mention this here to avoid carrying forward of this typo as was done in @xcite . a grid of 100 x 100 pixels was selected for calculating the field components . the magnitude of field strength at origin has been taken as 1000 g and the value of ` a ' is taken as 15 pixels ( below the photosphere , z = 0 ) . the simulated field components with corresponding contours are shown in the figure 1 . here we use the following function ( e.g. , @xcite ) for the generation of the field components ( b@xmath38 , b@xmath39 , b@xmath20 ) : @xmath40 results for the fields generated by different @xmath34 are quantitatively similar . in this way we generate a set of vector fields with known values of @xmath0 . most of the time one of the bipoles of a sunspot observed on the sun is compact ( leading ) and the other ( following ) is comparatively diffuse . observations of compact pole gives half of the total flux of the sunspot and is mostly used for analysis . one can derive the twist present in the sunspot using one compact pole of the bipolar sunspot for constant @xmath0 . thus we have selected a single polarity of the analytical bipole as shown in figure [ fig1 ] to calculate the twist . fine structure in real sunspots is difficult to model . our analysis applies to the large scale patterns of the magnetic field regardless of fine structure . all the following sections discuss the analysis and results obtained . using the analytical bipole method @xcite the non - potential force - free field components b@xmath38 , b@xmath39 & b@xmath20 in a plane have been generated and are given as in equations ( 8) , ( 9 ) & ( 10 ) . we have shown b@xmath38 , b@xmath39 , & b@xmath20 maps ( generated on a grid of 100 x 100 pixels ) in figure [ fig1 ] . from these components we have derived magnetic field strength ( b ) , inclination ( @xmath41 ) and azimuth ( @xmath42 : free from 180@xmath43 ambiguity ) . in order to simulate the effect of typical polarimetric noise in actual solar observations on magnetic field measurements and study the error in the calculation of @xmath0 and magnetic energy , we have generated the synthetic stokes profiles for each b , @xmath41 and @xmath42 in a grid of 100 x 100 pixels , using the he - line information extractor `` helix '' code @xcite . this code is a stokes inversion code based on fitting the observed stokes profiles with synthetic ones obtained by unno - rachkovsky solutions @xcite to the polarized radiative transfer equations ( rte ) under the assumption of milne - eddington ( me ) atmosphere @xcite and local thermodynamical equilibrium ( lte ) . however , one can also use this code for generating synthetic stokes profiles for an input model atmosphere . the synthetic profiles are functions of magnetic field strength ( b ) , inclination ( @xmath41 ) , azimuth ( @xmath42 ) , line of sight velocity ( @xmath44 ) , doppler width ( @xmath45 ) , damping constant ( @xmath46 ) , ratio of the center to continuum opacity ( @xmath47 ) , slope of the source function ( @xmath48 ) and the source function ( @xmath49 ) at @xmath50 = 0 . the filling factor is taken as unity . in our profile synthesis only magnetic field parameters b , @xmath41 , @xmath42 are varied while other model parameters are kept same for all pixels . the typical values of other thermodynamical parameters are given in table [ tbl-1 ] . we use the same parameters for all pixels . further , all the physical parameters at each pixel are taken to be constant in the line forming region . however , one must remember that real solar observations have often stokes v area asymmetries @xcite as a result of vertical magnetic and velocity field gradients present in the line forming region . this has not been taken into account in our simulations . a set of stokes profiles with 0.5% and 2.0% noise for a pixel is shown in figure [ fig2 ] . + the wavelength grid used for generating synthetic spectral profiles is same as that of hinode ( sot / sp ) data which are as follows : start wavelength of 6300.89 , spectral sampling 21.5 m / pixel , and 112 spectral samples . we add normally distributed random noise of different levels in the synthetic stokes profiles . typical noise levels in stokes profiles obtained by hinode ( sot / sp ) normal mode scan are of the order of 10@xmath1 of the continuum intensity , i@xmath2 @xcite . we add random noise of 0.5 % of the continuum intensity i@xmath2 to the polarimetric profiles . in addition , we also study the effect of adding a noise of 2.0% level to stokes profiles as a worst case scenario . we add 100 realizations of the noise of the orders mentioned above to each pixel and invert the corresponding 100 noisy profiles using the `` helix '' code . the guess parameters to initialize the inversion are generated by perturbing known values of b , @xmath41 and @xmath42 by 10% . thus after inverting 100 times we get 100 sets of b , @xmath41 & @xmath42 maps for the input b , @xmath41 & @xmath42 values from bipole data . in this way we estimate the spread in the derived field values for various field strengths , inclinations etc . first , the inversion is done without adding any noise in the profiles to check the accuracy of inversion process . we get the results retrieved in this process which are very similar to that of the initial analytical ones . the scatter plot of input field strength , inclination , azimuth against the corresponding retrieved strength , inclination , azimuth after noise addition and inversion is shown in figure [ fig3 ] ( upper panel ) . typical b@xmath38 , b@xmath39 & b@xmath20 maps with different noise levels are shown in the lower panel . as the noise increases b@xmath38 , b@xmath39 & b@xmath20 maps become more grainy . from the plots shown in figure [ fig3 ] we can see that the error in the field strength for a given noise level decreases for strong fields . this is similar to results of @xcite . as the noise increases in the profiles , error in deriving the field strength increases . we find that the error in the field strength determination is @xmath51 for 0.5% noise and @xmath52 for 2% noise in the profiles . inclination shows more noise near 0@xmath43 & 180@xmath43 than at @xmath53 . the error is less even for large noisy profiles for the inclination angles between @xmath54 . the reason for this may be understood in the following way . linear polarization is weaker near 0 and 180@xmath43 inclinations and is therefore more affected by the noise . the azimuth determination has inherent 180@xmath43 ambiguity due to insensitivity of zeeman effect to orientation of transverse fields . thus in order to compare the input and output azimuths we resolve this ambiguity in @xmath55 by comparing it with @xmath56 i.e. , the value of @xmath55 which makes acute angle with @xmath56 has been taken as correct . we can see azimuth values after resolving the ambiguity in this way show good correlation with input azimuth values . some scatter is due to the points where ambiguity was not resolved due to 90@xmath43 difference in @xmath56 and @xmath55 . first , the @xmath19 was calculated from the vector field components derived from the noise free profiles to verify the method of calculating global alpha and also the inversion process . we have used the single polarity to calculate global alpha present in sunspot as discussed in section 3 . we retrieved the same value of @xmath19 as calculated using the initial analytical field components . from the figure [ fig4 ] . we can see that the effect of noise on the field components is not much for the case of 0.5% noise but as the noise in the profiles is increased to 2.0% , the field components specially transverse fields show more uncertainty . the vertical field is comparatively less affected with noise . the scatter plot in figure [ fig4 ] . shows that the inversion gives good correlation to the actual field values . the points with large scatter are due to poor `` signal to noise '' ratio in the simulated profiles . the mean percentage error in the further discussions is given in terms of weighted average of error . we calculate the percentage error in global alpha each time after getting the inverted results , for both the cases when 0.5% and 2.0% ( of i@xmath57 ) noise is added in the profiles , by the following relation : @xmath58 where @xmath59 is calculated global alpha and @xmath19 is the analytical global alpha . the histogram of the results obtained is shown in figure [ fig5 ] . first , we inverted the profiles without adding any noise and calculated @xmath19 from retrieved results to compare it with the ` true ' @xmath19 calculated from the analytically generated vector field components . we get less than 0.002% difference in the both @xmath19 values . for the case of 0.5% noise in polarimetric profiles we get a mean error of 0.3% in the calculation of @xmath19 and error is never more than 1% . thus the calculation of @xmath19 is almost free from the effect of noise in this case . hence , by using data from hinode ( sot / sp ) , one can derive the accurate value of twist present in a sunspot . if 2.0% noise is present in the polarization , then maximum @xmath60 error is obtained . weighted average shows only 1% error . thus the estimation of alpha is not influenced very much even from the data obtained with old and ground based magnetographs . in any event it is unlikely that a realistic error will be large enough to create a change in the sign of @xmath61 . the magnetic energy has been calculated using virial theorem . one form of the general virial theorem @xcite states that for a force - free magnetic field , the magnetic energy contained in a volume v is given by a surface integral over the boundary surface s , @xmath62\cdot { \bf \hat{n}}\ \ ds\ ] ] where @xmath63 is the position vector relative to an arbitrary origin , and n is the normal vector at surface . let us adopt cartesian coordinates , taking as z=0 plane for photosphere . this assumption is reasonable because the size of sunspots are very small compared to the radius of the sun . if we make the further reasonable assumption that the magnetic field strength decreases with distance more rapidly than r@xmath64 whereas a point dipole field falls off as r@xmath1 , then the equation ( 15 ) can be simplified to @xcite @xmath65 where x and y are the horizontal spatial coordinates . b@xmath38 , b@xmath39 & b@xmath20 are the vector magnetic field components . this equation ( 16 ) is referred as the `` magnetic virial theorem '' . thus magnetic energy of an active region can be calculated simply by substituting the derived vector field components into the surface integral of equation ( 16 ) @xcite . magnetic field should be solenoidal and force - free as is the case for our analytical field . so the energy integral is independent of choice of the origin . if all the above conditions are satisfied then the remaining source of uncertainty in the magnetic energy estimation is the errors in the vector field measurements themselves . so , before the virial theorem can be meaningfully applied to the sun , it is necessary first to understand how the errors in the vector field measurements produce errors in the calculated magnetic energies . earlier , the efforts were made to estimate the errors @xcite for magnetographs like marshall space flight center ( msfc ) magnetograph . @xcite constructed a potential field from msfc data and computed its virial magnetic energy . then , they modified the vector field components by introducing random errors in b@xmath66 , b@xmath67 and b@xmath68 and recomputed the energy . they found the two energies differ by 11% . @xcite approached the problem differently . they introduced errors in the polarization measurements from which the field is derived instead of introducing errors to magnetic fields directly . in this way they were able to approximate reality , more closely and were able to include certain type of errors such as crosstalk which were beyond the scope of the treatment by @xcite . they found that the energy uncertainties are likely to exceed 20% for the observations made with the vector magnetographs present at that time ( e.g. msfc ) . here , our approach is very similar to that of @xcite except that we consider full stokes profile measurements to derive the magnetic fields like in the most of the recent vector magnetographs e.g. , hinode ( sot / sp ) , svm - uso etc . as mentioned earlier . we begin with an analytical field , determine polarization signal as explained in earlier parts , introduce the random noise of certain known levels ( 0.5% & 2.0% of i@xmath57 ) in the polarization profiles , infer an ` observed ' magnetic field after doing the inversion of the noisy profiles , compute an ` observed ' magnetic energy from the ` observed ' field and then compare this energy with the energy of the ` true ' magnetic field . the percentage error is calculated from the following expression : @xmath69 where e@xmath70 is ` observed ' energy and e@xmath3 is ` true ' energy . all the above processes have been described in detail in section 4 . figure [ fig6 ] shows the uncertainty estimated in the calculation of magnetic energy in two cases when error in the polarimetric profiles is 0.5% and 2.0% of i@xmath2 . needless to say , we first checked the procedure by calculating the magnetic energy from the vector fields derived from inverted results with no noise in the profiles . we found the same energy as calculated from the initial analytical fields . we can see that the magnetic energy can be calculated with a very good accuracy when less noise is present in the polarization as is observed in the modern telescopes like hinode ( sot / sp ) for which very small ( of the order of 10@xmath1 of i@xmath57 ) noise is expected in profiles . we find that a mean of 0.5% and maximum up to 2% error is possible in the calculation of magnetic energy with such data . so , the magnetic energy calculated from the hinode data will be very accurate provided the force - free field condition is satisfied . the error in the determination of magnetic energy increases for larger levels of noise . in the case of high noise in profiles ( e.g. 2.0% of i@xmath2 ) the energy estimation is very much vulnerable to the inaccuracies of the field values . we replaced the inverted value of the field parameters with the analytical value wherever the inverted values deviated by more than 50% of the ` true ' values . we then get the result shown in the right panel . we can see that the error is very small even in this case . the mean value of error is @xmath71 . we have discussed the direct method of estimating @xmath61 from vector magnetograms using the @xmath72 moment of minimization . the higher order moments also hold promise for generating an index for predicting the flare productivity in active regions . the global value of twist of an active region can be measured with a very good accuracy by calculating @xmath61 . accurate value of twist can be obtained even if one polarity of a bipole is observed . the magnetic energy calculation is very accurate as seen from our results . very less error ( approximately 0.5% ) is seen in magnetic energy with 0.5% noise in the profiles . thus we conclude that the magnetic energy can be estimated with very good accuracy using the data obtained from modern telescopes like hinode ( sot / sp ) . this gives us the means to look for magnetic energy changes released in weak c - class flares which release radiant energy of the order of 10@xmath73 ergs ( see appendix - c ) , thereby improving the statistics . these energy estimates are however subject to the condition that the photospheric magnetic field is force - free , a condition which is not always met with . we must then obtain the energy estimates using vector magnetograms observed at higher atmospheric layers where the magnetic field is force - free @xcite . the 180@xmath43 azimuthal ambiguity ( aa ) is another source of error for determining parameters like @xmath61 and magnetic free energy in real sunspot observations . the smaller the polarimetric noise , the smaller is the uncertainty in azimuth determination , thereby allowing us to extend the range of the acute angle method used in our analysis . on the other hand it is difficult to predict the level of uncertainty produced by aa . influence of aa is felt more at highly sheared regions which will anyway deviate from the global alpha value . thus , avoiding such pixels will improve determination of @xmath61 . magnetic energy calculation at such pixels could be done by comparing energy estimates obtained by ` flipping ' the azimuths and choosing the mean of the smallest and the largest estimate of the energy . here we assume that half the number of pixels have the true azimuth . this is the best one can do for a problem that really has no theoretical solution allowed by the zeeman effect ( but see also , @xcite and references therein ) . observational techniques such as use of chromospheric chirality @xcite or use of magnetograms observed from different viewing angles could perhaps resolve the aa . patches of both signs of alpha can be present in a single sunspot @xcite . in those cases the physical meaning of @xmath61 becomes unclear . efforts are needed to understand the origin of such complex variation of @xmath0 in a sunspot . real sunspots show filamentary structures . if this structure is accompanied by local variations of @xmath0 , then does the global @xmath0 result from correlations in the local @xmath0 values ? or , are the small scale variations due to a turbulent cascade from the large scale features ? the answers to these questions are beyond the scope of our present study . modeling sunspots with such complex fine structures is a great challenge . however , we plan to address the question of fine structure of twists in real sunspots observed from hinode ( sot / sp ) , in our forthcoming study . for the present , we demonstrate that the global twist present in an active region can be accurately measured without ambiguity in its sign . furthermore , the high accuracy of magnetic energy estimation that can be obtained using data from modern instruments will improve the probability for detecting the flare related changes in the magnetic energy of active regions . + + we thank professor e. n. parker for discussion leading to our understanding about the physical meaning of @xmath0 parameter during his visit to udaipur solar observatory in november 2007 . we also thank him for looking at an earlier draft of the manuscript and for making valuable comments to improve it . one of us ( jayant joshi ) acknowledge financial support under isro / cawses - india programme . we thank dr . a. lagg for providing the helix code . we are grateful for the valuable suggestions and comments of the referee which have significantly improved the manuscript . + ( derived from the discussions with professor eugene n. parker during his visit to udaipur solar observatory ) taking surface integral on both sides of eqn . ( 2 ) , we get @xmath74 in the cylindrical coordinate we can write eqn.(a2 ) as @xmath75 where z and @xmath76 are axial and radial distances from origin , respectively the equation of field lines in cylindrical coordinates is given as : @xmath77 or , @xmath78 using eqns . ( a3 ) & ( a5 ) , we get @xmath79 from equation ( a6 ) it is clear that the @xmath0 gives twice the degree of twist per unit axial length . if we take one complete rotation of flux tube i.e. , @xmath80 , and loop length @xmath81 , then @xmath82 comes out of the order of approximately @xmath83 per meter . ( 2 ) can be written as @xmath84 giving vector potential in terms of scalar potential @xmath85 as @xmath86 which is valid only for constant @xmath0 . + using this relation in eqn.(1 ) , we get magnetic helicity as @xmath87 second term in the right hand side of eqn . ( b3 ) can be written as , @xmath88 ( from gauss divergence theorem ) which is equal to zero for a closed volume where magnetic field does not cross the volume boundary ( @xmath89 ) provided that @xmath85 remains finite on the surface . therefore , we get magnetic helicity in terms of @xmath0 as @xmath90 which shows that the force free parameter @xmath0 has the same sign as that of the magnetic helicity . however , if @xmath91 , then the contribution of the second term in eqn . ( b3 ) remains unspecified . thus it is not correct to use alpha to determine the sign of magnetic helicity for the half space above the photosphere since @xmath91 at the photosphere . with the simplifying assumption that all classes of soft x - ray flares have a typical duration of 16 min @xcite , we can see that the energy released in the different classes of flares will be proportional to their peak power . since x - class flares typically release radiant energy of the order of @xmath92 ergs @xcite , therefore m - class , c - class , b - class and a - class flares will release radiant energy of the order of respectively 10@xmath93 , 10@xmath73 , 10@xmath94 and 10@xmath95 ergs . lr doppler velocity , v@xmath96 ( ms@xmath97 ) & 0 + doppler width , v@xmath98 ( m ) & 20 + ratio of center to continuum opacity , @xmath99 & 20 + source function , s@xmath100 & 0.001 + slope of the source function , s@xmath101 & 1.0 + damping constant , @xmath46 & 1.4 + , d. f. , et al . 1992 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 1746 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . d. h. goldstein & r. a. chipman , 22 , h. p. , harvey , j. w. , henney , c. j. , hill , f. , & keller , c. u. 2002 , in esa special publication , vol . 505 , solmag 2002 . proceedings of the magnetic coupling of the solar atmosphere euroconference , ed . h. sawaya - lacoste , 15 , c. u. , harvey , j. w. , & giampapa , m. s. 2003 , in presented at the society of photo - optical instrumentation engineers ( spie ) conference , vol . 4853 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . s. l. keil & s. v. avakyan , 194 , k. , et al . 2004 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol . 5171 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . s. fineschi & m. a. gummin , 207	the force - free parameter @xmath0 , also known as helicity parameter or twist parameter , bears the same sign as the magnetic helicity under some restrictive conditions . the single global value of @xmath0 for a whole active region gives the degree of twist per unit axial length . we investigate the effect of polarimetric noise on the calculation of global @xmath0 value and magnetic energy of an analytical bipole . the analytical bipole has been generated using the force - free field approximation with a known value of constant @xmath0 and magnetic energy . the magnetic parameters obtained from the analytical bipole are used to generate stokes profiles from the unno - rachkovsky solutions for polarized radiative transfer equations . then we add random noise of the order of 10@xmath1 of the continuum intensity ( i@xmath2 ) in these profiles to simulate the real profiles obtained by modern spectropolarimeters like hinode ( sot / sp ) , svm ( uso ) , asp , dlsp , polis , solis etc . these noisy profiles are then inverted using a milne - eddington inversion code to retrieve the magnetic parameters . hundred realizations of this process of adding random noise and polarimetric inversion is repeated to study the distribution of error in global @xmath0 and magnetic energy values . the results show that : ( 1 ) . the sign of @xmath0 is not influenced by polarimetric noise and very accurate values of global twist can be calculated , and ( 2 ) . accurate estimation of magnetic energy with uncertainty as low as 0.5% is possible under the force - free condition .
the work of professor sabina on modelling of composite media and , in particular , piezoelectric heterogeneous solids has made a significant impact and addressed highly challenging problems of wave propagation and homogenisation approximations , which are extremely important in a wide range of applications in physics , mechanics and engineering . the classical work by sabina and willis @xcite on scattering of sh waves by a rough half - space introduced a new modelling approach to important applications in problems of geophysics . widely used self - consistent analysis of wave propagation in composites was developed by sabina and willis in @xcite . significant results on homogenisation of magneto - electro - elastic composites and magneto - electric coupling have been published in @xcite and @xcite . piezoelectric composites that are made of piezoelectric ceramics are widely used in many applications in physics and engineering @xcite . many of these structures are inhomogeneous , _ i.e. _ they are made by using two or more types of materials . furthermore , piezoelectric materials are anisotropic . however , in certain types of polarised piezoelectric ceramics , symmetries around the polarisation axis can be found . in this case the mathematical treatment of piezoelectricity greatly simplifies and the formulation can be split in a 2d in - plane ( ip ) problem and a 1d out - of - plane ( op ) problem . many op problems are mathematically simple and can be often solved analytically thus clearly showing the physics involved . nevertheless op problems in piezoelectric materials possess effects of great technological interest as outlined in @xcite . + the analysis of piezoelectric structures is often based on the quasi - static approximation @xcite . as a result , in this theory , although the mechanical equations are dynamic , the electromagnetic equations are static , and the electric field and the magnetic field are not dynamically coupled . however , when electromagnetic waves are involved , the complete set of maxwell s equations has to be taken into account , coupled to the mechanical equations of motion @xcite . as an example , a.n . @xcite discussed the role of electromagnetic waves in the reflection of acoustic waves at the interface between two semi - infinite piezoelectric materials . at quasi - normal incidence , _ i.e. _ for an angle of incidence @xmath0 , where @xmath1 and @xmath2 are the typical speed of sound and light respectively , they found that according to the electromagnetic description , the acoustic wave must suffer total reflection while the quasi - electrostatic approximation predicts almost total transmission . + photonic and phononic crystals made of piezoelectric materials were discussed in @xcite . in particular , the article @xcite presents a model for the transmission problem in stratified media , emphasising applications in acoustics . the effects of electromechanical coupling in models of surface acoustic waves were discussed in @xcite . surface and bulk acoustic waves in two - dimensional phononic crystals were studied in ref . a plane - wave expansion method to study spectral problems in phononic piezoelectric crystals was presented in ref . sabina and movchan @xcite discussed the role of the electromechanical coupling on the dispersion properties of in - plane bloch waves within one- and two - dimensional phononic crystals . @xcite analytically derived and solved the dispersion equation of op bloch waves in 1d layered piezoelectric materials . the direction of polarisations , the electromechanical coupling and the aspect ratio of the unit cell have been investigated as potential tuning parameters for the dispersion properties . this paper analyses a class of spectral problems occurring in layered and doubly - periodic piezoelectric structures . the scattering by a layered piezoelectric finite stack is analysed first . we show that the dynamic response of such a structure depends on the frequency of the incident wave . in addition , the occurrence of piezoelectrically driven transmission resonances is analysed . we then proceed further with a more challenging setting within a 2d phononic crystal , consisting of a rectangular checkerboard whose unit cell contains piezoelectric materials . the article is organised as follows . in sec . [ sec : governing_equations ] we review the equations which govern the propagation of waves in a _ 6 mm _ symmetry class bulk piezoelectric material . the general 3d vector elasticity problem decouples into two problems : an in - plane vector problem and an out - of - plane problem @xcite . an analytical formalism based on the fully dynamic coupled theory @xcite is introduced in order to efficiently describe anti - plane shear displacement waves and the associated electromagnetic counterpart in a _ 6 mm _ piezoelectric material . the governing equations of the in - plane vector problem are outlined as well . in sec . [ sec:1d - periodic ] we study dispersion properties for elastic op and ip waves in piezoelectric phononic crystals . in sec . [ sec:1d - finite ] the finite stack of piezoelectric materials together with the governing model is introduced . several analytical results are presented for the transmission and reflection coefficients . particular attention is given to the occurrence of transmission resonances and to the pivotal role played by the piezoelectric effect . [ sec:2d - checkerboard ] the 2d periodic piezoelectric checkerboard model is introduced and its numerical description and implementation are discussed . the dispersion surfaces for 2d plane - strain elastic bloch waves exhibit a rich phenomenology . the low frequency regime is dominated by the in - plane pressure and shear modes which show dynamic anisotropy in the slowness curves . from a physical point of view localisation phenomena are likely to occur within checkerboard - like finite clusters along the preferential direction prescribed by the aforementioned slowness curves . at intermediate frequencies , it is shown that the dynamics of bloch waves is influenced by the dirac cone - like dispersion surfaces . at higher frequencies a total band gap is observed , which is important for filtering applications . in sec . [ sec : conclusions ] we draw our main conclusions and outline possible directions for future research . in this section we review several results concerning the propagation of elastic waves within bulk piezoelectric materials . we assume that the spontaneous electric polarisation of such a material points along the @xmath3 components of a given cartesian coordinate system . this choice specifies a plane of symmetry identified by the @xmath4 and @xmath5 basis vectors , and an out - of - plane direction @xmath3 . the in - plane components of the displacement field are fully decoupled from the out - of - plane shear component @xmath6 . + from classical electrodynamics and elasticity , the displacement and the electromagnetic vector fields are described by maxwell s equations and the second newton law . the second newton s law in its stress - displacement for is @xmath7 where @xmath8 is the stress tensor , @xmath9 is the density which is assumed to be uniform and @xmath10 is the elastic displacement vector field . the dynamics of the electromagnetic field can be taken into account through maxwell s equations @xmath11 where @xmath12 and @xmath13 are the electric displacement and electric field , @xmath14 and @xmath15 are the magnetic induction and magnetic fields . according to standard linear theory of piezoelectric phenomena ( see chap . 8 of auld @xcite for a detailed discussion ) , eqs ( [ eq : newton - law ] ) and ( [ eq : maxwell ] ) are coupled through the following _ stress - charge constituitive relations _ @xmath16 where we have introduced the strain field @xmath17 , the stiffness tensor @xmath18 , the piezoelectric tensor @xmath19 and the dielectric permittivity matrix @xmath20 . + we assume that the piezoelectric materials under consideration are orthotropic . for the sake of definiteness we assume that the materials which fill the unit cell belong to the @xmath21 symmetry class . as already mentioned , the 6-axis is oriented along the out - of - plane direction @xmath22 . given the problem which we have described , the electro - elastic parameters which appear in eq . ( [ eq : constitutive ] ) can be written in the following form @xcite @xmath23 capital indices in eqs ( [ eq : piezo - parameters ] ) denote the pairs of cartesian indices @xmath24 . + we now pass to the detailed investigation of the scalar ( out - of - plane ) and vector ( in - plane ) elasticity problems . ( [ eq : piezo - parameters ] ) , eqs ( [ eq : newton - law ] ) , ( [ eq : maxwell ] ) and ( [ eq : constitutive ] ) become respectively @xmath25 @xmath26 @xmath27 a plane elastic wave which propagates in a piezoelectric material is coupled to an electromagnetic wave , as is evident from the constitutive relations ( [ eq : constitutive ] ) . in the op shear configuration , the displacement @xmath6 and the magnetic field @xmath28 decouple . the magnetic field propagates at the speed of light and the displacement field propagates at piezoelectric - stiffened shear wave speed @xcite , _ i.e. _ @xmath29 the constituitive equations show that the electric field and the stress tensor depend on both the elastic displacement and the magnetic field . appropriate boundary conditions are set at the interface between different piezoelectric . + the following vector function is introduced @xmath30 in eq . ( [ eq : solution_phys_quantities ] ) , we implicitly assume plane - wave dependence of the physical fields on the spatial variable @xmath31 and on the time @xmath32 , with wave vector @xmath33 and frequency @xmath34 , respectively . using eqs ( [ eq : newton - law-6 mm ] ) , ( [ eq : maxwell-6 mm ] ) and ( [ eq : constitutive-6 mm ] ) we deduce the differential equation @xmath35 with @xmath36 and @xmath37 introduced in eq . ( [ eq : solution_phys_quantities]).the eigenvalues of @xmath38 are @xmath39 where @xmath40 and corresponding eigenvectors which are solutions of eq . ( [ eq : ode ] ) are @xmath41 the energy flux parallel to the @xmath4 direction is @xcite @xmath42 where the explicit form of @xmath43 is given in eq . ( [ eq : solution_phys_quantities ] ) and @xmath44 the eigen - modes are orthonormal with respect to the scalar product defined in eq . ( [ eq : ode_eigenmodes ] ) , _ i.e. _ @xmath45 on the other hand , denoting by @xmath46 the set of propagating modes @xmath47 @llllr@ & @xmath48 $ ] & @xmath49 $ ] & @xmath50 $ ] & @xmath51 $ ] + pzt & 2.56 & 12.7 & 646 & 7.6 + + @xmath52 & 4.3 & 11.6 & 1.264 & 5.7 + @llr@ & @xmath53 $ ] & @xmath54 $ ] + pzt & 2.58 & 11.10 + + @xmath52 & 43.30 & 250.91 + by requiring that both the wave - vectors in eq . ( [ eq : x - wavevectors - modulii ] ) are real , one can identify threshold frequencies above which electromagnetic and/or elastic waves _ can _ propagate within a given piezoelectric material . the elastic and electromagnetic threshold frequencies are respectively @xmath55 and table [ tab : threshold_frequencies ] shows several threshold frequencies ( see eq . ( [ eq : threshold_frequencies ] ) ) for different materials at @xmath56 . the smallest threshold elastic and electromagnetic frequency is the one associated with @xmath57 . therefore pzt will be chosen as an hosting material for the slabs . in addition , two frequency regimes naturally arise from eq . ( [ eq : threshold_frequencies ] ) . for frequencies such that @xmath58 only elastic displacement waves can propagate in the stack . we call this range _ elastic frequency regime_. for frequencies such that @xmath59 _ both _ elastic and electromagnetic waves can propagate in the stack . we will refer to this interval as the _ electromagnetic frequency regime_. the notations @xmath60 and @xmath61 refer to pzt-4 material and stand for elastic and electromagnetic threshold frequency , respectively . we outline now the governing equations of the in - plane vector elasticity . when the displacement field belongs to the @xmath62 , the newton law implies @xmath63 the maxwell equations become @xmath64 the constituitive equations become @xmath65 unit cell for the periodic piezoelectric structure or building block for the finite stack.,scaledwidth=60.0% ] a periodic piezoelectric structure whose unit - cell is depicted in fig . [ fig : system_unit_cell ] is considered in this section . sabina and movchan @xcite investigated the effect of the interfaces in a similar structure . they pointed out that in the case in which the piezoelectric polarisation is directed along the @xmath4 direction of fig . [ fig : system_unit_cell ] , the effect of piezoelectricity on the dispersion diagram of bloch waves is very small . when the polarisation vectors still belong to the @xmath4 axis but point in alternate directions , several partial gaps open . the formalism presented in sec . [ sec : governing_equations ] is compared with @xcite . in addition , we study the case in which the piezoelectric polarizations point in the out - of - plane direction but the elastic displacement is an in - plane vector . this analysis exhausts the 3d vector elasticity problem in 1d piezoelectric periodic structures where the piezoelectric polarization points in the out - of - plane direction . in both panels the full red lines are eigenfrequencies for the out - of - plane shear displacement waves while the blue dashed lines represent the in - plane shear and pressure modes . the plane here considered is the @xmath62 plane of fig . [ fig : system_unit_cell ] . with reference to the same picture , we fixed the parameters to be @xmath66 , @xmath67 . region ( 1 ) of the unit cell is occupied by pzt-4 while regions ( 2 ) and ( 3 ) are occupied by the @xmath52 . panel ( a ) has been obtained at normal incidence , _ i.e. _ setting @xmath68 while panel ( b ) has been obtained using @xmath56.,title="fig:",scaledwidth=50.0% ] in both panels the full red lines are eigenfrequencies for the out - of - plane shear displacement waves while the blue dashed lines represent the in - plane shear and pressure modes . the plane here considered is the @xmath62 plane of fig . [ fig : system_unit_cell ] . with reference to the same picture , we fixed the parameters to be @xmath66 , @xmath67 . region ( 1 ) of the unit cell is occupied by pzt-4 while regions ( 2 ) and ( 3 ) are occupied by the @xmath52 . panel ( a ) has been obtained at normal incidence , _ i.e. _ setting @xmath68 while panel ( b ) has been obtained using @xmath56.,title="fig:",scaledwidth=50.0% ] we begin our analysis by addressing the problem in which the displacement is a pure out - of - plane displacement vector . the vector of physical quantities within the @xmath69-th material in the unit cell has the form @xmath70 where @xmath71 are the eigenmodes given in eq . ( [ eq : ode_eigenmodes ] ) and @xmath72 is the amplitude of the @xmath73th eigenmodes . the continuity conditions within the unit cell are @xmath74 and the bloch - floquet conditions are @xmath75 for any @xmath76 from the elementary cell . the twelve equations for the twelve unknown wave amplitudes have non - trivial solutions if @xcite @xmath77 where we have introduced @xmath78 @xmath79 \nonumber \\ & - & 2 g \cos(bq_1 ) \cos(aq_2 ) \sin(br_1 ) \sin(ar_2 ) + qg \sin(bq_1 ) \sin(aq_2 ) \sin(br_1 ) \sin(ar_2 ) \nonumber \\ & + & p^4k_1k_2 \sin(bq_1)\sin(aq_2)\sin(br_1)\sin(ar_2),\end{aligned}\ ] ] and @xmath80 + the transmission conditions are discussed in appendix a. + we will assume that the polarisation vector lie along the @xmath3 direction of fig . [ fig : system_unit_cell ] as done in ref . @xcite . however we want to generalise those results , by allowing the structure to vibrate not only in the out of plane shear direction but also and at the same time , in the in - plane directions . + panels ( a ) and ( b ) in fig . [ fig:1d_plane_strain ] show the dispersion diagrams for bloch waves in a piezoelectric periodic structure with unit cell of length @xmath81 and made of pzt-4 and @xmath82 . panel ( a ) refers to normal incidence while panel ( b ) to oblique incidence ( @xmath56 ) . the solid lines are the dispersion curves for op bloch displacement waves while the dashed lines are the ip modes . in fig . [ fig:1d_plane_strain ] in addition , we observe formation of band gaps . in this section we study the reflection and transmission properties of a finite stack of binary piezoelectric cells . we restrict our analysis to the materials whose relevant parameters are listed in table [ tab : parameters ] . each binary cell is assumed to be composed of _ two _ different piezoelectric layers as depicted in fig . [ fig : system_unit_cell ] . the anti - plane shear configuration for the displacement field will be here assumed . in this section we will refer to this cell as _ building block _ of the finite stack to clearly distinguish it from the unit cell of the periodic structure . the total number of building blocks will be denoted as @xmath83 . + the interest in studying the problem stated above is twofold . first of all , when @xmath83 is sufficiently large , the reflection spectrum encapsulates some features of the well known periodic problem @xcite . in fact , the _ total reflection _ regions should precisely coincide with the stop band for bloch waves . the second and most important source of interest comes from the fact that a finite @xmath83 system exhibits a richer spectrum of phenomena . perhaps one of the most attractive are the resonances in the transmitted elastic and/or electromagnetic spectrum . this challenge is made more intriguing because of the piezoelectric effect , which has been demonstrated to have a substantial role in the determination of the band gaps width and position in periodic 1d stacks @xcite . it is natural to ask ourselves whether these transmission resonances do exist and what is the role of piezoelectric effect in the determination of their the position . + schematic representation of a stack of piezoelectric binary cells . panel ( a ) shows the stack with the corresponding incident ( i ) reflected ( r ) and transmitted ( t ) waves . in panel ( b ) is schematically shown the multiple reflection method which is at the basis of the recurrence procedure approach.,title="fig:",scaledwidth=50.0% ] schematic representation of a stack of piezoelectric binary cells . panel ( a ) shows the stack with the corresponding incident ( i ) reflected ( r ) and transmitted ( t ) waves . in panel ( b ) is schematically shown the multiple reflection method which is at the basis of the recurrence procedure approach.,title="fig:",scaledwidth=50.0% ] @xmath84 before studying the scattering properties of the stack , we focus on the reflection and transmission by a planar interface @xmath85 separating two semi - infinite piezoelectric materials @xcite . the material on the left and on the right of the interface will be denoted respectively by ( 2 ) and ( 1 ) . the materials are assumed to be perfectly matched at the interface . the continuity of fields at the boundary can be written in terms of the eigenmodes given in eq . ( [ eq : ode_eigenmodes ] ) . to distinguish material ( 1 ) from material ( 2 ) the notation @xmath86 is introduced , with @xmath87 the eigenvalue indices and @xmath88 the material indices . specifically for waves incident from the left @xmath89 and for waves incident from the right @xmath90 solving the systems associated with eqs ( [ eq : boundary - conditions - left ] ) and ( [ eq : boundary - conditions - right ] ) gives respectively @xmath91 and @xmath92 which are the reflection and transmission matrices which act on the two - dimensional subspace generated by the `` acoustic '' ( a ) and electromagnetic ( e ) eigenmodes which are incident from the left . similar scattering matrices for waves incident from the right can be derived using eqs ( [ eq : boundary - conditions - right ] ) . the proposed approach is equally valid for the case of heterogeneous structured layers . the representations of the entries of matrices ( [ eq : r&t - left ] ) and ( [ eq : r&t - right ] ) are given in appendix b. + let us first focus on the calculation of the scattering matrices for a single layer of piezoelectric material ( 1 ) immersed in a material ( 2 ) . the problem of waves moving from the left @xmath93 is discerned from that associated with waves moving from the right @xmath94 . in addition , the following notation is here introduced : the scattering matrices in eqs ( [ eq : r&t - left ] ) and ( [ eq : r&t - right ] ) evaluated at the boundary @xmath95 are non - primed ; the one evaluated at the boundary @xmath96 are primed . + the reflection and transmission matrices can be evaluated by summing up the contributions coming from the multiple bouncing between the interfaces at @xmath95 and @xmath96 . for waves incident from the left @xmath97 @xmath98 similarly the scattering matrices for waves incident from the right are @xmath99 @xmath100 the reflection and transmission coefficient associated with an arbitrary and finite number of building blocks , as the one depicted in fig . [ fig : system_finite ] panel ( a ) , can be calculated using a recurrence procedure @xcite-@xcite . the thick vertical line on the right hand side of fig . [ fig : system_finite ] ( b ) , schematically denotes @xmath101 layers made of material ( 1 ) , immersed in a material ( 2 ) environment . the associated @xmath101-stack scattering matrix for waves incident from the left will be denoted by @xmath102 and @xmath103 . the thin vertical line represent a single layer added to the left of the @xmath101-stack . the associated scattering matrices are given in eqs ( [ eq : r&t - left ] ) and ( [ eq : r&t - right ] ) . similarly to what we already done for the single layer in the previous section , the scattering matrices associated with a stack comprising @xmath104-layers are @xmath105 @xmath106 with @xmath107 the recurrence matrix relations above enable to evaluate the scattering matrices for a stack of ( s+1 ) layers , assuming that the @xmath101-stack matrices are known . since the output of the recurrence procedure has phase - shift origin at @xmath95 , before using the @xmath101-stack matrices in the next recurrence step , the phase origin must be shifted to @xmath108 . this is taken into account by the _ propagation matrix _ in eq . ( [ eq : propagation_matrix ] ) which appears in eqs ( [ eq : reflection_matrix_stack ] ) and ( [ eq : transmission_matrix_stack ] ) . the seed @xmath109 of this recurrence procedure gives the result for the @xmath110 double layer system . + transmittance ( blue dashed line ) and reflectance ( red solid line ) for a single @xmath52 slab in a pzt environment in the elastic frequency regime . we fix the parameters of the slab to @xmath111 and @xmath112 . panel ( a ) has been obtained at normal incidence ( @xmath68 ) while panel ( b ) setting @xmath56 . note that in panel ( b ) , the horizontal axis restricts to @xmath113 to avoid the non - physical region of frequencies.,title="fig:",scaledwidth=50.0% ] transmittance ( blue dashed line ) and reflectance ( red solid line ) for a single @xmath52 slab in a pzt environment in the elastic frequency regime . we fix the parameters of the slab to @xmath111 and @xmath112 . panel ( a ) has been obtained at normal incidence ( @xmath68 ) while panel ( b ) setting @xmath56 . note that in panel ( b ) , the horizontal axis restricts to @xmath113 to avoid the non - physical region of frequencies.,title="fig:",scaledwidth=50.0% ] + @xmath84 both panels are transmittance calculations of elastic waves through a piezoelectric slab surrounded by pzt . panel ( a ) shows transmittance curves for several values of @xmath33 , when the piezoelectric effect is ignored ( _ i.e. _ when the piezoelectric couplings @xmath114 in table [ tab : parameters ] are set to zero).panel ( b ) shows transmittance curves for several values of @xmath33 , when the piezoelectric effect is taken into account.,title="fig:",scaledwidth=50.0% ] both panels are transmittance calculations of elastic waves through a piezoelectric slab surrounded by pzt . panel ( a ) shows transmittance curves for several values of @xmath33 , when the piezoelectric effect is ignored ( _ i.e. _ when the piezoelectric couplings @xmath114 in table [ tab : parameters ] are set to zero).panel ( b ) shows transmittance curves for several values of @xmath33 , when the piezoelectric effect is taken into account.,title="fig:",scaledwidth=50.0% ] + @xmath84 [ equation ] reflectance ( red full lines ) and transmittance ( blue dotted lines ) obtained by the recurrence procedures derived in eqs ( [ eq : reflection_matrix_stack ] ) and ( [ eq : transmission_matrix_stack ] ) . the building block has the same parameters used in fig . [ fig : rt_el_single_oblique_normal ] . the thin black line proves the conservation of energy according to eq . ( [ eq : conservation_energy_flux_elr ] ) . the thick black segments represent stop bands for the infinite structure whose unit cell coincides with current building block . panel ( a ) refers to @xmath115 and panel ( b ) @xmath116.,title="fig:",scaledwidth=50.0% ] reflectance ( red full lines ) and transmittance ( blue dotted lines ) obtained by the recurrence procedures derived in eqs ( [ eq : reflection_matrix_stack ] ) and ( [ eq : transmission_matrix_stack ] ) . the building block has the same parameters used in fig . [ fig : rt_el_single_oblique_normal ] . the thin black line proves the conservation of energy according to eq . ( [ eq : conservation_energy_flux_elr ] ) . the thick black segments represent stop bands for the infinite structure whose unit cell coincides with current building block . panel ( a ) refers to @xmath115 and panel ( b ) @xmath116.,title="fig:",scaledwidth=50.0% ] + @xmath84 reflectance ( red solid line ) and transmittance ( blue dashed lines ) at oblique incidence ( @xmath56 ) and @xmath126 . we used the same building block as in fig . [ fig : t_el_single_piezo_no_piezo ] . panel ( a ) refers to an elastic incident wave while panel ( b ) refers to an electromagnetic incident wave . the frequency of the incident waves is in the electromagnetic frequency regime.,title="fig:",scaledwidth=50.0% ] reflectance ( red solid line ) and transmittance ( blue dashed lines ) at oblique incidence ( @xmath56 ) and @xmath126 . we used the same building block as in fig . [ fig : t_el_single_piezo_no_piezo ] . panel ( a ) refers to an elastic incident wave while panel ( b ) refers to an electromagnetic incident wave . the frequency of the incident waves is in the electromagnetic frequency regime.,title="fig:",scaledwidth=50.0% ] + @xmath84 cross - term contributions to the transmittance spectrum at oblique incidence ( @xmath56 ) . the parameters are the same as the one used in fig . [ fig : rt_el_single_oblique_normal ] . panel ( a ) refers to @xmath126 while panel ( b ) to @xmath115 . , title="fig:",scaledwidth=50.0% ] cross - term contributions to the transmittance spectrum at oblique incidence ( @xmath56 ) . the parameters are the same as the one used in fig . [ fig : rt_el_single_oblique_normal ] . panel ( a ) refers to @xmath126 while panel ( b ) to @xmath115 . , title="fig:",scaledwidth=50.0% ] + @xmath84 the left hand side of the figure represents the unit - cell of a checkerboard in real space . the right hand picture is a sketch of the first brillouin zone.,scaledwidth=70.0% ] in this section , we analyse a piezoelectric periodic structure whose geometry of the unit cell is schematically shown in the left - hand side of fig . [ fig : system - checkerboard ] . + the dispersion surfaces representing as a function of the wave - vector @xmath135 within the square first bz ( see the right - hand side of fig . [ fig : system - checkerboard ] ) are obtained . the unit cell in fig . [ fig : system - checkerboard ] is composed by pzt-4 ( blue area ) , a piezoelectric material , and pa66 ( grey area ) , a plastic dielectric material . + several polarisation directions for the piezoelectric material are taken into account . here we would like to remind the general transformation rules which apply to the compliance , coupling and dielectric matrices when the spontaneous polarisation of a piezoelectric material is rotated , _ i.e. _ respectively @xmath136 in eq . ( [ eq : alpha - rot - tensors ] ) we denoted with @xmath137 the three dimensional vector whose components @xmath138 with @xmath139 are three rotation angles about the @xmath4,@xmath5 and @xmath3 , respectively . furthermore , the @xmath140 second rank tensor @xmath141 and the @xmath142 second rank tensor @xmath143 represent an arbitrary @xmath144 rotation in the six dimensional _ reduced indices _ vector space and the three dimensional cartesian vector space , respectively ( see for example @xcite ) . first we analyse a square elementary cell , _ i.e. _ @xmath145 . the checkerboard - like structure is studied within plane strain assumptions . the aim is to investigate how the piezoelectric effect affects dispersion properties . materials used in this simulation are pzt-4 ( lead zirconium titanate ) and pa66 ( polyamide - nylon66-polymer ) whose physical properties are reported in tables [ tab : pzt4 ] and [ tab : pa66 ] , respectively . [ cols="<,<,<,<",options="header " , ] we begin our analysis by assuming that the polarisation vectors of the piezoelectric material within the unit cell are in - plane . we study several emblematic configurations of the polarisation vectors schematically shown in panel ( a ) fig . [ fig : dispersion_diagrams_square ] . the upper left configuration is referred to as `` parallel polarisation '' ; the upper right configuration is referred to as `` perpendicular polarisation '' . the calculations are performed along the path @xmath146 - see lower part of fig . [ fig : dispersion_diagrams_square ] ( a ) . in fig . [ fig : dispersion_diagrams_square ] ( b ) we show the dispersion diagrams corresponding to perpendicular , parallel and absence of polarisation . the high contrast between mechanical properties of the two materials produces a total band gap . for parallel polarisations ( solid red line ) , the dispersion curves along xb do not coincide with the ones along ba . this is due to the fact that the corresponding unit cell has only one symmetry property . specifically , it is symmetric with respect to the diagonal passing through the white squares . in addition to this symmetry , the perpendicular polarisations configuration exhibits @xmath147 rotational symmetry around the centre of its unit cell . finally , the configuration in which the piezoelectric effect has been ignored is symmetric with respect to both diagonals and is @xmath147rotationally invariant around the centre . these symmetries are responsible for the identity of the dispersion curves along xb and ba for unit cells with perpendicular polarisations and absence of polarisations . in addition we notice that piezoelectric effect rules the opening of partial gaps in the @xmath148 and @xmath149 - _ cfr . _ red solid and black dotted lines . the piezoelectric effect is more evident for perpendicular polarisations , but produces a significant changes only in the low frequency part of the dispersion diagram . + @xmath84 in fig . [ fig : mode_shapes ] we show examples of standing waves ( modulus of the displacement vector ) , supported by the checkerboard - like piezoelectric structure . the labels @xmath150,@xmath151 , @xmath152 with @xmath153 are used here to identify the three representative groups of bloch - floquet waves , which correspond to the states marked as 1,2,3 respectively in the dispersion diagram [ fig : dispersion_diagrams_square ] ( b ) . here @xmath150 states correspond to orthogonal polarisation configurations , @xmath151 states correspond to the case of parallel polarisation , whereas the @xmath152 wave forms are obtained in the absence of piezoelectric effect . for the case of lower frequency ( @xmath154 ) , the identified standing waves for non - polarised materials ( @xmath155 ) and for the checkerboard with the orthogonal polarisation ( @xmath156 ) , apart from piezoelectric stiffening resulting in the higher frequency of waveform @xmath156 . however , the case of parallel polarisation ( @xmath157 ) shows significant differences , clearly indicating preferential directions linked to the choice of the polarisation . it is also shown that the piezoelectric effect depends on the frequency range . in particular , the group @xmath158 , @xmath159 , @xmath160 includes waveforms of similar shape for parallel polarisations and absence of piezoelectric effect , whereas the change is more pronounced in the transition from the orthogonal polarisation ( @xmath158 ) to the parallel polarisation ( @xmath159 ) . the higher frequency group ( @xmath161 ) is important , as it shows an example of waveforms concentrated in the non - piezoelectric phase . although the displacement amplitudes in the piezoelectric phase is small , the transmission conditions , in the waveforms @xmath162 and @xmath163 affect the displacement amplitudes , as illustrated in the computations presented in fig . [ fig : mode_shapes ] . panel ( a ) and ( b ) are the dispersion surfaces for a rectangular checkerboard ( @xmath164 ) whose unit cell is made of pzt-4 and pa66 . in panel ( a ) the piezoelectric effect is considered while in panel ( b ) is ignored.,title="fig:",scaledwidth=50.0% ] panel ( a ) and ( b ) are the dispersion surfaces for a rectangular checkerboard ( @xmath164 ) whose unit cell is made of pzt-4 and pa66 . in panel ( a ) the piezoelectric effect is considered while in panel ( b ) is ignored.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) and ( b ) are the slowness curves of the dispersion surfaces in fig . [ fig : pzt / pa66_z_pol_sqrt2 ] panel ( a ) and ( b ) respectively . the slowness contours are made at constant @xmath165 frequency.,title="fig:",scaledwidth=50.0% ] panel ( a ) and ( b ) are the slowness curves of the dispersion surfaces in fig . [ fig : pzt / pa66_z_pol_sqrt2 ] panel ( a ) and ( b ) respectively . the slowness contours are made at constant @xmath165 frequency.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) ans ( b ) are details of fig . [ fig : pzt / pa66_z_pol_sqrt2 ] panel ( a ) and ( b ) respectively for a narrow frequency interval . the _ dirac cones _ in the dispersion surfaces are here clearly visible.,title="fig:",scaledwidth=50.0% ] panel ( a ) ans ( b ) are details of fig . [ fig : pzt / pa66_z_pol_sqrt2 ] panel ( a ) and ( b ) respectively for a narrow frequency interval . the _ dirac cones _ in the dispersion surfaces are here clearly visible.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) and ( b ) are slowness curves corresponding to the dispersion surface in panel ( a ) of fig . [ fig : pzt / pa66_z_pol_sqrt2 ] . panel ( a ) is obtained at @xmath166 and panel ( b ) at @xmath167 . the point - wise symbols within the two panels represent an approximation for the location of the dirac points.,title="fig:",scaledwidth=50.0% ] panel ( a ) and ( b ) are slowness curves corresponding to the dispersion surface in panel ( a ) of fig . [ fig : pzt / pa66_z_pol_sqrt2 ] . panel ( a ) is obtained at @xmath166 and panel ( b ) at @xmath167 . the point - wise symbols within the two panels represent an approximation for the location of the dirac points.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) is the dispersion surface for a rectangular checkerboard ( @xmath168 ) whose unit cell is made of pzt-4 @xmath3-polarised and pa66 . radian frequencies are given in rad / s . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath165.,title="fig:",scaledwidth=50.0% ] panel ( a ) is the dispersion surface for a rectangular checkerboard ( @xmath168 ) whose unit cell is made of pzt-4 @xmath3-polarised and pa66 . radian frequencies are given in rad / s . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath165.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) is the dispersion surface for a square checkerboard ( @xmath169 ) whose unit cell is made of pzt-4 @xmath3-polarised and pa66 . radian frequencies are given in @xmath170 . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath171.,title="fig:",scaledwidth=50.0% ] panel ( a ) is the dispersion surface for a square checkerboard ( @xmath169 ) whose unit cell is made of pzt-4 @xmath3-polarised and pa66 . radian frequencies are given in @xmath170 . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath171.,title="fig:",scaledwidth=50.0% ] + @xmath84 panel ( a ) is the dispersion surface for a rectangular checkerboard ( @xmath172 ) whose unit cell is made of pzt-4 @xmath4-polarised and pa66 . radian frequencies are given in @xmath170 . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath173.,title="fig:",scaledwidth=50.0% ] panel ( a ) is the dispersion surface for a rectangular checkerboard ( @xmath172 ) whose unit cell is made of pzt-4 @xmath4-polarised and pa66 . radian frequencies are given in @xmath170 . panel ( b ) is the slowness contour of panel ( a ) calculated at @xmath173.,title="fig:",scaledwidth=50.0% ] + @xmath84 we next assume that the piezoelectric materials within the the elementary cell are polarised in the out - of - plane direction . in fig . [ fig : pzt / pa66_z_pol_sqrt2 ] we calculate the dispersion surfaces for a periodic checkerboard whose unit cell is made of pa66 and pzt-4 . the side of the unit cell is @xmath164 . in panel ( a ) the full piezoelectric tensor is taken into account while in panel ( b ) the piezoelectric effect is absent . in fig . [ fig : pzt / pa66_z_pol_sqrt2_da ] ( a ) and ( b ) we show the slowness contour at @xmath171 corresponding to fig . [ fig : pzt / pa66_z_pol_sqrt2 ] ( a ) and ( b ) , respectively . these diagrams allow us to observe the dynamic anisotropy and the associated preferential directions for waves propagating in a piezoelectric checkerboard . suppression of the piezoelectric effect ( see panel ( b ) ) dramatically affects the slowness contour . in fact the preferential directions are now altered with respect to the full piezoelectric coupling case ( see panel ( a ) ) . figs [ fig : dirac_cones ] correspond to figs [ fig : pzt / pa66_z_pol_sqrt2 ] at higher frequencies . in panel ( a ) we used the full piezoelectric coupling while in panel ( b ) we suppressed the piezoelectric tensor . figs [ fig : dirac_cones ] show that dirac cones are present in checkerboard structures . furthermore , the position of the dirac point depends on the piezoelectric coupling . [ fig : dirac_cones_da ] are slowness contour of fig . [ fig : pzt / pa66_z_pol_sqrt2 ] ( a ) at @xmath174 and @xmath167 , respectively . this diagrams show an approximation for the locations of dirac points - see point - wise symbols within the panels . in fig . [ fig : pzt / pa66_z_pol_sqrt3 ] the aspect ratio of the checkerboard s unit cell is varied with respect to the one used in fig . [ fig : pzt / pa66_z_pol_sqrt2 ] . we here use @xmath175 while the materials are unchanged . panel ( a ) is the dispersion surface within the first brillouin zone of the checkerboard and panel ( b ) is the corresponding slowness contour . similar features can be observed in fig . [ fig : pzt / pa66_z_pol_square ] where a square unit cell with @xmath176 is used . + finally , in fig . [ fig : pzt / pa66_x_pol_sqrt2 ] we represent the dispersion surface and the corresponding slowness contour with the same materials and aspect ratio as fig . [ fig : pzt / pa66_z_pol_sqrt2 ] but with the piezoelectric polarisation pointing in the @xmath4 direction of fig . [ fig : system - checkerboard ] panel ( a ) . this brings quantitative differences to the dispersion surface . qualitatively , a preferential direction disappear from the slowness contour with respect to the one calculated with a @xmath177polarised material ( see fig . [ fig : pzt / pa66_z_pol_sqrt2_da ] panel ( a ) ) . in this article we have investigated several phenomena which occur in one- and two - dimensional piezoelectric structure . for the one - dimensional periodic piezoelectric structure we derived the bloch waves dispersion diagrams ( see fig . [ fig:1d_plane_strain ] ) for the full vector problem of elasticity . by assuming that the the polarization of the hortotropic @xmath21 piezoelectric materials is parallel to the interfaces , we compared the in - plane shear and pressure frequencies with the out - of - plane shear results @xcite . by using a recurrence procedure approach , we derived the transmission and reflection matrix for the out - of - plane shear displacement waves . moreover we identified a threshold frequency , linearly dependent on the parallel - to - interface component of the wave vector , which allowed us to distinguish two frequency regimes : below threshold ( elastic frequency regime ) only elastic waves can propagate while above threshold ( electromagnetic frequency regime ) both elastic and electromagnetic waves can propagate . in the latter regime , given an elastic ( electromagnetic ) incident wave an electromagnetic ( elastic ) response develops at the same frequency . the cross - term contributions to the propagating energy are driven by the piezoelectric effect and shows transmission resonance in the propagation spectrum ( see panel ( b ) fig . [ fig : t_cross_terms ] ) . it would be of interest to investigate the effect of an external electric field on the transmission spectrum . we plan to investigate as well different type of boundary conditions between piezoelectric layers , such as shortened boundary conditions . we studied the in - plane vector elasticity of a checkerboard - like periodic piezoelectric structure . the dispersion surfaces for bloch waves have been numerically evaluated and several dynamical effects have been clearly identified . in the low frequency regime the slowness curves show dynamic anysotropy , at intermediate frequencies the dispersion surfaces are dominated by dirac cones and in the high - frequency regime an absolute band gap develops . these three features of the dispersion surfaces which coexist in the same piezoelectric structure depend strongly on the piezoelectric effect . these will be the target of further investigations . specifically , the dynamic anysotropy indicates that localization phenomena for elastic waves are likely to occur along the prescribed preferential directions . piezoelectrically tunable dirac cones may be of interest in the field of acoustic cloaking . finally a piezoelectrically tunable band gap can be of interest for filtering applications . we believe that the solution of the transmission problem through a finite checkerboard like structure , will be essential in order to deeply understand the afore - mentioned physical phenomena . dt gratefully acknowledges the people programme ( marie curie actions ) of the european union s seventh framework programme fp7/2007 - 2013/ under rea grant agreement n. pitn - ga-2013- 606878 . nvm , abm and mc gratefully acknowledge support from the european union fp7 project intercer2 `` modelling and optimal design of ceramic structures with defects and imperfect interfaces '' under grant agreement n. 286110 . the transmission conditions for floquet of the up mentioned fields impose the following conditions @xmath178 + @xmath179 + @xmath180 + @xmath181 bloch - floquet conditions at the boundaries of the unit cell ( @xmath182 ) , hold for physical quantities because of the periodicity . for @xmath183,@xmath184 , @xmath185 and @xmath186 , bloch - floquet conditions lead respectively to @xmath187 + @xmath188 + @xmath189 @xmath190 the expressions of the entries for the reflection and transmission matrices in eqs ( [ eq : r&t - left ] ) and ( [ eq : r&t - right ] ) are @xmath191,\\ r_{\rm ae}^{(\ell)}&= & -\frac{2 i}{\mathcal{d } } e^{+i(q_2+r_2)x_1}p\sqrt{g_2 r_2}\sqrt{\frac{q_2}{\varepsilon_2 } } \left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right),\\ r_{\rm ee}^{(\ell)}&=&-\frac{i e^{2 i q_2 x_1}}{\mathcal{d } } \left[p^2\left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right)^2 + ( g_1 r_1 + g_2 r_2)\left(\frac{q_1}{\varepsilon_1}-\frac{q_2}{\varepsilon_2}\right ) \right],\\ r_{\rm ea}^{(\ell)}&=&\frac{2 i}{\mathcal{d } } e^{+i(q_2+r_2)x_1}p\sqrt{g_2 r_2}\sqrt{\frac{q_2}{\varepsilon_2 } } \left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right),\end{aligned}\ ] ] and @xmath192 where @xmath193 the explicit expressions of the entries are @xmath194,\\ r_{\rm ae}^{(r)}&= & \frac{2 i}{\mathcal{d } } e^{-i(q_1+r_1)x_1}p\sqrt{g_1 r_1}\sqrt{\frac{q_1}{\varepsilon_1 } } \left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right),\\ r_{\rm ee}^{(r)}&=&\frac{i e^{-2 i q_1 x_1}}{\mathcal{d } } \left[p^2\left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right)^2 - ( g_1 r_1 + g_2 r_2)\left(\frac{q_1}{\varepsilon_1}-\frac{q_2}{\varepsilon_2}\right ) \right],\\ r_{\rm ea}^{(r)}&=&-\frac{2 i}{\mathcal{d } } e^{-i(q_1+r_1)x_1}p\sqrt{g_1 r_1}\sqrt{\frac{q_1}{\varepsilon_1 } } \left(\frac{e_1}{\varepsilon_1}-\frac{e_2}{\varepsilon_2}\right),\end{aligned}\ ] ] and @xmath195 77 sabina fj , willis jr . scattering of sh waves by a rough half - space of arbitrary slop . _ geophys . j. international _ 1975 ; @xmath196(2):685 - 703 . sabina fj , willis jr . a simple self - consistent analysis of waves propagation in particulate composites . _ wave motion _ 1988 ; @xmath197(2):127 - 142 . bravo - castillero j , rodriguez - ramos r , mechkour h , otero ja , sabina fj . homogenization of magneto - electro - elastic multilaminated materials . _ qjmam _ 2008 , @xmath198(3):311 - 332 . camacho - montes h , sabina fj , bravo - castillero j , guinovart - diaz r. magnetolectric coupling and cross - property connections in a square array of a binary composite . _ international journal of engineering science _ 2009 , @xmath199(2):294 - 312 . yang j , wang j. dynamic anti - plane problems of piezoceramics and applications in ultrasonics - a review . _ acta mechanica solida sinica _ 2008 ; @xmath200(3):207 - 218 . wave propagation in piezoelectric crystals . _ the journal of the acoustical society of america _ 1949 ; @xmath200(3):159 - 167 . _ acoustic fields and waves in solids ( vol . 1 ) _ 1973 , wiley , new york , ( 1973 ) . yang j ( ed . ) , _ special topics in the theory of piezoelectricity_. springer , dorderecht , heidelberg , london , new york ( 2009 ) . darinskii an , le clezio e , feuillard g. _ the role of electromagnetic waves in the reflection of acoustic waves in piezoelectric crystals . _ _ wave motion _ 2008 , @xmath201:428 - 444 . zhang v , djafari - rouhani b. a general model for analysis of acoustic phonons in piezoelectric superlattices . application to ( 111)-alas / gaas super - lattice . _ j. phys : condensed matter _ 2007 , @xmath202:186209 ( 17pp ) . wu tt , huang zg , lin s , surface and bulk acoustic waves in two - dimensional phononic crystal consisting of materials with general anysotropy . _ physical review b _ 2004 ; @xmath203:094301 ( 10pp ) . laude v , wilm m , benchabane s and khelif sa . full band gap for surface waves in two - dimensional phononic crystal consisting of materials with general anisotropy . _ physical review e _ 2004 @xmath203 : 094301 ( 7pp ) . hou z , wu f and liu y. phononic crystals containing piezoelectric material . _ solid state communications _ 2004 ; @xmath204 . sabina fj and movchan ab . _ interfacial effects in electromagnetic coupling within piezoelectric phononic crystals_. _ acta mech . sin . _ , @xmath205:95 - 99 . piliposian gt , avetisyan as and ghazaryan kg . _ shear wave propagation in periodic phononic / photonic piezoelectric medium_. _ wave motion _ , @xmath206 , 125 , ( 2012 ) . platts sb , movchan nv , mcphedran rc , movchan ab . _ two dimensional phononic crystals and scattering of elastic waves by an array of voids . a _ 2002 ; @xmath207:2327 - 2347 . botten lc , nicorovici na , asatryan aa , mcphedran r , martijn de sterke c , robinson pa . formulation for electromagnetic scattering and propagation through grating stacks of metallic and dielectric cylinders for photonic crystal calculations . part i. method . am . _ 2000 ; @xmath208(12 ) : 2165 - 2176 . botten lc , nicorovici na , asatryan aa , mcphedran r , martijn de sterke c , robinson pa . formulation for electromagnetic scattering and propagation through grating stacks of metallic and dielectric cylinders for photonic crystal calculations . properties and applications . am . _ 2000 ; 17(12):2177 - 2190 . l. c botten , n. a. nicorovici , r. mcphedran , c. martijn de sterke and a. a. asatryan . photonic band structure calculations using scattering matrices . _ physical review e _ 2001 ; @xmath209:046603 ( 18pp ) .	in this article we discuss the modelling of elastic and electromagnetic wave propagation through one- and two - dimensional structured piezoelectric solids . + dispersion and the effect of piezoelectricity on the group velocity and positions of stop bands are studied in detail . we will also analyze the reflection and transmission associated with the problem of scattering of an elastic wave by a heterogeneous piezoelectric stack . special attention is given to the occurrence of transmission resonances in finite stacks and their dependence on a piezoelectric effect . + a 2d doubly - periodic piezoelectric checkerboard structure is subsequently introduced , for which the dispersion surfaces for bloch waves have been constructed and analysed , with the emphasis on the dynamic anisotropy and special features of standing waves within the piezoelectric structure .
hgmn stars are chemically peculiar stars for which periodic variability has not been found as of yet . searches for variability have been made mostly photometrically though some studies of spectral variability have also been attempted . historically , several hgmn stars have been claimed to be variable but variability as yet to be confirmed in any of them @xcite . a large number of hgmn stars were observed as part of the hipparcos mission but no periodic variability was detected . the maximum permitted amplitude can in many cases be expected to be at most a few mmag . recently , some spectral variability was claimed in @xmath0 andromed which were interpreted as possible surface chemical inhomogeneities @xcite . the authors argued that such variability would be the exception rather than the rule in hgmn stars . the pursuit of elusive evidence of variability , both spectroscopically and photometrically , is motivated by several unresolved questions : * pulsations is expected theoretically from current models , in other words confirmation of stability or the discovery of low amplitude pulsations can provide constraints on physical processes not accounted for in the models ( see turcotte & richard in these proceedings ) ; * rotational variability would provide evidence of surface inhomogeneities related to diffusion , mass loss and/or magnetism in the atmosphere of b stars ; * confirm or infirm that all hgmn stars are part of binary or multiple systems which could help answer the question as to how b stars can be slowly rotating in the absence of binarity or magnetism . in this short paper we present preliminary results of the search of line profile variability in a substantial series of echelle spectra of four bright hgmn stars of the southern hemisphere . these observations represent an unprecedented effort to study spectroscopic variability in hgmn stars and are expected to help put stronger constraints on pulsations in these stars . the four program stars were the brightest southern hgmn stars visible during the periods of observation ( see next section ) . three of the four are within the theoretical instability region for spb stars ( hd 11753 being right on the cool edge ) , the fourth ( hd 53244 ) being slightly too evolved ( figure [ fig : hrd ] ) . -t@xmath1 diagram showing the program stars and the theoretical limit of the spb instability region @xcite along with a sample of other hgmn stars @xcite . ] the spectra were taken over two campaigns of several days , from september 28@xmath2 to october 11th@xmath2 and from december 2@xmath3 to december 15@xmath2 2000 , with the coralie spectrograph at the 1.2 m telescope at la silla . the observations are summarized in table [ tab : obs ] . .summary of observations of the program stars [ cols="^,^,^,^,^,^ " , ] due to space constraints we henceforth discuss only the star for which the better results were obtained at this point in the analysis , hd221507 . the spectra selected for this star after bad data was removed are shown in figure [ fig : spec ] . we focused on the siii doublet at @xmath44128.053 and @xmath44130.884 for which the first moment was calculated , a procedure developed to study spb stars @xcite . the variability was studied using the pdm method . the models of hgmn stars suggest that they should pulsate in a similar way to spb stars , if at all . four phase plots are shown in figure [ fig : phase ] . the periods shown , 0.31 , 0.44 , 0.78 , 1.38 @xmath5 were the ones which would reproduce the best approximation to a sine wave . the periods are in the range expected for spbs . the scatter is evidently quite large in all cases and the variability , although somewhat suggestive , is far from clear . this work was performed in part under the auspices of the u.s . department of energy , national nuclear security administration by the university of california , lawrence livermore national laboratory under contract no.w-7405-eng-48 .	spectra of four non - magnetic chemically peculiar late b type stars ( hgmn ) stars are analysed to detect periodic spectral line variations ( lpvs ) . a procedure developed to study lpvs in slowly pulsating b stars has been adopted as pulsational properties of hgmn stars should be expected to be similar . in the preliminary results discussed here no conclusive evidence for periodic lpvs was uncovered . a more sensitive re - analysis of the data is under way .
there are two important phenomena observed in evolutionary dynamical systems of any kind : _ self - organization _ and _ emergence_. both phenomena are the exclusive result of endogenous interactions of the individual elements of an evolutionary dynamical system . emergence characterizes the patterns that are situated at a higher macro level and that arise from interactions taking place at the lower micro level of the system . self - organization , besides departing from the individual micro interactions , implies an increase in order of the system , being usually associated to the promotion of a specific functionality and to the generation of patterns . typically , complex patterns emerge in a system of interacting individuals that participate in a self - organizing process . self - organization is more frequently related to the process itself , while emergence is usually associated to an outcome of the process . although less frequently mentioned , the emergence of patterns from self - organizing processes may be strongly dependent on _ locality_. emergence and self - organization are not enough to distinguish between two important and quite different circumstances : the presence of an influence that impacts the system globally and , conversely , the absence of any global influence and the lack of information about any global property of the system . in the latter case , the system itself is the exclusive result of local interactions . such a global influence ( entity or property ) is often associated with the concept of _ environment_. noteworthy , the latter circumstance may be considered a case of the former : when that global entity does not exist , the environment for each agent is just the set of all the other agents . conversely , when the global entity exists , it is considered part of the environment and may have an inhomogeneous impact on the individual dynamics . regardless of the environmental type , economical , ecological and social environments share as a common feature the fact that the agents operating in these environments usually try to improve some kind of utility , related either to profit , to food , to reproduction or to comfort and power . a general concept that is attached to this improvement attempt is the idea of _ adaptation_. in the economy , adaptation may be concerned with the development of new products to capture a higher market share or with the improvement of the production processes to increase profits : that is , innovation . in ecology , adaptation concerns better ways to achieve security or food intake or reproduction chance and , in the social context , some of the above economical and biological drives plus a few other less survival - oriented needs . in all cases , adaptation aims at finding strategies to better deal with the surrounding environment ( @xcite ) . natural selection through fitness landscapes or geographic barriers are good examples how global influences are considered when modeling adaptation in an evolutionary process . on the other hand , adaptation also operates in many structure generating mechanisms that can be found in both physical and social sciences but that are built on the exclusive occurrence of local interactions . in biology , the ultimate domain of evolution and natural selection , we are confronted with tremendous organic diversity virtually infinite forms and shapes none of which found twice but the distribution is well structured in a way that allows us to order this diversity and to speak of species , families , orders etc . a quite illustrative description is given by the evolutionary geneticist theodusius dobzhanski ( @xcite : p.21 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ suppose that we make a fairly large collection , say some 10,000 specimens , of birds or butterflies or flowering plants in a small territory , perhaps 100 square kilometers . no two individuals will be exactly alike . let us , however , consider the entire collection . the variations that we find in size , in color , or in other traits among our specimens do not form continuous distributions . instead , arrays of discrete distributions are found . the distributions are separated by gaps , that is , by the absence of specimens with intermediate characteristics . we soon learn to distinguish the arrays of specimens to which the vernacular names english sparrow , chickadee , bluejay , blackbird , cardinal , and the like , are applied . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if we had to make a visual representation of this description of intra and interspecies variations it would perhaps look like the multi - modal distribution shown in figure [ fig : distribution01 ] . what we call a species , is in fact some norm or mean characteristics of a cluster of individuals . evolutionary theory is ultimately a theory about the history which led to such a pattern . and if the organic diversity we observe nowadays evolved in a way that is characterized by some kind of > > tree of live < < , then there must be events that may lead to the split of a connected set of individuals ( protospecies ) into ( at least ) two sets that are not connected any longer ( see figure [ fig : speciation ] ) . in biology , this is called _ speciation_. as we will see in this article , though , the generation of such a split with simple but well known evolutionary models in which `` natural selection impels and directs evolutionary changes '' ( ibid . p.2 ) is not straightforward . it so happens that constraints on the interaction behavior are required . the phenotype of living beings is not the only domain where patterns of structured diversity as illustrated in figure [ fig : distribution01 ] are observed . phenomena include certain phases of structure formation in physical cosmology , distribution of cultural behavior , languages and dialects , herd behavior in finance , among others . especially for the latter examples in the field of socio - cultural dynamics a variety of models has been proposed which do not rely on the evolutionary concept of ( natural ) selection . they are rather based on the idea of exclusively _ local interactions ( li ) _ implemented in form of a system of agents that interact locally according to simple rules like assimilation or conformity . in these systems , finding strategies to better deal with the surrounding environment ( and thus improving fitness ) is not constrained by any global property . it may , however , be constrained by local ( individual ) rules . as we shall see later in this paper , constraints on the mechanisms of selection , interaction and replacement and the way they are combined in the modeling of an evolutionary process have an important bearing on both adaptation and emergence of speciation . locality operating in each of these mechanisms seems to be the fundamental modeling principle by which emergence of a multi - modal distribution as shown in figure [ fig : distribution01 ] can be explained . on the basis of these observations about the > > modelability < < of speciation with evolutionary and self - organisatory models , we study in this paper the conditions and mechanisms required for speciation and the emergence of a multi - modal distribution . in this analysis , we use computational ( section [ sec : computation ] ) as well as mathematical ( section [ sec : mathematic ] ) arguments . our models simulate how a population of individuals evolves in time in an abstract attribute space @xmath0 that represent phenetic traits , attitudes , verbal behavior , etcetera . modeling agents as points in an attribute space of this kind is of course a highly artificial abstraction from the complexity and multi dimensionality of real agents . for the purposes of this paper , let us conceptualize an _ interaction event _ , defining the system evolution from one time step to the other , by the following three components : 1 . selection of agents , 2 . application of interaction rules , 3 . replacement of agents . any interaction event ( e.g. , mating , communication , ... ) that takes place in the course of a simulation of the model consists of the sequential application of these three steps . the reason to dissect the interaction events in this way is two fold : 1 . we want to look at the dynamical and structural effects of constraints applied to each of the three components independently ; 2 . the scheduling of interaction events may have a crucial effect on the model behavior , and with the distinction between selection and interaction on the one hand , and replacement on the other , we are able to make this effect explicit . the way interaction events are scheduled in the implementation of the models is not always given much importance in existing simulation studies . in the presence of constraints on the selection and interaction mechanisms , however , the outcome as well as the dynamical properties depend in a crucial way on the different choices . on the other hand , there are studies that do analyze the differences between synchronous and asynchronous update ( see , for instance , @xcite ) as well as studies on non overlapping ( nolg ) and respectively overlapping generations ( olg ) in biology and economics ( for instance , @xcite ) . here we show that especially when the interaction is constrained ( as in the case of assortative mating ) there emerges an important qualitative difference between olg and nolg models . namely , speciation is observed in the former , but not in the latter case , whereas adaptation is favored by the latter and hindered by the former . however , by the distinction of selection , interaction and replacement we are able to show that in fact the difference between local and non - local replacement plays the determinant role ( and not the distinction between olg and nolg ) . even though locality also impacts selection and interaction mechanisms , it is on the replacement mode where relies the fundamental difference with respect to the conditions required for either adaptiveness or speciation . this paper is organized as follows : section 2 addresses the main issues of both the fitness landscape and the self - organizing models from a computer simulation framework . in both cases , microscopic implementation rules are tested against their capability of reproducing adaptiveness and speciation . in section 3 , the emergence of speciation is analytically shown to be dependent on the choice of different replacement modes . this is accomplished through a probabilistic description of a minimal model of just three phenetic traits where the transition probabilities between traits follow a markov chain . section 4 is targeted at presenting concluding remarks and a framework that relates interaction events to the emergence of collective structures in adaptive and self - organizing complex systems . in biology , and population genetics in particular , adaptive walks on fitness landscapes have been studied intensively . the main questions addressed by fitness landscapes approaches are related to the possible structure of the landscapes ( e.g. , @xcite ) , to how populations climb an adaptive peak in the landscape ( e.g. , @xcite ) , and to the circumstances under which a population might wander from one peak to another by crossing adaptive valleys ( e.g. @xcite ) . one of the best known models for populations on fitness landscapes is the wright - fisher model with non overlapping generations ( sometimes called wright - fisher sampling and shortened in the sequel by wf model , see @xcite and also @xcite ) . consider a population of @xmath1 individuals which is said to constitute the original generation ( @xmath2 ) . we consider only the case of sexual reproduction in this paper , in which the genotype of a new born individual is obtained by the recombination of the genoms of two randomly chosen parent individuals . as noted above , the choice of two parents and the application of a recombination rule is referred to as interaction ( or mating ) event . in the wf model , @xmath1 such mating events are performed until a new generation of @xmath1 individuals is complete . as soon as it is complete , the parent generation is canceled and the process is repeated taking the new generation as parents . therefore , in the wf model the population size is always maintained at @xmath1 . we will denote the generation number by @xmath3 . we implemented this simple model and performed simulations on different toy fitness landscapes . the microscopic rules involved into the creation of a new individual , that is , the mating event , are as follows : 1 . selection of two individuals with a probability proportional to their fitness , 2 . application of recombination and mutation rules , 3 . replacement of an agent from the parent generation . in this toy model , we consider only one phenetic trait ( locus ) that takes discrete values ( from 0 to 99 ) . we denote the traits of the two chosen parent individuals @xmath4 and @xmath5 as @xmath6 and @xmath7 respectively and model recombination by taking the average of the two , @xmath8 . to model mutations we add a random value to @xmath9 . in the wf model , @xmath9 is stored at an arbitrary place in the children array and one of the main objectives of this paper is clarify that this has important consequences for the model dynamics . an adaptive landscape is introduced into the model by assigning a fitness value to each of the 100 traits . for the first analysis shown in figure [ fig : onepeak.wf ] , a single peaked fitness function with a peak at trait 75 is used and the fitness assigned to trait @xmath10 is given by @xmath11 we have used the normal distribution with @xmath12 and @xmath13 in the construction of the fitness landscape ( solid line in figure [ fig : onepeak.wf ] ) . in the iteration process , individuals are chosen as parents with a probability proportional to @xmath14 , @xmath10 being the trait of the respective individual . for the illustrative model realizations in this section , we set @xmath15 . initially , the 500 individuals are distributed in this space according to a normal distribution with mean @xmath16 and @xmath17 ( see first image of figure [ fig : onepeak.wf ] ) . this section is mainly thought as an illustration of the different behaviors and patterns generated by certain constraints on the interaction mechanism . as the qualitative effects of different assumptions become evident and comprehensible in single simulations of the model , there is no need for a rigorous statistical analysis of suites of simulations with varying initial conditions . moreover , a mathematical analysis of the model dynamics is presented in the second part of this paper ( section [ sec : mathematic ] ) . [ cols="^,^,^ " , ] the framework presented in table 1 schematically shows the consequences of adopting ( un)constrained mechanisms to the emergent outcome of a so process . it helps to emphasize that the emergence of some specific patterns may be strongly dependent on the way constraints dictate limitations on the selection , interaction and replacement mechanisms . more specifically , it shows that differently ( un)constraining the replacement mechanism of an so process provides the conditions required for either speciation ( the emergence of multi - modal distributions ) or adaptation , since these features appear as two opposing phenomena , not achieved by one and the same model . in the same way that random interbreeding leads to conservative dynamics , random replacement is also an opposing force to speciation since newcomers may take the place of former - distant agents . in so doing , at the macro level , random replacement sets aside the effect of bounded confidence and - like undirected genetic drift - may lead to the merging of subpopulations . even though we show in this paper that natural selection , operating as an external , environmental mechanism , is neither necessary nor sufficient for the creation of clustered populations , we do not want to argue against natural selection as an important mechanism in the biological domain and a substantive driving force in the speciation process . to the contrary , the concept of ( natural ) selection operating at a global level may provide us with plausible interpretations of the model results , even in disciplines where such interpretations are still lacking . in the words of t. dobzhanski ( @xcite , p.5 - 6 ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ... ] in biology nothing makes sense except in the light of evolution . it is possible to describe living beings without asking questions about their origins . the descriptions acquire meaning and coherence , however , only when viewed in the perspective of evolutionary development . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ financial support of the german federal ministry of education and research ( bmbf ) through the project _ linguistic networks _ is gratefully acknowledged ( http://project.linguistic-networks.net ) . this work has also benefited from financial support from the fundao para a cincia e a tecnologia ( fct ) , under the _ 13 multi - annual funding project of uece , iseg , technical university of lisbon_.	in this paper , we inspect well known population genetics and social dynamics models . in these models , interacting individuals , while participating in a self - organizing process , give rise to the emergence of complex behaviors and patterns . while one main focus in population genetics is on the adaptive behavior of a population , social dynamics is more often concerned with the splitting of a connected array of individuals into a state of global polarization , that is , the emergence of speciation . applying computational and mathematical tools we show that the way the mechanisms of selection , interaction and replacement are constrained and combined in the modeling have an important bearing on both adaptation and the emergence of speciation . differently ( un)constraining the mechanism of individual replacement provides the conditions required for either speciation or adaptation , since these features appear as two opposing phenomena , not achieved by one and the same model . even though natural selection , operating as an external , environmental mechanism , is neither necessary nor sufficient for the creation of speciation , our modeling exercises highlight the important role played by natural selection in the interplay of the evolutionary and the self organization modeling methodologies . _ keywords _ : * emergence,self - organization,agent based models * , * speciation * , * markov chains*. _ msc : _ 37l60 , 37n25 , 05c69 .
the general class of inventory - production systems is often associated to cost optimization problems . indeed , one must deal with three major matters : the storage of components , the possible random behavior of the manufacturing process and random clients demand @xcite . the controller must decide which production rate of the components fits best . a too slow production rate leads to low stock levels but it might not meet clients demand . on the opposite , a fast production rate does meet the demand , but may raise stock levels . one must then find a balance between both to minimize costs . this paper focuses on the optimization of a real - life industrial launcher integration process studied in collaboration with airbus defence and space . clients order a certain number of launches to be performed at specific dates . the controller has to determine the production rates in order to minimize costs . only storage and lateness costs are taken into account here . in general , the costs may also take into account several other constraints such as exploitation cost , workforce salary , the cost related to the unavailability of the structure including any penalty or the maintenance and inspection cost , among others . plus , a part of the architecture of the process is not set . indeed , the controller has to decide on the maximum capacity of one warehouse between two options . the originality of this problem is twofold . on the one hand , the optimization horizon is rather long , 30 years , but the controller can only make decisions once a year concerning the production rates . on the other hand , the launches must be performed according to a prescribed calendar corresponding to clients orders . our goal is to find an optimization procedure usable in practice . it should provide explicit decision rules applicable to each trajectory as a table giving the controller the best action to take according to the current state and time . a preliminary study was performed on a simplified process @xcite using petri nets @xcite . although they are easy to simulate , they are not suitable for performing dynamic decisional optimization . a more suitable framework is that of markov decision processes ( mdps ) @xcite . mdps are a class of stochastic processes suitable for cost and decision optimization . briefly , at each state , a controller makes a decision which has an influence on the transition law to the next state and on a cost function . the latter depends on the starting state and the decision made . the sequence of decisions is called a policy , and its quality is gauged thanks to a cost criterion ( typically , it is the sum of all the costs generated by the transitions ) . the first step to solve our problem is to implement an mdp - based simulator of the launcher integration process . simulation results were presented at the esrel conference in 2015 @xcite . this paper deals with the optimization itself . it is a non standard optimization problem within the mdp framework because the transition law is not analytically explicit , it is only simulatable . thus , standard optimization techniques for mdps such as dynamic programming @xcite , or linear programming @xcite do not apply . in addition , the cost function is unusual as the actual lateness can be computed only at the end of a year , and not at its beginning when the controller makes their decisions . as the launcher integration process can be simulated , we investigate simulation - based algorithms for mdps @xcite . these extensively use monte - carlo methods to estimate the performance of a policy . thus , they require a fast enough simulator for the algorithms to give a result within a reasonable time . new difficulties arise here . first , the state space of our mdp , though finite is huge . second , the first simulator in matlab is not fast enough . third , the algorithms require the computation of a product of numerous numbers between @xmath0 and @xmath1 , and although the output is non zero on paper , it is treated as zero numerically , leading to erroneous results . to overcome these difficulties , we reduce the state space by aggregating states in a manner that makes sense regarding our application , we use the c language and a special logarithmic representation of numbers . the results we obtained are presented and discussed . this paper is organized as follows . section [ lauint ] is dedicated to the description of the assembly line under study and the statement of the optimization problem . in section [ mardec ] , we present how the optimization problem for the assembly line fits into the mdp framework . section [ optlau ] presents the main difficulties encountered while trying to optimize our mdp , and solutions to bypass them . in section [ numres ] , we present and comment the numerical results obtained . finally a last section gives some concluding remarks . technical details regarding the implementation of algorithms are provided in the appendix . airbus defense and space ( airbus ds ) as prime contractor is in charge of launchers and ground facilities design . this paper is dedicated to the optimization of an assembly line representative of a launcher integration process managed by airbus ds . for confidentiality matters , all parameter values and random distributions given in this paper are arbitrary but realistic . the launcher integration process we study in this paper is depicted on figure [ process ] . this assembly line is composed of several workshops and storage facilities that are described in detail in the following sections , and is operational typically for 30 years . the subassemblies are the input of the assembly line . a launcher needs four types of subassemblies to be manufactured . these are * the insulated motor cases ( imcs ) , which are powder - free boosters , * the lower liquid propulsion modules ( llpms ) and the upper liquid propulsion modules ( ulpm ) , which are the lower part of a launcher , * the upper launchers , which are the fairings of the launchers . the upper launchers are always available when needed . so in the following , their production rate and storage cost are not taken into account . the production time of the other subassemblies ( imc , llpm , ulpm ) is assumed to be random . its distribution is symmetric around a mean @xmath2 and takes 5 different values with probabilities given in table [ lawsub ] . the average production time @xmath2 is computed by dividing the total number of workdays by the target number of subassemblies the controller decides to produce in a year , and taking its integer part . the number of workdays in a year is set at 261 . so for instance , if the controller wants an average production of 12 llpms a year , the value of @xmath2 for llpms will be @xmath3 = 21 $ ] days , where @xmath4 $ ] stands for the integer part of a real number @xmath5 . .distribution of the production time of a subassembly with mean @xmath2 [ cols="<,<",options="header " , ] [ third ] storing 4 srms at most induces a lower cost . however , the gain compared to an 8-unit storage is only 0.61% ( it is 14.36% when compared to the nave policy ) . taking into account the variance of the costs , one may conclude that , in this case , the two scenarios lead to similar performances . with these three examples , one sees that the question of the optimal srm storage capacity is not trivial at all . indeed , it seems to be impossible to answer this question with only prior knowledge . one may consider static costs to have a more accurate answer to this problem ( for instance exploitation costs , a larger warehouse being naturally more expensive ) . however , it seems that a maximum capacity of 4 srms is a better choice in general . using the mdp framework together with a simulation based algorithm , we performed the optimization of the launcher integration process . several problems had to be addressed , such as different time scales , state space reduction computation speed or numerical representation of numbers . given a launch calendar , optimal policies are computed and stored in the form of a matrix . to apply the optimal policy in practice , the controller looks up the best action to select in the matrix , given the current state of the process and the current time . such policies do not have a trivial form and can not be easily explained . they lead to up to 70% gain compared to trivial policies prescribing a manufacturing rate corresponding to the exact number of launches to be performed in the year . to address real life optimization problem regarding launcher operations , one should now work with a calendar that is known only two years advance . one option that would fit the present framework would be to consider that the calendar of year @xmath6 is randomly drawn at the beginning of year @xmath7 . further exchange with practitioners is required to derive realistic distributions for such calendars . another possible extension of interest to airbus ds is to model the production of subassemblies with more detail , inducing longer delays for selecting the production rates . instructions for the mras are given on algorithm [ mrasa ] and for the asa on algorithm [ asaa ] . they use the following notation . for some set @xmath8 , @xmath9 denotes its indicator function , that is @xmath10 if @xmath11 and @xmath12 otherwise . the upper integer part of @xmath5 is denoted @xmath13 . let @xmath14 and define the function @xmath15 on @xmath16 by @xmath17 for all states @xmath18 , all decisions @xmath19 and all times @xmath20 , let @xmath21 the set of policies which prescribe action @xmath19 in state @xmath18 at time @xmath20 . for a policy @xmath22 , let @xmath23 and , for @xmath24 , let @xmath25 with @xmath26 a given probability matrix . initial 3-dimensional probability matrix @xmath26 , @xmath27 $ ] , @xmath14 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath24 , @xmath32 $ ] , initial state @xmath33 , iteration count @xmath34 , limit number of iterations @xmath35 . draw policy @xmath36 from matrix @xmath26 with probability @xmath37 and matrix @xmath38 with probability @xmath39 . simulate @xmath40 trajectories with @xmath33 as initial state using policy @xmath36 . compute the cost @xmath41 generated for the trajectories @xmath42 . @xmath43 . arrange the @xmath44 in descending order to get a sequence @xmath45 such that @xmath46 . @xmath47 . @xmath48 . @xmath49 . @xmath50 . @xmath51 policy @xmath36 such that @xmath52 . @xmath53 . @xmath54 . @xmath50 . @xmath55 policy @xmath36 such that @xmath56 . @xmath57 where @xmath58 . @xmath49 . @xmath59 . @xmath60 . @xmath61 . @xmath62 + @xmath63 @xmath64 . @xmath65 . @xmath66 . initial 3-dimensional probability matrix @xmath26 , @xmath68 , @xmath69 , temperature parameter @xmath70 , @xmath71 , @xmath72 , initial state @xmath33 , iteration count @xmath34 , limit number of iterations @xmath35 . draw policy @xmath36 from matrix @xmath26 with probability @xmath73 and matrix @xmath38 with probability @xmath74 . simulate @xmath40 trajectories with @xmath33 as initial state using policy @xmath36 . compute the cost @xmath41 generated for the trajectories @xmath42 . @xmath43 . @xmath75 @xmath76 . @xmath77)$ ] . @xmath78)$ ] . @xmath79 . @xmath80 . @xmath81 . @xmath66 . elegbede c , brard - bergery d et al . dynamical modelling and stochastic optimization for the design of a launcher integration process . in _ safety , reliability and risk analysis : beyond the horizon _ ( eds rdjm steenbergen et al ) , amsterdam , the netherlands , 29 september2 october 2013 , pp . 3039 3046 , london : taylor and francis group , crc press . puterman ml . _ markov decision processes : discrete stochastic dynamic programming_. wiley series in probability and mathematical statistics : applied probability and statistics . new - york : john wiley & sons , inc , 1994 . nivot c , de saporta b , dufour f et al . modeling and optimization of a launcher integration process . in _ safety and reliability of complex engineered systems : esrel 2015 _ ( eds l podofillini et al ) , zrich , switzerland , 7 september10 september 2015 , pp . 2281 2288 , london : taylor and francis group , crc press .	this paper is dedicated to the numerical study of the optimization of an industrial launcher integration process . it is an original case of inventory - production system where a calendar plays a crucial role . the process is modeled using the markov decision processes ( mdps ) framework . classical optimization procedures for mdps can not be used because of specificities of the transition law and cost function . two simulation - based algorithms are tuned to fit this special case . we obtain a non trivial optimal policy that can be applied in practice and significantly outperforms reference policies .
transport properties of one - dimensional spin-@xmath0 systems are currently at the focus of active research . this has been motivated by the experimental manifestation of significant contributions to the thermal conductivity originating from magnetic excitations @xcite , stimulating intensive theoretical work@xcite . strong theoretical efforts@xcite have been devoted to the question of possible ballistic thermal transport in generic spin models such as spin ladders , frustrated chains , and dimerized chains . such ballistic transport would be characterized by a finite thermal drude weight . recent numerical and analytical studies indicate that in pure but _ nonintegrable _ spin models , the thermal drude weight scales to zero in the thermodynamic limit implying that the thermal current is likely to have a finite intrinsic life - time@xcite . in addition , the effects of extrinsic magnon scattering by phonons and/or impurities have been addressed in several works@xcite . for the integrable @xmath1 model , the energy current operator is a conserved quantity@xcite , leading to a finite thermal drude weight . its temperature dependence has been studied with exact diagonalization@xcite and bethe ansatz techniques@xcite and is well understood for arbitrary values of the exchange anisotropy at zero magnetic field . in this paper , we address the issue of thermal transport in the @xmath1 model in the presence of a finite magnetic field @xmath2 . in this case , magnetothermal effects become important and must be accounted for . the magnetothermal response itself has been studied by louis and gros in the limit of small magnetic fields@xcite and recently also by sakai and klmper in the low - temperature limit@xcite . here , we consider magnetic fields of arbitrary strength and we discuss the temperature dependence of the thermal drude weight under the condition of zero spin - current flow . + the hamiltonian of the @xmath1 model reads @xmath3 where @xmath4 is the number of sites , @xmath5 are spin-@xmath0 operators acting on site @xmath6 , and @xmath7 denotes the exchange anisotropy . the exchange coupling @xmath8 is set to unity in our numerical calculations . we focus on @xmath9 and periodic boundary conditions are imposed . + . magnetic phases of the @xmath1 model ( see , e.g. , ref . ) and leading term of the thermal drude weight @xmath10 at low temperatures . @xmath11 is the average local magnetization at @xmath12 . @xmath13 denotes the gap in either the polarized state ( i ) or the massive antiferromagnetic regime ( iii ) . in the polarized state ( i ) , @xmath14 . [ cols="<,^,^,^,^",options="header " , ] the quantum phases and the spectrum of ( [ eq : m2 ] ) are well understood , both as a function of exchange anisotropy @xmath7 and magnetic field @xmath2 . the reader is referred to ref . for a detailed summary and further references . here we only repeat the main points . at zero magnetic field , the spectrum of the hamiltonian eq . ( [ eq : m2 ] ) is gapless for @xmath15 and gapped for @xmath16 . the situation at finite magnetic fields is summarized in the first four columns of table [ tab:1 ] . three different cases are found : ( i ) the ferromagnetic gapped state for @xmath17 ( fm ) ; ( ii ) the gapless or massless phase for @xmath18 and @xmath19 ; and ( iii ) the antiferromagnetic , gapped state for @xmath20 and @xmath21 ( afm ) . the line @xmath22 starts at the @xmath23 symmetric point @xmath24 and @xmath25 grows exponentially slowly in the region @xmath20 , @xmath26 . + the fifth column of table [ tab:1 ] is a first account of our main findings for the low - temperature behavior of the thermal drude weight , denoted by @xmath10 in this paper . these results are now briefly summarized . one can expect qualitative changes in the low - temperature behavior of the thermal drude weight as the transition lines @xmath22 and @xmath27 are crossed . in particular , we focus on the transition from the gapless phase to the ferromagnetic state . in sec . [ sec:3 ] , we will argue that for @xmath28 , first , @xmath29 in the ferromagnetic state , @xmath30 being the gap , and @xmath31 temperature ; second , @xmath32 in the massless phase ; and third , @xmath33 along the line @xmath27 . + regarding the antiferromagnetic state , there is certainly also an exponentially suppressed drude weight ; see for instance refs . and for @xmath34 . however , the low - temperature region in this case and for @xmath22 is difficult to reach with the methods of the present paper . for a discussion of the low - temperature limit at vanishing magnetic field , we refer the reader to refs . and . apart from the low - temperature behavior , this paper studies the field dependence of the thermal drude weight in the phases ( i ) and ( ii ) at finite temperatures . the plan of this paper is the following . first , we discuss the expressions for the transport coefficients and the current operators in sec . [ sec:2 ] . second , in sec . [ sec:3 ] we perform an analysis of the transport coefficients based on a jordan - wigner mapping of the spin system onto spinless fermions . in this case , interactions at @xmath35 will be treated by a hartree - fock approximation . third , we present our results from exact diagonalization for @xmath36 in sec . [ sec:4 ] and compare them to the results from the jordan - wigner approach . the field and temperature dependence of the thermal drude weight is discussed with a particular focus on the case of the heisenberg chain . a summary and conclusions are given in sec . [ sec:5 ] . within linear response theory , the thermal and the spin current are related to the gradients @xmath37 and @xmath38 of the field @xmath2 and the temperature @xmath31 by@xcite @xmath39 where @xmath40 is either the thermodynamic expectation value of the spin current @xmath41 or the thermal current operator @xmath42 , respectively . @xmath43 denote the transport coefficients . at finite frequencies @xmath44 , the coefficients @xmath45 depend on the time - dependent current - current correlation functions via @xcite @xmath46 in this equation , @xmath47 for @xmath48 and @xmath49 for @xmath50 . @xmath51 is the inverse temperature and @xmath52 denotes the thermodynamic expectation value . note that @xmath53 due to onsager s relation@xcite . the real part of @xmath45 can be decomposed into a @xmath54-function at @xmath55 with weight @xmath56 and a regular part @xmath57 : @xmath58 this equation defines the drude weights @xmath56 , for which a spectral representation can be given@xcite @xmath59 here , @xmath60 is the boltzmann weight and @xmath61 denotes the partition function . in the exponent , @xmath62 has to be chosen in the same way as in eq . ( [ eq : k100 ] ) . + let us now introduce the appropriate definitions of the current operators . the local current operators @xmath63 and @xmath64 satisfy the continuity equations @xmath65 where @xmath66 is the local magnetization density and @xmath67 is the local energy density , respectively , with @xmath68 . at zero magnetic field , the total currents @xmath69}}=\sum_l j_{\mathrm{th[s]},l}$ ] are given by@xcite @xmath70 with the definition @xmath71to achieve a compact representation , while @xmath72 is defined as usual . note that subscripts in brackets @xmath73 refer to spin transport . + at finite magnetic field , the proper set of current operators is@xcite @xmath74 now , the crucial point is that , while the _ spin _ current @xmath75 is only conserved in the @xmath76 case ( @xmath77 ) , the current @xmath78 is conserved for all fields @xmath2 and values of @xmath7 , i.e. , @xmath79 ( refs . and ) . thus , it immediately follows from eqs . ( [ eq : k100 ] ) and ( [ eq : k1 ] ) that the drude weights @xmath80 , @xmath81 , and @xmath82 are finite for arbitrary fields @xmath2 . + furthermore , one can show that the spin drude weight @xmath83 is also finite in the thermodynamic limit for @xmath84 . we briefly outline the proof along the lines of ref . . given a set of all conserved observables @xmath85 , the spin drude weight @xmath83 can be written as @xmath86 where @xmath87 is the projection operator in the liouville space on the subspace spanned by all conserved quantities @xmath85 . the brackets @xmath88 denote mori s scalar product ; see , e.g. , ref . for details . restricting to a subset @xmath89 , one obtains an inequality@xcite @xmath90 providing a lower bound for the drude weight @xmath91 . in the literature , this relation is often referred to as mazur s inequality@xcite . several authors@xcite have used eq . ( [ eq:21 ] ) to infer a finite spin drude weight for the heisenberg chain , assuming broken particle - hole symmetry , or the presence of a finite magnetic field , respectively . more explicitly , only one conserved quantity is often considered in eq . ( [ eq:21 ] ) , namely @xmath92 , which has a finite overlap @xmath93 with the spin current for @xmath84 . this finally proves @xmath94 for @xmath84 . + the main focus of this paper is on the case of purely thermal transport with a vanishing spin current , i.e. , @xmath95 . we therefore arrive at a thermal conductivity @xmath96 which is described by @xmath97 where @xmath98 in terms of the drude weights @xmath56 reads @xmath99 exactly the same result for @xmath98 is obtained if a different choice of current operators and corresponding forces is made , e.g. , @xmath75 and @xmath78 from eqs . ( [ eq:3 ] ) and ( [ eq:4 ] ) ( see ref . ) . the expression for @xmath98 , being fully equivalent to eq . ( [ eq : k5 ] ) , is then given by @xmath100 note that for @xmath34 , @xmath101 . therefore , two competing terms contribute to @xmath98 in eq . ( [ eq : k5a ] ) : the `` pure '' thermal drude weight @xmath102 and the `` magnetothermal correction '' , @xmath103 . note that the magnetothermal correction might be suppressed by external scattering or spin - orbit coupling , breaking the conservation of the total magnetization of the spin system ( ref . ) . this is an open issue which may depend crucially on the particular material investigated in experimental transport studies . + let us now give spectral representations for the quantities @xmath104 , @xmath105 , and @xmath106 ( refs . ) @xmath107}}^i(h , t ) & = & \frac{\pi \beta^{2[1]}}{n } \sum_{m , n\atop e_m = e_n } p_n \|\langle m\| j_{\mathrm{th[s]}}\|n\rangle\|^2,\label{eq : k3}\\ d_{\mathrm{s}}^{ii}(h , t ) & = & \hspace{-0.1cm}\frac{\pi}{n}\left\lbrack \langle - \hat t\rangle -2 \hspace{-0.3cm}\sum_{m , n\atop e_m\not = e_n}\hspace{-0.2 cm } p_n\frac{\|\langle m \| j_{\mathrm{s}}\|n\rangle\|^2}{e_m - e_n } \right\rbrack\hspace{-0.1cm},\label{eq : k8}\\ d_{\mathrm{th , s}}(h , t)&= & \frac{\pi\beta}{n } \sum_n p_n \langle n\| j_{\mathrm{th } } j_{\mathrm{s}}\|n\rangle . \label{eq : k4}\end{aligned}\ ] ] the operator @xmath108 is the kinetic energy . in the eqs . ( [ eq : k3 ] ) , ( [ eq : k8 ] ) , and ( [ eq : k4 ] ) , the magnetic field only enters via the boltzmann weights @xmath109 . the two expressions @xmath110 and @xmath111 are equivalent in the thermodynamic limit , but exhibit differences at low temperatures for finite system sizes@xcite . in this context , note that @xmath112 is the so - called meissner fraction , which measures the _ superfluid _ density in the thermodynamic limit and in a transverse vector - field@xcite . this quantity vanishes for @xmath113 in one dimension , but it can be nonzero for finite systems@xcite . in ref . , we have performed a study of the finite - size scaling of both quantities for the @xmath1 chain , showing that @xmath114 already holds at sufficiently high temperatures . at low temperatures and zero magnetic field , @xmath115 is always exponentially suppressed for even @xmath4 due to finite - size gaps ; thus a finite value of @xmath116 can only be found for @xmath117 . on the contrary , since @xmath118 at low temperatures , @xmath111 correctly results in a _ finite _ value at @xmath12 in the massless regime . depending on the context , one should carefully check which of these two quantities exhibits the more reliable finite - size behavior , and in fact , in the present case of finite magnetic fields we will argue in sec . [ sec:4 ] that @xmath119 should preferably be used . for a more detailed discussion of the relation between @xmath110 and @xmath111 , we refer the reader to ref . and references therein . + in our numerical analysis , we will evaluate @xmath102 , @xmath120 , and @xmath121 while the coefficients @xmath56 from eq . ( [ eq : k5 ] ) can be derived if desired as they are linear combinations of @xmath122 , and @xmath121 : @xmath123 the @xmath1 model is integrable and solvable via the bethe ansatz . therefore one expects all quantities in eqs . ( [ eq : k5 ] ) and ( [ eq : k5a ] ) to be accessible by analytical techniques . yet , for the spin drude weight @xmath124 at zero magnetic field , partly contradicting results can be found in the literature regarding both its temperature dependence and the question whether it is finite or not for the heisenberg chain ( @xmath125 ) in the thermodynamic limit . see refs . and further references therein . we now discuss a hartree - fock type of approximation to the hamiltonian eq . ( [ eq : m2 ] ) , which we use to compute the drude weights @xmath56 . the spin operators @xmath126 are first mapped onto spinless fermions via the jordan - wigner transformation@xcite @xmath127 here , @xmath128 destroys(creates ) a fermion on site @xmath6 . the string operator @xmath129 reads @xmath130 with @xmath131 . next , the interaction term @xmath132 appearing in the fermionic representation is treated by a hartree - fock decomposition leading to an effective mean - field hamiltonian @xmath133 with the mean - field dispersion @xmath134 the quantities to be determined self - consistently are @xmath135 and @xmath136 where the latter is related to the average local magnetization @xmath137 via @xmath138 . the drude weights can then be obtained from @xmath139 the current operators read @xmath140 with @xmath141 . + while this approach is exact for @xmath77 , fair results for @xmath142 are even obtained for @xmath143 ; see refs . and . from eqs . ( [ eq : mf3 ] ) to ( [ eq : mf5 ] ) , the leading contribution at low temperatures can be derived . + we start with the free fermion case @xmath77 , for which we find at the saturation field @xmath144 @xmath145 with @xmath146 @xmath147 being the riemann - zeta function . note that the spin drude weight at @xmath12 is finite for @xmath148 and vanishes for @xmath149 . at low temperatures and for @xmath27 , we find @xmath150 and a divergence of the pure thermal drude weight @xmath102 with @xmath151 to leading order in temperature , which follows from eqs . ( [ eq : k9b ] ) and ( [ eq : k9c ] ) . we mention that the result @xmath152 at the critical field was also found within a continuum theory suggested to describe transport properties of two - leg spin ladders.@xcite + in the intermediate regime , i.e. , the gapless state ( ii ) [ see table [ tab:1 ] ] , @xmath153 holds at low temperatures , because the dispersion is linear in the vicinity of the fermi level for @xmath154 . note that @xmath155 for small @xmath31 in this regime . equation ( [ eq : mf9a ] ) results in @xmath156 for @xmath34 , which is , e.g. , known from ref . . + for @xmath157 , both @xmath158 and the second term in eq . ( [ eq : k5 ] ) , i.e. , @xmath159 , are given by @xmath160 to leading order in temperature and for @xmath161 , where @xmath162 is the gap . this implies that @xmath98 is strongly suppressed at low temperatures due to the cancellation of the contributions to @xmath98 in eq . ( [ eq : k5 ] ) . in fact , such cancellation occurs in the next - to - leading order in @xmath31 as well . one can further show , taking into account the first non - vanishing contribution to @xmath10 in eq . ( [ eq : k5 ] ) , that @xmath163 describes the low - temperature behavior of the thermal drude weight above @xmath144 . in ref . , it has been argued that @xmath164 is a generic feature of gapped systems with a finite thermal drude weight . + we further point out that the ratio of the thermal drude weight @xmath10 and the spin drude weight @xmath120 fulfills a wiedemann - franz type of relation in the low - temperature limit in all three cases , i.e. , in the massless and the fully polarized state as well as for @xmath27 : @xmath165 the constant @xmath166 takes different values in the regimes ( i ) and ( ii ) , but for the free - fermion case ( and within mean - field theory as well ) it is independent of the magnetic field _ in _ the massless and fully polarized state , respectively . + before turning to the mean - field theory for @xmath167 , let us briefly discuss which results can be expected from conformal field theory for the massless state . the expressions for the spin and thermal drude weight @xmath158 and @xmath168 have the same structure as at zero magnetic field , i.e. , @xmath169 and @xmath170 with field - dependent velocity @xmath171 and luttinger parameter @xmath172 ( see , e.g. , refs . and ) . this implies that the constant @xmath166 appearing in eq . ( [ eq:100 ] ) is field dependent in the massless regime ( see ref . for @xmath34 ) : @xmath173 furthermore , @xmath81 vanishes in the continuum limit due to particle - hole symmetry . while a finite magnetic field initially breaks this symmetry for the original bosonic fields , the original form of the luttinger - liquid hamiltonian is restored by introducing a shifted bosonic field@xcite . this has an interesting consequence for the low temperature behavior of the pure thermal drude weight @xmath102 . namely , by solving eqs . ( [ eq : k9b ] ) and ( [ eq : k9c ] ) for @xmath102 , one obtains @xmath174 which implies that @xmath102 diverges at low temperatures with @xmath175 in the massless regime , consistent with results of ref . . + additionally , one obtains @xmath10 in the massless regime and in the low - temperature limit @xmath176 both parameters , i.e. , @xmath177 and @xmath178 , can be computed exactly by solving the bethe - ansatz equations@xcite . the velocity @xmath179 has been calculated for @xmath125 in ref . . further numerical values for these parameters can be found in , e.g. , ref . . + let us next dicuss the results from the mean - field approximation ( mf ) for @xmath167 . figure [ fig:1 ] shows @xmath98 for @xmath180 and @xmath181 ( thick lines ) . the main features are : ( i ) a suppression of the thermal drude weight by the magnetic field ; ( ii ) a shift of the maximum to higher temperatures for @xmath182 compared to @xmath34 ; ( iii ) a change in the low - temperature behavior which will be discussed in more detail below in this section . + for comparison , the results from exact diagonalization ( ed ) for @xmath183 sites are included in fig . [ fig:1 ] ( thin lines ) and we find that the agreement is very good . deviations at low temperatures for @xmath184 and @xmath185 are due to finite - size effects , i.e. , the ed results are not yet converged to the thermodynamic limit . for larger fields @xmath186 , deviations between ed and mf are negligibly small . + from eq . ( [ eq : mf2 ] ) , we can derive the critical field @xmath144 within the hartree - fock approximation . at @xmath12 and @xmath27 , the ground state is the fully polarized state with @xmath187 , i.e. , the parameter @xmath188 from eq . ( [ eq : mf2 ] ) vanishes . consequently , we find @xmath189 in accordance with the exact result@xcite . indeed , the low - energy theories along the line @xmath190 and for @xmath77 are equivalent in the sense that they are characterized by the same luttinger parameter@xcite . within bosonization , the line @xmath27 is particular since the velocity of the elementary excitations vanishes here . + regarding the low - temperature behavior of the thermal drude weight we can then conjecture that it is given by eqs . ( [ eq : mf7 ] ) and ( [ eq : mf9 ] ) for @xmath27 , independently of @xmath7 . we will come back to this issue in sec . [ sec:4 ] where we discuss the results from exact diagonalization for @xmath167 . the case of @xmath191 and @xmath34 , however , seems to be an exception as we have found indications for @xmath192 at low temperatures before@xcite . here , the existence of many low - lying excitations might complicate the situation . + in the ferromagnetic state and for low temperatures , the parameter @xmath188 from eq . ( [ eq : mf2 ] ) is exponentially suppressed and the average local magnetization is @xmath193 . thus , to leading order in @xmath31 the low - temperature dependence of @xmath98 is independent of @xmath7 , similar to the case of @xmath27 , and the thermal drude weight is exponentially suppressed @xmath194 with @xmath195 . + in the gapless state our mean - field theory results confirm that @xmath196 for @xmath167 and low temperatures . however , the mean - field prefactor @xmath197 will be renormalized if interactions are fully accounted for ; see eq . ( [ eq : xx ] ) . + in summary , we have obtained the leading low - temperature contributions to @xmath10 in the regimes ( i ) and ( ii ) of table [ tab:1 ] using mean - field theory and conformal field theory . mean - field theory provides a reasonable quantitative description of the transport coefficients for small @xmath7 and @xmath2 as well as for @xmath198 . in this section , we first present numerical results for the thermal drude weight of the heisenberg chain ( @xmath125 ) . second , the field dependence of @xmath98 for intermediate temperatures @xmath31 is analyzed . next , @xmath98 for @xmath27 is discussed for different choices of the anisotropy @xmath199 and finally , we make some remarks on the lower bound for the spin drude weight @xmath200 given in eq . ( [ eq:21 ] ) . while @xmath104 still eludes an exact analytical treatment for arbitrary temperatures , analytically exact results for @xmath105 and @xmath106 of the heisenberg chain have very recently been reported in ref . . + let us first address a technical issue , namely the appropriate choice for @xmath104 in eq . ( [ eq : k5a ] ) . for the case of zero magnetic field , we know from our previous study ref . that @xmath201 and @xmath202 exhibit a different finite - size behavior at @xmath34 . this is similar to the situation at finite fields . the inset of fig . [ fig:2](a ) shows both @xmath201 and @xmath202 for @xmath125 and @xmath185 , and we see that first , @xmath202 is well converged at low temperatures ; and second , a large difference between @xmath203 and @xmath202 is visible at low temperatures . the thermal drude weight @xmath98 , resulting from either inserting @xmath201 or @xmath202 in eq . ( [ eq : k5a ] ) , is shown in fig . [ fig:2](a ) . we have decided to use @xmath110 in the numerical study for consistency reasons , since then , all drude weights entering in eq . ( [ eq : k5a ] ) have a similar finite - size dependence at low temperatures , characterized by the exponential suppression at low temperatures due to the finite - size gap . on the contrary , using @xmath202 leads to an artificial double peak structure in @xmath98 ; seen in fig . [ fig:2](a ) . we have checked that a similar scenario arises for @xmath77 for finite systems . however , for this case the drude weight can be computed exactly in the thermodynamic limit and we find that one of the two maxima disappears . thus we expect an analogous behavior for @xmath167 , supporting the choice of @xmath119 instead of @xmath111 . + further numerical results for @xmath98 of the heisenberg chain are provided in fig . [ fig:2](b ) for @xmath204 . the main features of the thermal drude weight can be summarized as follows : ( i ) for @xmath148 , finite - size effects are small for @xmath205 see fig . [ fig:2](a ) ; ( ii ) for @xmath149 , finite - size effects are negligible ; ( iii ) the position of the maximum depends on the magnetic field ; ( iv ) @xmath98 is strongly suppressed as the magnetic field is increased . + as both @xmath105 and @xmath106 converge rapidly to the thermodynamic limit at high temperatures , the small finite - size effects observed for @xmath205 [ see fig . [ fig:2](a ) ] are due to @xmath104 [ see the inset of fig . [ fig:2](a ) ] . at low @xmath31 , @xmath98 increases with system size @xmath4 while it decreases with growing @xmath4 at high temperatures . the vanishing of pronounced finite - size effects upon approaching the line @xmath27 from below can be ascribed to the fact that a description in terms of free fermions with parameters independent of @xmath7 is valid here , as was already evidenced in the previous section . for the ferromagnetic state ( @xmath206 ) , the curves shown in fig . [ fig:2](b ) for @xmath207 are indistinguishable from the corresponding ones for @xmath183 ( not included in the figure ) within the line width . + regarding the position of the maximum , there is evidence that it is first shifted to higher temperatures when the field is switched on as compared to the case of @xmath34 ; see fig . [ fig:2](a ) . a precise determination of its position in the intermediate gapless phase is somewhat complicated as typically , the numerical data converge well down to roughly only the peak temperature . still , there are indications that at strong fields @xmath208 , the maximum tends to be located at lower temperatures than for @xmath34 . this can be seen , for instance , in the case of @xmath209 in fig . [ fig:2](b ) . in the polarized state , @xmath98 definitely peaks at larger temperatures than at vanishing field due to its exponential suppression at low temperatures . + the decrease of @xmath98 as a function of increasing magnetic field as mentioned in the preceding discussion of the heisenberg chain is also observed for other choices for the anisotropy @xmath7 . this is demonstrated for @xmath210 at @xmath211 in the main panel of fig . [ fig:3 ] , where @xmath98 is shown as a function of the magnetic field @xmath2 and plotted versus @xmath212 . in contrast to @xmath98 , @xmath105 grows with increasing magnetic field at intermediate temperatures . for illustration , @xmath105 is plotted versus @xmath212 at @xmath211 in the inset of fig . [ fig:3 ] for the same choice of parameters as in the main panel . it exhibits a maximum at large fields , which increases and its position approaches @xmath27 when the temperature is lowered . thus , indications of the transition to the ferromagnetic phase are visible in @xmath105 , but not present in @xmath98 . note , however , that all three curves in the main panel of fig . [ fig:3 ] almost pass through the same point for @xmath213 . for @xmath214 , @xmath105 is enhanced by the magnetic field and we can therefore conclude that the decrease of @xmath98 as a function of magnetic field is due to a cancellation of @xmath105 and the magnetothermal correction in eq . ( [ eq : k5a ] ) . + along the critical line @xmath215 , further evidence for universal low - temperature behavior can be found by ed . this can be seen in fig . [ fig:4 ] , where we present @xmath98 for @xmath216 and @xmath183 . the curve for @xmath77 is also included in the figure ; this one , however , is exact in the thermodynamic limit . below @xmath217 , the curves lie on top of each other . small deviations at lowest temperatures visible in the plot can be ascribed to the presence of finite - size gaps . this supports our conclusion from sec . [ sec:3 ] that eqs . ( [ eq : mf7 ] ) and ( [ eq : mf9 ] ) hold for arbitrary @xmath199 and further numerical data ( not included in the figure ) show that it is also correct for @xmath218 . + finally , let us turn to the inequality eq . ( [ eq:21 ] ) for the spin drude weight @xmath219 introduced in sec . [ sec:2 ] . here , we want to discuss to which extent the inequality eq . ( [ eq:21 ] ) is exhausted by @xmath78 at finite magnetic fields and finite temperatures . an analogous analysis in the limit of @xmath220 can be found in ref . . to this end we compare @xmath201 and @xmath221 in fig . [ fig:5 ] . note that first , the relation @xmath222 is equivalent to the positivity of the thermal drude weight @xmath223 . second , @xmath224 implies a very small thermal drude weight and thus , the comparison provided in fig . [ fig:5 ] also reveals the relative size of the two contributions to @xmath98 in eq . ( [ eq : k5a ] ) , namely @xmath105 and the magnetothermal correction @xmath225 . in fig . [ fig:5 ] , results are shown for @xmath125 , @xmath207 sites , and @xmath226 . for the sake of clarity , data for smaller system sizes are not included in the figure . differences between the curves for @xmath183 and @xmath207 are anyway only pronounced for temperatures @xmath227 and become smaller as the magnetic field @xmath2 increases . + figure [ fig:5 ] allows for three major observations : ( i ) @xmath228 at low temperatures and for all cases shown in the figure ; ( ii ) @xmath229 approximates @xmath203 the better the larger the magnetic field is ; ( iii ) significant deviations are present for high temperatures implying that for a quantitative description of @xmath104 using eq . ( [ eq:21 ] ) , more conserved quantities need to be considered in eq . ( [ eq:21 ] ) . + our comparison provides , at least for finite system sizes , a quantitative measure of the temperature range where @xmath230 . point ( i ) indicates that analytical approaches can make use of @xmath229 for a quantitative description of @xmath104 at low temperatures as it has been done by fujimoto and kawakami within a continuum theory in ref . . the quantities that appear on the right hand side of eq . ( [ eq:22 ] ) are less involved than eqs . ( [ eq : k3 ] ) and ( [ eq : k8 ] ) for @xmath104 , as the former are static correlators . furthermore , for finite magnetic fields , we suggest to compute @xmath104 analytically from eq . ( [ eq:21 ] ) , taking into account some of the conserved quantities @xmath231 , which are in principle known ( see , e.g. , ref . ) . such a procedure is applicable to @xmath84 and might circumvent the ambiguities in the results encountered in recent computations of @xmath232 ( refs . ) . the latter have used eq . ( [ eq : k8 ] ) directly or kohn s formula@xcite , equivalently . regarding the relative size of @xmath105 and the magnetothermal correction , we see that the latter becomes more relevant the larger the magnetic field is which leads to the strong suppression of @xmath98 . this is consistent with results of the previous sections of this paper . in this paper we have studied the thermal drude weight of the @xmath1 model with exchange anisotropy @xmath199 in finite magnetic fields using mean - field theory and exact diagonalization . magnetothermal effects have been taken into account and the condition of zero magnetization current flow has been applied . let us now summarize the main findings and relate them to experiments . + we have discussed the low - temperature limit of the thermal drude weight @xmath98 and we have given arguments that it changes from an algebraic behavior for @xmath233 to an exponentially activated behavior in the polarized state for @xmath206 . in addition , the leading term at low temperatures along the critical line @xmath234 , is universally given by @xmath235 , where the prefactor @xmath236 , given in eqs . ( [ eq : mf7 ] ) and ( [ eq : mf9 ] ) , is independent of @xmath7 . in the gapless phase , the leading contribution to @xmath98 is linear in the temperature with a field- and anisotropy dependent prefactor . in consequence , the thermal drude weight @xmath98 can be expected to be proportional to the specific heat in the gapless state in the low - temperature limit , where the velocity of elementary excitations is constant . + further , the drude weight is suppressed by the magnetic field , which can be ascribed to the increase of the magnetothermal correction relative to the pure thermal drude weight @xmath105 . as a third result , the position of the maximum of @xmath98 depends non - monotonically on the magnetic field . while in the present paper , we have focused on the thermal drude weight @xmath10 under the condition of zero spin - current flow , our analysis of the full transport matrix eq . ( [ eq : k0 ] ) can easily be extended to a variety of other transport situations , which would be characterized by different combinations of the drude weights . + turning now to experiments , we emphasize that for a description of realistic materials external scattering has to be accounted for . in a simple picture , one may expect the drude peak to be broadened by external scattering mechanisms . the behavior in magnetic fields that one may observe in experiments will likely depend both on external scattering rates as well as on the thermal drude weight . nevertheless , one may speculate that qualitative trends of the field dependence of the thermal drude weight are reflected in thermal transport experiments . + in recent experiments on quasi one - dimensional magnetic materials@xcite , the thermal conductivity has often been found to be insensitive to the application of an external magnetic field . this is , however , explained by the large absolute value of the exchange coupling in these materials , being typically of the order of magnitude of @xmath237 k. it would therefore be desirable to perform measurements on materials with a moderately small exchange coupling to check our results . still , it might be very difficult to reach the saturation field @xmath144 in realistic materials , but at least the qualitative features like the suppression of the thermal conductivity or a shift of the maximum could be verified . in particular , experiments in relatively large magnetic fields may provide an indirect probe of spin currents in one - dimensional quantum magnets , which have so far not yet been observed directly in experiments . conclusions about the low - temperature limit would require reliable methods to separate the magnetic contribution from the phonon part , which is a challenging task in the interpretation of experiments . nevertheless , we believe that further experiments on the field dependence of the thermal conductivity could hint at the nature and mechanisms of magnetic transport properties . this work was supported by the dfg , schwerpunktprogramm 1073 . it is a pleasure to thank b. bchner , d.c . cabra , and c. hess for fruitful discussions . we are indebted to c. hess for a careful reading of the manuscript and valuable suggestions . we acknowledge support by the rechenzentrum of the tu braunschweig where parts of the numerical computations have been performed on a compaq es45 . 99 a.v . sologubenko , k. giann , h.r . ott , u. ammerahl , and a. revcolevschi , phys . rev . lett . * 84 * , 2714 ( 2000 ) . c. hess , c. baumann , u. ammerahl , b. bchner , f. heidrich - meisner , w. brenig , and a. revcolevschi , phys . b * 64 * , 184305 ( 2001 ) . k. kudo , s. ishikawa , t. noji , t. adachi , y. koike , k. maki , s. tsuji , and ken - ichi kumagai , j. phys . jpn . * 70 * , 437 ( 2001 ) . sologubenko , e. felder , k. giann , h.r . ott , a. vietkine , and a. revcolevschi , phys . b * 62 * , r6108 ( 2000 ) ; a.v . sologubenko , k. giann , h.r . ott , a. vietkine , and a. revcolevschi , _ ibid . _ * 64 * , 054412 ( 2001 ) . sologubenko , h.r . ott , g. dhalenne , and a. revcolevschi , europhys . lett . * 62 * , 540 ( 2003 ) . c. hess , h. elhaes , b. bchner , u. ammerahl , m. hcker , and a. revcolevschi , phys . * 93 * , 027005 ( 2004 ) . j.v . alvarez and c. gros , phys . lett . * 89 * , 156603 ( 2002 ) ; c. gros and j.v . alvarez , _ ibid . _ * 92 * , 069704 ( 2004 ) . f. heidrich - meisner , a. honecker , d.c . cabra , and w. brenig , phys . b * 66 * , 140406(r ) ( 2002 ) ; phys . * 92 * , 069703 ( 2004 ) ; j. mag . mag . mat . * 272 - 276 * , 890 ( 2004 ) . f. heidrich - meisner , a. honecker , d.c . cabra , and w. brenig , phys . b * 68 * , 134436 ( 2003 ) . e. orignac , r. chitra , and r. citro , phys . b * 67 * , 134426 ( 2003 ) . k. saito , phys . b * 67 * , 064410 ( 2003 ) . e. shimshoni , n. andrei , and a. rosch , phys . b * 68 * , 104401 ( 2003 ) . x. zotos , phys . lett . * 92 * , 067202 ( 2004 ) . f. heidrich - meisner , a. honecker , d.c . cabra , and w. brenig , to appear in physica b , cond - mat/0406378 ( unpublished ) . a.v . rozhkov and a.l . chernyshev , cond - mat/0407257 ( unpublished ) . a. klmper and k. sakai , j. phys . a * 35 * , 2173 ( 2002 ) . k. sakai and a. klmper , j. phys . a * 36 * , 11 617 ( 2003 ) . k. louis and c. gros , phys . rev . b * 67 * , 224410 ( 2003 ) . k. sakai and a. klmper , cond - mat/0410192 ( unpublished ) . niemeijer and h.a . w. van vianen , phys . * 34a * , 401 ( 1971 ) . x. zotos , f. naef , and p. prelovek , phys . b * 55 * , 11029 ( 1997 ) . cabra , a. honecker , and p. pujol , phys . b * 58 * , 6241 ( 1998 ) ; d.c . cabra and p. pujol , _ field - theoretical methods in quantum magnetism _ , lect . notes phys . * 645 * , 253 - 305 ( 2004 ) . g. d. mahan , _ many - particle physics _ ( plenum press , new york , 1990 ) . k. kawasaki , prog . theor . * 29 * , 801 ( 1963 ) . shastry and b. sutherland , phys . 65 * , 243 ( 1990 ) . d. forster , _ hydrodynamic fluctuations , broken symmetry , and correlation functions _ ( benjamin , reading , mass . p. mazur , physica * 43 * , 533 ( 1969 ) ; m. suzuki , _ ibid . _ * 51 * , 277 ( 1971 ) . s. fujimoto and n. kawakami , phys . * 90 * , 197202 ( 2003 ) . w. kohn , phys . 133 * , a171 ( 1964 ) . h. castella , x. zotos , and p. prelovek , phys . 74 * , 972 ( 1995 ) . narozhny , a.j . millis , and n. andrei , phys . b * 58 * , r2921 ( 1998 ) ; d.a . rabson , b. n. narozhny , and a.j . millis , phys . b * 69 * , 054403 ( 2004 ) . d. scalapino , s.r . white , and s. zhang , phys . b * 47 * , 7995 ( 1993 ) . t. giamarchi , and b. s. shastry , phys . b * 51 * , 10915 ( 1995 ) . s. kirchner , h. g. evertz , and w. hanke , phys . b * 59 * , 1825 ( 1999 ) . x. zotos and p. prelovek , phys . b * 53 * , 983 ( 1996 ) . x. zotos , phys . lett . * 82 * , 1764 ( 1999 ) . j.v . alvarez and c. gros , phys . * 88 * , 077203 ( 2002 ) ; phys . rev . b * 66 * , 094403 ( 2002 ) . long , p. prelovek , s. el shawish , j. karadamoglou , and x. zotos , phys . b * 68 * , 235106 ( 2003 ) . bogoliubov , a.g . izergin , and v.e . korepin , nucl . b * 275 * , 687 ( 1986 ) ; v.e . korepin , n.m . bogoliubov , and a.g . izergin , _ quantum inverse scattering method and correlation functions _ ( cambridge university press , cambridge , england , 1993 ) . p.r . hammar , m.b . stone , d.h . reich , c. broholm , p.j . gibson , m.m . turnbull , c.p . landee , and m. oshikawa , phys . b * 59 * , 1008 ( 1999 ) . s. qin , m. fabrizio , lu yu , m. oshikawa , and i. affleck , phys . b * 56 * , 9766 ( 1997 ) . dzhaparidze and a.a . nersesyan , jetp lett * 27 * , 334 ( 1978 ) ; v.l . pokrovsky and a.l . talapov , phys . * 42 * , 65 ( 1979 ) . grabowski and p. matthieu , ann . phys . * 243 * , 299 ( 1996 ) . a. klmper , private communication .	we study the heat conduction of the spin-@xmath0 @xmath1 chain in finite magnetic fields where magnetothermal effects arise . due to the integrability of this model , all transport coefficients diverge , signaled by finite drude weights . using exact diagonalization and mean - field theory , we analyze the temperature and field dependence of the thermal drude weight for various exchange anisotropies under the condition of zero magnetization - current flow . first , we find a strong magnetic field dependence of the drude weight , including a suppression of its magnitude with increasing field strength and a non - monotonic field - dependence of the peak position . second , for small exchange anisotropies and magnetic fields in the massless as well as in the fully polarized regime the mean - field approach is in excellent agreement with the exact diagonalization data . third , at the field - induced quantum critical line between the para- and ferromagnetic region we propose a universal low - temperature behavior of the thermal drude weight .
the optical or conductivity properties of generally anisotropic material are described by ( generally complex ) permittivity @xmath11 or conductivity @xmath10 tensor , being mutually related by ( in si units ) @xmath13 , where @xmath14 being kronecker delta and @xmath15 light frequency ( i.e. @xmath16 corresponds to d.c . current ) . onsager relations require antisymmetry in magnetization direction , @xmath17 . as various magnetization direction causes small perturbation to the crystal optical properties , @xmath11 , they can be expressed as taylor series of permittivity tensor in direction of @xmath18 @xcite @xmath19 where superscripts denote zeroth , first , second etc . order in magnetization . optical permittivity independent on magnetization direction is @xmath20 and in case of cubic crystal , it has unitary form @xmath21 . the term linear in magnetization @xmath22 is described by third - order tensor @xmath23 , where onsager relations assures @xmath24 . in case of cubic crystals , it has simple form @xmath25 , where @xmath26 is levi - civita symbol and @xmath27 is linear magnetooptic element . the term quadratic in magnetization @xmath28 is described by fourth order tensor @xmath29 , with onsager relations assuring @xmath30 . in case of cubic crystal with @xmath31 $ ] , @xmath32 $ ] , @xmath33 $ ] , the @xmath29 tensor can be written in compact form as @xcite @xmath34 = \left [ \begin{array}{cccccc } g_{11 } & g_{12 } & g_{21 } & 0 & 0 & 0 \\ g_{21 } & g_{11 } & g_{12 } & 0 & 0 & 0 \\ g_{12 } & g_{21 } & g_{11 } & 0 & 0 & 0 \\ 0 & 0 & 0 & 2g_{44 } & 0 & 0 \\ 0 & 0 & 0 & 0 & 2g_{44 } & 0 \\ 0 & 0 & 0 & 0 & 0 & 2g_{44 } \end{array } \right ] \ , \left [ \begin{array}{c } m_x^2 \\ m_y^2 \\ m_z^2\\ m_ym_z\\ m_zm_x\\ m_xm_y \end{array } \right]\ ] ] in order to express @xmath35-elements for crystals with cubic symmetry , the elements can be rearranged defining @xmath36 and @xmath37 . note that alternative way used to express anisotropy of the permittivity is @xmath38 @xcite . in case of cubic crystals owning point o@xmath39 symmetry , @xmath40 and hence @xmath41 , @xmath42 @xcite . in case of isotropic medium , then both @xmath43 , @xmath42 . furthermore , notice , that spectra of @xmath44 element can not be determined ( only value of @xmath45 ) as for the ferromagnetic materials , the magnitude of magnetization vector @xmath46 is constant for any magnetization direction and hence @xmath44 becomes inseparable part of @xmath47 . hence , spectra of any permittivity element @xmath11 ( up to second order in magnetization ) can be expressed as weighted sum of principal spectra , where the weights are determined by crystallographic and magnetization directions . table [ t : scans ] demonstrates those weights for @xmath48 $ ] , @xmath32 $ ] and @xmath33 $ ] oriented crystal for two magnetization scans , used later in ab - initio calculations . namely , the angular dependence of @xmath11 permittivity elements is expressed for two magnetization scans , in - plane ( @xmath49 , @xmath50 ) and out - of - plane ( @xmath51 , @xmath52 ) scans as the magnetization passes through all high symmetry directions [ 001 ] , [ 110 ] and [ 111 ] . angles @xmath53 , @xmath54 define magnetization direction , being @xmath55 = [ \sin\theta \cos\varphi , \sin\theta \sin\varphi , \cos\theta ] $ ] . see ref . @xcite providing weighting factors of principal spectra for general magnetization direction and several crystallographic orientations . cc in - plane scan : @xmath56 , @xmath57 : & out - of - plane scan : @xmath58 , @xmath59 : + [ cols="^,^",options="header " , ] to determine all principal spectra we first separate linear and quadratic contributions , by extracting off - diagonal linear in magnetization ( @xmath60 , @xmath61 ) , off - diagonal quadratic in magnetization ( @xmath62 , @xmath61 ) and diagonal quadratic in magnetization ( @xmath63 , @xmath61 ) contributions from which the principal spectra are determined . the examples of angular dependencies of those contributions are demonstrated in fig . [ f : epsdiff ] for fe at photon energies 1.3ev and 56ev . in figs . [ f : specfe][f : specni ] we show principal spectra of @xmath47 , @xmath27 , @xmath64 , @xmath2 for bcc fe , fcc co and fcc ni . as those materials own point o@xmath39 symmetry @xmath65 . the parts of the spectra related with light absorption are shown in first four lines , whereas last four lines are related to light dispersion ( determined by kramers - kronig relations @xcite ) . two approaches to determine principal spectra are used . ( i ) we determine principal spectra from ab - initio calculated @xmath11 for in - plane and out - of - plane magnetization scans . for both scans , the angular dependences as predicted by symmetry arguments for cubic crystals are presented in tab . [ t : scans ] . knowing those dependences we can extract all principal spectra . two pairs of spectra of @xmath45 and @xmath2 are determined , each pair from one scan . for computational details , see appendix [ app : comp ] . ( ii ) we determine principal spectra solely from @xmath11 calculated for magnetization pointing in fundamental symmetry directions such as [ 100 ] , [ 110 ] or [ 111 ] being sufficient to determine all principal spectra . [ t : fund ] summarizes analytical expressions of principal spectra from known permittivity elements with magnetization at fundamental magnetization directions . two pairs of spectra are determined , spectra of @xmath45 from [ 100 ] and [ 110 ] magnetization directions , and spectra of @xmath2 from [ 110 ] and [ 111 ] magnetization directions . obviously , when calculated angular dependences follow relations predicted by symmetry arguments , and when we neglect contributions of higher order effects , both approaches providing four pairs of spectra should be identical . all those principal spectra are shown in figures [ f : specfe]-[f : specni ] for @xmath45 and @xmath2 . one can see that although all elements and spectral ranges were calculated from equal electronic structure , there is very good agreement between all approaches in case of 2@xmath0- and 3@xmath0-edges spectral range . on the other hand , agreement between @xmath35-spectra in visible range is much smaller ( particularly visible for @xmath45 in fe ) . this can be also seen on fig . [ f : epsdiff ] , where demonstrated example ( fe at 1.3ev ) where , in case of both in - plane scan and out - of - plane scans , the values of e.g. @xmath66 or @xmath67 are shifted . the discrepancy stems from the numerical precision of the brillouin zone integration technique used to calculate optical properties for photon energies below about 12ev . for more details see the appendix [ app : comp ] . finally , recall , for isotropic materials @xmath68 . however , in 2@xmath0-edge and 3@xmath0-edge spectra , the peaks of @xmath2 and @xmath64 corresponding to l@xmath4 and m@xmath4 edges have roughly opposite signs , demonstrating very strong ( dominating ) anisotropy of quadratic effects in cubic bcc fe , fcc co and fcc ni in their respective spectral ranges . very strong anisotropy signal in xmld is demonstrated and analysed by both experimental and theoretical calculations at both 2@xmath0-edge and 3@xmath0-edge in case of the bcc fe and fcc co @xcite and compare the corresponding experimental spectral shapes with spectral shape of here predicted g - tensor elements ( @xmath69 ) related to the quadratic contributions of magneto - optical effects . note also the determined im@xmath70 , element related to the linear magneto - optical effects ( middle column of fig . [ f : specfe ] ) has the same spectral shape as recently measured t - moke spectra of bcc fe at the 3@xmath0 edges @xcite . furthermore , fig . [ f : specfe ] demonstrates ( visible particularly for 3@xmath0-edges ) that light absorption @xmath71 has form of sum of lorentzian peaks . then the absorption of linear element @xmath27 is roughly proportional to the first derivation of @xmath47 : @xmath72 , and quadratic elements are approximately proportional to second derivative of @xmath47 : @xmath73 @xcite . notice also different spectral features in both @xmath35-spectra , e.g. , in case of fe at 4.7ev ( or ni at 1.9ev ) , there is a strong peak in @xmath64 spectra , but no such a feature in @xmath2 spectra . similar can be told about feature for ni at 0.4ev , presented only in @xmath2 spectra and absent in @xmath64 spectra . finally notice that for ( 001 ) oriented crystals , ( a ) the real / imaginary part of @xmath64 is proportional to magnetic linear birefringence / dichroism ( mlb / mld ) , respectively , with @xmath74 $ ] . namely , it corresponds to difference of light speed / absorption for @xmath75 $ ] and @xmath76 $ ] . similarly , @xmath2 is proportional to mlb / mld with @xmath77 $ ] , i.e. @xmath78 $ ] and @xmath79 $ ] . once more let us stress that spectral shapes of g - tensors can be directly compared with recorded xmld spectra w.r.t . various m and e vectors of refs . this can be seen also seen on expressions of @xmath80 and @xmath2 using kubo formula ( [ eq : sigs][eq : sig44 ] ) below . ( b ) in case of both @xmath18 and @xmath81 vectors laying in ( 001)-plane ( i.e. in - plane magnetization and normal light incidence ) , the isotropic part of mlb / mld , is proportional to the term @xmath82 , whereas the anisotropic part of mlb / mld ( i.e. varying with crystallographic orientation ) is given by @xmath83 @xcite . we express analytically all principal spectra ( @xmath47 , @xmath27 , @xmath64 , @xmath2 ) by quantum mechanical description ( i.e. by modified kubo formula ) within the electric dipole approximation . in cartesian system defined @xmath31 $ ] , @xmath32 $ ] , @xmath33 $ ] and for magnetization along @xmath84 $ ] direction , the absorption ( dissipative ) part is the well - known non - magnetic ( isotropic ) term @xmath85 is @xcite @xmath86 \left\|\langle i\|\vec{\pi}\cdot\hat{z}\|f\rangle\right\|^2 \delta(\omega-\omega_{fi})\ , . \label{eq : sig0}\ ] ] where @xmath87 is matrix elements for light polarization ( electric field ) along @xmath88 with @xmath89 , @xmath90 being initial and final electron states , respectively . kinetic momentum operator @xmath91 is sum of canonical momentum @xmath92 , vector potential of the magnetic field @xmath93 and spin - orbit coupling between electron potential @xmath94 and its spin @xmath95 , @xmath96 , @xmath97 are occupation functions . the dispersive part of the dielectric functions are determined by kramers - kronig relations @xcite . note that according to tab . [ t : scans ] , eq . ( [ eq : sig0 ] ) calculates imaginary part of @xmath98 , not solely @xmath99 . however , @xmath44 can not be determined , as length of magnetization vector is constant for ferromagnetic materials . hence , @xmath44 is inseparable part of @xmath47 and we neglect it compared to @xmath47 . when both magnetization and light propagation is along @xmath88 axis , @xmath84 $ ] , then eigenmodes of light propagation are circularly left and circularly right polarizations . therefore , the linear term @xmath27 is given as a difference of light absorption for circularly - left and circularly - right polarizations as determined by bruno _ et al _ @xcite @xmath100 \left ( \|\langle i\|\vec{\pi}\cdot(\hat{x}+i\hat{y})\|f\rangle\|^2 - \|\langle i\|\vec{\pi}\cdot(\hat{x}-i\hat{y})\|f\rangle\|^2 \right ) \delta(\omega-\omega_{fi})\ , . \label{eq : sigk}\ ] ] in case magnetization is along @xmath101 $ ] direction ( i.e. along @xmath102 ) , the symmetry arguments predict @xmath103 ( tab . [ t : scans ] ) . by employing eq . ( [ eq : sig0 ] ) , @xmath45 can be expressed as @xmath104 \left ( \|\langle i\|\vec{\pi}\cdot\hat{x}\|f\rangle\|^2 - \|\langle i\|\vec{\pi}\cdot\hat{y}\|f\rangle\|^2 \right ) \delta(\omega-\omega_{fi})\ , . \label{eq : sigs}\ ] ] this equation corresponds to the case when light propagation is along @xmath88 axis and @xmath105 , then eigenmodes of light propagation are linearly polarized waves parallel and perpendicular to magnetization direction . hence , the quadratic term @xmath45 is given by difference of absorptions for those two light polarizations . similarly , in cartesian system defined @xmath106 $ ] , @xmath107 $ ] , @xmath108 $ ] and for magnetization @xmath109 $ ] , the element @xmath110 is expressed formally in the same way as @xmath45 @xcite . transforming back to cartesian system @xmath31 $ ] , @xmath32 $ ] , @xmath33 $ ] , quadratic term @xmath2 is expressed by difference of absorption of linear light polarization along [ 110 ] and @xmath111 $ ] directions when magnetization is along @xmath112\parallel \hat{x}+\hat{y}$ ] @xmath113 \left ( \|\langle i\|\vec{\pi}\cdot(\hat{x}+\hat{y})\|f\rangle\|^2 - \|\langle i\|\vec{\pi}\cdot(-\hat{x}+\hat{y})\|f\rangle\|^2 \right ) \delta(\omega-\omega_{fi})\ , . \label{eq : sig44}\ ] ] finally note , that here expressed @xmath45 and @xmath2 corresponds to principal spectra calculated from [ 100 ] and [ 110 ] magnetization directions , as denoted in tab . [ t : fund ] and figs . [ f : specfe][f : specni ] the permittivity tensor phenomenologically describes optical properties of matter . in this section , we provide a brief relation between elements of permittivity tensor and measured effects in common magnetooptical techniques . from experimental point of view , the measured quantities are straightforwardly related with the reflectivity matrix , usually expressed in jones formalism @xcite . then , measured optical quantities , such as kerr effect , magnetic circular dichroism ( mcd ) , mld , reflectivity , magnetoreflectivity , ellipsometry etc . are directly related to this reflection matrix . in case of multilayer structure , the relation between reflection matrix and permittivities of constituent layers are routinely described by yeh @xcite or berreman @xcite formalisms , based on propagation of light eigenmodes in the layer , where eigenmodes intensities are determined by boundary conditions on the interfaces ( continuity of transversal component of electric and magnetic fields through the interfaces ) . as optical anisotropies are usually week , the perturbation optical approach has been developed as well @xcite , handling optical anisotropy as a perturbation from optical isotropic response . analytical solution of optical response of multilayer structure with generally optically anisotropic layers is very complicated . hence , analytical solution is mostly expressed for two simple structures : ( i ) reflection on interface of bulk material . however , even in this simple case the analytical expression of reflection matrix from generally optically anisotropic media is still very complicated @xcite . ( ii ) ultrathin approximation , where thickness of anisotropic layer is assumed to be much thinner than wavelength of light in the material @xcite . in this approximation , optical response can be expressed analytically rather simply . although different multilayer structures provides different relations between constituent permittivity tensor and reflection matrix , some relations can be generalized . in following we assume one of the following simplifications : * many magnetooptical technique ( e.g. moke , transverse moke ( tmoke ) or xmcd ) are selectively sensitive to off - diagonal permittivity elements of the fm layer . in this case , to describe effect measured by the setup , we neglect difference between the diagonal permittivity elements , and assume all diagonal permittivity elements to be equal . * on contrary , mld employs change in reflection / transmission of linearly polarized light . hence , when handling intensities of those reflected or transmitted linearly polarized light , we can neglect influence of the off - diagonal permittivity elements and assume them vanished . in case of ( a ) approximation , the reflection matrix element can be written in general form ( neglecting higher orders in @xmath11 , @xmath61 ) as @xmath114 where @xmath115 , @xmath116 , @xmath117 , @xmath118 and @xmath119 are optical terms given by details of optical and geometrical properties of the multilayer structure , being independent on the off - diagonal permittivity of the fm layer @xmath11 , @xmath61 . note that for zero incidence angle ( @xmath120 ) , @xmath121 as experimentally measured magnetooptical quantities are directly related to reflection matrix , we demonstrate relation between off - diagonal permittivity elements and few magnetooptic techniques : the complex moke angle @xmath122 ( @xmath53 is kerr rotation , @xmath123 is kerr ellipticity ) can be expressed as @xmath124 where @xmath125 , @xmath126 are again optical terms depending on all other optical properties of the constituent layers @xcite . note that for @xmath120 , @xmath127 and @xmath128 , and for small incidence angle @xmath129 . tmoke is change of p - reflectivity due to off - diagonal elements is described by term @xmath130 , being usually induced by transverse magnetization . then , tmoke writes @xmath131 where again @xmath132 is optical term depending on optical details of the multilayer . mcd ( and obviously also xmcd ) is difference of light absorption for circularly left and right polarized light . to be consistent with previous part , we express mcd in reflection , mcd=@xmath133 , where @xmath134 , @xmath135 is reflection for circularly right , left polarized light , respectively : @xmath136 where @xmath137 denotes complex conjugation . in any case , the relations between off - diagonal permittivity elements and mcd in transmission are formally identical . assuming small incidence angle ( @xmath138 ) , and hence @xmath139 , we got @xmath140 where again @xmath141 and @xmath142 are optical terms depending on optical details of the structure . in case of mld ( or xmld ) , the approximation ( b ) is employed , assuming off - diagonal term vanished . then , off - diagonal reflectivity coefficients vanishes as well ( @xmath143 ) and diagonal reflectivities can be written as @xmath144 where @xmath145 , @xmath146 and @xmath147 are again optical terms and @xmath148 , @xmath149 @xmath150 are differences of diagonal permittivity elements from from @xmath47 ( i.e. differences between anisotropic and isotropic permittivities ) . note in case @xmath120 term @xmath147 vanish and @xmath151 . also , in case of metals , term @xmath147 can be neglected as well even for non - zero incidence angle due to large refractive index of metals . mld is defined as a difference of reflectivity ( or transmittivity ) between incident @xmath152 and @xmath0 polarized light . further assuming @xmath120 , mld writes @xmath153 where @xmath94 is again an optical term . in conclusion , we provide principal spectra , _ i_.e . decomposition of isotropic , linear and quadratic contributions in magnetization of magnetooptic effects up to the second order in magnetization . the values of magnetooptic permittivity elements are determined as weighted sum of principal spectra , where weights are determined solely by magnetization direction . although in x - ray range , this spectroscopy is already established , it was not introduced for visible range . in case of ferromagnetic cubic crystals owning point symmetry , all magneto - transport up to second - order effects are described by spectra of four complex principal spectra , denoted @xmath154 , @xmath27 , @xmath64 and @xmath2 , as predicted by symmetry arguments . this approach expresses second - order magnetooptic properties as bulk material property , independent on sample crystallographic orientation , layer thickness , structure details etc . the principal spectra were expressed from ab - initio calculations for bcc fe , fcc co and fcc ni . to extract principal spectra from ab - initio , we have used two methods ( i ) by comparing ab - initio calculated angular dependence of the permittivity tensor , with dependences predicted by symmetry arguments and ( ii ) using permittivity elements calculated for fundamental symmetry magnetization directions . we also express principal spectra analytically using modified kubo formula . although the spectroscopy presented here is for cubic crystals owning @xmath155 point symmetry , it can also be simply modified for materials with different crystal symmetries , higher orders in magnetizations or even different magneto - transport effects . we expect that experimental and theoretical determination of principal spectra will be routinely used in magnetooptic spectroscopy of ferromagnetic materials . financial support by czech grant agency ( 13 - 30397s ) , by it4innovations centre of excellence project ( cz.1.05/1.1.00/02.0070 ) , funded by the european regional development fund and the national budget of the czech republic via the research and development for innovations operational programme , as well as czech ministry of education via the project large research , development and innovations infrastructures ( lm2011033 ) and student grant competitions of vsb - tu ostrava ( sp2015/61 , sp2015/71 and sp2016/182 ) as well as helpful discussions with peter m. oppeneer and karel vborn are well acknowledged . the angular dependence of @xmath11 on magnetization direction with respect to the crystal axes is determined by ab - initio calculations . the electronic structure is determined by full potential linearised augmented plane wave method ( wien2k code @xcite ) , where the exchange energy is based on local spin density approximation @xcite . the spin - orbit coupling was included in a second - variation scheme @xcite . the following parameters were employed : the energy cutoff , given as the product of the muffin - tin radius and the maximum reciprocal space vector was @xmath156 , the largest reciprocal vector in the charge fourier expansion , @xmath157 , was set to 14 ry@xmath158 , and the maximum value of partial waves inside the muffin - tin spheres , @xmath159 . the brillouin zone sampling was done on grid of 27@xmath16027@xmath16027 @xmath161-points , whereas for the optical transition matrix elements @xcite a finer grid of 81@xmath16081@xmath16081 was used . the optical calculations are based on kubo formula @xcite , with omitted drude term , hence we do not present dc transport . at present , also core - hole effects are omitted . the lattice constants of studied materials are fixed being @xmath162(bcc fe ) , @xmath163(fcc co ) and @xmath164(fcc ni ) @xcite . the numerical precision shows to be limited for the beginning of the optical energy range below about 12ev,_i.e . _ the states in vicinity of the fermi level . there may be two reasons for it : ( 1 ) the calculated transition probabilities are calculated for each pair of energy levels and for each @xmath161-point of the mesh of the reciprocal space . the integration of the kubo formula is implemented by blchl - enhanced tetrahedron integration @xcite , which determines mean value of a given property ( e.g. transition probability ) inside each tetrahedron . this mean value is calculated by weighted sum of a given property at each corner of the tetrahedron , where the weights for each stationary state ( i.e. for each energy level ) are calculated from eigenenergy of each k - point and value of the fermi energy . this works well , when either initial or final tetrahedron are fully above or below fermi level . however , in case when both initial and final states are within tetrahedrons cutting fermi level , this approach may not work properly . this may explain , why dependence of permittivity elements on magnetization direction for photon energy below 12ev are rather different from expected symmetry dependence compared to those calculated for higher photon energies . ( 2 ) in blchl - enhanced tetrahedron integration , the irreducible @xmath161-points are selected to fulfil that the symmetry operation copies those irreducible @xmath161-points to all @xmath161 points inside the unit cell of the reciprocal space . obviously , in the case of high symmetry structure , such as cubic crystal , the position of @xmath161-points in the unit cell follows most of symmetry operations . however , the integration employs tetrahedrons , which does not follow so many symmetry operations . for example , in case of cubic crystal , the rotation by 90@xmath165 copy @xmath161-points to themselves , but the tetrahedrons are not copied to themselves . therefore , the numerical error of the integration from different parts of the unit cell does not follow structural symmetry , which may increase error described in the previous paragraph . showcase of imaginary part of the permittivity for bcc fe at 1.3ev ( left columns ) and 56ev ( right columns ) for differences of permittivity tensor elements on magnetization direction for ( top line ) in - plane scans and ( bottom line ) out - of - plane scans . line is fit to dependence as predicted by symmetry arguments ( tab . [ t : scans ] ) . diagonal elements ( blue ) corresponds to difference of diagonal permittivity elements @xmath166 , @xmath61 , being proportional to @xmath41 . lower triangle ( green ) @xmath62 is proportional to linear magnetooptic @xmath27 element . upper triangle ( red ) @xmath60 is proportional to @xmath2 element . ] 65ifxundefined [ 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 * * , ( ) link:\doibase 10.1103/physrev.117.689 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.87.047401 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.104.187401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.052401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.65.054415 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.196602 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.187201 [ * * , ( ) ] @noop ( ) @noop * * , ( ) @noop * * , ( ) , http://stacks.iop.org/0038-5670/18/i=6/a=r02 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1002/pssb.2220490128 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase dx.doi.org/10.1103/physrevlett.83.1862 [ * * , ( ) ] link:\doibase 10.1126/science.287.5455.1014 [ * * , ( ) ] link:\doibase 10.1103/physrevb.67.024431 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.82.184415 [ * * , ( ) ] link:\doibase 10.1103/physrevb.67.024431 [ * * , ( ) ] link:\doibase 10.1103/physrevb.74.094407 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.98.197201 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.100.067403 [ * * , ( ) ] link:\doibase 10.1103/physrevb.78.064427 [ * * , ( ) ] link:\doibase 10.1103/physrevb.82.094403 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.89.140404 [ * * , ( ) ] link:\doibase 10.1109/tmag.2014.2321632 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.83.064409 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.53.9214 [ * * , ( ) ] @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.64.235421 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * ( ) link:\doibase 10.1103/physrevb.43.6423 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevb.67.155411 [ * * , ( ) ] link:\doibase 10.1103/physrevb.52.1090 [ * * , ( ) ] @noop ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.63.205111 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( )	we provide unified phenomenological description of magnetooptic effects being linear and quadratic in magnetization . the description is based on few principal spectra , describing elements of permittivity tensor up to the second order in magnetization . each permittivity tensor element for any magnetization direction and any sample surface orientation is simply determined by weighted summation of the principal spectra , where weights are given by crystallographic and magnetization orientations . the number of principal spectra depends on the symmetry of the crystal . in cubic crystals owning point symmetry we need only four principal spectra . here , the principal spectra are expressed by _ ab - initio _ calculations for bcc fe , fcc co and fcc ni in optical range as well as in hard and soft x - ray energy range , _ i.e. _ at the 2@xmath0- and 3@xmath0-edges . we also express principal spectra analytically using modified kubo formula . there is a vast number of physical phenomena proportional to quadratic form of magnetization . in case of dc transport phenomena , the most well - known examples are anisotropy magnetoresistance ( amr ) @xcite or longitudinal hall effect @xcite . within the magnetooptic community , the field of magnetooptic effects quadratic in magnetization is complicated by incredible number of nomenclature , being called cotton - mouton effect , voigt effect , quadratic magnetooptic kerr effect ( qmoke ) @xcite , magnetic linear birefringence , x - ray magnetic linear dichroism ( xmld ) @xcite , magnetic double refraction , magnetooptic orientation effect , magnetooptic anisotropy , hubert - schfer effect or magnetorefractive effect @xcite . the nomenclature is not strictly defined , however it refers either to type of samples ( liquid , gas , solid state ) or it refers to experimental configurations of the setup ( namely detecting change of light intensity or detecting change of polarization state upon variation of magnetization direction ) . although those effects are usually not considered as single phenomena , they originate from equal parts of permittivity tensors ( i.e. from equal symmetry breaking ) . notice , that recently new types of quadratic - in - magnetization effects arose , for example anisotropic magneto - thermopower @xcite in spin - caloritronics . within some generalization , one can expect that any magneto - transport linear in magnetization will have its quadratic - in - magnetization counterpart . the first observation of magnetooptic effects quadratic in magnetization dates back more than century ago in works of kerr @xcite , majorana @xcite and cotton and moutton @xcite , where magnetic birefringence was observed in liquids and colloids . later , those quadratic magnetooptic response was observed in gases , solid state materials and obviously also in ferromagnetic materials . see large reviews of smolenskii _ et al _ @xcite and ferr , gehring @xcite from 80 s . the anisotropy of magnetooptic effects ( i.e. dependence of qmoke on crystal and field orientations ) was investigated for various systems . however , the investigation were mostly done for a single photon energy , for example fe @xcite or antiferromagnetic rbmnf@xmath1 @xcite . buchmaier _ et al _ have even determined values of principal spectra ( namely values of @xmath2 and @xmath3 ) , however also only for a single photon energy @xcite . the spectral investigations were rare , for example spectroscopy of qmoke for different orientations of ni surfaces by krinchik @xcite and parker @xcite . the evolution of quadratic magnetooptic effects in x - ray photon energy range was rather independent on those in extended visible range . the x - ray magnetic linear dichroism were theoretically predicted by thole and van der laan @xcite . due to experimental restrictions , the quadratic - in - magnetization effects in x - ray range were measured in absorption , _ i.e. _ dichroic effects @xcite and later in polarization , voigt effect @xcite . the xmld was analytically predicted for [ 100 ] and [ 111 ] magnetization axis by kunes _ et al _ @xcite . they showed that for the correct shape of the xmld spectra at l@xmath4 edges the exchange splitting of the core levels is necessary @xcite . later , with direct comparison with the experiment the angular dependence of the xmld of bcc fe shown that the role of the valence spin - orbit interaction is much weaker @xcite . the simple model of kunes _ et al _ that accounts for the difference of the xmld spectra for the high symmetry crystal directions , _ et . al _ with magnetization along [ 001 ] and [ 111 ] axis showed that the origin is expressed as different weighted spin densities of @xmath5 and @xmath6 orbitals @xcite . shortly after , arenholz _ et . al _ @xcite have shown that in case of cubic crystals having point symmetry , xmld along any quantization axis can be expressed as weighted sum of two fundamental spectra . this approach was extended to tetragonal crystal field by laan _ et al _ , extending number of fundamental spectra to four , demonstrated on cofe@xmath7o@xmath8 @xcite . krug _ et al _ @xcite have generalize this approach from xmld fundamental spectra to fundamental spectra of scattering form - factor @xmath9 , where imaginary part of @xmath9 corresponds to light absorption . later , havekort _ et al _ have expressed symmetry arguments also for other crystal symmetries @xcite . recently , the quadratic effects in magnetization like voigt in reflection or xmld were detected and theoretically explained also in the soft x - ray energy range , _ i. e. _ at the 3@xmath0 edges of transition metals @xcite . from phenomenological point of view , all magnetooptic effects are described by conductivity @xmath10 , permittivity @xmath11 or scattering @xmath12 tensors , being simply convertible between each other . in following , we use tensor of permittivity , being mostly employed in optical community . from quantum - mechanical point of view , the fundamental part of the permittivity tensor is optical absorption , related to optical dispersion by kramers - kronig relations @xcite . the relations of magnetooptic response ( or in general to any magneto - transport ) to crystallographic orientation is given by symmetry arguments , predicting dependence of the permittivity tensor elements on magnetization direction for effects being independent , linear , quadratic , cubic etc in magnetization . in case of magnetooptic effects quadratic - in - magnetization , the form of such tensors can be found in works of birss @xcite , nye @xcite and bhagavantam @xcite and visnovsky @xcite , whose notation is used in this article . symmetry arguments predict , that the permittivity tensor is given by a weighted sum of constituent single dielectric tensor elements , where their weights are given solely by direction of magnetization . in this article , we express principal spectra , _ i.e. _ the non - magnetic , linear and quadratic contributions in magnetization for permittivity tensors , from which the complete spectra of permittivity tensor can be obtained for any magnetization direction . then , we express principal spectra calculated from first principles of cubic material bcc fe , fcc co and fcc ni for three spectral ranges , extended visible range , soft and hard x - rays ( at the 2@xmath0-edges and 3@xmath0-edges , respectively ) . also , we express principal spectra directly using modified kubo formula . finally , we briefly express relations between signal measured by various magnetooptical techniques and off - diagonal permittivity elements .
solar - cycle prediction , _ i.e. _ forecasting the amplitude and/or the epoch of an upcoming maximum is of great importance as solar activity has a fundamental impact on the medium - term weather conditions of the earth , especially with increasing concern over the various climate change scenarios . however , predictions have been notoriously wayward in the past @xcite . there are basically two classes of methods for solar cycle predictions : empirical data - analysis - driven methods and methods based on dynamo models . most successful methods in this regard can give reasonably accurate predictions only when a cycle is well advanced ( _ e.g. , _ three years after the minimum ) or with the guidance from its past @xcite . hence , these methods show very limited power in forecasting a cycle which has not yet started . the theoretical reproduction of a sunspot series by most current models shows convincingly the illustrative nature " of the existing record @xcite . however , they generally failed to predict the slow start of the present cycle 24 @xcite . one reason cited for this is the emergence of prolonged periods of extremely low activity . the existence of these periods of low activity brings a big challenge for solar - cycle prediction and reconstruction by the two classes of methods described above , and hence prompted the development of special ways to evaluate the appearance of these minima @xcite . moreover , there is increasing interest in the minima since they are known to provide insight for predicting the next maximum @xcite . some earlier authors have both observed and made claims for the chaotic or fractal features of the observed cycles , but the true origin of such features has not yet been fully resolved . for instance , the hurst exponent has been used as a measure of the long - term memory in time series @xcite an index of long - range dependence that can be often estimated by a rescaled range analysis . the majority of hurst exponents reported so far for the sunspot numbers are well above @xmath1 , indicating some level of predictability in the data . nonethteless , it is not clear whether such predictability is due to an underlying chaotic mechanism or the presence of correlated changes due to the quasi-11-year cycle @xcite . it is the irregularity ( including the wide variations in both amplitudes and cycle lengths ) that makes the prediction of the next cycle maximum an interesting , challenging and , as yet , unsolved issue . in contrast to the 11-year cycle _ per se _ , we concentrate on the recently proposed hypothetical long - range memory mechanism on time scales shorter than the quasi - periodic 11-year cycle @xcite . in this work , we provide a distinct perspective on the strong maximal activities and quiescent minima by means of the so - called visibility graph analysis . such graphs ( mathematica graphs , in the sense of networks ) have recently emerged as one alternative to describe various statistical properties of complex systems . in addition to applying the standard method , we generalize the technique further making it more suitable for studying the observational records of the solar cycles . both the international sunspot number ( isn ) and the sunspot area ( ssa ) series @xcite are used in this work , and we have obtained consistent conclusions in either case . the length of the data sets are summarized in table [ tab : tspan ] . we perform a visibility - graph analysis using both monthly and daily sunspot series , which yields , respectively , month - to - month and day - to - day correlation patterns of the sunspot activities . note that we depict the annual numbers _ only _ for graphical visualization and demonstration purposes ( we use the annual numbers to demonstrate our method the actual analysis is performed in daily and monthly data ) . we discuss the results with the isn ( in figs . [ sn_sa_data ] , [ ts_deg_maxmin_cp ] ) in the main text and illustrate the results for the ssa ( in figs . [ sa_nasa_data ] , [ ts_deg_maxmin_cpnasa ] ) with notes in the captions . moreover , we compare our findings based on observational records to the results obtained from data produced by simulations from computational models @xcite . .temporal resolution and the length of the data sets . values in parentheses are the number of points of each series . note that the annual isn is used _ only _ for graphical visualization purposes and to provide a reference time interval for models . [ cols="<,^,^",options="header " , ] recently a variety of methods have been proposed for studying time series from a complex networks viewpoint , providing us with many new and distinct statistical properties @xcite . in this work , we restrict ourselves to the concept of the visibility graph ( vg ) , where individual observations are considered as vertices and edges are introduced whenever vertices are visible . more specifically , given a univariate time series @xmath2 , we construct the 01 binary adjacency matrix @xmath3 of the network . the algorithm for deciding non - zero entries of @xmath4 considers two time points @xmath5 and @xmath6 as being mutually connected vertices of the associated vg if the following criterion @xmath7 is fulfilled for all time points @xmath8 with @xmath9 @xcite . therefore , the edges of the network take into account the temporal information explicitly . by default , two consecutive observations are connected and the graph forms a completely connected component without disjoint subgraphs . furthermore , the vg is known to be robust to noise and not affected by choice of algorithmic parameters most other methods of constructing complex networks from time series data are dependent on the choice of some parameters ( _ e.g. _ the threshold @xmath10 of recurrence networks , see more details in @xcite ) . while the inclusion of these parameters makes these alternative schemes more complicated , they do gain the power to reconstruct ( with sufficient data ) the underlying dynamical system . for the current discussion we prefer the simplicity of the visibility graph . the vg approach is particularly interesting for certain stochastic processes where the statistical properties of the resulting network can be directly related with the fractal properties of the time series @xcite . figure [ visi_intro ] illustrates an example of how we construct a vg for the sunspot time series . it is well known that the solar cycle has an approximately 11-year period , which shows that most of the temporal points of the decreasing phase of one solar cycle are connected to those points of the increasing phase of the next cycle ( figure [ visi_intro](a ) ) . therefore , the network is clustered into communities , each of which mainly consists of the temporal information of two subsequent solar cycles ( figure [ visi_intro](b ) ) . when the sunspot number reaches a stronger but more infrequent extreme maximum , we have inter - community connections , since they have a better visibility contact with more neighbors than other time points hence , forming hubs in the graph . the inter - community connections extend over several consecutive solar cycles encompassing the temporal cycle - to - cycle information . ( open circles ) and the negatively inverted series @xmath11 ( filled circles ) , respectively.,title="fig : " ] ( open circles ) and the negatively inverted series @xmath11 ( filled circles ) , respectively.,title="fig : " ] ( open circles ) and the negatively inverted series @xmath11 ( filled circles ) , respectively.,title="fig : " ] depending on various notions of `` importance '' of a vertex with respect to the entire network , various centrality measures have been proposed to quantify the structural characteristics of a network ( _ c.f . _ @xcite ) . recent work on vgs has mainly concentrated on the properties of the degree and its probability distribution @xmath12 , where degree @xmath13 measures the number of direct connections that a randomly chosen vertex @xmath14 has , namely , @xmath15 . the degree sequence reflects the maximal visibility of the corresponding observation in comparison with its neighbors in the time series ( figure [ visi_intro](c ) ) . based on the variation of the degree sequence @xmath16 , we consider @xmath17 , @xmath18 , @xmath19 , and @xmath20 as hubs of the network , which can be used to identify the approximately 11-year cycle reasonably well ( figure [ visi_intro](b ) ) . furthermore in the case of sunspot time series , one is often required to investigate what contributions local minimum values make to the network something that has been largely overlooked by the traditional vgs . one simple solution is to study the negatively inverted counterpart of the original time series , namely , @xmath11 , which quantifies the properties of the local minima . we use @xmath21 and @xmath22 to denote the case of @xmath11 . here , we remark that this simple inversion of the time series allows us to create an entirely different complex network this is because the vg algorithm itself is extremely simple and does not attempt to reconstruct an underlying dynamical system . as shown in figure [ visi_intro](c ) , @xmath21 captures the variation of the local minima rather well . we will use this technique later to understand the long - term behavior of strong minima of the solar cycles . the degree distribution @xmath12 is defined to be the fraction of nodes in the network with degree @xmath13 . thus if there are @xmath23 nodes in total in a network and @xmath24 of them have degree @xmath13 , we have @xmath25 . for many networks from various origins , @xmath12 has been observed to follow a power - law behavior : @xmath26 . in the case of vgs , @xmath12 is related to the dynamical properties of the underlying processes @xcite . more specifically , for a periodic signal , @xmath12 consists of several distinct degrees indicating the regular structure of the series ; for white noise , @xmath12 has an exponential form ; for fractal processes , the resulting vgs often have power law distributions @xmath27 with the exponent @xmath28 being related to the hurst exponent @xmath29 of the underlying time series @xcite . it is worth pointing out that when one seeks to estimate the exponent @xmath28 it is often better to employ the cumulative probability distribution @xmath30 so as to have a more robust statistical fit . estimating the exponent @xmath28 of the hypothetical power law model for the degree sequence of vg can be done rather straightforwardly , but , the statistical uncertainties resulting from the observability of sunspots are a challenge for reliable interpretation . meanwhile , fitting a power law to empirical data and assessing the accuracy of the exponent @xmath28 is a complicated issue . in general , there are many examples which had been claimed to have power laws but turn out not to be statistically justifiable . we apply the recipe of to analyze the hypothetical power - law distributed data , namely , ( i ) estimating the scaling parameter @xmath28 by the maximum likelihood ( ii ) generating a @xmath31-value that quantifies the plausibility of the hypothesis by the kolmogorov smirnov statistic . aside from the aforementioned degree @xmath16 and degree distribution @xmath12 , in the appendix [ sapp : bet ] , some alternative higher order network measures are suggested which may be applied to uncover deeper dynamical properties of the time series from the vg . figure [ sn_sa_data](a , b ) show the degree distributions @xmath12 of the vgs derived from the isn @xmath32 with heavy - tails corresponding to hubs of the graph , which clearly deviates from gaussian properties . in contrast , @xmath22 of the negatively inverted sunspot series @xmath11 shows a completely different distribution , consisting of a bimodal property ( figure [ sn_sa_data]c , d ) , extra large degrees are at least two orders of magnitude larger than most of the vertices ( figure [ sn_sa_data](d ) ) . of vgs from monthly ( a , c ) and daily data ( b , d ) . ( a , b ) is for @xmath32 , and ( c , d ) @xmath11 . upper insets of all plots are @xmath12 in linear scale , while lower inset of ( a , b ) shows cumulative distribution @xmath33 in double logarithmic scale , where a straight line is expected if @xmath33 would follow a power law @xmath34 . in ( a , b ) @xmath28s could be suspected to be in the range @xmath35 $ ] , while a fit to the first part of @xmath22 yields that the slope of dashed line in ( c ) is 1.79 , and that of ( d ) is 3.61 , _ but _ all @xmath31-values are @xmath36 , rejecting the hypothetical power laws . ] , while circles ( strong minima ) are that of the vg from @xmath11 . ( b , c ) cumulative probability distribution of the time intervals between subsequent strong maxima ( b ) , and minima ( c ) . a cumulative exponential distribution is plotted as a dashed line in ( b ) , while the dashed line in ( c ) is a linear fit with slope being equal to @xmath37 . the corresponding @xmath31-value of ( c ) is @xmath38 , indicating that the power law is a plausible hypothesis for the waiting times of strong minima . ] since well - defined scaling regimes are absent in either @xmath12 or @xmath22 ( nor do they appear in the cumulative distributions as shown in the insets , see captions of figure [ sn_sa_data ] for details of the statistical tests which we apply ) , we may reject the hypothetical power laws in contrast to what has been reported in other contexts @xcite . in the context of studying grand maxima / minima over ( multi-)millennial timescale using some particular indirect proxy time series , the main idea lies in an appropriately chosen threshold , excursions above which are defined as grand maxima , respectively , below which are defined as grand minima @xcite . in this work , we use a similar concept but here in terms of degrees of the corresponding vg . we define a strong maximum if its degree @xmath13 is larger than @xmath39 ( figure [ sn_sa_data]a , b ) , a strong minimum if its degree @xmath21 is over @xmath40 ( figure [ sn_sa_data]c , d ) . note that , in general , our definition of strong maxima / minima coincides _ neither _ with the local maximal / minimal sunspot profiles since degree @xmath13 takes into account the longer term inter - cycle variations _ nor _ with those defined for an individual cycle . our definition avoids the choice of maximum / minimum for one cycle , which suffers from moving - average effects ( _ e.g. _ , @xcite ) . in contrast , our results below are robust with respect to the choice of the threshold degrees , especially in the case of the definition of strong minima , since large degrees are very well separated from others . the gray line in figure [ ts_deg_maxmin_cp](a ) shows the sunspot numbers overlaid by the maxima / minima identified by the large degrees . we find that the positions of strong maxima are substantially homogeneously distributed over the time domain , while that of the strong minima are much more clustered in the time axis although irregularly ( figure [ ts_deg_maxmin_cp](a ) ) . we emphasize that the clustering behavior of the strong minima on the time axis as shown in figure [ ts_deg_maxmin_cp](a ) and figure [ ts_deg_maxmin_cpnasa](a ) does _ not _ change if threshold degrees are varied in the interval of ( 200 , 8000 ) to define strong minima , _ i.e. _ , @xmath41 , as shown in figure [ ts_degk7000](a , b ) . therefore , the bimodality as observed in figure [ sn_sa_data](c , d ) and figure [ sa_nasa_data](c , d ) are not due to the finite size effects of time series . the hidden regularity of the time positions of maxima / minima can be further characterized by the waiting time distribution @xcite : the interval between two successive events is called the waiting - time . the statistical distribution of waiting - time intervals reflects the nature of a process which produces the studied events . for instance , an exponential distribution is an indicator of a random memoryless process , where the behavior of a system does not depend on its preceding states on both short or long time scales . any significant deviation from an exponential law suggests that the underlying event occurrence process has a certain level of temporal dependency . one representative of the large class of non - exponential distributions is the power laws , which have been observed in many different contexts , ranging from the energy accumulation and release property of earthquakes to social contacting patterns of humans @xcite . in the framework of vgs , the possible long temporal correlations are captured by edges that connect different communities ( the increasing and decreasing phases of one solar cycle belong to two temporal consecutive clusters ) . as shown in figure [ ts_deg_maxmin_cp](b ) , the distribution of the waiting time between two subsequent maxima sunspot deviates significantly from an exponential function , although the tail part could be an indicator of an exponential form . in contrast , we show in figure [ ts_deg_maxmin_cp](c ) that the waiting times between subsequent strong minima have a heavy - tail distribution where the exponent @xmath42 is estimated in the scaling regime . this suggests that the process of the strong minima has a positive long term correlation , which might be well developed over the time between 151000 days where a power - law fit is taken . waiting time intervals outside this range are due to either noise effects on shorter scales or the finite length of observations on longer time scales . again , the power law regimes identified by the waiting time distributions are robust for various threshold degree values in the interval of ( 200 , 8000 ) ( for instance , the case of @xmath41 is shown in figure [ ts_degk7000](c , d ) ) . of vgs from monthly ( a , c ) and daily data ( b , d ) . ( a , b ) is for @xmath32 , and ( c , d ) @xmath11 . upper insets of all plots are @xmath12 in linear scale , while lower inset of ( a , b ) shows cumulative distribution @xmath33 in double logarithmic scale , where a straight line is expected if @xmath33 would follow a power law @xmath34 . in ( c , d ) , @xmath28s could be fit by dashed lines ( slopes : c , @xmath43 ; d , @xmath44 respectively ) , _ but _ all @xmath31-values are @xmath36 rejecting the hypothetical power laws . ] , while circles ( strong minima ) are that of the vg from @xmath11 . ( b , c ) cumulative probability distribution of the time intervals between subsequent strong maxima ( b ) , and minima ( c ) . a cumulative exponential distribution is plotted as a dashed line in ( b ) , while the dashed line in ( c ) is a linear fit with slope being equal to @xmath37 . the corresponding @xmath31-value of ( c ) is @xmath45 , indicating that the power law is a plausible hypothesis for the waiting times of strong minima . ] ) of vgs reconstructed from negatively inversed series . ( a ) isn , and ( b ) ssa . cumulative probability distribution of the time intervals between subsequent strong minima , ( c ) isn , and ( d ) ssa . the dashed lines in ( c , d ) are linear fits which are obtained in the same way as shown in figs . [ ts_deg_maxmin_cp]c , [ ts_deg_maxmin_cpnasa]c , respectively . ] in contrast to the computations described above with observational data , we now demonstrate the inadequacy of two models of solar cycles . by applying the vg methods to model simulations we demonstrate that the observed data has , according to the complex network perspective , features absent in the models . we first choose a rather simple yet stochastic model which describes the temporal complexity of the problem , the barnes model @xcite , consisting of an autoregressive moving average arma(2 , 2 ) model with a nonlinear transformation @xmath46 where @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath52 are identically independent distributed gaussian random variables with zero mean , and standard deviation sd = 0.4 . the second model is a stochastic relaxation van der pol oscillator which is obtained from a spatial truncation of the dynamo equations @xcite . the equations read @xmath53,\end{aligned}\ ] ] where @xmath54 , @xmath55 , @xmath56 , @xmath57 is gaussian noise with zero mean and sd = 1 , and @xmath58 is adjustable but often chosen to be @xmath59 . the variable @xmath60 is associated with the mean toroidal magnetic field , and therefore the sunspot number is considered to be proportional to @xmath61 , which prompts us to construct vgs from @xmath61 ( @xmath62 respectively ) . both models reproduce the rapid growth in the increasing phase and slow decay in the decreasing phase of the activity cycles adequately . for the truncated model of the dynamo equations , a statistical significant correlation between instantaneous amplitude and frequency has been established , while the barnes model shows virtually no correlation @xcite , which are generally termed as the waldmeier effect . from both models , we generate @xmath63 independent realizations , each of them has a one month temporal resolution and the same time span as we have for the observations ( namely , over @xmath64 years ) . we then construct vgs from both @xmath65 and @xmath66 from each realization in the same way as we processed for the original observation . as shown in figure [ barnes_mininni_model ] , neither @xmath12 can mimic the heavy tails of the distributions we observed in figure [ sn_sa_data](a , b ) , nor @xmath22 can capture the bimodality of the large degrees for strong minima as we have observed in the case of observational raw records in figure [ sn_sa_data](c , d ) . this does not occur even if the parameter @xmath58 of the second model is adjusted ( figure [ barnes_mininni_model](c , d ) ) . one reason for the absence of the bimodality of @xmath12 in the nonlinear - oscillator model is the fact that the model was designed to reproduce smoothed sunspot - number time series . the often - used 13-month running - average method in the literature is known to suppress the maximum / minimum amplitudes of the series @xcite . therefore , the visibility condition for each time point is changed if the 13-month smoothing technique is applied to the original data . we show the degree distribution @xmath12 of vgs reconstructed from smoothed isn series in appendix ( figure [ pdf_smooth](b ) ) , where the bimodality is absent . of vgs constructed from : ( a , b ) @xmath32 and @xmath11 of barnes model , and ( c , d ) @xmath61 and @xmath62 of mininni s model . all @xmath12 and @xmath22 are estimated by an average over @xmath63 independent realizations using a kernel smoother . in ( c , d ) @xmath67 ( filled circles ) , @xmath68 ( open triangles ) . ] of vgs reconstructed from smoothed monthly isn data . ( a ) is for @xmath32 , and ( b ) @xmath11 . ] as shown in figure [ barnes_mininni_model ] , strong maxima / minima are _ not _ well separated . consequently we are prevented from identifying unique waiting time sequences . the corresponding analysis then depends significantly on the choice of threshold degrees . dynamo theory provides several hints that might explain the features observed in the long - term evolution of the solar activity . the two theoretical models tested in this work can reproduce qualitative features of the system reasonably well , however , within the context performed in this study , can not yield a complete and conclusive rendition of the statistical properties of @xmath12 . the power - law regimes obtained from waiting time sequences suggest that the interaction patterns for two subsequent minima can be much more complicated than what has been previously described as the instantaneous amplitudes frequencies correlation using rather simple models @xcite . in this work , we apply a recently proposed network approach , namely the visibility graph , to disclose the intricate dynamical behavior of both strong maximal and minimal sunspot numbers with observational records . more specifically , we show that : 1 . there is a strong degree of memory af the time scale of 151000 days in the occurrence of low - activity episodes , observed as clusters of inactive days . the identified persistence time scale of the strong minima agrees with the recently proposed hypothetical long range memory on time scales shorter than the 11-year cycle @xcite . the occurrence of high activity episodes is nearly random , _ i.e. _ strong active regions appear more or less independently of each other . the distinctive long - term correlations of the strong maxima and strong minima are reflected by the structural asymmetries of the degree distributions @xmath12 of the respective vgs . 3 . there is no evidence for a long term inter - cycle memory . this is in agreement with the present paradigm based on alternative methods ( see , _ e.g. _ , reviews by @xcite ) , and provides an observational constraint for solar - activity models . since the long term intra - cycle memory is relatively easy to establish but inter - cycle memory remains largely unclear , therefore we propose that our results could be used for evaluating models for solar activity at this time scale because they reflect important properties that are not included in other measures reported in the literature . from the methodological perspective , we propose an interesting generalization for the construction of vgs from the negatively - inverted time series . this has been seen , via our analysis of sunspot observations , to show complementary aspects of the original series . note that the negatively - inverse transformation is crucial for understanding when asymmetry is preserved in the time series . therefore , it is worth analyzing the dependence of the resulted vgs on an arbitrary monotonic nonlinear transformation . furthermore , as presented in appendix [ sapp : cor ] , a systematic investigation of the general conclusion as to whether large sampled points correspond to hubs of vgs will be a subject of future work especially in the presence of cyclicity and asymmetry . it is worth stressing that we construct vgs directly based on the raw sunspot series without any preprocessing . many researchers prefer to base their studies on some kind of transformed series since most common methods of data analysis in the literature rely on the assumption that the solar activity follows gaussian statistics @xcite . it is certain that the conclusions will then show some deviations depending on the parameters chosen for the preprocessing . the pronounced peaked and asymmetrical sunspot - cycle profiles prompt one to develop techniques such that the possible bias due to the unavoidable choice of parameters should be minimized . the complex network perspective offered by vg analysis has the clear advantage of being independent of any priori parameter selection . the procedure for network analysis outlined here can be directly applied to other solar activity indicators , for instance , the total solar irradiance and the solar flare index . we compared our results to two rather empirical models and showed that the distinctive correlation patterns of maximal and minimal sunspots are currently absent from these two models . using our analysis for more refined dynamo - based models ( _ e.g. _ @xcite ) would be straightforward . certainly , further work on this line of research will examine any differences given by the particular quantity and strengthen the understanding of the hypothetical long - range memory process of the solar activity from a much broader overview . this work was partially supported by the german bmbf ( projects progress ) , the national natural science foundation of china ( grant no . 11135001 ) , and the hong kong polytechnic university postdoctoral fellowship . ms is supported by an australian research council future fellowship ( ft110100896 ) . besides the fact that the maximal sunspot numbers are identified as hubs of vgs by the degree sequence @xmath16 , convincing links between further network - theoretic measures and distinct dynamic properties can provide some additional interesting understanding for the time series @xcite . in this study , we provide a graphical visualization on the relationship between degree @xmath16 and node betweenness centrality @xmath69 , which characterizes the node s ability to transport information from one place to another along the shortest path . using the annual isn as a graphical illustration , here only the relationship between some relatively large degrees ( @xmath70 ) and betweenness centrality values ( @xmath71 ) is highlighted in figure [ year_sspndegbet ] for the entire series available . for instance in the annual series , time points @xmath72 , @xmath73 , @xmath17 , @xmath20 , @xmath74 , @xmath75 , and @xmath76 are all identified simultaneously as large degrees and high betweenness , indicating strong positive correlations . ) , and ( b ) high betweenness centrality ( @xmath77 ) . ] certainly many other measures can be directly applied to the sunspot numbers , however , providing the appropriate ( quantitative ) interpretations of the results in terms of the particular underlying geophysical mechanisms remains a challenging task and is largely open for future work . note that there is in general a strong interdependence between these different network structural quantities . a general rule of understanding the scale - free property of the degree distribution of complex networks is the effects of a very few hubs having a large amount of connections . in the particular case of vgs , hubs are related to maxima of the time series since they have better visibility contact with more vertices than other sampled points . however , this result can not be generalized to all situations , for instance , it is easy to generate a time series such that its maxima are not always mapped to hubs in the vgs . one simple way to better explore this correlation is to use scatter plots between the degree sequence and the sunspot time series . as we show in figure [ deg_ts_rhob ] , the spearman correlation coefficients @xmath79 as very small ( still significantly larger than zero ) in the case of vgs reconstructed from the original time series . this provides an important cautionary note on the interpretation of hubs of vgs by local maximal values of the sunspot numbers . on the contrary if the network is reconstructed from @xmath80 , hubs of vgs could be better interpreted by local minimal values of the sunspot data since the correlations become larger . these results hold for both the isn and ssa series ( figure [ deg_ts_rhog ] ) . one reason for the lack of strong correlation between the degree @xmath16 and @xmath78 is because of the ( quasi-)cyclicity of the particular time series , which has a similar effect as the conway series @xcite . it is this concave behavior over the time axis ( although quasi - periodic from cycle to cycle ) that prevents the local maxima from having highly connected vertices . it remains unclear how local maxima of a time series are mapped to hubs of vgs we defer this topic for future work especially in the presence of cyclicity . this situation becomes even more challenging if some sort of asymmetric property is preserved in the data , as we have found for the sunspot series . @xmath78 for vg constructed from isn series based on ( a ) monthly and ( b ) daily data respectively . spearman @xmath79 is indicated . ( c , d ) are based on the vgs constructed from negatively inverted series @xmath81 . ] @xmath78 for vg constructed from ssa series based on ( a ) monthly and ( b ) daily data respectively . spearman @xmath79 is indicated . ( c , d ) are based on the vgs constructed from negatively inverted series @xmath81 . ] : 2011 , the international sunspot number & sunspot area data . _ monthly report on the international sunspot number , http://www.sidc.be/sunspot-data/ , royal observatory greenwich , http://solarscience.msfc.nasa.gov / greenwch.shtml/_.	complex network approaches have been recently developed as an alternative framework to study the statistical features of time - series data . we perform a visibility - graph analysis on both the daily and monthly sunspot series . based on the data , we propose two ways to construct the network : one is from the original observable measurements and the other is from a negative - inverse - transformed series . the degree distribution of the derived networks for the strong maxima has clear non - gaussian properties , while the degree distribution for minima is bimodal . the long - term variation of the cycles is reflected by hubs in the network which span relatively large time intervals . based on standard network structural measures , we propose to characterize the long - term correlations by waiting times between two subsequent events . the persistence range of the solar cycles has been identified over 151000 days by a power - law regime with scaling exponent @xmath0 of the occurrence time of the two subsequent and successive strong minima . in contrast , a persistent trend is not present in the maximal numbers , although maxima do have significant deviations from an exponential form . our results suggest some new insights for evaluating existing models . the power - law regime suggested by the waiting times does indicate that there are some level of predictable patterns in the minima .
point - contact spectroscopy ( pcs ) has long been known as a method to study the interactions of electrons with other excitations in metals . @xcite the interpretation of the observed characteristics in point - contact ( pc ) spectra is usually difficult because most often contacts are made by the needle - anvil or shear technique and are not microscopically well - defined with respect to contact size and geometry , structure , and local electronic parameters . recently , andreev reflection at point contacts was used to extract values of the transport spin polarization @xmath1 out of spectra measured on superconductor / ferromagnet ( s / f ) contacts . @xcite however , different models @xcite used to describe the transport through s / f interfaces yielded varying values for @xmath1 , also depending on the contact fabrication and the transport regime , @xcite an issue that is not yet understood in detail . @xcite therefore , a key issue in pcs is to determine the pc parameters , such as the form and diameter of the metallic nanobridge and the mean free path in the immediate contact region , so that one is able to identify the relevant transport regime . usually sharvin s @xcite or wexler s @xcite formulae for the ballistic and diffusive transport regime , respectively , are used to infer the pc diameter from the measured pc resistance . only very few experimental studies deal with the question whether these formulae - especially the interpolation formula in the diffusive regime - yield correct values for the pc diameter . in this paper , we employ e - beam lithography to structure a nanometer - sized orifice into a free - standing insulating si@xmath2n@xmath3 membrane followed by metallization of both sides of the membrane to get a pb / fe contact with well - defined orifice size . a detailed characterization of heterocontacts with respect to contact size and geometry , structure , and local electronic parameters allows a direct comparison of the measured pc parameters to different contact - size estimates . we find that the theoretical approximation of the contact size is appropriate if the measured electronic mean free path of each individual contact region is used . the current assignment of the contact regime is facilitated by the analysis of the pc spectrum with features due to electron - phonon interaction or the pair - breaking critical current through the orifice . when considering electron transport through a circular microscopic constriction with diameter @xmath0 between two equal metallic reservoirs , different transport regimes have to be distinguished . in the ballistic regime where @xmath0 is much smaller than the elastic and inelastic electron mean free paths @xmath4 and @xmath5 ( @xmath6 ) , the electrons pass the constriction mostly without scattering . the resistance of a pc can be calculated according to sharvin @xcite as @xmath7 where @xmath8 is a material constant with fermi momentum @xmath9 , elementary charge @xmath10 , electron density @xmath11 , and the total mean free path @xmath12 obtained using matthiessen s rule . in the opposite case @xmath13 known as thermal limit where elastic and inelastic scattering takes place in the immediate contact region , the resistance can be calculated after maxwell @xcite as @xmath14 when the transport is mainly diffusive with @xmath15 , wexler @xcite derived an interpolation formula between the two regimes for the contact resistance @xmath16 where the maxwell term ( [ eq : maxwellres ] ) is weighted with a slowly varying , non - analytical function @xmath17 . this function can be approximated by using the pad fit . @xcite based on these formulae , an estimate of the pc diameter @xmath0 can be obtained as long as the transport regime , the resistance @xmath18 and the pc parameters of the individual contact such as the local @xmath12 , and the local resistivity @xmath19 in the contact region are known . usually these parameters are not determined experimentally but the bulk values found in literature are used instead . for geometrically symmetric heterocontacts of two different metals 1 and 2 , @xmath19 and @xmath20 are replaced by @xmath21 and by @xmath22 , respectively . @xcite an alternative method to determine the local pc parameters and transport regimes _ individually _ for _ each _ pc arises from differentiation of eq.([eq : wexlerres ] ) . for a diffusive pc and under the assumption of dominant phonon scattering one gets @xmath23 valid in a region where both the pc resistance in the normalconducting state @xmath24 and the phonon contribution to the resistivity @xmath25 have the same functional temperature dependence . under further assumption that wexler s formula ( [ eq : wexlerres ] ) is valid and @xmath26 , we gain an estimate for the at low temperatures dominant elastic scattering length in the _ immediate _ contact region and therefore an independent estimation of the spectroscopic regime : @xmath27^{-1}. \label{eq : akimenko2}\ ] ] this approach should yield more reliable values for @xmath0 and @xmath4 than eq . ( [ eq : wexlerres ] ) where the intrinsic resistivity can differ from that determined on reference samples . @xcite the method was first experimentally verified by akimenko _ et al . _ @xcite for cu - cu - homocontacts and the values for @xmath0 and @xmath4 obtained from eqs . ( [ eq : akimenko1 ] ) and ( [ eq : akimenko2 ] ) could reproduce the intensity of theoretical phonon spectra . we fabricate pcs by structuring nanobridges between the two metallic reservoirs by means of e - beam lithography . this technique originally developed by ralls @xcite and used in a number of publications @xcite offers a variety of advantages compared to standard pc techniques like the needle - anvil or the shear method . first , mechanically stable contacts with small contact diameters down to a few nanometers can be obtained . furthermore , _ in situ _ preparation of the two metal reservoirs yields clean metal interfaces without oxide barriers . the main advantage is , however , that the geometry and size of the contacts are well defined through the orifice size which can be measured by scanning electron microscopy ( sem ) . in brief , we follow the process described in ref . to fabricate pb / fe point contacts , including an additional experimental characterization step . a bowl - shaped hole is structured into a @xmath28 nm thick insulating non - stoichiometric silicon - nitride membrane ( si@xmath2n@xmath3 ) by using electron - beam lithography followed by isotropic reactive ion etching with sf@xmath29 . this leads to a smaller orifice in the membrane than originally structured into the pmma resist mask . the nanoholes in the membranes are analyzed prior to the pb / fe metallization by scanning electron microscopy . an approximately @xmath30 nm thick pt layer is sputtered onto the membranes . this step is inevitable because charging effects would otherwise destroy the insulating , free - standing membrane and because of the necessary high resolution . we measured the hole diameters in the membrane directly by using a zeiss supra 55 vp sem at small acceleration voltages of @xmath31kv , a @xmath32 m aperture and the secondary electron in - lens detector for high surface resolution and contrast . the silicon - nitride membranes are typically pierced for diameters less than @xmath33 nm , with the smallest orifices having a diameter of @xmath34 nm . [ fig : nanoholes](c ) and ( d ) show exemplarily two nanoholes with diameters of @xmath35 and @xmath36 nm . in these sem pictures , we can clearly distinguish a border between pt covered areas showing a granular structure and the interior part of the ` bowl ' , which we assume not to be pt covered . however , further investigations including a focused - ion - beam ( fib ) lateral cut through a smaller pc revealed that for smaller nanoholes ( @xmath37 nm ) pt might reduce the original hole diameter in the membrane by forming a ring of adsorbed material close to the rim of the hole . fig . [ fig : fibcut ] shows a fib lateral cut through a pc with a nominal diameter in the membrane of 24 nm , performed by successive @xmath31nm - distance lateral cuts with a ga - ion fib system . one can clearly see the layer geometry with the continuous pb layer , the `` bowl '' in the membrane and the corresponding fe / cu layer on the back side of the membrane . in addition , one can identify the possible effect of pt in narrowing the original constriction of the si@xmath2n@xmath3 membrane . thus the pt film is removed completely prior to metallization by immersing the sample in a @xmath38c hot bath of _ aqua regia _ for @xmath39h . possible organic residues are removed by using a commercial diener o@xmath40 plasma cleaner . according to literature , @xcite neither the silicon substrate nor the silicon - nitride membrane are etched by _ aqua regia _ so that the size determination of the nanoholes remains valid . in a final step , a @xmath41 nm thick pb layer is deposited on the bowl - shaped side of the membrane by e - beam evaporation at room temperature under ultra - high vacuum ( @xmath42 mbar ) , followed by a @xmath43 _ in - situ _ rotation of the sample and the deposition of a @xmath44 nm thick fe layer , topped by a @xmath45 nm thick cu layer for a good ohmic contact on the flat side of the membrane ( see fig . [ fig : nanoholes](b ) ) . due to the evaporation at room temperature , the high mobility of the pb atoms leads to a stranski - krastanov - like island - growth mode . further sem analysis gave clear evidence for a continous pb film with average island sizes of @xmath46 nm , i.e. much larger than the typical pc size . each sample was characterized by measuring its resistance at room temperature in order to check whether a conductive nanobridge has been realized in the metallization process . it turned out that a nanobridge was only established in samples with hole diameters in the membrane larger than @xmath33 nm reliably . 9 point contacts with hole diameters ranging from @xmath47 nm and resistances @xmath48 in the range of @xmath49 to @xmath50 were investigated ( see also table [ table : contactradius ] ) . nm hole in the membrane . recorded under @xmath51 tilt with se2 detector and @xmath52kv acceleration voltage . ] for an independent determination of the resistivity of the two metallic layers that eventually form the metallic nanobridge we structured reference samples consisting of pb and fe layers , respectively , in a usual 4-point measurement geometry by using optical lithography methods and a standard lift - off process . since the resistivity of thin metallic layers strongly depends on the layer thickness and the growth conditions , the same silicon nitride substrate , identical evaporation conditions and layer thicknesses as for the pcs were chosen . the reference samples had the form of meander - like bars with the dimensions of @xmath53 mm length , @xmath54 width ( see inset fig.[fig : reference ] ) , and @xmath55 nm or @xmath56 nm thickness for the fe and pb layer , respectively . in addition , the fe layer was covered _ in - situ _ with a @xmath57 nm thick insulating sio@xmath40 layer to avoid surface oxidation in air . resistance measurements were performed instantly after fabrication to reduce oxidation . sample geometries were determined afterwards by using sem for the width and length measurements and an ambios technology xp-2 profilometer in combination with layer - thickness monitors . the resistance @xmath18 of all samples was measured in the range of room temperature down to @xmath58k in a @xmath59he cryostat by using a lr-700 resistance bridge . the differential resistance @xmath60 vs applied bias voltage @xmath61 was recorded at low temperatures between @xmath62k and @xmath63k via lock - in technique . some measurements were done in an external magnetic field of @xmath64mt applied parallel to the current through the pc to drive the superconducting pb to the normalconducting state . from the @xmath65 measurements down to @xmath66k on the reference samples and pcs ( shown in fig.[fig : reference ] exemplarily for the pb reference sample and the pc sample 4 ) one obtains all necessary contact parameters . for the pb reference sample we obtain a residual resistance ratio @xmath67 , a critical temperature @xmath68k and the resistance in the normal - conducting state @xmath69 at @xmath70k . with the known geometry of the sample , we calculate the resistivity of our @xmath41 nm thick pb layer at @xmath70k as @xmath71 m . for our @xmath44 nm thick fe layer we obtain @xmath72 , the averaged resistance @xmath73 and hence @xmath74 m at @xmath70k . both values agree well with those obtained on similar samples in the literature . @xcite according to ref . , the resistivity of a geometrical symmetric pb / fe heterocontact ( hc ) at @xmath70k is @xmath75 of course , @xmath76 is chiefly determined by the highly resistive fe layer . for further analysis we also need the material constants @xmath20 for the corresponding metals . here , we use the value @xmath77 for pb . @xcite from these values we obtain an electron mean free path in the pb layer of @xmath78 at @xmath70k , which is of the order of the island structure of the pb film . for fe , @xmath79 is calculated using the drude model with the charge carrier density @xcite @xmath80 obtained from band - structure and de - haas - van - alphen measurements , and the fermi velocity @xcite @xmath81 m/s . the magnitude of the electron mean free path @xmath82 nm clearly demonstrates that interface scattering plays the most important role in the @xmath44 nm thick fe layer . with these values , we calculate the arithmetic average of @xmath20 for a heterocontact as @xmath83 hence , the electron mean free path through the contact is estimated as @xmath84 nm . these values are usually used to allocate the transport regime . to calculate pc diameters as suggested by akimenko _ et al . _ @xcite ( cf . eq.([eq : akimenko1 ] ) ) one needs additionally @xmath85 and @xmath86 of each individual pc as a function of @xmath87 . @xmath86 is determined by a linear fit for @xmath88k to the resistance data of the two reference samples whence @xmath89 in the range where @xmath90 . similarly , @xmath85 was determined individually for each pc for @xmath88k where @xmath91 as well . using equations ( [ eq : sharvinres ] ) to ( [ eq : wexlerres ] ) and the parameters given in section [ analysisdata ] , we can determine the contact radius @xmath92 and allocate a transport regime to each individual contact . however , this method is problematic because it is based on the assumption of one universal electron mean free path @xmath93 nm fixed for all pcs . for example , the calculation would result in much larger contact radii @xmath94 nm respectively 58 nm for samples 3 and 4 , respectively , than the measured upper limits @xmath95 nm and accordingly 35 nm . in reality , the electron mean free path in the immediate region of the nanobridge may differ from sample to sample due to the granular structure of the films and the growth process of the nanobridge which usually is not controlled on the atomic level . the advantage of the second approach following akimenko _ et al . _ @xcite is that the t dependence of the contact resistance @xmath48 enters the calculation . it provides a more realistic insight into the dominant scattering processes that determine the pc resistance . we therefore employ akimenko s @xcite approach using eq . ( [ eq : akimenko1 ] ) which can be applied nearly independently for contacts in the diffusive regime and eq . ( [ eq : akimenko2 ] ) to calculate the pc parameters @xmath96 and @xmath4 for each individual pc with the parameters given in subsection [ analysisdata ] . table [ table : contactradius ] summarizes those values and the allocated transport regimes for all measured samples together with @xmath48 and the measured hole radii @xmath97 . in addition , the last column displays the pc radii @xmath98 calculated by equations ( [ eq : sharvinres])-([eq : wexlerres ] ) when taking the correct contact regime and elastic mean free path @xmath4 determined by akimenko s approach into account which furnishes a self - consistency check of the method . indeed , the elastic mean free path in the immediate contact region varies from sample to sample and for some samples differs significantly from @xmath99 calculated in the former paragraph . the main reason for the variation is probably a difference in grain structure of the immediate contact region caused by the island growth of the pb film which tends to increase the mean free path in larger contacts . the analysis also shows that the low - resistive pb part of the heterocontact is decisive for the pc properties and prevails over the highly resistive fe part . most of our samples seem to belong to the diffusive ( ` d ' ) regime with some being closer to the ballistic limit labeled as ` b ' in table [ table : contactradius ] - some being closer to the thermal limit - indicated as ` t ' . for very small @xmath96 wexler s equation ( [ eq : wexlerres ] ) is no longer valid and eq . ( [ eq : akimenko2 ] ) may yield unphysical negative values of @xmath4 . the corresponding samples can be assigned to the pure ballistic regime ( samples 6 and 7 ) . for these samples , the sharvin formula does indeed yield reliable values for @xmath96 . [ cols="^,^,^,^,^,^,^,^ " , ] according to akimenko _ _ @xcite and @xmath98 according to the consistency check with the experimentally determined hole radius in the membranes @xmath97 . dashed line indicates the `` ideal '' expectation @xmath100 . solid line presents a guide to the eye of experimental data . ] as can be seen from fig . [ fig : comparisonradius ] , the contact radii @xmath98 derived from the self - consistency check yield nearly the same values for the pc radius @xmath96 when taking the correct transport regime and the locally determined @xmath4 into account . the calculated radii are always somewhat smaller than those of the holes in the membranes , but do seem to follow the same general experimental trend . when recalling the sample fabrication process , it seems obvious that the metals will not completely fill the holes in the membranes during the evaporation process - thus leading to narrower _ metallic _ nanobridges . the sem measurement yields an upper limit for the effective pc radius . statistical variations of the contact size can be attributed due to the rather uncontrolled nature of the aggregation process at the atomic level following the evaporation . hence , the exact geometry of establishing a nanobridge can not be controlled . finally , we want to emphasize that experimentally determined values only enter into the determination of the pc radius . the very good agreement of those calculations with the experimental trend shows _ directly _ for metallic heterocontacts that frequently used size estimates for pc diameters do agree with the experimental data and yield reasonable values for those parameters . however , extreme care has to be taken when characterizing the samples . pc spectra have been recorded at different temperatures and in an applied magnetic field @xmath64mt which proved to be sufficient to drive the superconducting pb into the normalconducting state at lowest @xmath58k . these spectra reveal clear nonlinearities at voltages @xmath101 in the normal and superconducting states of pb , where @xmath102 denotes the superconducting energy gap . [ fig : overviewspectr ] displays a typical spectrum with pb in the superconducting state , where we have marked the regimes of nonlinearities . besides the well - known andreev double - minimum structure ( 1 ) at @xmath103mv for pb , we observe an overall rise in the differential resistance and changing slopes which can be interpreted as due to electron - phonon scattering in pb ( 2 ) , and a rather sharp peak in the superconducting state of pb at higher energies @xmath104 that scales with temperature is attributed to the current through the contact exceeding the pair - breaking critical current ( 3 ) . features of andreev reflection with the possibility to extract @xmath102 can be found even for rather ill - defined pcs . the nonlinearities caused by ( 2 ) and ( 3 ) , on the other hand , depend sensitively on the quality of the pc . k. in addition to the andreev signature ( 1 ) in the middle part of the spectrum , we observe a rise and change of slope ( 2 ) of @xmath105 as well as a sharp peak ( 3 ) on this rise . ] [ fig : phononspectra ] exemplarily shows the second derivative @xmath106 vs. @xmath61 spectra of samples 2 and 4 , recorded at @xmath58k and @xmath107mt where pb is in the normal state . @xmath106 vs. @xmath61 spectra are obtained by numerical differentiation of the measured @xmath108 data . typically , the data have to be averaged over 20 - 30 data points to reduce the noise . the second derivative @xmath106 is proportional to the eliashberg function @xmath109 , where for heterocontacts the spectrum is a sum of contributions of both pb and fe . @xcite for low energies , predominantly the phonon excitations of pb are expected while the contributions of fe will be significant at higher energies @xmath110mev only . @xcite indeed , we identify features indicated by arrows for sample 4 at @xmath111mv , which are ascribed to transverse acoustic ( ta ) phonons of pb . @xcite in contrast , only a broad feature is seen in the spectrum of sample 2 . the broadening is a direct consequence of the reduced mean free path , corresponding to a knudsen ratio @xmath112 , of this particular sample . in the diffusive regime , the intensity of the peaks depends on the knudsen ratio . @xcite indeed , for sample 4 which is closer to the ballistic regime with much larger @xmath113 the phonon peak is much more pronounced . k and @xmath107mt in the normalconducting state , obtained by numerical differentiation @xmath114 . arrows indicate the position of ta phonon peaks in pb . ] we note that we observe a shift of the peak positions to higher energies on the order of the energy gap @xmath115mev and a peak intensity which is nearly independent of the contact regime when pb is in the superconducting state , in agreement with theory @xcite and earlier experiments . @xcite the second feature which supports the allocation of the transport regimes is the sharp peak in the differential resistance spectra at higher bias @xmath116 when pb is in the superconducting state ( see fig . [ fig : overviewspectr ] , ( 3 ) ) . it is observed for all our pcs . the dc current @xmath117 at the peak position , simultaneously recorded as a function of @xmath87 , scales with the superconducting order parameter in bcs theory ( see inset in fig . [ fig : critcurrent ] , where the @xmath87 dependence is studied for samples 1 - 4 ) . therefore , we identify the peak current @xmath118 with the critical current @xmath119 . the sharp peaks arise when the current through the contact exceeds the pair - breaking critical current density of pb , leading to a sudden rise in the differential resistance . as a function of point - contact radius @xmath96 at @xmath120k and temperature ( inset ) . the dashed line indicates the expected linear dependence according to silsbee s rule , the solid red line is a quadratic least - squares fit of the data of six point contacts in the ballistic or diffusive limit . the inset compares the temperature dependence of the peak current for four samples with the bcs temperature dependence of the order parameter . the experimental data point at lowest temperature of each sample has been scaled on the bcs curve . ] the appearance of those peaks in the spectra of pcs with a superconducting counterelectrode has been analyzed in a number of publications . @xcite the peaks have been interpreted , e.g. , as being akin to pcs with large contact dimensions , @xcite i.e. contacts in the thermal regime . however , that this is not necessarily the case . the position @xmath121 of the peaks varies depending on the contact geometry and can often be found at voltages as high as @xmath122mv at lowest @xmath87 . as discussed above , our fabrication process results in a controlled pc geometry , where in our specific sample geometry pb establishing the contact grows in a cylindrical shape into the nanohole ( see fig . [ fig : fibcut ] ) . according to silsbee s rule for a cylindrical superconductor with radius @xmath123 ( @xmath124 : london penetration depth ) superconductivity is destroyed if the self - field at the surface produced by the current through the wire reaches the thermodynamic critical field @xmath125 . it follows that @xmath126 , where @xmath127 is the permeability of free space , i.e. @xmath128 ( ref . ) . for @xmath129 , which is the case for our contacts , one would expect a geometry independent and constant pair - breaking current density @xmath130 . macdonald and leavens @xcite showed that the current distribution through the contact depends on the ratio @xmath131 . while for clean contacts with @xmath132 the current density is practically constant over the cross - section , it increases abruptly at the periphery for dirty contacts with @xmath133 . indeed , several experimental studies @xcite have shown that the critical current for contacts in the thermal transport regime scales as @xmath128 , while for clean contacts @xmath134 was found in accordance with ref . . [ fig : critcurrent ] shows the critical current @xmath118 at @xmath58k vs. the pc radius @xmath96 for six of our nearly ballistic or diffusive contacts . the critical current of our contacts does not follow silsbee s rule shown as a dashed line@xcite but rather shows a quadratic dependence . a least - squares fit to the data via @xmath135 yields an universal critical current density for all contacts of @xmath136 . this value is of the same order of magnitude as the calculated @xcite bcs value for pb @xmath137 and an experimentally determined value @xcite on @xmath138 thick pb layers of @xmath139 . the observed behavior and the geometry independent critical current density for all contacts with @xmath140 confirms our assumption that the observed peaks arise from reaching the pair - breaking current in the immediate contact region and supports our assignment of the transport regimes and the calculated pc radii . in conclusion , we have presented an experimental study of size estimates for heterocontacts in pc spectroscopy in the different transport regimes . a direct sem measurement of the nanocontact size allows a comparison with theoretical models for contact - size estimates of heterocontacts in the semiclassical approach . due to the good agreement between experimental and calculated values , we conclude that the semiclassical models yield reasonable values for the pc diameter @xmath0 as long as the samples are carefully characterized and the correct transport regime is determined taking the local transport parameters of the individual contact into account . our assignment of the samples to different transport regimes is corroborated by the analysis of further features in the spectra such as phonon peaks and the critical pair - breaking current of pb . the authors would like to thank e. mller , s. khn and t. peichl for stimulating discussions of the sample fabrication process and c. reiche for the support in the fabrication of the reference samples . we acknowledge the financial support provided within the dfg - center for functional nanostructures . due to experimental limitations of our measurement setup ( change in the measuring range of our precision current source ) , some pcs with lower contact resistances could only be measured up to voltages of @xmath141mv . however , since phonon excitations for pb should already be visible at relatively low energies , this data seems to be sufficient for a qualitative discussion . the longitudinal acoustic ( la ) phonon which is expected as a peak around @xmath142mv is barely visibly for sample 2 but often covered by the additional raise of the critical - current peak . for the calculation according to silsbee s rule we used @xmath143mt and @xmath144tm / a , for the calculation of @xmath145 we used @xmath146 nm and @xmath147mt , taken from c. poole jr . , _ handbook of superconductivity _ ( academic press , san diego , ca , 2000 ) .	nanostructured superconductor / ferromagnet heterocontacts are studied in the different transport regimes of point - contact spectroscopy . direct measurements of the nanocontact size by scanning electron microscopy allow a comparison with theoretical models for contact - size estimates of heterocontacts . our experimental data give evidence that size estimates yield reasonable values for the point - contact diameter @xmath0 as long as the samples are carefully characterized with respect to the local electronic parameters .
in this work we adopt a chemical evolution model ( see chiappini , matteucci , & romano 2000 ) that assumes two main accretion episodes for the formation of the galaxy : the first one forming the halo and bulge in a short timescale followed by a second one that forms the thin - disk , with a timescale which is an increasing function of the galactocentric distance ( being of the order of 7 gyrs at the solar neighborhood ) . the present model takes into account in more detail than previously the halo density distribution and explores the effects of a threshold density in the star formation process , both during the halo and disk phases . the model also includes the most recent nucleosynthesis prescriptions concerning supernovae of all types , novae and single stars dying as white dwarfs . in the comparison between model predictions and available data , we have focused our attention on abundance gradients as well as gas , star and star formation rate distributions along the disk , since this kind of model has already proven to be quite successful in reproducing the solar neighborhood characteristics . we suggest that the mechanism for the formation of the halo leaves heavy imprints on the chemical properties of the outer regions of the disk , whereas the evolution of the halo and the inner disk are almost completely disentangled . this is due to the fact that the halo and disk densities are comparable at large galactocentric distances and therefore the gas lost from the halo can substantially contribute to build up the outer disk . we also show that the existence of a threshold density for the star formation rate , both in the halo and disk phase , is necessary to reproduce the majority of observational data in the solar vicinity and in the whole disk . in particular , a threshold in the star formation implies the occurrence of a gap in the star formation at the halo - disk transition phase , in agreement with recent data . @xmath0 the outer gradients are sensible to the halo evolution , in particular to the amount of halo gas which ends up into the disk . this result is not surprising since the halo density is comparable to that of the outer disk , whereas is negligible when compared to that of the inner disk . therefore , the inner parts of the disk ( @xmath1 @xmath2 @xmath3 ) evolve independently from the halo evolution . @xmath0 we predict that the abundance gradients along the galactic disk must have increased with time . this is a direct consequence of the assumed `` inside - out '' scenario for the formation of the galactic disk . moreover , the gradients of different elements are predicted to be slightly different , owing to their different nucleosynthesis histories . in particular , fe and n , which are produced on longer timescales than the @xmath4-elements , show steeper gradients . unfortunately , the available observations can not yet confirm or disprove this , because the predicted differences are below the limit of detectability . @xmath0 our model guarantees a satisfactory fit not only to the elemental abundance gradients but it is also in good agreement with the observed radial profiles of the sfr , gas density and the number of stars in the disk . @xmath0 our best model suggests that the average @xmath5fe]@xmath6 ratios in stars slightly decrease from 4 to 10 kpcs . this is due to the predominance of disk over halo stars in this distance range and to the fact that the `` inside - out '' scenario for the disk predicts a decrease of such ratios . on the other hand we predict a substantial increase ( @xmath7 dex ) of these ratios in the range 1018 kpcs , due to the predominance , in this region , of the halo over the disk stars . finally , we conclude that a relatively short halo formation timescale ( @xmath8 0.8 gyr ) , in agreement with recent estimates for the age differences among galactic globular clusters , coupled with an `` inside - out '' formation of the galactic disk , where the innermost regions are assumed to have formed much faster than the outermost ones , represents , at the moment , the most likely explanation for the formation of the milky way . this scenario allows us to predict abundance gradients and other radial properties of the galactic disk in very good agreement with observations . more observations at large galactocentric distances are needed to test our predictions .	we present theoretical results on the galactic abundance gradients of several chemical species for the milky way disk , obtained using an improved version of the two - infall model of chiappini , matteucci , & gratton ( 1997 ) that incorporates a more realistic model of the galactic halo and disk . this improved model provides a satisfactory fit to the elemental abundance gradients as inferred from the observations and also to other radial features of our galaxy ( i.e. , gas , star formation rate and star density profiles ) . we discuss the implications these results may have for theories of the formation of the milky way and make some predictions that could in principle be tested by future observations . # 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
to uncover the neural circuit mechanisms underlying animal behavior , e.g. , working memory or decision making , is a fundamental issue in systems neuroscience @xcite . recent developments in multi - neuron recording methods make simultaneous recording of neuronal population activity possible , which gives rise to the challenging computational tasks of finding basic circuit variables responsible for the observed collective behavior of neural populations @xcite . the collective behavior arises from interactions among neurons , and forms the high dimensional neural code . to search for a low dimensional and yet neurobiologically plausible representation of the neural code , thus becomes a key step to understand how the collective states generate behavior and cognition . correlations among neurons spiking activities play a prominent role in deciphering the neural code @xcite . various models were proposed to understand the pairwise correlations in the population activity @xcite . modeling these correlations sheds light on the functional organization of the nervous system @xcite . however , as the population size grows , higher order correlations have to be taken into account for modeling synchronous spiking events , which are believed to be crucial for neural information transmission @xcite . in addition , the conclusion drawn from small size populations may not be correct for large size populations . theoretical studies have already proved that high order interactions among neurons are necessary for generating widespread population activity @xcite . however , introduction of high order multi - neuron couplings always suffers from a combinatorial explosion of model parameters to be estimated from the finite neural spike train data . to account for high order correlations , various models with different levels of approximation were proposed , for example , the reliable interaction model @xcite with the main caveat that the rare patterns are discarded during inference of the coupling terms , the dichotomized gaussian model @xcite in which correlations among neurons are caused by common gaussian inputs to threshold neurons , the k - pairwise model @xcite in which an effective potential related to the synchronous firing of @xmath0 neurons was introduced , yet hard to be interpreted in terms of functional connectivity , and the restricted boltzmann machine @xcite where hidden units were shown to be capable of capturing high order dependences but their number should be predefined and difficult to infer from the data @xcite . one can also take into account part of the statistical features of the population activity ( e.g. , simultaneous silent neural pattern ) and assume homogeneity for high order interactions among neurons due to the population size limitation @xcite . in this paper , i provide a low dimensional neurobiological model for describing the high order correlations and extracting useful information about neural functional organization and population coding . in this study , i interpret correlations in terms of population coupling , a concept recently proposed to understand the multi - neuron firing patterns of the visual cortex of mouse and monkey @xcite . the population coupling characterizes the relationship of the activity of a single neuron with the population activity ; this is because , the firing of one neuron is usually correlated with the firing pattern of other neurons . i further generalize the original population coupling to its higher order form , i.e. , the relationship of pairwise firing with the population activity . i then derive the practical dimensionality reduction method for both types of population couplings , and test the method on different types of neural data , including ganglion cells in the salamander retina onto which a repeated natural movie was projected @xcite , and layer 2/3 as well as layer 5 cortical cells in the medial prefrontal cortex ( mpc ) of behaving rats @xcite . in this paper , i develop a theoretical model of population coupling and its advanced form , to explain higher order correlations in the neural data . methodologically , i propose the fast mean field method not only to learn the population couplings but also to evaluate the high order correlations . note that this is computationally hard in a traditional maximum entropy model by using sampling - based method . conceptually , i generalize the normal population coupling by introducing the second - order population coupling , which reveals interesting features from the data . first , it can explain a significant amount of three - cell correlations , and it works much better in cortical data than in retinal data . second , the second - order population coupling matrix has distinct features in retinal and cortical data . the cortical one shows clear stripe - like structure while the retinal one has no such apparent structure . altogether , this work marks a major step to understand the low - order representation of complex neural activity in both concepts and methods . for a neuronal population of size @xmath1 , the neural spike trains of duration @xmath2 are binned with temporal resolution @xmath3 , yielding @xmath4 samples of @xmath1-dimensional binary neural firing patterns . i use @xmath5 to denote firing state of neuron @xmath6 , and @xmath7 for silent state . neural responses to repeated stimulus ( or the same behavioral tasks ) vary substantially ( so - called trial - to - trial variability ) @xcite . to model the firing pattern statistics , i assign each firing pattern @xmath8 a cost function ( energy in statistical physics jargon ) @xmath9 , then the probability of observing one pattern @xmath8 can be written as @xmath10 , where @xmath11 this is the first low dimensional representation to be studied . high energy state @xmath8 corresponds to low probability of observation . @xmath12 is the firing bias constraining the firing rate of neuron @xmath6 , while @xmath13 characterizes how strongly neuron @xmath6 s spiking activity correlates with the population activity measured by the sum of other neurons activity . i name @xmath13 the first order population coupling ( @xmath14 ) . thus , only @xmath15 parameters needs to be estimated from the neural data . this number of model parameters is much less than that in conventional maximum entropy model @xcite . to model the high order correlation ( e.g. , three neuron firing correlation ) , i further generalize @xmath14 to its advanced form , i.e. , the second order population coupling , namely @xmath16 , describing the relationship of pairwise firing with the population activity , and the corresponding energy is given by @xmath17 where @xmath18 characterizes how strongly the firing state of the neuron pair @xmath19 correlates with the firing activities of other neurons . this term is expected to increase the prediction ability for modeling high order correlations in the neural data . under the framework of @xmath16 , the total number of parameters to be estimated from the data is @xmath20 . @xmath14 and @xmath16 have a clear neurobiological interpretation ( for @xmath14 , see a recent study @xcite , and the results obtained under the @xmath16 can also be experimentally tested ) , and moreover they can be interpreted in terms of functional interactions among neurons ( as shown later ) . to find the model parameters as a low dimensional representation , i apply the maximum likelihood learning principle corresponding to maximizing the log - likelihood @xmath21 with respect to the parameters . the learning equation for @xmath14 is given by [ le01 ] @xmath22 where @xmath23 and @xmath24 denote the learning step and learning rate , respectively . the maximum likelihood learning shown here has a simple interpretation of minimizing the kullback - leibler divergence between the observation probability and the model probability @xcite . in an analogous way , one gets the learning equation for @xmath16 , @xmath25 in the learning equations eq . ( [ le01 ] ) and eq . ( [ le02 ] ) , the data dependent terms can be easily computed from the binned neural data . however , the model expectations of the firing rate ( magnetization in statistical physics ) and correlations are quite hard to evaluate without any approximations . here i use the mean field method to tackle this difficulty . first , i write the energy term into a unified form , @xmath26 where @xmath27 denotes the interaction index and @xmath28 denotes the neuron set involved in the interaction @xmath27 . @xmath29 for @xmath14 and @xmath30 for @xmath16 . therefore , @xmath14 introduces the pairwise interaction as @xmath31 , while @xmath16 introduces the triplet interaction as @xmath32 . the multi - neuron interaction in the conventional ising model is decomposed into first order or second order population coupling terms . this decomposition still maintains the functional heterogeneity of single neurons or neuron pairs , but reduces drastically the dimensionality of the neural representation for explaining high order correlations . in principle , one can combine @xmath14 and @xmath16 to predict both pairwise and triplet correlations . however , in this work , i focus on the pure effect of each type of population coupling . in fact , the conventional ising model @xcite can be recovered by setting @xmath33 , which is pairwise interaction . the learning equation is derived similarly , and is run by reducing the deviation between the model pairwise correlation and the clamped one ( computed from the data ) @xcite . second , the statistical properties of the model ( eq . [ energyising ] ) can be analyzed by the cavity method in the mean field theory @xcite . the self - consistent equations are written in the form of message passing ( detailed derivations were given in refs @xcite , see also appendix [ app : deriv ] ) as [ bp ] @xmath34 where @xmath35 denotes the member of interaction @xmath36 except @xmath6 , and @xmath37 denotes the interaction set @xmath6 is involved in with @xmath27 removed . @xmath38 is interpreted as the message passing from the neuron @xmath6 to the interaction @xmath27 it participates in , while @xmath39 is interpreted as the message passing from the interaction @xmath36 to its member @xmath6 . this iteration equation is also called the belief propagation ( bp ) , which serves as the message passing algorithm for the statistical inference of the model parameters . iteration of the message passing equation on the inferred model would converge to a fixed point corresponding to a global minimum of the free energy ( in the cavity method approximation @xcite ) @xmath40 where @xmath41 is the normalization constant of the model probability @xmath42 . the free energy contribution of one neuron is @xmath43 and the free energy contribution of one interaction is @xmath44 . i define the function @xmath45 . at the same time , the model firing rate and multi - neuron correlation can be estimated as [ magcorre ] @xmath46 magnetization and correlation are defined as @xmath47 and @xmath48 , respectively . a brief derivation of eq . ( [ magcorre ] ) is given in appendix [ app : deriv ] . here the multi - neuron correlation is calculated directly from the cavity method approximation and expected to be accurate enough for current neural data analysis @xcite . this is because , correlations under the model are evaluated taking into account nearest - neighbor interactions , rather than naive full independence among neurons . this approximation is expected to work well in a weakly - correlated neural population @xcite , where long - range strong correlations do not develop . a similar application of this principle revealed a non - trivial geometrical structure of population codes in salamander retina @xcite . another advantage is the low computation cost . both the free energy and the pairwise correlations can be estimated by the time complexity of the order @xmath49 for @xmath14 , and @xmath50 for triplet correlations in @xmath16 , which is one order of magnitude lower than the tractable model of @xmath14 recently proposed in ref . a more accurate expression could be derived from linear response theory @xcite with much more expensive computational cost ( increased by an order of magnitude ( @xmath51 ) . to estimate the information carried by a neural population , one needs to compute the entropy , which is defined as @xmath52 , and it measures the capacity of the neural population for information transmission . the more obvious variability the neural responses have , the larger the entropy value is . the entropy of the model can be estimated from the fixed point of the message passing equation . based on the standard thermodynamic relation , @xmath53 , where @xmath54 is the energy of the neural population and given by [ energ ] @xmath55\\ & \times\prod_{a\in\partial i\backslash b}\cosh\gamma_a(1+x\hat{m}_{a\rightarrow i } ) . \end{split}\end{aligned}\ ] ] the basic procedure to infer population couplings is given as follows . at the beginning , all model parameters are assigned zero value . it is followed by three steps : ( @xmath6 ) messages are initialized randomly and uniformly in the interval @xmath56 . ( @xmath57 ) eq . ( [ bp ] ) are then run until converged , and the magnetizations as well as multi - neuron correlations are estimated using eq . ( [ magcorre ] ) . ( @xmath58 ) the estimated magnetizations and correlations are used at each gradient ascent learning step ( eq . ( [ le01 ] ) or eq . ( [ le02 ] ) ) . when one gradient learning step is finished , another step starts by repeating the above procedure ( from ( @xmath6 ) to ( @xmath58 ) ) . to learn the higher order population coupling , the damping technique is used to avoid oscillation behavior , i.e. , @xmath59 where @xmath60 is the damping factor taking a small value . the inferred model can also be used to generate the distribution of spike synchrony , i.e. , the probability of @xmath0 simultaneous spikes . this distribution can be estimated by using monte carlo ( mc ) simulation on the model . the standard procedure goes as follows . the simulation starts from a random initial configuration @xmath61 , and tries to search for the low energy state , then the energy is lowered by a series of elementary updates , and for each elementary update , @xmath1 proposed neuronal state flips are carried out . that is , the transition probability from state @xmath8 to @xmath62 with only @xmath63 flipped ( @xmath64 ) is expressed as @xmath65 where @xmath66 . the equilibrium samples are collected after sufficient thermal equilibration . these samples ( a total of @xmath67 samples in simulations ) are finally used to estimate the distribution of spike synchrony . on the retinal data ( @xmath68 , one typical example ) . ( a ) firing rate in the data is reproduced by the model . ( b ) two - cell correlations are explained partially by @xmath14 ( @xmath69 ) . a monte - carlo ( mc ) sampling of the model yields similar results to belief propagation ( bp ) , which is much faster . ( c ) from the mc samples , three - cell correlations can also be estimated . ( d ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] by using the mean field method , i first test both types of population couplings on the retina data , which is the spike train of @xmath70 ganglion cells in a small patch of the salamander retina @xcite . the retina was stimulated with a repeated natural movie . the spike train data is binned with the bin size equal to @xmath71 reflecting the temporal correlation time scale , yielding about @xmath72 binary firing patterns for data modeling . i then test the same concepts on the cortical data of behaving rats . rats performed the odor - place matching working memory during one task session , and spiking activities of @xmath73 cells in both superficial layer @xmath74 and deep layer ( layer @xmath75 ) of medial prefrontal cortex were simultaneously recorded ( for detailed experiments , see ref . one task session consists of about @xmath76 trials , yielding a spike train of these cortical cells binned with the temporal resolution @xmath77 ( a total of @xmath78 firing patterns ) . [ retinapc1 ] reports the inference result on a network example of @xmath76 neurons selected randomly from the original dataset . the firing rate is predicted faithfully by the model using either mc or bp ( fig . [ retinapc1 ] ( a ) ) . inferring only @xmath14 , one could predict about @xmath69 of entire pairwise correlation ( a precision criterion is set to @xmath79 in this paper ) ( fig . [ retinapc1 ] ( b ) ) . this means that @xmath69 of the whole correlation set have the absolute value of the deviation between the predicted correlation and measured one ( @xmath80 ) smaller than the precision criterion . using the sampled configurations of neural firing activity from the mc simulation , one could also predict three - cell correlations ( fig . [ retinapc1 ] ( c ) ) , whereas , the prediction fraction can be improved by a significant amount after introducing @xmath16 , as i shall show later . in addition , fitting only @xmath15 model parameters in @xmath14 analysis could not predict the tail of spike synchrony distribution ( fig . [ retinapc1 ] ( d ) ) ; this is expected as no higher order interaction terms are included in the model , and rare events of large @xmath0 spikes are also difficult to observe in a finite sampling during mc simulations . on the retinal data ( @xmath68 , one typical example ) . ( a ) firing rate in the data is reproduced by the model . ( b ) three - cell correlations are explained partially by @xmath16 ( @xmath81 ) . ( c ) interaction matrix for @xmath16 . ( d ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] the inference results of @xmath16 are given in fig . [ retinapc2 ] . note that , by considering the correlation between the pairwise firing activity and the global population activity , i.e. , the second order population coupling , the three - cell correlation could be predicted partially ( @xmath81 ) , and this fraction is much larger than that of @xmath14 ( fig . [ retinapc1 ] ( c ) ) . this is due to the specific structure of @xmath16 , which incorporates explicitly three - cell correlations into the construction of couplings ( eq . ( [ le02 ] ) ) . technically , the mean - field theory for @xmath16 avoids the slow sampling and evaluates the high order correlations in a fast way . alternatively , one could fit the data using the conventional ising model @xcite with the same number of model parameters as @xmath16 , whereas , the three - cell correlations are hard to predict using mc samplings , and a similar phenomenon was also observed in a previous work for modeling pairwise correlations @xcite . therefore i speculate that @xmath16 acts as a key circuit variable for third order correlations . the interaction matrix of @xmath82 reveals how important each pair of neurons is for the entire population activity ( emergent functional state of the whole network ) . as shown in fig . [ retinapc2 ] ( c ) , @xmath16 matrix has no apparent structure of organization , i.e. , each neuron can be paired with both positive and negative couplings . some pairs have large negative @xmath16 , suggesting that these components are anti - correlated with the population activity . that is to say , the activity of these neuron - pairs is not synchronized to the population activity characterized by the summed activity over all neurons except these pairs . in the network , there also exist positive @xmath16s , which shows that these neuron - pairs are positively correlated with the population in neural activity . the interaction matrix shown here may be related to the revealed overlapping modular structure of retinal neuron interactions @xcite . in this structure , neurons interact locally with their adjacent neurons , and in particular this feature is scalable and applicable for larger networks . it seems that one individual neuron does not impact directly the entire population , and a small group of neighboring neurons have similar visual feature selectivity @xcite . this result is also consistent with two - neuron interaction map of the conventional ising model ( fig . [ retinaising ] ( a ) ) . note that in functional interpretation , these two - neuron interactions are inherently different from @xmath16 , which is designed to explain high - order correlations by using less model parameters than necessary . , one typical example ) compared with population coupling . ( a ) two - neuron interaction matrix . ( b ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] @xmath16 behaves better than @xmath14 in predicting the spike synchrony distribution ( fig . [ retinapc2 ] ( d ) ) in the small @xmath0 regime ( the prediction is improved from @xmath83 for @xmath14 to @xmath84 for @xmath16 ) . an intuitive explanation is that @xmath16 introduces equivalently triplet interactions among neurons , and it is known that high order interactions are necessary for generating widespread population activity @xcite . however , @xmath16 overestimates the distribution when rare events of synchronous spiking are considered . this may be related to the difficulty of obtaining sufficient equilibrium samples of the model , especially those samples with large population activity . the spike synchrony distribution is also compared with that obtained under ising model ( fig . [ retinaising ] ( b ) ) . different performances are related to the multi - information measure of neural population explained below . network samples for each @xmath1 . ( a ) multi - information ( in bits ) versus the network size @xmath1 . ( b ) the prediction fraction @xmath85 versus the network size . @xmath14 is used to predict two - cell correlations , and @xmath16 is used to predict three - cell correlations . , title="fig : " ] .1 cm network samples for each @xmath1 . ( a ) multi - information ( in bits ) versus the network size @xmath1 . ( b ) the prediction fraction @xmath85 versus the network size . @xmath14 is used to predict two - cell correlations , and @xmath16 is used to predict three - cell correlations . , title="fig : " ] .1 cm the amount of statistical structure in the neural data due to introducing interactions among neurons can be measured by the multi - information @xcite . i first introduce an independent model where only the firing rates of individual neurons are fitted and the corresponding entropy is defined as @xmath86 . the multi - information is then defined as @xmath87 , in which @xmath88 , where @xmath89 , and @xmath90 is assumed to be an upper bound to the true entropy . the true entropy for large populations is difficult to estimate since it requires including all possible interactions among neurons . however , the model entropy with low order interaction parameters could be an approximate information capacity for the neural population , which depends on how significant the higher order correlations are in the population . [ retinaif ] ( a ) shows the multi - information as a function of the network size . @xmath14 and @xmath16 are compared with the ising model @xcite , which reconstructs faithfully the pairwise correlations . @xmath16 improves significantly over @xmath14 in capturing the information content of the network , but its multi - information is still below that of the ising model , which is much more evident for larger network size . this is expected , because only part of third order correlations are captured by @xmath16 , while the ising model describes accurately the entire pairwise correlation profile which may be the main contributor to the collective behavior observed in the population . however , @xmath16 provides us an easy way to understand the higher order correlation , while in the ising model , it is computationally difficult to estimate the higher order correlations . the average prediction fraction of correlations by @xmath14 and @xmath16 is plotted in fig . [ retinaif ] ( b ) . @xmath14 predicts more than @xmath91 of the pairwise correlations , while @xmath16 predicts more than @xmath92 of the triplet correlations . the prediction fraction changes slightly with the network size . on the cortical data ( @xmath68 , one typical example ) . ( a ) firing rate in the data is reproduced by the model . ( b ) two - cell correlations are explained partially by @xmath14 ( @xmath93 ) . a monte - carlo ( mc ) sampling of the model yields similar results to belief propagation ( bp ) . ( c ) from the mc samples , three - cell correlations can also be estimated . ( d ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] on the cortical data ( @xmath68 , one typical example ) . ( a ) firing rate in the data is reproduced by the model . ( b ) three - cell correlations are explained partially by @xmath16 ( @xmath94 ) . ( c ) interaction matrix for @xmath16 . ( d ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] to show the inference performance of both types of population couplings on the cortical data , i randomly select a typical network example of @xmath76 neurons from the original dataset , and then apply the computation scheme to this typical example . results are shown in fig . [ mpcpc1 ] . surprisingly , the simplified @xmath14 is able to capture as high as @xmath93 of pairwise correlations , implying that when a rat performed working memory tasks , there exists a simplified model to describe emergent functional states in the medial prefrontal cortical circuit . moreover , mc sampling of the @xmath14 model also predicts well the spike synchrony distribution ( fig . [ mpcpc1 ] ( d ) ) . this is very different from that observed in the retinal data . in this sense , the mpc circuit is simple in its functional states when the subject is performing specified tasks . more interesting circuit features are revealed by @xmath16 , which is shown in fig . [ mpcpc2 ] . about @xmath94 of three - cell correlations are explained by @xmath16 in the mpc circuit . the interaction matrix of @xmath16 in fig . [ mpcpc2 ] ( c ) shows a clear non - local structure in the cortical circuit ( stripe - like structure ) . that is , some neurons interact strongly with nearly all the other neurons in the selected population , and these interactions have nearly identical strength of @xmath16 . such neurons having stripe - like structure in the @xmath16 matrix may receive a large number of excitatory inputs from pyramidal neurons @xcite , and thus play a key role in shaping the collective spiking behavior during the working memory task . the non - local effects are consistent with findings reported in the original experimental paper ( cross - correlogram analysis ) @xcite and the two - neuron interaction map under ising model ( fig . [ cortexising ] ( a ) ) . thus , to some extent , @xmath16 may reflect intrinsic connectivity in the cortical circuit , although the relationship between functional connections and anatomical connections has not yet been well established @xcite . lastly , @xmath16 overestimates the tail of the spike synchrony distribution ( fig . [ mpcpc2 ] ( d ) ) , which may be caused by the sampling difficulty of the inferred model ( a model with triplet interactions among its elements ) . the spike synchrony distribution of ising model is also compared ( fig . [ cortexising ] ( b ) ) . , one typical example ) compared with population coupling . ( a ) two - neuron interaction matrix . ( b ) probability of @xmath0 synchronous spiking under the model is compared with that of the data . ] network samples for each @xmath1 . ( a ) multi - information ( in bits ) versus the network size @xmath1 . ( b ) the prediction fraction @xmath85 versus the network size . @xmath14 is used to predict two - cell correlations , and @xmath16 is used to predict three - cell correlations . , title="fig : " ] .1 cm network samples for each @xmath1 . ( a ) multi - information ( in bits ) versus the network size @xmath1 . ( b ) the prediction fraction @xmath85 versus the network size . @xmath14 is used to predict two - cell correlations , and @xmath16 is used to predict three - cell correlations . , title="fig : " ] .1 cm multi - information versus the cortical network size is plotted in fig . [ mpcif ] ( a ) . in the cortical circuit , @xmath16 behaves comparably with the ising model ; even for some network size ( @xmath95 ) , it reports a higher information content than the ising model in the randomly selected subpopulations , which may be caused by the nature of the selected neurons ( e.g. , inhibitory interneurons @xcite , and they have stripe - like structure in the @xmath16 matrix ) . note that @xmath14 gives an information close to zero for small network sizes , suggesting that by introducing @xmath14 , one could not increase significantly the amount of statistical structure in the network activity explained by the model . however , the multi - information of @xmath14 grows with the network size , indicating that the role of @xmath14 would be significant for larger neural populations . [ mpcif ] ( b ) reports the prediction fraction of the correlation profile by applying @xmath14 and @xmath16 . both population couplings can capture over @xmath96 of correlations , which is significantly different from that observed in the retinal data . the emergent properties of the neural code arise from interactions among individual neurons . a complete characterization of the population activity is difficult , because on the one hand , the number of potential interactions suffers from a combinatorial explosion , on the other hand , the collective behavior at the network level would become much more complex as the network size grows . in this paper , i develop a theoretical framework to understand how pairwise or higher order correlations arise and the basic circuit variables corresponding to these correlation structures . the model is based on the concept of population coupling , characterizing the relationship between local firing activity of individual neuron or neuron - pair and the global neural activity . an advantage is that , it provides a low dimensional and neurobiologically interpretable representation to understand the functional interaction between neurons and their correlation structures . in particular , the concept of population coupling and the associated mean field method used in this paper offer an easy way to evaluate higher order correlations , while the usual sampling method is computationally hard and traditional models ( e.g. , ising model ) lack a direct interpretation of higher order correlations in terms of simplified ( population ) couplings . with the mean field method , the concept of population coupling is tested on two different types of neural data . one is the firing neural activities of retinal ganglion cells under natural movie stimuli . the other is the population activities of medial prefrontal cortex when a rat was performing odor - place matching working memory tasks . for the retinal data , on average @xmath14 accounts for more than @xmath91 of pairwise correlations , and @xmath16 accounts for over @xmath92 of three - cell correlations . the interaction matrix of @xmath16 contains information about the functional interaction features in the retinal circuitry . it seems that a retinal neuron can be paired with not only negatively strong couplings , but also slightly positive couplings . only a few pairs of neurons have strong correlations with the global activity of the population . to describe the spike synchrony distribution , @xmath16 performs better than @xmath14 , nevertheless , both of them could not capture the trend of the tail ( rare events related to higher order interactions existing in the network ) . this is not surprising , because @xmath14 and @xmath16 are simplified descriptions of the original high dimensional neural activity , taking the trade - off between the computation complexity and the description goodness . to extract the statistical structure embedded in the neural population , @xmath16 improves significantly over @xmath14 , and has further additional benefit of describing the third - order correlations observed in the data , as @xmath16 could be used to construct triplet interactions among neurons , although direct constructing all possible triplet interactions is extremely computationally difficult . unlike the retinal circuit , the cortical circuit yields a much smaller absolute value of the multi - information , implying that no significant higher order correlations ( interactions ) were present in the neural circuit when the circuit was carrying out task - related information processing rather than encoding well - structured stimuli ( as in the retinal network ) this also explains why a simplified description such as @xmath14 and @xmath16 is accurate enough to capture the main features of the population activity , including the spike synchrony distribution . the inferred model on the cortical data reveals a different interaction map from that of the retinal circuit . in the cortical circuit , neurons form the stripe - like structure in the interaction matrix , suggesting that these neurons may receive a large number of excitatory inputs @xcite . these inputs may come from different layers of cortex , and they can execute top - down or bottom - up information processing , thus modulate the global brain state in the target cortex during behavioral tasks . before summary of this work , i made some discussions about two relevant recent studies on population coupling ( see _ notes added _ ) . @xcite modeled directly the joint probability distribution of individual neural response and population rate ( the number of neurons simultaneously active ) by linear coupling and complete coupling models . the linear coupling reproduces separately the distribution of individual neuronal state and the population rate distribution , and their couplings , while @xmath14 introduced in my work reproduces mean firing rate and the correlation between individual neuronal state and the background population activity ( except the neuron itself ) . note that @xmath14 does not model population rate distribution explicitly , which is hard to interpret in terms of functional connectivity . the complete - coupling model reproduces the joint probability distributions between the response of each neuron and the population rate , from which it is hard to conclude that the high - order interactions responsible for high - order correlations can be interpreted and tested . however , @xmath16 reproduces mean firing rate and the correlation between neuron - pair activities and the background population activity ( except the neuron - pair itself ) , and thus explains high - order correlations by an energy model . furthermore , in this sense , this work overcame a weakness pointed out in another independent later work of population coupling @xcite , which fitted directly the population rate distribution and the firing probability for each neuron conditioned on the population rate , and analogously the corresponding model parameters can not be readily interpretable in a biological setting . due to intrinsic difference in model definitions , these two relevant works have nice properties of studying tuning curves of individual neurons to the population rate , and sampling from the model to reproduce the population synchrony distribution . in summary , i develop a theoretical model of population coupling and its advanced form , to relate the correlation profile in the neural collective activity to the basic circuit variables . the practical dimensional reduction method is tested on different types of neural data , and specific features of neural circuit are revealed . this model aiming at describing high order correlations with a low order representation , is expected to be useful for modeling big neural data . note that the interaction matrices shown in fig . [ retinapc2 ] and fig . [ mpcpc2 ] are qualitatively robust to changes of the data size to only the first half ( data not shown ) , verifying that the revealed features are not an artifact of overfitting . however , it still deserves further studies by introducing regularization in the learning equation . it is also very interesting to incorporate more physiologically plausible parameters to explain how the collective spiking behavior arises from the microscopic interactions among the basic units . another interesting study is to clarify the role of higher order correlations in decoding performances based on maximum likelihood principles @xcite . i am grateful to shigeyoshi fujisawa and michael j berry for sharing me the cortical and retinal data , respectively . i also thank hideaki shimazaki and taro toyoizumi for stimulating discussions . this work was supported by the program for brain mapping by integrated neurotechnologies for disease studies ( brain / minds ) from japan agency for medical research and development , amed . _ note added._after i submitted this work to arxiv:1602.08299 , i became aware of ref . @xcite ( arxiv:1606.08889 ) , and later ref . @xcite ( biorxiv , 2016 ) . discussions about these two recent relevant works are made in the last section of this paper . in this appendix , i give a simple derivation of mean - field equations given in sec . [ model ] . more details can be obtained from refs @xcite . first , after removing an interaction @xmath27 , one defines the cavity probability @xmath97 , where the product comes from the physical meaning of the second kind of cavity probability , namely @xmath98 defined as the cavity probability when only the connection from @xmath36 to @xmath6 is retained while other neighbors of @xmath6 are removed ( so - called cavity probability ) . @xmath98 is thus formulated as @xmath99 . with these two probabilities , it follows that the cavity magnetization @xmath100 , where @xmath101 is named cavity bias in physics @xcite . it is related to @xmath39 by @xmath102 . [ biassm02 ] @xmath108+e^{-\gamma_b}[p_{j\rightarrow b}(+1)p_{k\rightarrow b}(-1)+p_{j\rightarrow b}(-1)p_{k\rightarrow b}(+1 ) ] } { e^{-\gamma_b}[p_{j\rightarrow b}(+1)p_{k\rightarrow b}(+1)+p_{j\rightarrow b}(-1)p_{k\rightarrow b}(-1)]+e^{\gamma_b}[p_{j\rightarrow b}(+1)p_{k\rightarrow b}(-1)+p_{j\rightarrow b}(-1)p_{k\rightarrow b}(+1)]}\\ & = \frac{1}{2}\ln\frac{1+e^{-2\gamma_b}\frac{1-m_{j\rightarrow b}m_{k\rightarrow b}}{1+m_{j\rightarrow b}m_{k\rightarrow b}}}{e^{-2\gamma_b}+\frac{1-m_{j\rightarrow b}m_{k\rightarrow b}}{1+m_{j\rightarrow b}m_{k\rightarrow b}}}\\ & = \tanh^{-1}\left(\tanh\gamma_b m_{j\rightarrow b}m_{k\rightarrow b}\right ) . \end{split}\end{aligned}\ ] ] similarly , the single neuron magnetization @xmath109 is obtained via @xmath110 , where @xmath111 is a normalization constant , and the multi - neuron correlation @xmath112 , where @xmath113 is a normalization constant . note that @xmath111 and @xmath113 are also related to the free energy contribution of single neuron and neuronal interaction @xcite , respectively . the full ( non - cavity ) magnetization can be derived in a similar manner to eq . ( [ magsm ] ) , as @xmath114 . in detail , the two - point correlation @xmath115 ( @xmath116 ) is computed as follows , where i have used @xmath121 and the magnetization parameterization of @xmath122 . for triplet interaction , @xmath123+e^{-\gamma_b s_i}[p_{j\rightarrow b}(+1)p_{k\rightarrow b}(-1)+p_{j\rightarrow b}(-1)p_{k\rightarrow b}(+1)]$ ] , and the corresponding @xmath124 . following the same line , the partition function @xmath113 is evaluated for pairwise interaction as m. okun , n. a. steinmetz , l. cossell , m. f. iacaruso , h. ko , p. bartho , t. moore , s. b. hofer , t. d. mrsic - flogel , m. carandini , and k. d. harris . diverse coupling of neurons to populations in sensory cortex . , 521:511 , 2015 . michael okun , pierre yger , stephan l marguet , florian gerard - mercier , andrea benucci , steffen katzner , laura busse , matteo carandini , and kenneth d harris . population rate dynamics and multineuron firing patterns in sensory cortex . , 32:17108 , 2012 .	to understand the collective spiking activity in neuronal populations , it is essential to reveal basic circuit variables responsible for these emergent functional states . here , i develop a mean field theory for the population coupling recently proposed in the studies of visual cortex of mouse and monkey , relating the individual neuron activity to the population activity , and extend the original form to the second order , relating neuron - pair s activity to the population activity , to explain the high order correlations observed in the neural data . i test the computational framework on the salamander retinal data and the cortical spiking data of behaving rats . for the retinal data , the original form of population coupling and its advanced form can explain a significant fraction of two - cell correlations and three - cell correlations , respectively . for the cortical data , the performance becomes much better , and the second order population coupling reveals non - local effects in local cortical circuits .
patsis and zachilas [ 1994 ] proposed a method to visualize 4d spaces of section . it is based on rotation of the 3d projections of the figures in order to understand the geometry of the projections and on color for understanding the distribution of the consequents in the 4th dimension . this combined information from the rotated , colored , 3d projections allows us to associate specific structures with the dynamical behavior in the neighborhood of periodic orbits . recently the method has been applied successfully in a series of papers that studied the dynamics in the neighborhood of periodic orbits in a 3d galactic potential [ katsanikas and patsis 2011 , katsanikas et al 2011a , katsanikas et al 2011b ] . the results of these papers , together with those of the original one by patsis and zachilas [ 1994 ] , as well as further results from work in progress , find a consistency between the shapes of the encountered structures in the 4d spaces of section and the kind of ( in)stability of the periodic orbit . despite the fact that until now no exceptional behavior has been found , the results can not be characterized as generic . the up to now studied systems are 3d autonomous hamiltonians with potentials suitable to describe stellar motion in 3d rotating galactic disks . they can be used to investigate observed morphological features out of the equatorial planes of disk galaxies [ patsis et al 2002 ] . the motivation for investigating the phase space structure of a 4d symplectic map is to check whether or not the behavior encountered in the hamiltonian systems is model dependent . so , we apply the same technique in the case of a 4d symplectic map and we study the structure of the phase space in the case of stability and in cases of instabilities studied in the katsanikas papers . our goal is to compare the dynamics of the 4d map with those found in the hamiltonian system , testing in this way the ubiquity of the results of the latter studies . the paper is structured as follows : in sec . [ meth ] we briefly describe the method , in sec . [ map ] we present the map we use in our study , in sec . [ res ] we describe our results and finally we enumerate our conclusions in sec . [ concl ] . we consider the map @xmath0 and follow the evolution of the array @xmath1 . a set of three coordinates , e.g. @xmath2 , are used for the 3d projection , while the fourth coordinate ( e.g. @xmath3 ) determines the color of the consequents . there is a normalization of the color values in the [ min(@xmath3 ) , max(@xmath3 ) ] interval , which is mapped to [ 0,1 ] . in order to understand the shape of each 3d projection we rotate the figures on the screen of our computer . for presentations on paper we chose a set of projection angles that help the reader understand the shape of the object we describe in the text . we use in our applications the `` mathematica '' package . following the intrinsic `` mathematica '' subroutines our viewpoint is given in spherical coordinates . the unit for the distance @xmath4 of the consequents of the surface of section from the observer is given by `` mathematica '' in a special scaled coordinate system , in which the longest side of the bounding box has length 1 . for all figures we use @xmath5 . the method associates the smooth distribution or the mixing of colors , with specific types of dynamical behavior in the 4th dimension [ patsis and zachilas 1994 , katsanikas and patsis 2011 , katsanikas et al 2011a , katsanikas et al 2011b ] . for a more detailed description see [ katsanikas and patsis 2011 ] the map we have chosen , @xmath6 , belongs to a family of nonlinear symplectic 4d mappings in @xmath7 that is a generalization of the standard map . the definition in the general form is : @xmath8 several cases of the maps of this family have been used in the past to study the dynamics in the neighborhood of fixed points . already froeschl [ 1972 ] used ( 1 ) with @xmath9 and tried to visualize the 4d surfaces of section . this was an additional motivation for choosing the particular map to apply our visualization technique . furthermore studies accomplished by pfenniger [ 1985 ] , oll and pfenniger [ 1999 ] , jorba and oll [ 2004 ] , guarantee that hamiltonian hopf bifurcations are happening in this system . a study of a 4d symplectic map by means of gali indicators can be found in manos et al . [ this allows us to compare also the behavior of the galactic type hamiltonian with that of the map at transitions from stability to complex instability . following pfenniger [ 1985 ] and jorba and oll [ 2004 ] , we examine the case with @xmath10 having as conjugated variables @xmath11 and @xmath12 . we also restrict the parameter space by taking @xmath13 . then , we rename @xmath14 as @xmath15 , and @xmath16 as @xmath17 . it is evident from the stability diagrams in pfenniger [ 1985 ] , that there is always a critical value @xmath18 , for @xmath19 , for which we have a transition from stability to complex instability . in order to include the case of such a transition in our calculations , we take here @xmath20 . we performed our calculations in double , as well as in quadruple precision and we found the same results , except in a case of simple instability ( see sect . [ res ] below ) , where the results are only qualitatively similar . in that case we present the results found with quadruple precision . we studied the dynamics close to the fixed point ( 0,0,0,0 ) by varying @xmath15 . in the neighborhood of stable fixed points @xmath21 we find in the 4d phase space invariant tori . such invariant tori have been presented in projections already by froeschl [ 1972 ] . however , with the method of color and rotation we could verify that these tori are 4d objects . their 3d projections are of the kind called `` rotational tori '' by vrahatis et al . [ their 4d representation , in all cases we examined , is in perfect agreement with the rotational invariant tori with smooth color variation on their surfaces found by katsanikas and patsis [ 2011 ] ( see their figure 13 ) . in fig . [ stab ] we give a typical example for @xmath22 . the initial conditions are @xmath23 . we use the @xmath2 3d projection for the spatial representation of the consequents and @xmath3 to give the color that represents the fourth dimension . in this and in all subsequent figures the color bar on the right hand side gives the range of the color variation that corresponds to the normalized values of the fourth coordinate . even the change of the smooth color variation from the external to the internal side of the torus described by katsanikas and patsis [ 2011 ] in their study of the autonomous hamiltonian system is observed in the case of the rotational torus around the stable fixed point of the symplectic map . the detailed description of the color variation on the surface of the rotational torus should be sought in katsanikas and patsis [ 2011 ] . these structures characterize the dynamics of ordered behavior in the case of our map . calculating lyapunov characteristic numbers for all these orbits on the tori we find that they tend very fast to zero . case . ( a ) the characteristic toroidal figure associated with stability . the arrows indicate the regions where the smooth color variation changes from the external to the internal side of the torus . the point of view is @xmath24 ( b ) an enlargement of ( a ) at one of the transition regions of the color from the external ( `` _ _ ext _ _ '' ) to the internal ( `` _ _ int _ _ '' ) side of the torus ( cf . with figure 12 in katsanikas and patsis [ 2011 ] ) . we use less consequents in order to show how the structure of the colors varies . the point of view is now @xmath25 to show better how the color change side . ] simple instability is encountered for @xmath26 as the @xmath27 value increases . in the neighborhood of simple unstable ( 0,0,0,0 ) fixed points ( two eigenvalues on the unit circle and two on the real axis ) we find consequents defining a ribbon of the form of a double loop . this characteristic double loop structure is encountered only close to a critical value of the parameter ( the value of @xmath15 ) for which the stability of the fixed point changes and it becomes simple unstable . a typical example for @xmath28 and initial conditions @xmath29 is presented in fig . it gives the first @xmath30 consequents . such structures are associated with simple instability in 3d galactic potentials by patsis and zachilas [ 1994 ] and are extensively investigated by katsanikas et al . [ 2011c ] . the current situation resembles the dynamical behavior encountered in the neighborhood of simple unstable periodic orbits of the z - axis family in 3d rotating hamiltonian systems [ katsanikas et al . 2011c ] , when we have a surface of section defined by @xmath31 . in the intersection of two branches of the double loop structure we have the same color , indicating that it belongs to the cases where the double loops correspond to a real 4d `` 8''-like structure . case . we observe the characteristic double loop figure associated with simple instability . the point of view is @xmath32 . ] after a large number of iterations the consequents depart from the surface of the ribbon - like object and build a cloud of points filling finally all available phase space . as @xmath27 increases the consequents leave the `` 8''-like structure faster . in the example of fig . [ u2 ] , @xmath33 and the initial conditions are as in the previous case . we observe that the consequents have left the `` 8''-like structure after @xmath34 iterations and start visiting larger volumes of the phase space ( fig . this is a case where we encountered quantitative differences when we use quadruple precision in our calculations . however , also in this case the results are qualitative the same . this behavior is depicted in the evolution of the `` finite time '' lyapunov characteristic number @xmath35 , where @xmath36 and @xmath37 are the distances of two nearby points at @xmath38 and after @xmath39 iterations [ see e.g. skokos 2010 ] . the @xmath40 for the example of fig . [ u2]a is given in fig . the evolution of the @xmath40 index has been followed for @xmath41 iterations . projection of the phase space close to a simple unstable fixed point for @xmath33 . the consequents depart from the double loop structure after about @xmath34 iterations . ( b ) the @xmath40 evolution . we observe that lcn levels off at a positive value about 0.355 . ] the index has an overall decreasing part that ends after about @xmath42 iterations ( when the consequents start departing from the 4d `` 8''-structure ) . then it increases and finally it levels off at a value about 0.355 . the fixed point ( 0,0,0,0 ) becomes double unstable for @xmath43 . in all cases we have examined , close to double unstable fixed points we observe clouds of scattered consequents with mixing of colors , even close to the transitions from simple to double instability . in the example of fig . [ du]a we give an orbit for @xmath44 with initial conditions @xmath45 . in this case @xmath40 levels off almost immediately at a value about 0.85 ( fig . this is the usual behavior also in most cases studied by katsanikas et al [ 2011c ] close to double unstable periodic orbits . . in this case we observe a cloud of points with scattered colors . the point of view is again @xmath46 . ( b ) the _ _ lcn(n)__,title="fig : " ] . in this case we observe a cloud of points with scattered colors . the point of view is again @xmath46 . ( b ) the _ _ lcn(n)__,title="fig : " ] for @xmath20 , the eigenvalues of the characteristic polynomial of the monodromy matrix become complex , off the unit circle , for @xmath47 . the critical value at which happens the transition from stability to complex instability is @xmath48 , i.e. for this value we have a hamiltonian hopf bifurcation . studying the dynamics close to the ( 0,0,0,0 ) fixed point , for @xmath49 , just beyond the critical value , we find the disky structure ( confined torus ) that characterizes the dynamics close to complex unstable periodic orbits or fixed points . this structure has been found in the 3d projections of the 4d phase space by pfenniger [ 1985 ] , jorba and oll [ 2004 ] and oll et al . [ it has been shown by katsanikas et al . [ 2011a ] that this structure is a 4d object with a smooth color variation as we move from one side of the `` disk '' to the other . for the case of the map ( 1 ) we can see it in fig . [ co1 ] for @xmath50 iterations . case . we observe a disky structure with smooth color variation in the 4d phase space of our symplectic map . the point of view is @xmath51 . ] these `` disks '' have an internal spiral structure [ contopoulos et al . 1994 , papadaki et al . 1995 , contopoulos 2002 ] , the 4d structure of which has been shown by katsanikas et al [ 2011a ] . a similar internal spiral structure as the one described by katsanikas et al [ 2011a ] in the 4d confined tori ( see their figures 2,3,4 and 5 ) has been encountered in all cases of map ( 1 ) we have studied in this paper . in the present case the spiral has six arms . the behavior of the @xmath52 indices in the two cases are also similar . fig . [ liadisk ] depicts the variation of @xmath40 up to @xmath53 and is similar to the @xmath54 variation of the corresponding case of the autonomous hamiltonian system ( cf . figure 6 in katsanikas et al 2011a ) . in both case an initial decrease of the indices is followed by a leveling off . evolution in the case of fig . we observe that @xmath40 levels off at a positive value about 0.009 , after an initial decreasing part . ] also the behavior in the neighborhood of a complex unstable fixed point away from the critical @xmath55 value resembles that in the neighborhood of a complex unstable periodic orbit away from the jacobi constant value for which we have the transition from stability to complex instability . we have clouds of points with mixed colors . in the latter case the @xmath40 index reaches a positive value , larger than that of the confined torus case . the whole variation of the index remains in the cases we examined larger than the values we obtained for the disky structure for the same number of iterations . in this paper we compared the dynamical behavior in the neighborhood of fixed points in a 4d symplectic map and we compared the results with those found in the neighborhood of periodic orbits in a 3d autonomous hamiltonian system with a potential of galactic type . in our study we used the method of color and rotation . we found that the structures encountered in the spaces of section in the hamiltonian system , which characterize the kind of ( in)stability of the periodic orbits are also found characterizing the dynamics in the neighborhood of fixed points in the 4d symplectic map . these structures are : ( 1 ) the tori with the smooth color variation for stability , ( 2 ) the double loop ribbons with smooth color variation for the simple unstable case close to transitions from stability to simple instability , ( 3 ) the clouds of scattered points in the neighborhood of double unstable periodic orbits , and ( 4 ) the disky structures with smooth color variation close to a complex unstable fixed point near the transition from stability to complex instability . these results indicate that the 4d visualization of the surfaces of section in 3d autonomous hamiltonian systems and of the phase space in 4d symplectic maps associates the dynamics in the neighborhood of a fixed point , or a periodic orbit , with specific structures . we have not found any exception even in the case of the symplectic map we used in the present study . thus , these structures seem to reflect a generic behavior in a larger class of dynamical systems . our study also indicates that the color and rotation method allows us to know the kind of instability of the fixed point by a simple inspection of the 4d phase space and gives in a direct way an insight of the dynamics in such regions , not only in galactic type potentials , but in a broader spectrum of dynamical systems . _ acknowledgments _ we thank prof . contopoulos for fruitful discussions , as well as both referees for their constructive comments that improved the paper . 1 . contopoulos g. [ 2002 ] _ order and chaos in dynamical astronomy _ springer - verlag , new york berlin heidelberg . 2 . contopoulos g. , farantos s.c . , papadaki h. and polymilis c. [ 1994 ] `` complex unstable periodic orbits and their manifestation in classical and quantum dynamics '' _ phys . e _ * 50 * , 4399 - 4403 . 3 . froeschl , c.[1972 ] , `` numerical study of a four - dimensional mapping '' _ astron . astrophys . _ * 16 * , 172 - 189 4 . jorba a. , oll m. [ 2004 ] `` invariant curves near hamiltonian hopf bifurcations of four - dimensional symplectic maps '' _ nonlinearity _ * 17 * , 691 - 710 . 5 . katsanikas m. , patsis p.a . [ 2011 ] `` the structure of invariant tori in a 3d galactic potential '' _ int . . chaos _ * 21 * , no.2,467 - 496 6 . katsanikas m. , patsis p.a . , contopoulos g. [ 2011a ] `` the structure and evolution of confined tori near a hamiltonian hopf bifurcation '' _ int * 21 * , no.8 , 2321 - 2330 7 . katsanikas m. , patsis p.a . , pinotsis a.d . [ 2011b ] `` chains of rotational tori and filamentary structures close to high multiplicity periodic orbits in a 3d galactic potential '' _ int . * 21 * , no.8 , 2331 - 2342 8 . katsanikas m. , patsis p.a . , contopoulos g. [ 2011c ] `` instabilities and stickiness in a 3d rotating galactic potential '' - submitted . 9 . manos t. , skokos ch . and antonopoulos ch . [ 2012 ] `` probing the local dynamics of periodic orbits by means of generalized alignment index ( gali ) method '' , _ int . . chaos _ * 22 * , 1250218 . 10 . oll m. , pfenniger d. [ 1999 ] `` bifurcation at complex instability '' in _ hamiltonian systems with three or more degrees of freedom _ , nato asi c , c.sim ( ed ) , pp 518 - 522 , kluwer , dordrecht 11 . oll m. , pacha j.r . and villanueva j. [ 2004 ] `` motion close to the hopf bifurcation of the vertical family of periodic orbits of @xmath56 '' _ celest . astr . _ * 90 * , 89 - 109 . 12 . papadaki h. , contopoulos g. , polymilis c. [ 1995 ] `` complex instability '' in : _ from newton to chaos _ ed by a.e . roy , b.a . steves , plenum press , new york , pg . patsis p.a . , skokos c. , athanasoula e. [ 2002 ] `` orbital dynamics of three - dimensional bars . boxy / peanut edge - on profiles '' , _ mon . not . _ , * 337 * , 578 - 596 14 . patsis p.a . and zachilas l. [ 1994 ] `` using color and rotation for visualizing four - dimensional poincar cross - sections : with applications to the orbital behavior of a three - dimensional hamiltonian system '' _ int . j. bif . chaos _ * 4 * , 1399 - 1424 . pfenniger d. [ 1985 ] `` numerical study of complex instability : i mappings '' _ astron . astrophys . _ * 150 * , 97 - 111 . 16 . skokos ch . [ 2010 ] `` the lyapunov characteristic exponents and their computation '' , _ lect . not . phys . _ * 790 * , 63 - 135 17 . vrahatis , m.n . , isliker , h. & bountis , t.c . [ 1997 ] `` structure and breakdown of invariant tori in a 4-d mapping model of accelerator dynamics '' , _ int . j. bif . chaos . _ * 7 * , 2707 - 2722 .	we study the dynamics in the neighborhood of fixed points in a 4d symplectic map by means of the color and rotation method . we compare the results with the corresponding cases encountered in galactic type potentials and we find that they are in good agreement . the fact that the 4d phase space close to fixed points is similar to the 4d representations of the surfaces of section close to periodic orbits , indicates an archetypical 4d pattern for each kind of ( in)stability , not only in 3d autonomous hamiltonian systems with galactic type potentials but for a larger class of dynamical systems . this pattern is successfully visualized with the method we use in the paper .
beams of light can carry quantized angular momentum , @xmath0 , that are typically referred to as having a photonic topological charge ( ptc ) of @xmath1 @xcite . circularly polarized light carries a spin angular momentum ( sam ) of @xmath2 and so has ptc of @xmath3 , but beams can also be created that twist about their axis of propagation @xcite causing them to be referred to as _ vortex beams _ or simply _ twisted light _ @xcite . at the macro - scale , their interaction with matter generates torque which can be used to manipulate rings of atoms @xcite and particles @xcite , carry out fabrication @xcite , provide control for spintronics @xcite , and for many other applications as well @xcite . at the atomic scale , the absorption of such photons can change the angular momentum state of bound electrons , only recently confirmed experimentally @xcite . the ptc of twisted light also amounts to an additional degree of freedom for carrying information content , and ptc values in excess of @xmath4 have been experimentally realized @xcite . this information perspective on ptc is particularly relevant to emerging technologies in communications , computing and quantum cryptography @xcite , classical data transfer @xcite , quantum key distribution@xcite , quantum entanglement@xcite , and quantum cloning@xcite . there are a variety of ways in which photonic angular momentum can be generated . linearly polarized light can be transformed to a circular polarization by exploiting material birefringence in the form of a quarter - wave plate giving it sam and a topological charge of @xmath3 . only slightly more complicated to carry out , light beams with orbital angular momentum ( oam)i.e . twisted light can be generated using standard optical elements such as spiral phase plates @xcite , computer generated holograms @xcite , and spatial light modulators @xcite . beams with sam can also be converted to those with oam . staged spiral phase plates @xcite , uniaxial crystals @xcite , liquid crystals @xcite , q - plates@xcite , and even metamaterials @xcite have all been demonstrated to produce such conversions . changing the angular momentum of photons that already have a topological charge is less standard though . such manipulations currently exploit higher - order susceptibilities to up / down convert oam along the same lines as is carried out to alter the wavelength of light . for instance , topological charge doubling via second harmonic generation ( shg ) has been demonstrated @xcite , and ptc addition operations have also been realized using shg@xcite . topological charge can even be transferred from a pump beam to a down - converted idler beam via optical parametric oscillation ( opo ) @xcite . along the same lines , four - wave - mixing ( fwm ) with atomic vapors can result in output beams that have either the sum of the input ptcs or no ptc at all @xcite . in each of these nonlinear processes , charge conversion is essentially instantaneous and involves two vortex beam inputs that must illuminate the target material simultaneously . a variation of fwm , applied to bulk crystals , generates output beams with ptcs that are linear combinations of those of the input beams @xcite . intriguingly , there is a time delay between pump and probe implying that ptc is being temporarily stored in the electronic structure of the bulk crystal , but the nature of this structure was not elucidated . here we consider an alternative means of manipulating ptc that exploits the properties of a special type of molecular exciton . the requisite structures have a discrete rotational symmetry group of either @xmath5 or @xmath6 with respect to the propagation direction of light . such n - arm molecules have been previously identified as having excitonic states that can relax by emitting photons with ptc @xcite . we extend this work by first introducing the notion of an _ excitonic topological charge _ ( etc ) that can be generated with azimuthally polarized light @xcite . the resulting electronic state of the molecule will be referred to as a _ twisted exciton_. since both spin and circularly polarized vortex beams can be decomposed into a linear combination of radial and azimuthal vector vortices@xcite , they can create twisted excitons too . the idea is then generalized to consider the sequential excitation of molecules with pulses of light that carry either sam or oam . it is shown that the sum of ptc and etc is conserved in such processes resulting in a simple algebra in which etc can be raised and lowered using photons . subsequent emission , either spontaneous or stimulated , produces photons with the accumulated topological charge of the excitons . in this way , the excitons play the role of an angular momentum bank in which photonic charge can be deposited and withdrawn in different increments . this is illustrated in figure [ calculator ] . within a linear setting , the fact that the real momentum of absorbed photons is transferred to crystal momentum is a foundational plank of solid - state physics @xcite . an analogous discrete - space application of noether s first theorem has also be used to show that real angular momentum can be transferred to crystal angular momentum @xcite . this provides a conceptual basis for understanding the topological charge algebra that can be carried out with molecules . in the molecular setting , periodicity is azimuthal and lattice points are associated with identical molecular arms . etc quantifies the molecular analog of crystalline angular momentum . we first use a tight binding ( tb ) paradigm to show that topological charge is conserved in the molecular setting . this paradigm is used to demonstrate topological charge algebra using sequential laser pulses . the associated hamiltonian is subsequently replaced with one that does not rely on prescribed transition dipoles and for which electron excitations are treated as many - body events with exchange energies and correlation effects included . twisted excitons are no longer just superpositions of two - level excitations on each molecular arm , but topological charge conservation emerges none - the - less . this time - domain density functional theory ( td - dft ) setting is used to simulate the dynamics of topological charge transfer between photonic and excitonic manifestations , and simple charge additions and subtractions are once again demonstrated . in both tb and td - dft settings , the underlying processes are linear in the sense that only first - order electric dipole interactions are necessary . this is in contrast to the nonlinear optics strategy for converting up / down converting angular momentum using higher order susceptibilities . first consider a molecular hamiltonian in the absence of light - matter coupling . the requisite @xmath5 or @xmath6 symmetry is provided by an n - arm molecule in which arm @xmath7 supports two energy levels : ground state @xmath8 and excited state @xmath9 . the tight - binding ( tb ) hamiltonian is taken to be @xmath10 here @xmath11 is the excited state energy of each arm , @xmath12 is the coupling between nearest arms , and @xmath13 is the creation operator for arm @xmath7 . it is straightforward to show @xcite that the ground state is @xmath14 while the n excited states are @xmath15 with @xmath16 and @xmath17 . the etc , @xmath18 , is an integer ranging from @xmath19 to @xmath20 . the corresponding energies are @xmath21 with @xmath22 for @xmath23 and @xmath24 for @xmath25 . a hollow @xmath26 will be used to distinguish exciton energy from electric field , @xmath27 . now introduce semi - classical light - matter coupling via two hamiltonians : @xmath28 that governs light - mediated interactions between the ground state and each molecular eigenstate , while @xmath29 governs the analogous laser interactions that can cause transitions between eigenstates . the angular momentum of incident electric fields may be manifested as a circular polarization , a vector vortex , or linear polarization with a scalar vortex , but we restrict attention to the first two types . an electric dipole approximation is made , and the discrete rotational symmetry ensures that a rotation of the molecule about its axis by @xmath30 maps one dipole into the next . under these conditions , the details of electric field structure and dipole orientations are irrelevant , and the light - matter interactions are well - captured by the following hamiltonians , which are functions of the topological charge of the incident light , @xmath31 : @xmath32 the _ mod _ function returns its argument modulo @xmath33 and use has been made of the fact that @xmath34 . the scalars , @xmath35 and @xmath36 , represent the inner product of electric transition dipole moments with a time - dependent electric field . the total hamiltonian , @xmath37 , is then applied to the schr " odinger equation with solutions assumed to be of the form @xmath38 this results in a set of @xmath39 coupled ode s that can be solved numerically for a prescribed electric field and initial state . the evolving state can then be projected onto each excitonic eigenstate to determine the population of each topological charge state , @xmath18 , as a function of time : @xmath40 within the tb setting , conservation of energy implies conservation of topological charge in association with light - matter interactions . to see this , suppose that the molecule is initially in eigenstate @xmath41 , where subscript @xmath42 indicates an initial etc of @xmath43 . a beam with ptc of @xmath31 , is incident on the molecule , and the molecule is excited into eigenstate @xmath44 as a result . subscripts @xmath45 and @xmath46 delineate photonic and excitonic manifestations while @xmath47 is the etc of the final state . a necessary condition for this transition to occur is that @xmath48 . assuming that @xmath49 , equations [ tbstates ] and [ h12 ] imply that @xmath50 which can be easily reduced to @xmath51 the following cyclic sum orthogonality property of periodic exponentials is then useful : @xmath52 it is applied to both terms in equation [ twoterms ] to give @xmath53 energy conservation implies that exactly one of the two kronecker delta functions will be nonzero . we therefore have the following statements of topological charge conservation : @xmath54 if the initial exciton energy is lower than that of the final state , crystal angular momentum will be removed by the light . the converse is also true . since the sign of the ptc can be either positive or negative , this allows for a number of ways in which photonic angular momentum quanta can be used to manipulate the level of excitonic angular momentum . it also offers a strategy for withdrawing angular momentum from the molecule in a range of denominations . this is next illustrated in two applications . as a first proof - of - concept , a 7-arm molecule with an etc of @xmath55 is subjected to windowed , continuous wave ( cw ) , azimuthally polarized lasers with angular momenta of @xmath56 and the following scalar waveform : @xmath57 as usual , @xmath42 and @xmath58 are the initial and final topological charges of the exciton with corresponding energies , @xmath59 and @xmath60 , given by equation [ tbenergies ] . all tb simulations used the following parameters in atomic units : @xmath61 and a radial distance to the center of each arm of 0.6 . here @xmath62 is the strength of the transition dipole . proportional changes to these parameters do not affect the results of course . the top plots of figure [ tb_am2_up_down ] show how the eigenstate populations evolve as a function of time , and the associated charge conservations are listed above each plot . the bottom plot is a composite of results from both simulations which shows the phase relationship between two adjacent arms before and after application of the laser pulses . the horizontal axis was obtained by rigidly translating sections of the plots of amplitude versus time so that they appear in the same time interval . then time was mapped into phase through multiplication with the associated frequencies of light . in this form , the phase relation between two arms can be easily measured by comparing the amplitude of one arm ( solid ) with its neighbor ( dashed ) . the initial phase progression should be @xmath63 , from equation [ tbstates ] , and this is confirmed in the black and dashed black curves . the addition of ptc results in a @xmath64 etc and the measured phase between the red and dashed red curves exhibits the expected progression of @xmath65 . likewise , the subtraction of ptc leaves the molecule with @xmath66 and the anticipated phase progression of @xmath67 between arms , shown in blue and dashed blue . a second tb simulation , figure [ tb_addition ] , carries out a sequence of charge addition and subtraction that starts and ends in the ground state ( gs ) . the same 7-arm molecule is subjected to three windowed , cw laser pulses . the first laser @xmath68 has @xmath69 and an energy equal to that of the eigenstate for an @xmath70 . the second laser @xmath71 has the same ptc but with an energy equal to the difference of excitonic charge states one and two . the ptc of the third laser is @xmath72 with an energy equal to that of the eigenstate for which @xmath73 . the topological charge balances associated with each laser are given in the figure to more easily interpret the data plotted . these plots also show how the excitonic energy evolves as a function of time , becoming asymptotic to the appropriate eigenenergies after each laser pulse is applied . photons are absorbed and one @xmath74 photon is emitted . the color legend identifies the populations of each etc state.,scaledwidth=70.0% ] many of the idealizations associated with the tb paradigm can be removed by reconsidering the light - matter dynamics using time - domain density functional theory ( td - dft ) . unlike standard ground state density functional theory ( dft ) , td - dft captures the non - equilibrium response of material to an externally applied , time - varying electric field . such real - time simulations are made possible through runge - gross ( rg ) reformulation of the time - dependent schr " odinger equation @xcite . a methodology was developed so that td - dft can be used to quantify topological charge transfers as detailed in the methods section . td - dft calculations are computationally intense and amount to carrying out a standard dft calculation for a series of very small time steps . once again using atomic units , the requisite time step for the calculations of this study , in particular , is @xmath75 au for simulations covering approximately @xmath76 au in total . to reduce the computational cost in this initial proof - of - concept , a ring of radially aligned @xmath77 molecules was used as an idealized n - arm system . the ring was given a radius of @xmath78 bohr with the h - h bond lengths taken to be @xmath79 bohr . for each simulation , a computational domain was constructed as the sum of spheres of radius @xmath80 bohr around each atom . this domain was discretized with a spacing of @xmath81 bohr . the generalized gradient approximation ( gga ) parametrized by perdew , burke , and ernzerhof ( pbe)@xcite was adopted to account for exchange and correlation , and a troullier - martins pseudopotential was used . since the wavelength of laser field is much larger than the dimension of the n - arm system , a point dipole approximation of light - matter interaction is applied in our td - dft simulations . because the limitations on the type of external field that can be inputted to octopus , an approximation scheme was used to create twisted light . specifically , it was taken to have n piecewise homogeneous components . transition dipoles from the ground state were found to be maximal on each arm and so the associated field components were constructed to be arm - centered . in contrast , the transition dipoles between two excited states were maximal at the midpoints between arms , so the associated external field components were centered between the arms . a 5-arm configuration of @xmath77 dimers was considered all of the td - dft simulations . casida s perturbative td - dft methodology @xcite was first performed to obtain excitation energies . this was necessary in order to design laser pulses with the frequency need to excite a given excitonic state . the energies , dominant determinants , and the corresponding topological charges of the first five excited states are given in table [ casidaexs ] . .dominant determinant , @xmath82 , and topological charge , @xmath18 , of the first five lowest excited states as calculated from casida perturbation within td - dft . here @xmath82 is the spin - adapted singlet so that one electron is excited from @xmath83 occupied ks orbital to the @xmath84 unoccupied ks orbital . energies are in hartrees ( ha ) . [ casidaexs ] [ cols="<,^,^,^,^,>",options="header " , ] the determinants listed in table [ casidaexs ] are those that dominate each excited state , representing approximately @xmath85 of the respective wavefunction . approximating an excited state with only a single determinant makes it possible to find the population of etc states , equation [ population ] , as detailed in the methods section . it was found that the more realistic many - body setting is still able to carry out topological charge algebra . three cases were considered which each involve a sequence of two laser pulses . the following radial vector vortex field , with a gaussian envelope , was utilized : @xmath86 as with the tb analyses , conservation of energy determines whether or not the ptc is added to or subtracted from the molecular assembly . first consider a scenario in which a 0.367 ha laser pulse of @xmath72 is used to create one of the two lowest - energy twisted excitons , @xmath87 , of table [ casidaexs ] . a second laser pulse , with an energy equal to the difference between that of the first and second excitonic states , 0.0220 ha , is subsequently applied as shown in figure [ etcm2tom1 ] . this requires that the photon be absorbed since there is no excitonic state between the ground and first - excited energy levels . the associated statement of charge conservation is thus @xmath88 . in contrast , nothing happens if the @xmath87 state is subjected to the same laser energy but of the opposite sam since this would be violate topological charge conservation . the lower - right panel of figure [ etcm2tom1 ] demonstrates this . as is clear in the figure , the second laser needs to be much stronger than the first because the transition dipole between excited states is in the mid - arm region , where the relevant state densities are small , while the transition dipoles from the ground state are located in the arms where the relevant state densities are much larger . . upper panel shows sequenced laser pulses of 0.367 ha and 0.0220 ha , respectively . the system is initially given a charge of @xmath87 using a laser of the same charge . application of second laser with charge @xmath89 transfers the system to a charge of @xmath90 ( lower - left panel ) . the same laser , but with a charge of @xmath56 , does not cause the state to evolve ( lower - right panel ) . envelop parameters are @xmath91 and @xmath92 in atomic units , respectively . , scaledwidth=90.0% ] a second case shows how etc can be either increased or decreased in response to the application of pulses with identical ptc . the system is first given a charge of @xmath64 and , in both cases , a second laser pulse with a ptc of @xmath56 is then applied . in the evolution shown at lower - right , the energy of the second pulse is equal to the difference between the states @xmath64 and @xmath93 , 0.0158 ha ( table [ casidaexs ] ) . this results in the absorption of a photon because the @xmath93 state is of higher energy with the charge balance equation of @xmath94 . on the other hand , tuning the second pulse to an energy of 0.0257 ha causes the system to transition to the @xmath55 state because this is the energy difference between the @xmath64 and @xmath55 states ( table [ casidaexs ] ) . in this case a photon is emitted because the @xmath55 state is of lower energy , and the charge balance is @xmath95 . the oscillation in the population of the @xmath64 state ( blue curve ) is an artifact associated with the single - determinant approximation in concert with our piecewise homogeneous construction of the incident beam . this field approximation results in a larger contribution of the non - primary determinant , and the artificial oscillation in population is an indicator that there is the relative weighting of these determinants is time dependent . in contrast , the analogous curve associated with circularly polarized light ( figure [ stoc ] ) shows no such oscillation because the non - primary determinants make almost no contribution . ( left ) and @xmath96 ( right ) . upper panels shows two laser sequences . in both processes , the parameters for first laser are @xmath97 . the parameters for second lasers are ( left ) : @xmath98 and ( right ) : @xmath99 in atomic units . , scaledwidth=90.0% ] a third scenario , shown in figure [ etc0tom1orp1 ] , demonstrates how photonic charges of opposite sign can be emitted from the same initial etc state . the system is placed its highest energy state , @xmath93 , after application of an appropriate laser pulse . a second laser with ptc of @xmath56 stimulates photon emission and changes the system to an etc of @xmath90 . the associated statement of charge conservation is @xmath100 . on the other hand , illuminating the system with a ptc of @xmath56 also stimulates photon emission but changes the system to an etc of @xmath101 . the associated statement of charge conservation is @xmath102 . ( left ) and @xmath103 ( right ) . upper panels shows two laser sequences . lower panels show that the system is initially given a @xmath93 charge which is subsequently changed to @xmath90 ( left ) by emission of @xmath89 or to @xmath101 by emission of @xmath56 . the parameters for first laser are @xmath104 and those for the second laser are @xmath105 in atomic units.,scaledwidth=90.0% ] all the cases considered so far have focused on excitonic charge arithmetic . when purposed as a photonic charge converter , though , it is important to be able to transfer the final charge into a photonic state . such an operation , already considered in within the tb setting , is shown to hold in for td - dft as well in figure [ etcemission ] . the system is first given a @xmath87 charge which is then changed to @xmath90 using a second laser . taken together , these first two steps can be viewed as ptc addition : @xmath106 . a third laser then stimulates the emission of photonic charge of @xmath56 leaving the system in its ground state . the overall photonic arithmetic is therefore @xmath107 . . upper panels shows two laser sequences . lower panel shows the system is excited into @xmath87 state by the first laser . subsequent absorption of a @xmath108 photon then transfers the system into @xmath90 state . this etc is converted to ptc with a third laser which brings system back to its ground state after the photon is emitted . the laser parameters are : ( first ) @xmath109 ; ( second ) @xmath110 ; and ( third)@xmath111 in atomic units . , scaledwidth=70.0% ] the excited state excitonic populations are small because very weak and short lasers are applied in all td - dft simulations this is a pragmatic step taken to avoid the population of extraneous eigenstates that result from stronger or longer laser pulses . this is purely a computational issue that results because the vector vortex beams were approximated with five piecewise - homogeneous components a computational work - around to the limitations of the input fields allowed by the octopus td - dft software . the result is unphysical light - matter interactions in the regions between each arm that can be remedied by dividing the region into more than five piecewise - homogeneous components . then stronger and/or longer laser illumination can be applied to increase the population of twisted excitons . as an alternative to applying twisted light , photonic topological charge can be inputted in the form of spin angular momentum i.e . using circularly polarized light . twisted excitons can also be used to convert photons with spin angular momentum to those with angular orbital to momentum . this just amounts to a change of basis for describing the molecular dipoles since circularly polarized light can be mathematically decomposed into a combination of radial and azimuthal vector vortices@xcite : @xmath112 here @xmath113 and @xmath114 are the basis vectors in cartesian and polar representations . in the 5-arm @xmath77 system , only the radial vortex components are absorbed . subsequent photon emission would be in the form of a vector vortex , though , so the molecular assembly serves to convert light from spin to orbital angular momentum ( stoc ) . since multiple photons can be absorbed , pulses of circularly polarized light can be used to generate vector vortices with an arbitrary topological charge . the stoc concept is demonstrated in figure [ stoc ] , where a sequence of @xmath115 and @xmath116 spin laser pulses are applied . the photonic subscript @xmath117 highlights the spin nature of light while @xmath118 subscripts indicate that the photonic charge is in the form of a vector vortex . the first laser pulse causes photon absorption and the following charge conservation relation : @xmath119 . the second laser , of opposite spin , results in photon emission : @xmath120 . . upper panels shows two laser sequences . lower panel shows the system is sequentially provided with two topological charges resulting in @xmath121 . subsequent emission would produce a @xmath122 photon . the laser parameters are : ( first ) @xmath123 ; and ( second ) @xmath124 in atomic units.,scaledwidth=70.0% ] figure [ stoc ] shows that the strength of first laser is the same as those used for the vortex beams of figure [ etcm2tom1 ] , [ etc1to2or0 ] , and [ etc0tom1orp1 ] , but the duration is three times longer . this allows a much larger population of proper excited states without involvement of extraneous eigenstates . as compared with figure [ etc1to2or0 ] , we now have a very smooth population of the @xmath64 state ( blue curve ) . this confirms that the origin of the oscillations in our former td - dft simulations is actually the approximation of vector vortex with five piecewise - homogeneous components . beams of light can transfer their angular momentum quanta to the quasi - angular momentum associated with the electronic structure of molecules with @xmath5 or @xmath6 point group symmetry . strictly speaking , this is a transformation of real angular momentum into a molecular analog to the crystal angular momentum of solid - state physics . topological charge is then a convenient bookkeeping scheme since it is conserved in such light - matter interactions . in concert with conservation of energy , it is possible to design processes in which sequential laser pulses are used to increase or decrease the excitonic topological charge . subsequent emission results in photons with a different orbital angular momentum than the input beams . these molecules can be used to convert either spin polarized or vortex beams into vector vortex beams with a range of angular momentum values . unlike existing approaches , this molecular strategy offers a means of manipulating the angular momentum of light which does not rely on the nonlinear optical properties of a mediating crystal . topological charge conservation can be elucidated using simple tight - binding hamiltonians , a setting in which it is straightforward to demonstrate charge algebra . however , time - domain dft gives consistent results within a many - electron setting with the effects of electron correlation and exchange accounted for . the level of tight - binding analysis employed does not require any details of the molecular structure beyond point group , and the time - domain dft analyses adopted , as a computational expedient , molecular arms composed of hydrogen dimers . there exist a panoply of molecules which exhibit @xmath5 or @xmath6 symmetry , though , and several examples are shown in figure [ molecules ] . or @xmath6 symmetry . _ , scaledwidth=90.0% ] these may radiate outward as chiral spoke , such as triphenylphosphene or hexaphenylbenzene , be in a planar , non - spoke arrangement such as cyclohexane , or even be in a three - dimensional band such as cycloparaphenylene @xcite . another intriguing possibility is to functionalize inert scaffolds that have the requisite point symmetry . in the present work , stimulated emission causes photons to be generated that have the polarization of the stimulating beam . however , it is clear that charge conversions may also culminate in spontaneous emission . the resulting photons will then exhibit a vector vortex in the far field with radial and azimuthal components that depend upon the orientation of the electric dipoles of each arm . real - time simulations are made possible through the runge - gross ( rg ) reformulation of time - dependent schr " odinger equation @xcite : @xmath125(\vec r , t ) + \nu_{xc}[\rho](\vec r , t)\big]\psi_i(\vec r , t ) . \label{rg}\ ] ] here the spin - reduced electronic density , @xmath126 , is expressed in terms of the time - dependent kohn - sham ( tdks ) orbitals , @xmath127 , as @xmath128 these orbitals , in turn , can be represented in the basis of their counterparts at time zero , @xmath129 , so that the time - propagated multi - electron wavefunction is constructed from a linear combination of determinants built from these initial orbitals@xcite : @xmath130 ket @xmath131 is the ground state and @xmath132 is a determinant with the @xmath83 occupied kohn - sham ( ks ) orbital replaced by the @xmath84 unoccupied orbital . the first summation is over all occupied ks orbitals , five occupied ks orbitals in the case of 5-arm @xmath77 system , and the second summation is over all unoccupied ks orbitals . in our case , because the frequency of laser is chosen to only access the first five lowest excited states , the only unoccupied ks orbital is the sixth as shown in table [ casidaexs ] . if it was possible to express equation [ tpstate ] in form of equation [ tbstates ] , the associated excitonic charge could be determined directly . such a simple expansion of @xmath133 in basis of @xmath134 does not exist , though , since it is a many - body wavefunction . this is remedied , albeit in an approximate way , by working only with the dominant determinant for which only lowest unoccupied molecular orbital ( lumo ) is involved in equation [ tpstate ] . this makes it possible to combine the determinants corresponding to each etc subspace @xmath135 with @xmath136 , as in equation [ rearranged ] , allowing the etc of @xmath133 to be obtained via ks orbitals using the method detailed below . in all simulations , twisted excitons are a constructed as a linear combination of corresponding pairs of degenerate excited states , consistent with the tb model . focusing on the 5-arm system , if a laser with @xmath137 is applied , then the resulting excited state can be approximated with a determinant involving only @xmath138 . likewise , the application of @xmath139 results in excited states that can be approximated with determinants involving only @xmath140 and @xmath141 , and @xmath142 yields states that are well - approximated with only @xmath143 and @xmath144 . the time - propagated wavefunction for each etc subspaces @xmath145 , @xmath146 and @xmath147 can therefore be expressed , respectively , as : @xmath148 system . _ the five occupied ks orbitals have been expressed in the basis of arm wavefunctions , @xmath134 , with their coefficients labeled on the corresponding arms . the red ( blue ) isosurfaces indicate positive ( negative ) values of the wavefunctions . the lumo is given in order to show that it is symmetric . homo = highest occupied molecular orbital . , scaledwidth=80.0% ] the only difference among these three equations is that the ground state determinant is modified as follows : @xmath149 replaced by @xmath150 ; @xmath151 and @xmath152 are replaced by @xmath153 ; and @xmath154 , @xmath155 and @xmath156 are replaced by @xmath157 and @xmath154 . the ground state determinant @xmath158 is the @xmath93 state of course . these replacement orbitals must therefore be responsible for the etc of excited states . figure [ ksorbitals ] gives the isosurface and decomposition in the basis of @xmath159 with @xmath160 of all the relevant ks orbitals . as shown in figure [ ksorbitals ] , the lumo is symmetrically distributed across all five arms . therefore @xmath154 will not introduce a phase difference among arms in the right side of equation [ rearranged ] . this implies that @xmath153 and @xmath157 will introduce a phase dependence corresponding to @xmath161 and @xmath162 , respectively . the population of each twisted exciton state is therefore given by : @xmath163 here @xmath164 is the eigenstate associated with an etc of @xmath18 , from equation [ tbstates ] , and the factor of two in each expression accounts for the fact that the electron spin can be either up or down . we are grateful to profs . david l. andrews , guillermo f. quinteiro and mark e. siemens for extended discussions on the generation of twisted light . all calculations were carried out using the high performance computing resources provided by the golden energy computing organization at the colorado school of mines . r. baltz and c. f. klingshirn , in _ ultrafast dynamics of quantum systems : physical processes and spectroscopic techniques _ , edited by b. di bartolo and g. gambarota ( springer us , boston , ma , 2002 ) , pp . 381396 .	the molecular absorption of photons with angular momentum can result in _ twisted excitons _ with a well - defined quasi - angular momentum . although they represent different physical properties , photonic and excitonic quanta can both be described in terms of topological charge , a conserved quantity . multiple absorption events can be used to create a wide range of excitonic topological charges . subsequent emission produces photons that exhibit this same range . the molecule can thus be viewed as a mediator for changing the orbital angular momentum of light . this sidesteps the need to exploit nonlinear light - matter interactions based on higher - order susceptibilities . a tight - binding paradigm is used to establish topological charge conservation and demonstrate how it can be exploited to combine , subtract , and change the nature of the angular momentum of light . the approach is then extended to a time - dependent density functional theory setting where the key results are shown to hold in a many - body , multi - level setting .
our understanding of distant galaxies and the history of galaxy formation has undergone a revolution in the past decade . galaxies are now routinely discovered and studied out to redshifts @xmath8 ( e.g. , dickinson et al . 2004 ; yan et al . 2005 ; bouwens et al . 2007 ; bouwens et al . 2010 ) . samples of a few dozen objects have been found at even higher redshift , back to the era of reionization ( @xmath9 ) , and perhaps some galaxies have been discovered at even higher redshifts , @xmath10 ( e.g. , bouwens et al . 2010 ; finkelstein et al . this relatively rapid advance in our discovery of the earliest galaxies is the direct result of technical advances in spectroscopy and imaging over the past decade , in which the _ hubble space telescope _ ( _ hst _ ) has played a leading role . historically , distant galaxies are found within deep optical imaging surveys , and are confirmed as high redshift galaxies with large multi - object spectrographs on 8 - 10 meter telescopes , which came online in the mid-1990s . it can be argued however that some of the most important advances in our understanding of galaxies has come about from very deep imaging , especially from _ hst_. the _ hubble _ space telescope has played a key role in high - redshift discoveries and our understanding of galaxy evolution through large blank field and targeted programs such as the _ hubble _ deep and ultra deep fields , goods , egs , and cosmos , among others ( e.g. , williams et al . 1996 ; giavalisco et al . 2004 ; beckwith et al . 2006 ; davis et al . 2007 ; scoville et al . 2007 ) . this _ hubble _ imaging has proven invaluable for two primary reasons . one is simply due to the depth that can be achieved with a high photometric fidelity , ensuring that exquisite photometry of distant galaxies can be obtained . whilst ground based telescopes can reach the depths of _ hst _ at optical wavelengths , in principle the accuracy and precision of this photometry is not nearly as good , due to a higher background , and importantly , the large and variable psf . this makes accurate measurements of light difficult , particularly for colours which require exact apertures for accurate measures . furthermore , _ hst _ data have proven important for the discovery of the most distant galaxies in the universe through the use of the lyman - break method of looking for drop - out galaxies in bluer bands . many filter choices within multi - colour deep imaging programmes were in fact selected to facilitate optimal drop - out searches ( e.g. , giavalisco et al . 2004 ) . _ hubble _ imaging furthermore has facilitated a renaissance in the study of galaxy structure in the distant universe , which provides a key observable for understanding how distant galaxies form and evolve ( e.g. , conselice et al . 2003 ; conselice et al . 2008 , 2009 ; lotz et al . 2008 ; buitrago et al . 2008 ; jogee et al . 2009 ; bluck et al . 2009 ; casatta et al . these structural measurements have proven critical for determining how galaxy morphologies , sizes , and merger / kinetic states have evolved through time ( e.g. , trujillo et al . 2007 ; ravindranath et al . 2004 ; conselice et al . this allows us to examine how the merger history of galaxies has changed ( e.g. , conselice et al . 2003 , 2008 ; lotz et al . 2008 ; jogee et al . 2009 ) , and thus we can begin to derive how galaxies form , as opposed to simply when . it is not currently straightforward to measure the structures of distant galaxies with ground based imaging even with adaptive optics , and thus _ hubble _ has and continues to provide a key aspect for tracing evolution using these methods . however , one key aspect of parameter space that has not yet been explored with _ hst _ , or other space - based telescopes in any depth over large areas , is deep infrared imaging over a relatively large area . previously there exists deep nic3 imaging over the _ hubble _ deep field ( dickinson et al . 2000 ) and hubble ultra - deep field ( thompson et al . 2004 ) , as well as very deep nicmos imaging over a small area of the hdf - n ( thompson et al . 1999 ) . these areas are however very small , and while nic3 parallel data exists over the cosmos and egs fields , it is quite shallow at @xmath11 orbit depth . _ hst _ imaging data however has a distinct advantage over ground - based imaging not only in terms of the higher quality photometric fidelity and higher resolution , but also the depth which can be achieved in the near infrared ( nir ) with _ hst _ as opposed to the ground - based optical where comparable depths to _ hst _ can be reached . within one or two orbits , the _ hst _ can reach a depth in the nir which is difficult to obtain from the ground even with an 8 - 10-m class telescope , and which will not have the same photometric quality , nor resolution as the _ hubble _ data . we thus designed and carried out the goods nicmos survey ( gns ) , which is a large _ hst _ programme intended to remedy this situation by providing through an initial 180-orbit program of nic3 imaging in the goods fields , a data set designed to examine a host of problems requiring very deep nir data . the gns data consist of 60 nicmos nic3 pointings , centred on the most massive ( @xmath12@xmath13 ) galaxies at @xmath14 . the depth of each image is 3 orbits / pointing within the @xmath15-bandpass over a total area of @xmath16 arcmin@xmath17 ( buitrago et al . 2008 ; bluck et al . 2009 , casey et al . 2009 , bouwens et al . 2009 , 2010 present results using this data ) . with these nicmos data we are able to explore the rest - frame optical features of galaxies at @xmath18 in detail . this allows a few measurements to be made that can not be easily reproduced with optical imagining and/or deep nir imaging from the ground . this includes : filling in the important near - infrared gap in galaxy spectral energy distributions ( seds ) ; sampling the rest - frame optical structures and sizes of galaxies out to @xmath19 ( buitrago et al . 2008 ) ; and the detection and characterization of the population of massive @xmath20 galaxies and agn , and determining the relation of agn evolution to that of massive galaxies ( e.g. , bluck et al . 2010 ) . in this paper we present the basic outline , background , and results from this survey . we discuss the design of the observations , our field selection , as well as the selection for our initial massive galaxy sample which has guided the centres for our nicmos nic3 pointings . we also discuss the various methods for locating the massive galaxy population at higher redshifts , and the connection of these massive galaxies to those at @xmath4 . we show that no one colour method is able to identify the massive galaxy population at high redshifts , and that a combination of methods and photometric redshifts are needed to construct a semi - complete massive galaxy sample at higher redshifts . in this paper we construct as complete as possible sample of massive galaxies within our fields , and discuss the properties of these galaxies , as well as some features of lower mass galaxies . this paper is organised as follows : 2 gives a summary of our observations and the design of the gns , including how the initial sample of galaxies was selected . 3 gives a description of the derived parameters from the @xmath21band imaging , including photometric redshifts and catalogue matching . 4 describes our initial analysis of the survey data , including how the various selections for massive galaxies at high redshifts compare , while finally 5 is our summary . we use a standard cosmology of _ km s@xmath23 mpc@xmath23 , and @xmath24 = 0.3 throughout . the gns selection and field coverage is based on the previous optical acs and ground - based imaging from the original goods program ( giavalisco et al . the goods programme is a multi - wavelength campaign to obtain a coherent collection of deep imaging and spectroscopy in two 150 arcmin@xmath17 areas in the northern and southern hemispheres ( goods - n and goods - s ) . these two fields are centred around the _ hubble _ deep field - north ( hdf - n ) and _ chandra _ deep field - south ( cdf - s ) , which are areas of very low dust extinction , and minimal stellar and radio contamination . the existing goods / acs fields match the coverage of the goods _ spitzer _ program and cover the 2 msec exposure of the _ chandra _ deep field south and the 2 msec exposure of the _ chandra _ deep field north ( luo et al . large ongoing campaigns to obtain spectroscopy for the goods fields have also been carried out , including 3000 spectra as part of the keck treasury redshift survey ( wirth et al . another @xmath25 redshifts in goods - s have been measured from various eso programs ( e.g. , vanzella et al . 2008 ; le fevre et al . 2005 ; popesso et al . 2009 ; balestra et al . 2010 ) . the co - moving volume probed by goods at high redshifts , @xmath26 , is similar to the co - moving volume covered by the cosmos field ( e.g. , scoville et al . 2007 ) at @xmath27 . furthermore , due to its depth at all wavelengths the goods fields are thus an ideal location for examining the formation and evolution of early galaxies . deep nir imaging of these fields however is lacking , although some deep nir imaging has been obtained with eso telescopes using sofi and isaac for the goods - south , as well as deep cfht wircam imaging , subaru imaging , and some keck imaging over the goods north ( e.g. , kajisawa et al . 2009 ; retzlaff et al . 2010 ; wang et al . however , these data only reach modest depths of @xmath28 compared to our _ hst _ imaging . the depth of our suvey is only comparable to previous nicmos deep programmes covering the hdf - north and hdf - south fields , as well as new near - ir data obtained with wfc3 ( cassata et al . 2010 ) . ideally , one would want to cover both goods north fields completely , yet given the small nicmos field of view it is not practical to cover the entire goods fields any deeper than one orbit with nicmos . the wfc3 camera will , however , soon cover these fields to an even great depth with the candels programme . our strategy is not to map out a continuous area , but to collect 60 pointed observations directed towards the most massive galaxies at @xmath29 found in the goods fields ( 2.2 ) , maximised to obtain the largest number of galaxies based on our selection methods . to obtain the most unique and useful science we therefore constructed a program which covers a sixth of the area of a single goods field ( in total 43.7 arcmin@xmath17 ) in three orbits depth in the @xmath15 band , in areas of the deepest _ spitzer _ , _ chandra _ , and acs imaging , and where the greatest amount of spectroscopy already exists . some of these fields were then observed in the @xmath30-band ( @xmath31 ) with nicmos or wfc3 as part of a follow up programme to obtain near infrared seds to look for high redshift drop - out galaxies ( bouwens et al . 2010 ) . our nicmos pointings were chosen to target a set of objects selected to be known massive galaxies at high redshift , identified using a variety of color selection methods . these include `` distant red galaxies '' ( drg : franx et al . 2003 ; papovich et al . 2006 ) , irac extremely red objects ( yan et al . 2006 ) , and bzk color - selected galaxies ( daddi et al . 2004 , 2007 ) . all of these methods are designed to find red , dusty or passively evolving older galaxies at @xmath32 . in practice , we utilised all three of these colour - selections separately , in order to obtain as much as possible a complete sample of massive galaxies at @xmath33 . to optimise our field placement , we also used catalogues of lyman - break selected bm / bx objects ( reddy et al . 2008 ) , as well as high redshift drop - outs and sub - mm galaxies . however , the primary field selection was done in terms of the massive galaxy selection through the three primary colour criteria as described further below . colour selection of distant galaxies has a long history dating back to the early work of finding lyman - break galaxies through image drop - outs in blue bands ( e.g. , guhathakurta , tyson , majewski 1990 ; steidel & hamilton 1992 ) . it is generally accepted that no single method can find all galaxies at a given redshift , and some of these methods are better at finding star - forming objects , as opposed to those which are more passive and evolved . in fact , it is generally agreed that no method or combination of methods can identify an obviously complete sample of high - z galaxies . one of the methods we use for finding likely passively evolving and dusty red galaxies is to find distant red galaxies ( drgs ) defined by a nir colour cut ( e.g. , franx et al . 2003 ; papovich et al . 2006 ; conselice et al . the selection we use to find drgs , and to be included within our sample , is galaxies at @xmath29 with ( @xmath34 ) @xmath35 mag in vega magnitudes ( or @xmath36 in ab mags ) . the selection for these galaxies is based on ground - based imaging from isaac on the very large telescope ( vlt ) . this selection is only used for choosing systems in goods - s , as deep nir imaging over the entire goods - n field was not available when the target selection was carried out . this goods - s drg sample is approximately complete for @xmath0 drg selected galaxies at @xmath37 ( papovich et al . 2006 ) . another selection we use to construct our initial massive sample is the _ spitzer _ selected extremely red objects ( eros ) , otherwise known as infrared eros ( ieros ) . these were first described in yan et al . ( 2004 ) , based on nir and _ spitzer _ data within the goods fields . the selection for these objects is s@xmath38(3.6 @xmath39m)/s@xmath40 . these objects were found by yan et al . ( 2004 ) , based on sed fits , to have a mixture of old and younger populations . note that selecting galaxies in this way ensures that they are massive given their brightness in the ir . however , because they are selected with _ spitzer _ imaging , which has a large psf , resulting in potential confusion from neighboring objects , this selection can have issues with contamination from other galaxies . hence any galaxies which would satisfy the criteria but are too close to another galaxy will not be included simply due to the problem of confusion . another method we use to select distant galaxies is through the bzk approach , which is described in daddi et al . ( 2007 ) in terms of selection within the goods fields . the selection for these objects is slightly more complicated than that of the drgs or ieros , since they are selected through colours using the @xmath41 , @xmath42 and @xmath43-bands together . this method proposes to separate evolved galaxies or passive pbzks , and those which are star forming , or sbzks . the selection for these galaxies is done through the quantity @xmath44 , defined using these three bands by : star forming galaxies at @xmath46 are proposed to have @xmath47 . the redder , possibly more evolved galaxies , are found through the selection @xmath48 and @xmath49 . for the bzk sample we use , the selection is somewhat more limited than for the other colour selections as these sources were selected down to @xmath50 vega in the north and @xmath51 in the south . we utilise photometric redshifts and stellar masses of the galaxies selected through these methods to identify and study these colour selected populations taken directly from papovich et al . ( 2006 ) , yan et al . ( 2004 ) and daddi et al . ( 2007 ) . our initial massive galaxy sample from which we optimise our nicmos pointings are selected through these three methods , with a further photometric redshift cut of @xmath14 , and with a stellar mass cut of @xmath0 . in practice our final pointings were chosen by finding the locations within the goods fields where the number of these massive galaxies was maximised within the nic3 fields . in total we imaged 45 pre - selected massive @xmath0 galaxies at @xmath14 in the goods - n , and 35 in the goods - s . galaxies selected in other ways were also used to optimise the number of galaxies in each nic3 pointing , although each pointing was designed to have at least one massive galaxy with the properties above . these ` additional ' galaxies are selected through the lyman - break drop - out method utilising @xmath41 , @xmath52 and @xmath53 drop - outs , the bx / bm selection , as well as sub - mm galaxies from greve et al . each nic3 pointing contained between four to 19 of each of these galaxy types . figure 1 shows our field layout within the goods fields with the different galaxy types shown as different colours and symbols , and figure 2 shows a typical nicmos nic3 pointing of one of our fields . tables 1 and 2 list the statistics and positions of our 60 pointings , with 30 nic3 pointings in the goods - north field , and 30 in goods - south . we also list the number of various other types of galaxies within each of these fields . tables 3 and 4 list the initial massive galaxies for which we picked our fields , along with basic information such as their photometric redshifts , stellar masses , and information on the optical and @xmath21band magnitudes for these systems . nicmos and acs images of ten of these massive galaxies are shown in figure 3 . this data , including catalogs of sources , redshifts and stellar masses , as well as the original reduced nic3 imaging itself is available at * http://www.nottingham.ac.uk / astronomy / gns/*. + notes . ( a ) the values of the errors on @xmath54 and sersic @xmath55 are representative of the 1 @xmath56 model errors from galfit ( see buitrago et al . this does not take into account many possible sources of error that may bias these measurements , including magnitude of galaxy , concentration of its light profile , etc . the uncertainty in these structural parameters increase by on order of 10 percent for @xmath54 and 20 percent for @xmath55 due to changes in the psf across the nicmos nic3 field of view . also listed is the fitted axis ratios for these galaxies ( b / a ) , and position angles ( p.a . ) . the effective surface brightness ( @xmath57 ) is listed , as is the spectroscopic redshift ( @xmath58 ) , if available . the value of ` sep ' is the difference between the position of an object and the identification of the spectroscopic target , in arcsec . finally , the calculated photometric redshift , @xmath59 , is shown . balestra , i. et al . 2010 , a&a , 512 , 12 barger , a. et al . , 2008 , apj 689 , 687 beckwith , s. , et al . 2006 , aj , 132 , 1729 benitez , n. 2000 , apj 536 , 571 benitez , n. 2000 , apj , 536 , 571 bertone , s. , conselice , c.j . 2009 , mnras , 396 , 2345 bluck , a.f.l . , conselice , c.j . , bouwens , r.j . , daddi , e. , dickinson , m. , papovich , c. , yan , h. 2009 , mnras , 394 , 51l bluck , a.f.l . , conselice , c.j . , almaini , o. , laird , e.s . , nandra , k. , gruetzbauch , r. mnras in press , arxiv:1008.2162 bohlin , r.c . , mack , j. , hartig , g. , sirianni , m. 2005 , instrument science report acs , 2005 - 12 bolzonella , m. et al . 2000 , a&a 363 , 476 bouwens , r.j . , illingworth , g.d . , franx , m. , ford , h. 2007 , apj , 670 , 928 bouwens , r.j . , 2010 , arxiv:1003.1706 bouwens , r.j . , 2009 , apj , 705 , 936 bruzual , g. , charlot , s. 2003 , mnras , 344 , 1000 buitrago , f. , trujillo , i. , conselice , c.j . , bouwens , r.j . , dickinson , m. , yan , h. 2008 , apj , 687 , 61l bundy , k. , et al . 2006 , apj , 651 , 120 bundy , k. , et al . 2008 , apj , 681 , 931 calzetti et al . 2000 , apj 533 , 682 casey , c.m . , chapman , s.c . , muxlow , t.w.b . , beswick , r.j . , alexander , d.m . , conselice , c.j . 2009 , mnras , 395 , 1249 cassata , p. , et al . 2010 , apj , 714 , 79 conselice , c.j . , bershady , m.a . , dickinson , m. , papovich , c. 2003 , aj , 126 , 1183 conselice , c.j . bundy , k. , ellis , r.s . , brichmann , j. , vogt , n.p . , philips , a.c . 2005 , apj , 628 , 160 conselice , c.j . , et al . 2007a , mnras , 381 , 962 conselice , c.j . , 2007b , apj , 660 , 55l conselice , c.j . , rajgor , s. , myers , r. 2008 , mnras , 386 , 909 conselice , c.j . , bundy , k. , u , v. , eisenhardt , p. , lotz , j. , newman , j. 2008 , mnras , 383 , 1366 conselice , c.j . , yang , c. , bluck , a. , 2009 , mnras , 394 , 1956 daddi , e. , et al . 2004 , apj , 617 , 746 daddi , e. , et al . 2007 , apj , 670 , 156 davis , m. , et al . 2007 , apj , 660 , 1l de jong , r.s . 2006 , instrument science report nicmos 2006 - 003 dickinson , m. , et al . 2000 , apj , 531 , 624 dickinson , m. , et al . 2004 , 600 , 99l faber , s.m . , et al . 2007 , apj , 665 , 265 finkelstein , s.l . , papovich , c. , giavalisco , m. , reddy , n.a . , ferguson , h.c . , koekemoer , a.m. , dickinson , m. 2010 , apj , 719 , 1250 foucaud , s. , conselice , c.j . , hartley , w.g . lane , k.p . , bamford , s.p . , almaini , o. , bundy , k. 2010 , mnras , 406 , 147 franx , m. , et al . 2003 , apj , 587l , 79 giavalisco , m. , et al . 2004 , apj , 600 , 93l grazian , a. et al . 2007 , a&a , 465 , 393 greve , t.r . , pope , a. , scott , d. , ivison , r.j . , borys , c. , conselice , c.j . , bertoldi , f. 2008 , mnras , 389 , 1489 guhathakurta , p. , tyson , j.a . , majewski , s.r . 1990 , apj , 357 , 9l jogee , s. , et al . 2009 , apj , 697 , 1971 kajisawa , m. et al . 2009 , apj , 702 , 1393 kriek , m. et al . 2008 , apj , 677 , 218 lane , k. , et al . 2007 , mnras , 379 , 25l le fevre , o. , et al . 2005 , a&a , 439 , 845 lotz , j.m . , 2008 , apj , 672 , 177 luo , g. et al . 2008 , apjs , 179 , 19 magee , d. , bouwens , r. , illingworth , g. 2007 , aspc , 376 , 261 maraston , c. 2005 , mnras , 362 , 799 metcalfe , n. shanks , t. , weilbacher , p.m. , mccracken , h.j . , fong , r. , thompson , d. 2006 , mnras , 370 , 1257 papovich , c. , et al . 2006 , apj , 640 , 92 papovich , c. , finkelstein , s.l . , ferguson , h.c . , lotz , j.m . , giavalisco , m. 2010 , arxiv:1007.4554 popesso , p. et al . 2009 , a&a , 494 , 443 ravindranath , s. , et al . 2006 , apj , 652 , 963 reddy , n. , et al . 2005 , apj , 633 , 748 reddy , n. , steidel , c.c . , pettini , m. , adelberger , k.l . , shapely , a.e . , erb , d.k . , dickinson , m. 2008 , apjs , 175 , 48 retzlaff , j. et al . 2010 , a&a , 511 , 50 scoville , n. et al . 2007 , apjs , 172 , 1 thompson , r. et al . 1999 , apj , 523 , 100 thompson , r. et al . 2005 , aj , 130 , 1 trujillo , i. , conselice , c.j . , bundy , k. , cooper , m.c . , eisenhardt , p. , ellis , r.s . 2007 , mnras , 382 , 109 steidel , c.c . , & hamilton , d. 1992 , aj , 104 , 941 van dokkum , p.g . , 2009 , pasp , 121 , 2 vanzella , f. , et al . 2008 , a&a , 478 , 83 wang , w .- h . 2010 , apjs , 187 , 251 williams , r. , et al . 1996 , aj , 112 , 1335 wirth , g. , et al . 2004 , aj , 127 , 3121 wuyts , s. et al . 2008 , apj 689 , 653 yan , h. , et al . 2004 , apj , 616 , 63 yan , h. , et al . 2005 , apj , 634 , 109	we present the details and early results from a deep near - infrared survey utilising the nicmos instrument on the _ hubble space telescope _ centred around massive @xmath0 galaxies at @xmath1 found within the great observatories origins deep survey ( goods ) fields north and south . the goods nicmos survey ( gns ) was designed to obtain deep f160w ( h - band ) imaging of 80 of these massive galaxies , as well as other colour selected objects such as lyman - break drop - outs , bzk objects , distant red galaxies ( drgs ) , extremely red objects ( eros ) , _ spitzer _ selected eros , bx / bm galaxies , as well as flux selected sub - mm galaxies . we present in this paper details of the observations , our sample selection , as well as a description of features of the massive galaxies found within our survey fields . this includes : photometric redshifts , rest - frame colours , and stellar masses . we furthermore provide an analysis of the selection methods for finding massive galaxies at high redshifts , including colour selection methods and how galaxy populations selected through these colour methods overlap . we find that a single colour selection method can not locate all of the massive galaxies , with no one method finding more than 70 percent we however find that the combination of these colour methods finds nearly all the massive galaxies , as selected by photometric redshifts with the exception of apparently rare blue massive galaxies . by investigating the rest - frame @xmath2 vs. m@xmath3 diagram for these galaxies we furthermore show that there exists a bimodality in colour - magnitude space at @xmath4 , driven by stellar mass , such that the most massive galaxies are systematically red up to @xmath5 , while lower mass galaxies tend to be blue . we also discuss the number densities for galaxies with stellar masses @xmath0 , whereby we find an increase of a factor of eight between @xmath6 and @xmath7 , demonstrating that this is an epoch when massive galaxies establish most of their mass . we also provide an overview of the evolutionary properties of these galaxies , such as their merger histories , and size evolution . [ firstpage ] galaxies : evolution , formation , structure , morphology , classification
assuming general relativity to be valid on all scales , data ranging from the cosmic microwave background to individual galaxies point towards a universe dominated by dark energy and dark matter , the nature of these being most certainly amongst the deepest problems of modern physics . while dark energy is well - represented by a cosmological constant @xmath0 in einstein s field equations , the currently preferred dark matter candidate is a collection of stable , neutral , elementary particles that condensed from the thermal bath of the early universe , and which are known as ` cold dark matter ' ( cdm ) particles ( see , e.g. , frenk & white 2012 , strigari 2012 for recent reviews ) . on galaxy scales , however , predictions of this standard @xmath0cdm cosmological model , although plagued by the enormous complications of baryonic astrophysics , are difficult to reconcile with observations . we hereafter review these challenges for the @xmath0cdm model , and point out that some of them hint at a richer and more complex physics of the dark sector than currently assumed . in particular , many observed scaling relations involve the ubiquitous appearance of an acceleration constant @xmath1 , whose origin is a deep mystery in the standard context . surprisingly , most of these scaling relations can be summarized by the empirical formula of milgrom ( 1983 ) . the success of this formula means that the gravitational field in galaxies mimicks , for whatever reason , an effectively modified force law on galaxy scales , known as modified newtonian dynamics ( mond ) . this formula however fails to account for dynamics and lensing of galaxy clusters ( e.g. , clowe et al . 2006 , angus et al . 2007 ) , meaning that if the formula would be reflecting a true modification of gravity as an alternative to galactic dark matter , it should rely at cluster scales on residual missing mass , which could be in baryonic or non - baryonic form ( milgrom 2008 , angus et al . 2007 ) , or the formula should be extended ( zhao & famaey 2012 ) . it is also unclear how the angular power spectrum of the cosmic microwave background ( cmb ) could be accounted for without resorting to some form of non - baryonic dark matter ( e.g. , slosar et al . 2005 , angus 2009 ) . nevertheless , the main motivation for studying alternatives to @xmath0cdm based on the mond phenomenology is _ not _ necessarily to get rid of any form of ` dark matter ' , but rather to explain why the observed gravitational field in galaxies is apparently mimicking a universal force law generated by the baryons alone . the simplest explanation is of course _ a priori _ not that dark matter particles arrange themselves ( by a hitherto unknown physical mechanism ) in order to mimick a fake force law , but rather that the force law itself is modified . however , at a fundamental level , covariant theories of modified gravity often have to include new fields in the dark sector to reproduce this effective force law ( fields with an energy density nevertheless subdominant to the baryonic one , and a role completely different from that of cdm : they would mediate the force rather than sourcing it ) , or even introduce what could be called a ` dark matter medium ' ( with an energy density outweighing the baryonic one ) exhibiting a kind of new fundamental interaction with baryons : this makes the confrontation between mond and dark matter much less clear than often believed , since the former implies a more complex structure of the dark sector than the currently assumed cdm , but does not necessarily imply the absence of a dark sector . in mond , the new fields responsible for the effects usually attributed to cdm would in fact be somewhat closer to dark energy fields than to cdm . it would of course be even more exciting if one would manage to find a physical connection between these putative new fields and the actual dark energy sector . we herefater list a ( non - exhaustive ) series of problems for @xmath0cdm ( sect . 2 ) , then show how theories based on the mond phenomenology might address a lot of them ( sect . 3 ) , before listing a ( non - exhaustive ) series of covariant theories currently proposed in this vein ( sect . 4 ) , together with their own questionmarks and internal problems . the reading of this short proceeding can be completed by the reading of the exhaustive review paper recently published in living reviews in relativity ( famaey & mcgaugh 2012 ) . * * the missing satellites challenge . * @xmath0cdm simulations predict vast numbers of subhaloes that are satellites to the main halo hosting a galaxy like the milky way . one would naively expect each of these subhaloes to host their own miniature satellite galaxy . the local group looks nothing like this prediction , having only a small handful of dwarfs around each of the giant galaxies . taking into account stellar feedback and heating processes ( mainly at re - ionisation ) in the galaxy formation process , the predicted number of faint satellites around a milky way - like galaxy is @xmath2 100 to 600 . since the majority of the 24 known satellites of the milky way have been largely discovered with the sloan digital sky survey ( sdss ) , and since this survey covered only one fifth of the sky , it has been argued that the problem was solved . however , models that successfully explain the properties of the milky way dwarf satellites predict unobserved dwarfs beyond the virial radius ( bovill & ricotti 2011a ) . moreover , the models tend to produce an overabundance of bright dwarf satellites ( @xmath3 ) with respect to observations ( bovill & ricotti 2011b ) . a rather discomforting way - out is to simply state that the milky way must be a statistical outlier , but this is actually contradicted by the study of strigari & wechsler ( 2012 ) on the abundance of bright satellites around milky way - like galaxies in the sdss survey . a somewhat related but nevertheless distinct problem is that simulations predict that the most massive subhaloes of the milky way are too _ dense _ to host any of the observed bright satellite galaxies ( boylan - kolchin et al . 2011 , 2012 ) . this is the ` too big to fail ' aspect of the missing satellites challenge . a potential solution might be warm rather than cold dark matter . * * the satellites phase - space correlation challenge ( disks of satellites ) . * the distribution of dark subhalos around milky way - sized halos is also predicted by @xmath0cdm to be roughly isotropic . however , the milky way satellites are currently observed to be highly correlated in phase - space : they lie within a seemingly rotationally supported , relatively thin disk ( fig . 1 , see kroupa et al . this differs from the prediction for subhalos in both coordinate and momentum space . since the sdss survey covered only one fifth of the sky , it will of course be most interesting to see whether ongoing surveys such as pan - starrs will confirm this state of affairs . whether or not such a satellite phase - space correlation would be unique to the milky way should also be carefully checked : the milky way could be a statistical outlier , but if the @xmath0cdm model is a realistic description of nature , then the average satellite configurations in external galaxies should be only moderately flattened , and generally non - rotating . in this respect , the recent findings of ibata et al . ( 2013 ) that half of the satellites of m31 define an extremely thin and extended rotating structure makes this phase - space correlation challenge even more severe ( see sect . 3 for a possible solution ) . + spatial distribution of the milky way ( mw ) satellites ( from kroupa et al . 2010 ) . both axes in kpc . the 11 classical satellites are shown as large ( yellow ) circles , the 13 recently discovered satellites are represented by the smaller ( green ) dots , pisces i and ii being the two southern dots . the two open squares near the mw are seg 1 and 2 . the obscuration - region from the mw disk is given by the horizontal gray area . ] * * the density - morphology relation of dwarf ellipticals . * more dwarf elliptical galaxies are observed in denser environments ( e.g. kroupa et al . this relation , observed in the field , in galaxy groups and in galaxy clusters , is not yet understood . * * the angular momentum challenge . * both the merger history of galaxy disks in a hierarchical formation scenario and the associated transfer of angular momentum from the baryonic disk to the dark halo cause the specific angular momentum of the baryons to end up being much too small in simulated disks . these simulated disks in turn end up much smaller than the observed ones , and elliptical systems end up too concentrated too . to address this challenge within the standard paradigm , one needs to form disks through late - time quiescent gas accretion from large - scale filaments , with less late - time mergers than currently predicted in @xmath0cdm . * * the pure disk challenge . * related to the previous problem , large _ thin disk galaxies are very difficult to produce in simulations . this is because major mergers typically create bulges . this is fundamental to the hierarchical nature of galaxy formation in @xmath0cdm , wherein galaxies are built by the merger of smaller galaxies . bulgeless galaxies should therefore represent the rare , quiescent tail of a distribution of merger histories for galaxies of the local volume . however , these bulgeless disk galaxies are quite common , representing more than half of large galaxies ( with @xmath4 ) in the local volume . * * the stability challenge . * quasi - spherical cdm halos stabilize very low surface density disks against the formation of bars and spirals , due to a lack of disk self - gravity . the observation ( mcgaugh et al . 1995 ) of low surface brightness ( lsb ) disk galaxies with prominent bars and spirals is thus challenging in the absence of a significant disk component of dark matter . moreover , in the absence of such a disky dark matter component , the lack of disk self - gravity prevents the creation of observed large razor thin lsb disks . this suggests more disk self - gravity than meets the eye ( mcgaugh & de blok 1998 ) . in the standard context , this would tend to point towards an additional disk - specific dm component , either a cdm - one linked to in - plane accretion of satellites , or a baryonic one , for instance in the form of molecular gas or due to a bottom - heavy imf . * * the bulk flow challenge . * in the @xmath0cdm model , peculiar velocities of galaxy clusters are predicted to be of order @xmath5 km / s . these velocities can actually be measured through the fluctuations in the cmb generated by the scattering of the cmb photons by the hot x - ray - emitting gas inside galaxy clusters . these observations yield an observed coherent bulk flow of order @xmath6 km / s on scales out to at least 400 mpc ( kashlinsky et al . 2012 ) . this bulk flow challenge also appears in galaxy studies ( watkins et al . a related problem is the collision velocity of the merging bullet cluster 1e0657 - 56 at @xmath7 , which is larger than 3100 km / s , too high a value to be accounted for by @xmath0cdm . these observations seem to indicate that the attractive force is enhanced compared to what @xmath0cdm predicts . * * the high - z clusters challenge . * the observation of a single very massive cluster at high redshift can falsify @xmath0cdm . in this respect the existence of galaxy clusters like el gordo with a mass of @xmath8 at @xmath9 ( menanteau et al . 2012 ) and xmmu j2235.3 - 2557 with a mass of of @xmath10 at @xmath11 is surprising . though not sufficient to rule out the model , these clusters certainly push the envelope : there is only about a 50:50 chance that a cluster as massive as el gordo exists at such high redshift . this has been the empirical experience : for every generation of surveys , structures are found further away , and appear sooner than expected . * * the local void challenge . * the 562 known galaxies at distances smaller than 8 mpc from the center of the local group define the ` local volume ' . within this volume , the region known as the ` local void ' hosts only 3 galaxies . this is much less than the expected @xmath12 galaxies for a typical similar void in the @xmath0cdm model ( peebles & nusser 2010 ) . moreover , in the local volume , large luminous galaxies are 6 times more abundant than predicted in the underdense regions . this could just mean that the local volume is a statistical outlier , but it could also point , in line with the two previous problems , towards more rapid structure formation in the early universe , allowing sparse regions to form large galaxies cleaning their environment in a shorter time , thus making the voids emptier and the galaxies larger at early times . * * the cusp - core challenge . * a long - standing problem of @xmath0cdm is the fact that the numerical simulations of the collapse of dark matter halos lead to a ` cuspy ' density distribution that rises steeply as a decreasing function of radius , while rotation curves of external galaxies imply nearly constant density cores in the central parts . the state - of - the - art solution to this problem is to enforce strong supernovae outflows that move large amounts of low - angular - momentum gas from the central parts and that pull on the central dark matter concentration to create a core , but this is still a relatively fine - tuned process , which fails to account for cored profiles in the smallest galaxies , and fails to produce their observed baryon fractions * * the missing baryons challenge(s ) : baryon fraction defined by the acceleration in units of @xmath13 . * constraints from the cmb imply @xmath14 and @xmath15 . however , our inventory of known baryons in the local universe comes up short of the total . for example , bell et al . ( 2003 ) estimate that the sum of cold gas and stars is only @xmath2 5% of @xmath16 . while it now seems that many of the missing baryons are in the form of highly ionized gas in the warm - hot intergalactic medium ( whim ) , we are still unable to give a confident account of where all the baryons reside . but there is another missing baryons challenge within each cdm halo : one would indeed naively expect each halo to have the same baryon fraction as the universe , @xmath17 . on the scale of clusters of galaxies , this is approximately true ( but still systematically low ) , but for galaxies , observations depart from this in a systematic way which remains unexplained in the standard context . in fig . 2 ( mcgaugh et al . 2010 ) , the ratio of the detected baryon fraction over the cosmological one , @xmath18 , is plotted as a function of the potential well of the systems . there is a clear correlation : at large radii , the baryon fraction is actually directly equivalent to the acceleration in units of @xmath19 . if we adopt a rough relation @xmath20 , we get that the acceleration at @xmath21 , and thus the baryon fraction , is @xmath22 . divided by the cosmological baryon fraction , this explains the trend for @xmath23 with potential ( @xmath24 ) in fig . this missing baryons challenge is actually closely related to the baryonic tully fisher relation . + the fraction of detected over expected baryons @xmath25 as a function of potential well and mass . each point represents many objects . orange squares are local group dwarf satellites , light blue circles gas - dominated disk galaxies , dark blue circles star - dominated spiral galaxies , grey triangles galaxy clusters . ] * * the baryonic tully fisher relation ( with @xmath26 defining zero point ) . * the baryonic tully fisher relation is an observed relation between the asymptotic circular velocity and baryonic mass in galaxies ( fig . recent work extending the relation to low mass , typically low surface brightness and gas rich galaxies , extends the dynamic range of the relation to five decades in baryonic mass . over this range , the relation has remarkably little intrinsic scatter ( consistent with zero given the observational errors ) and is well described as a power law , or equivalently , as a straight line in log - log space @xmath27 , with slope @xmath28 . this slope is consistent with a constant acceleration scale @xmath29 such that the normalization constant @xmath30 . all rotationally - supported galaxies are observed to follow the relation , irrespective of their size , surface brightness , or gas fraction . * * tidal dwarf galaxies . * even young tidal dwarf galaxies ( tdgs ) formed in the collision of larger galaxies appear to obey the baryonic tully fisher relation ( gentile et al . this is surprising because these galactic collisions should be very effective at segregating dark and baryonic matter : the rotating gas disks of galaxies that provide the fodder for tidal tails and the tdgs that form within them initially have nearly circular , coplanar orbits . in contrast , the dark matter particles are on predominantly radial orbits in a quasi - spherical distribution . this difference in phase space leads to tidal tails that themselves contain very little dark matter . when tdgs form from tidal debris , they should be largely devoid of dark matter . nevertheless , tdgs do appear to contain enough dark matter to obey the same baryonic tully - fisher relation as galaxies that have not suffered this baryon - dark matter segregation process . + the baryonic tully fisher relation for galaxies with well - measured asymptotic velocities @xmath31 . the dark blue points have @xmath32 , while the light blue points have @xmath33 and are generally less precise in velocity , but more accurate in terms of mass . the dotted line has the observed slope 4 , the dashed line has slope 3 , the difference between these two lines being the origin of the variation in the detected baryon fraction in fig . 2 . ] * * @xmath13 as a transition acceleration . * the mass discrepancy in galaxies always appears ( transition from baryon dominance to dark matter dominance ) when @xmath34 , yielding a clear mass - discrepancy acceleration relation ( fig . this , again , is the case for every single rotationally supported system irrespective of its formation mechanism and history . for high surface brightness ( hsb ) galaxies , where there exists two distinct regions where @xmath35 in the inner parts and @xmath36 in the outer parts , locally measured mass - to - light ratios show no indication of hidden mass in the inner parts , but rise beyond the radius where @xmath37 . it can never be emphasized enough that the role played by @xmath13 in the zero - point of the baryonic tully - fisher relation and this role as a transition acceleration have strictly no intrinsic link with each other , they are fully independent of each other within the cdm paradigm . there is nothing built in @xmath0cdm that stipulates that these two relations ( the existence of a transition acceleration and the baryonic tully - fisher relation ) should exist at all , and of course nothing that implies that these should exhibit an identical acceleration scale . + the mass discrepancy is defined as @xmath38 where @xmath39 is the observed velocity and @xmath40 is the velocity attributable to baryonic matter . many hundreds of individual resolved measurements along the rotation curves of @xmath41 spiral galaxies are plotted . the mass discrepancy is plotted as a function of radius ( top panel ) , centripetal acceleration @xmath42 ( middle panel ) , and newtonian acceleration @xmath43 ( bottom panel ) . ] + the fractional contribution of baryons ( @xmath44 , top ) and dark matter ( @xmath45 , bottom ) to the rotation velocity @xmath46 at the radius @xmath47 where the contribution of the baryons peaks , plotted as a function of the characteristic surface density ( @xmath48 ) . as the baryonic surface density increases , the fractional contribution of dark matter to the total gravitating mass decreases . this fine - tuning persists for any choice of stellar mass - to - light ratio . ] * * @xmath49 as a transition central surface density . * the acceleration @xmath13 defines the transition from lsb galaxies to hsb galaxies : in lsb galaxies whose _ central _ surface density is much smaller than some critical value of order @xmath50 , dm dominates everywhere , and the magnitude of the mass discrepancy is given by the inverse of the acceleration in units of @xmath13 . on the other hand , baryons dominate in the inner parts of hsb galaxies whose central surface density is higher than @xmath51 . there is an anti - correlation between the baryonic surface density and the fractional contribution of dm to the rotation curve ( fig . 5 ) . to explain this , there must be a strong fine - tuning between dark and baryonic surface densities , a sort of repulsion between them , a repulsion which is however contradicted by the correlations between baryonic and dark matter bumps and wiggles in rotation curves ( known as ` renzo s rule ' ) . the shapes of rotation curves also depend on surface density : hsb galaxies have rotation curves that rise steeply then become flat , or even fall somewhat to the not - yet - reached asymptotic flat velocity , while lsb galaxies have rotation curves that rise slowly to the asymptotic flat velocity . finally the _ total _ ( baryons+dm ) acceleration declines with the mean _ baryonic _ surface density of galaxies , in the form @xmath52 ( fig . * * features in the baryonic distribution imply features in the rotation curve . * this is known as ` renzo s rule ' ( sancisi 2004 , see also swaters et al . 2012 ) . while the effect of non - axisymmetric motions should be investigated in more detail , this is currently most easily interpreted by either a heavy ` dark baryonic ' component ( e.g. , in the form of molecular h2 ) scaling with hi in galaxy disks , or by a theory where the baryons effectively act as the main source of gravity . * * @xmath49 as a critical mean surface density for stability . * disks with mean baryonic surface density @xmath53 are extremely rare ( fig . 7 ) and unstable . if stability is provided by a dm halo , there is no reason for this stability threshold to be related to the same acceleration scale that appears elsewhere ( so this is again an independent occurence of the @xmath13-scale ) . dense halos predicted by @xmath0cdm could ( and should ) in principle host higher surface brightness disks than are observed . + the characteristic centripetal acceleration @xmath54 in units of @xmath13 plotted against the characteristic baryonic surface density . the data do show a correlation ( @xmath55 ) , clearly indicating a dynamical role for the baryons , whilst cdm dominance should normally erase any such correlation . ] + the characteristic surface density of baryons vs. their dynamical scale length @xmath56 ( left ) , and disk - only central surface density vs. disk scale - length ( right ) . high characteristic surface densities at low @xmath56 in the left panel are typical of bulge - dominated galaxies , and disks with characteristic surface density @xmath57 are extremely rare . ] as we have seen , the @xmath13 acceleration - scale is ubiquitous in galaxies . it defines the zero point of the tully - fisher relation and the amplitude of the mass - discrepancy in the weak - gravity limit , it appears as a transition - acceleration above which no dm is needed , and below which dm appears , and it defines the maximum surface density of pure disks . it would obviously be interesting if these observational occurences could be summarized and empirically unified in some way : such a unification _ is _ actually feasible through the rather old idea of milgrom ( 1983 ) that the total gravity @xmath58 is related to the baryon - generated one @xmath59 in a way analogous to coulomb s law in a dieletric medium : @xmath60 where the interpolating function @xmath61 this law is at the basis of the modified newtonian dynamics ( mond ) paradigm . it does _ not _ simply make rotation curves flat , but predicted _ all _ the challenging observations of sect . 2.4 in a quick and efficient way ( see sect . 5.2 of famaey & mcgaugh 2012 ) . this formula was not constructed to explain already known observational facts , but predicted almost all of them _ a priori_. its salient properties can be elegantly summarized by simply stating that the underlying theory , whatever its deep nature , has to become scale - invariant under transformations @xmath62 in the regime of small acceleration @xmath63 ( see , e.g. , milgrom 2012 ) . addressing all the other challenges , from sect . 2.1 to 2.3 , would need a full - fledged theory predicting milgrom s law and the associated mond phenomenology . as we shall see in the next section , such a full - fledged theory does not exist at the moment . one can however speculate on how these problems could be addressed by such a theory . for instance , if the fields generating the mond behavior on galaxy scales have an energy density subdominant to the baryonic one , effects such as dynamical friction with dm would disappear . galaxy disks would then easily get more extended thanks to the absence of angular momentum transfer to the dark halo , but mergers would also take longer , and be subject to multiple passages ( tiret & combes 2008 ) . while mond would naturally provide added stability compared to newtonian dynamics without dm , it does not over - stabilize disks : features like bars and spiral arms are a natural result of disk self - gravity , and numerical simulations based on mond appear to do a good job at reproducing the range of observed morphologies of spiral galaxies ( tiret & combes 2007 , 2008 ) . concerning structure formation , the mond paradigm has _ a priori _ no cosmology , providing analogs for neither the friedmann equation nor the robertson - walker metric . for these , one must appeal to specific hypotheses for the covariant parent theory of mond ( sect . 4 ) , which is far from unique . what is more , none of the candidates is theoretically satisfactory , as none emerges from first principles . it is thus not clear whether a compelling candidate mond cosmology will ever emerge . one can however speculate on what would be a reasonable scenario , based on simulations using mond in the weak - field regime at their basis ( e.g. , sanders 1998 , angus & diaferio 2011 ) . in the absence of cdm , and with a mond - like force driving structure formation , after initially lagging behind , structure formation would rapidly speed up compared to @xmath0cdm once the influence of radiation declines and perturbations begin to enter the mond regime . large galaxies could form by @xmath64 ( sanders 2008 ) and clusters by @xmath65 , considerably earlier than in @xmath0cdm . by @xmath66 , the voids would become more empty than in @xmath0cdm , but otherwise mond simulations ( of collisionless particles , which is of course not the best representation of the baryon fluid ) show the same qualitative features of the cosmic web ( llinares et al . 2008 ) . the main difference is in the timing of when structures of a given mass would appear ( assembling a large mass early being easier in mond ) . this means that mond is promising in addressing the high - z clusters challenge and local void challenge , as well as the bulk flow challenge and high collisional velocity of the bullet cluster ( llinares et al . 2009 ) , due to the much larger than newtonian mond force in the structure formation context . what is more , it could allow large massive galaxies to form early ( @xmath64 ) from monolithic dissipationless collapse ( sanders 2008 ) , with well - defined relationships between the mass , radius and velocity dispersion . consequently , there would be less mergers than in @xmath0cdm at intermediate redshifts , but since these mergers would last longer , the number of interacting galaxies could remain unchanged . this could explain the observed abundance of large thin bulgeless disks unaffected by major mergers , and in those rare mergers between spirals , many tidal dwarf galaxies ( tdgs ) would be formed . this could lead to the possibility that many dwarf galaxies are not primordial but have been formed tidally in these encounters ( kroupa et al . these populations of satellite galaxies , associated with globular clusters that formed along with them , would naturally appear in closely related planes , thereby providing a natural solution to the satellites phase - space correlation problem ( ibata et al . 2013 ) . to explain why these tdg dwarf satellites appear dark matter - dominated , while tdgs should not if dm is collisionless ( see also problem 2.4.d hereabove ) , mondian dynamics would be the only possible explanation . what is more , the density - morphology relation for dwarf ellipticals could also find a natural explanation ( dabringhausen & kroupa 2012 ) . however , all this would require a covariant parent theory of mond , which will also have extreme challenges to overcome , and most notably the missing baryon problem that mond suffers in rich clusters of galaxies and the high third peak in the acoustic power spectrum of the cmb . we provide hereafter a non - exhaustive list of the most popular covariant theories that have been proposed to reproduce the mond phenomenology on galaxy scales . more details and references ( as well as a more complete list including proposals based on non - local theories and modified inertia ) can be found in famaey & mcgaugh ( 2012 ) , as well as in bruneton & esposito - farse ( 2007 ) and skordis & zlosnik ( 2012 ) . we also point out hereafter some of the main weaknesses of all these approaches . it should be highlighted again that these theories should _ not _ be confused with the mond paradigm itself , which is not a theory _ per se _ ( see e.g. milgrom 2012 ) . * main ingredients * : an ` einstein ' metric @xmath67 ( with einstein - hilbert action ) , a k - essence scalar field @xmath68 , and normal matter from the standard model coupling to a combination of both . + * action * : the matter couples to the ` physical metric ' : @xmath69 and the scalar field is given a so - called k - essence action , with no potential and a non - linear , aquadratic , kinetic term : @xmath70 where @xmath71 is a dimensionless constant , @xmath72 is a length - scale , and @xmath73 . the function @xmath74 yields the mond dynamics in the approriate limit @xmath75 ( bekenstein & milgrom 1984 ) . + * main weakness * : does not enhance gravitational lensing in galaxies because of the conformal relation between the einstein and physical metrics , a priori enough to reject the theory . * main ingredients * : an einstein metric @xmath67 ( with einstein - hilbert action ) , a k - essence scalar field @xmath68 , a dynamical unit - norm vector field @xmath76 allowing a disformal relation between the einstein and physical metrics ( thus allowing to enhance gravitational lensing ) , and normal matter coupling to a combination of these fields . + * action * : the matter couples to the physical metric : @xmath77 the scalar field is given a k - essence action ( eq . [ scalaraction ] ) , and the vector field ( bekenstein 2004 , skordis 2008 ) : @xmath78 } , \ ] ] where @xmath79 for a set of constants @xmath80 , and @xmath81 is a lagrange multiplier enforcing the unit - norm . + * main weaknesses * : ( i ) possible stability issues ( especially inside matter ) , ( ii ) requires a fine - tuned form of @xmath74 to evade strong - field observational constraints in the solar system and binary pulsars , ( iii ) can not produce a high third peak in the angular power spectrum of the cmb without leading to an exaggerated integrated sachs - wolfe ( isw ) effect , or without resorting to additional collisionless hot dark matter ( appealing to which might overproduce superclusters and massive galaxy clusters as pointed out by angus & diaferio 2011 ) , etc . * main ingredients * : normal matter coupling to the physical metric @xmath82 ( with einstein - hilbert action ) , and a unit - norm dynamical vector field @xmath76 also coupling to the physical metric . + * action * : zlosnik et al . ( 2006 ) showed teves to be expressible as a pure tensor - vector theory in the matter frame , with the _ physical _ metric at the same time satisfying the einstein - hilbert action and coupling minimally to the matter fields , just like in general relativity ( gr ) . the gea approach makes use of this fact to devise a simpler tensor - vector action in the matter frame , where the einstein - hilbert and matter actions remain precisely as in gr , but with an additional unit - norm vector field ( zlosnik et al . 2007 ) : @xmath83 } , \ ] ] where ( see eq . [ k ] and replacing @xmath84 by @xmath85 ) @xmath86 and @xmath87 in the static weak - field limit , the unit - norm constraint fixes the vector field in terms of the metric , and from there we have that , in the weak - field limit , @xmath88 . + * main weaknesses * : mostly the same as teves for the cmb and isw ( zuntz et al . 2010 ) and for the additional hot dark matter , _ but _ , contrary to teves , it can be chosen to approach gr as fast as needed for high accelerations ( with no fine tuning ) so it avoids all the problems teves has in binary pulsars , and the unacceptably large preferred - frame effects of teves in the solar system . * main ingredients * : two metrics @xmath82 and @xmath89 , normal matter coupling to @xmath82 , and a new form of ` twin matter ' coupling to the second metric @xmath89 . + * action * : the heart of this class of theories ( milgrom 2009 ) is an interaction term between the two metrics , through a function of scalars constructed from @xmath90 where @xmath91 . for instance , one can write : @xmath92 + \vphantom{\int}s_{\rm m}[{\rm twin \ , matter } , \hat{g}_{\mu\nu } ] + \frac{c^4}{16\pi g } \int{d^4 x [ \alpha \sqrt{-\hat{g } } \hat{r } + \beta \sqrt{-g } r - 2 ( g \hat{g})^{1/4 } l^{-2 } f(x ) ] } , \label{bimondact}\ ] ] where @xmath93 , and @xmath94 judiciously chosing the @xmath95 and @xmath96 parameters yields a whole class of bimond theories with various phenomenological behaviors . for instance , in matter - twin matter symmetric versions ( @xmath97 ) , and within a fully symmetric matter - twin matter system , a cosmological constant is given by the zero - point of the function @xmath98 , naturally of the order of 1 , thereby naturally leading to @xmath99 for the large - scale universe . matter and twin matter would not interact at all in the high - acceleration regime , and would repel each other in the mond regime ( i.e. , when the acceleration difference of the two sectors is small compared to @xmath13 ) , thereby possibly playing a crucial role in the universe expansion and structure formation . + * main weaknesses * : ( i ) should still be checked against the existence of ghost modes , ( ii ) structure formation etc . to be studied in more details , ( iii ) not clear how these theories can explain the angular power spectrum of the cmb ( even when appealing to twin matter ) . * main ingredients * : normal matter coupling to the gr physical metric @xmath82 , dark matter also coupling to the physical metric , but carrying a space - like four - vector gravitational dipole moment @xmath100 . + * action * : the dipolar dark matter medium ( blanchet 2007 , blanchet & le tiec 2009 ) is described as a fluid with mass current @xmath101 endowed with the dipole moment vector @xmath102 ( which will affect the total density in addition to the above mass density @xmath103 ) , with the following action : @xmath104 } , \label{ddmact}\ ] ] where @xmath105 is the norm of the projection perpendicular to the four - velocity of the polarization field @xmath106 . the poisson equation in the weak - field limit is then recovered as : @xmath107 from there , in order to reproduce the mond phenomenology in galaxies , one appeals to a ` weak - clustering hypothesis ' , namely the fact that , in galaxies , the dark matter fluid does not cluster much ( @xmath108 ) and is essentially at rest ( @xmath109 ) because the internal force of the fluid precisely balances the gravitational force , in such a way that the polarization field @xmath110 is precisely aligned with the gravitational one @xmath58 , and @xmath111 . at the cosmological level , the monopolar density of the dipolar atoms @xmath103 will play the role of cdm , while the minimum of the potential @xmath112 naturally adds a cosmological constant term @xmath99 , thus explaining this numerical coincidence and making the theory precisely equivalent to the @xmath0cdm model at linear order for the expansion , for large scale structure formation , and for the cmb ( naturally explaining the high third peak ) . + * main weaknesses * : ( i ) not clear that the weak - clustering hypothesis in galaxies would be the only natural outcome of structure formation within this model , ( ii ) the weak clustering hypothesis in itself might be problematic for explaining the residual missing mass in galaxy clusters , due to the fact that this residual mass should essentially be concentrated in the central parts of these objects , thus leaving the model to rely on baryonic dark matter or additional hot dark matter to explain galaxy clusters . if one is familiar with cosmology and large scale structure , it must seem rather peculiar that anyone would seriously consider alternatives to @xmath0cdm based on the mond phenomenology . but if one is more concerned with the observed phenomenology in a wide range of galaxy data , it seems just as odd to invoke non - baryonic and collisionless cdm together with fine - tuned feedback to explain the appearance of an effective force law that appears to act with only the observed baryons as a source in galaxies ( see sect 2.4 and 3 , see also famaey & mcgaugh 2012 for an extensive review ) . accepting this , and building a theory to account for this phenomenology , could perhaps help resolving ( see sect . 3 ) many of the other challenges ( sects . 2.12.3 ) currently faced by @xmath0cdm . nevertheless , consistent covariant theories that have currently been proposed along these lines are at best effective , and bring with them their own challenges ( see sect . 4 ) . indeed , the most important aspect before one rejects any model is to have a ` simpler ' model at hand , that still reproduces the successes of the earlier favored model but also naturally predicts the discrepant data . while mond , as a paradigm , has _ a priori _ predicted a lot of observations which @xmath0cdm can not explain in galaxies , it is absolutely fair to say that there is currently no alternative which does better _ overall _ than @xmath0cdm , and in favor of which ockham s razor would be . it would however probably be a great mistake to persistently ignore the fine - tuning problems for @xmath0cdm and the related uncanny successes of the mond paradigm on galaxy scales , as they could very plausibly point at a hypothetical better new theory of the dark sector . it would for instance be extremely exciting if one would manage to find a physical connection between the dark energy sector and the galactic mond phenomenology in the weak - field limit through the @xmath99 coincidence . in this sense , ideas based on entropic gravity are an interesting line of thought ( verlinde 2011 , ho et al . 2010 , klinkhamer & kopp 2011 , and others , all somewhat inspired by milgrom 1999 ) , but one should nevertheless keep in mind the observational challenges that any such new theory of the dark sector could face , especially regarding the third peak of the acoustic power spectrum of the cmb . it is , on the other hand , conceivable that a fully successful alternative theory does not exist , and that the apparent mondian behavior of galaxies will be explained through small compensatory adjustments of the current @xmath0cdm paradigm . but one has to realize that this is far from trivial , and one would have to demonstrate how this could possibly occur . one also has to realize that most currently proposed mond - inspired effective theories ( see sect . 4 ) do include new fields , so that the mond paradigm does not necessarily imply the absence of a dark sector . in any case , the existence of a characteristic acceleration @xmath113 playing various roles in many apparently independent galactic scaling relations is by now an empirically established fact , and it is thus mandatory for _ any _ successful model of galaxy formation and evolution to explain it , together with all the other challenges mentioned in sect . 2 . angus , g.w . , 2009 , _ mon . not . r. astron . _ , * 394 * , 527	the universe on large scales is well described by the @xmath0cdm cosmological model . there however remain some heavy clouds on our global understanding , especially on galaxy scales , which we review here . while some of these clouds might perhaps disappear through small compensatory adjustments of the model , such as changing the mass of the dark matter particles or accounting better for baryonic physics , others should rather be taken as strong indications that the physics of the dark sector is , at the very least , much richer and complex than currently assumed , and that our understanding of gravity and dynamics might also be at play . for instance , the empirically well - tested mond phenomenology in galaxies , whatever its final explanation , should be understood in any model of galaxy formation and dynamics . current alternatives to @xmath0cdm however bring with them many unsolved questions and challenges .
the remarkable results of the lhc experiments @xcite , @xcite culminating with the discovery of a higgs boson with a mass @xmath1 gev has limited the parameter space of many beyond the standard model theories @xcite without eliminating them completely . among the most popular candidates of this type are the supersymmetric models with their minimal version the mssm . the fact that the supersymmetry breaking scale has been pushed further up by the experimental constraints might question if the naturalness @xcite should be one of the main driving reason for low scale supersymmetry . in this context it would be interesting to explore other avenues supersymmetry related like that of a supersymmetric model with strong dynamics . in what follows we shall consider this point of view and sketch a possible picture in this direction . inspired by the low energy qcd and by models with dynamical symmetry breaking we shall hypothesize that the squarks of one generation ( assume the top and bottom quarks ) form bound states and condense thus leading to both electroweak and supersymmetry breaking . the corresponding operators exist already in the mssm lagrangian but instead on dwelling on them we will consider a simple effective lagrangian which parameterizes our ignorance of the underlying strong dynamics . we start with the mssm model without the soft supersymmetry breaking terms . we then set the gauge kinetic term plus the gauge interaction terms for the up and down higgs doublets to zero . we can do that without altering the supersymmetric nature of the model as these terms are invariant under the supersymmetry transformation by themselves . the superpotential is that of the mssm , specifically : @xmath2 for simplicity in what follows we will consider only one generation of the standard model fermions and sfermions denoted generically by up and down . from the absence of the f terms for the up and down higgs doublets we deduce : @xmath3 this leads to : @xmath4 \nonumber\\ & & h_d=-\frac{1}{\mu}y_u\tilde{u}^{}_r\tilde{q}_u . \label{rez56}\end{aligned}\ ] ] replaced in the superpotential this leads to a new expression : @xmath5y_u\tilde{u}^{}_r\tilde{q}_u . \label{superpot56}\end{aligned}\ ] ] note that the chiral superfields corresponding to the up and down higgs doublets have been eliminated completely and we obtain an effective superpotential . if we neglect the terms of order @xmath6 this superpotential will lead to the following lagrangian : @xmath7 where for simplicity we adopted the notation : @xmath8 this lagrangian contains the standard couplings of the higgs bosons with the standard model fermions . however since it is hard to justify the formation of a two lepton composite state we shall adopt a picture where the composite higgses correspond to one up and one down quark such that a dynamical mass will be generated for these . we introduce the up higgs supermultiplet where for example the neutral states have the structure : @xmath9 all other fermions and sfermions then couple as in the regular mssm with the up and down composite higgses . we further check that these composite states still respect the supersymmetry transformation . we will show the invariance under the regular supersymmetry transformation only for the neutral parts of the up supermultiplet : @xmath10= \frac{1}{\mu}y_d\epsilon(d_r^{\dagger}\tilde{d}_l+\tilde{d}_r^d_l)= \epsilon \tilde{h}_u^0 \nonumber\\ & & \delta\tilde{h}_u^0=\frac{1}{\mu}[-i[((\sigma^{\mu}\epsilon^{\dagger}_{\dot{\alpha}})\delta \tilde{d}_r^)\tilde{d}_l+((\sigma^{\mu}\epsilon^{\dagger}_{\dot{\alpha}})\delta \tilde{d}_l)\tilde{d}_r^]+\delta(\tilde{d}_r^)d_l+d_r^{\dagger}\delta(\tilde{d}_l ) ] \nonumber\\ & & = -i\frac{1}{\mu}[(\sigma^{\mu}\epsilon^{\dagger}_{\dot{\alpha}})\delta ( \tilde{d}_r^\tilde{d_l})]-\frac{1}{\mu}\epsilon ( d_r^{\dagger}d_l)=-i(\sigma^{\mu}\epsilon^{\dagger}_{\dot{\alpha}})\delta(h_u^0)-\frac{1}{\mu}\epsilon ( d_r^{\dagger}d_l ) . \label{finalrez56478}\end{aligned}\ ] ] the last term in the second equation of ( [ finalrez56478 ] ) would correspond to an @xmath11 contribution or better to a @xmath12 term associated with a possible superpotential . note that the regular mssm lagrangian with the elementary states already contains the seeds of possible dynamical electroweak and supersymmetry breaking through the contributions from the d terms : @xmath13 , \label{foursq45}\end{aligned}\ ] ] such that two scalar condensates may form . here @xmath14 are the hypercharges corresponding to each particle . this is not unnatural for low energy qcd where it is hypothesized that the scalar states are an admixture of two quark and four quark states @xcite . moreover the four quark states may be regarded as molecule formed out from two quark states . for example if the two quark state is given by , @xmath15 there is a possibility that the four quark state have the form : @xmath16 then the low energy dynamics @xcite suggests that the vacuum expectation expectation values of the scalar mesons and the two and four quark condensates are related as in : @xmath17 here @xmath18 and @xmath19 are the vacuum expectation values for the two quark and four quark scalar states whereas the quantities in the brackets are the two quark and four quark vacuum condensates respectively . moreover @xmath20 which is adimensional parameter is determined to be ( from the instanton dynamics ) in the range @xmath21 . low energy spectroscopy of scalar and pseudoscalar indicate that @xmath22 . note that the existence of possible two squarks bound states has been explored earlier in @xcite , @xcite with the strong dynamics related to a large squark - squark - higgs coupling . however here we adopt a picture in which the strong mechanism is similar to low energy qcd and may require the introduction of an additional strong gauge interaction . the resulting effective theory has thus in terms of matter particles only the minimal content of the mssm with the regular two higgses . we shall describe a strongly interaction supersymmetric model through an effective theory theory which by itself in the absence of quantum corrections should lead to the correct mass of the higgs boson found at the lhc . as an illustration we assume that the usual mssm higgs potential is generated together with two beyond mssm effective terms introduced first in @xcite . one of them is supersymmetric and stems from the effective superpotential , @xmath23 the other one is supersymmetry breaking term . together they bring the following contribution to the higgs potential : @xmath24 here @xmath25 and @xmath26 are small coefficients which parameterize the ratio of the supersymmetry scale to that of possible new physics in general but for our case at hand we consider them to parameterize the unknown strong dynamics . in particluar @xmath27 where @xmath28 is the usual mssm scale . for simplicity we shall take the parameters @xmath25 and @xmath26 real although there is no need to do so . the full higgs potential : @xmath29 will lead to the minimum equations : @xmath30 we introduce the notation @xmath31 . moreover we express the two higgs doublets in terms of the mass eigenvalues in the usual way . @xmath32 here , @xmath33 and , @xmath34 we shall ignore the charged states in what follows . the mass matrix for the scalar higgs boson is calculated to be : @xmath35 whereas that of the pseudoscalar states is : @xmath36 in order to see if the above mass matrices are compatible with the latest lhc data @xcite , @xcite and especially with a light higgs boson with a mass at @xmath1 gev we will use the results of the phenomenological study done in @xcite . there it is shown that the best fit points for the ratio of the higgs signal strenghts ( the ratios @xmath37 and @xmath38 , @xmath39 ) is for @xmath40 and for @xmath18 very close to the decoupling limit @xmath41 . we shall use these value together with the mass of the higgs boson found at the lhc as inputs . in fig [ s1 ] we plot the parameter @xmath26 as a function of the parameter @xmath25 for five values of the parameter b in the range 1 - 5 tev . in the limit considered the mass of heaviest higgs boson h and that of the pseudoscalar a are degenerate . in fig [ s2 ] we plot the square of these masses as a function of the parameter @xmath25 in the same range of masses for b. the experimental lower bounds that exist on @xmath42 and @xmath43 agree better with those region of the parameter b is lower in value whereas @xmath25 and @xmath26 are higher in absolute values although still much lesser than 1 . = 10 cm = 10 cm we consider the possibility that the supersymmetric standard model has an underlying strong dynamics where the higgses are bound states of two squarks whereas the corresponding higgsinos are composite states of an squark and a quark . correct supersymmetry transformations can be preserved for these states . the formation of two scalar vacuum condensates due to the color group or possibly an additional gauge sector may trigger both electroweak and supersymmetry breaking . since it is hard to describe such an intricate model we parameterize the underlying strong dynamics through the introduction of two effective terms in the scalar potential , one supersymmetric the other one with supersymmetry breaking . this new potential can lead by itself to the correct mass of the light higgs boson found at the lhc . we further show that this picture is consistent with the latest lhc data and the corresponding experimental constraints lead to degenerate masses for the h and a bosons which may take a large range of values , depending on the choice of parameters . -.5 cm the work of r. j. was supported by a grant of the ministry of national education , cncs - uefiscdi , project number pn - ii - id - pce-2012 - 4 - 0078 . we would like to thank salah nasri for useful comments about the manuscript . 15 the atlas collaboration , phys . b 716 * , 1 ( 2012 ) . the cms collaboration , phys . b * 716 * , 30 ( 2012 ) . j. d. lyyken , cern yellow report cern-2010 - 002 , arxiv:1005.1676 ( 2010 ) . n. arkani - hamed , a. gupta , d. e. kaplan , n. weiner and t. zorawski , arxiv:1212.6971 ( 2012 ) . s. p. martin , arxiv : hep - ph/9709356 ( 1997 ) . ( note that we used everywhere the notations in this paper ) . a. h. fariborz , r. jora and j. schechter , phys . d * 77 * , 094004 ( 2008):arxiv:0801.2552 . j. m. cornwall , a. kusenko , l. pearce , r. d. peccei , arxiv:1210.6433 ( 2012 ) . l. pearce , a. kusenko and r. d. peccei , arxiv:1307.6157 ( 2013 ) . m. dine , n. seiberg and s. thomas , phys . d * 76 * , 095004 ( 2007):arxiv:0707.0005 . a. djouadi , l. maiani , g. moreau , a. polosa , j. quevillon and v. riquer , arxiv:1307.5205 ( 2013 ) .	we discuss the supersymmetric standard model from the perspective that the up and down higgs supermultiplets are composite states . we show that a higgs multiplet in which the scalar states are bound states of two squarks and the corresponding higgsinos are bound states of a quark and an squark has the correct supersymmetry transformations and may lead to an alternative model which displays dynamical symmetry breaking . we describe this model through an effective higgs potential which by itself may lead to the correct mass of @xmath0 gev for the lightest higgs boson and to other neutral scalar masses respecting the experimental constraints .
in @xcite , falk and randell introduced the notion of a formal arrangement . an arrangement is formal iff every linear dependency among the defining forms of the hyperplanes can be expressed as linear combination of dependencies among exactly 3 defining forms . many interesting classes of arrangements are formal : in @xcite , falk and randell proved that @xmath7 arrangements and arrangements with quadratic orlik - solomon ideal are formal and , in @xcite , yuzvinsky showed that free arrangements are also formal ; and gave an example showing that formality does not depend on the intersection lattice . in @xcite , brandt and terao generalized the notion of formality to @xmath0formality , proving that every free arrangement is @xmath0formal . for this they introduced the concept of higher relation spaces , which capture the dependencies among dependencies. in the first section of this paper we briefly recall the notions of relation spaces and @xmath0formality . by rewriting the definitions , we obtain a lemma characterizing @xmath0formality topologically . then we apply this criteria for graphic arrangements to obtain a description of @xmath0formality in terms of the homology of a certain chain complex . with this it is easy to produce examples of graphic arrangements which are @xmath0formal but not @xmath8formal , for any given @xmath9 . in what follows we adopt all the notation from @xcite . let @xmath10 be an arrangement of @xmath11 hyperplanes in a vector space @xmath12 over a field @xmath13 . for each @xmath14 we fix the defining form @xmath15 . define a map @xmath16 , by @xmath17 , where @xmath18 is the vector space with basis @xmath19 . let @xmath20 be the kernel of this map . then @xmath21 where @xmath22 is the rank of @xmath10 . the vector space @xmath20 describes which linear forms are linearly dependent , as well as the dependency coefficients ( up to scalar multiplication ) . we will refer to elements of @xmath20 as _ let @xmath23 be the subspace of @xmath20 generated by the relations corresponding to dependencies of exactly 3 linear forms . @xmath10 is formal iff @xmath24 . for @xmath25 , recursively define @xmath26 to be the kernel of the map @xmath27 where @xmath28 is the lattice of intersections of @xmath10 and @xmath29 is the sum of the inclusion maps @xmath30 . we identify @xmath31 with @xmath20 . to simplify notation , for @xmath32 we will denote with @xmath33 the vector space @xmath34 . we define 1 . an arrangement is 2-formal if it is formal . 2 . for @xmath35 , @xmath10 is @xmath0formal iff it is @xmath36formal and the map @xmath37 is surjective . .2 in for any arrangement @xmath10 , the following sequence of vector spaces and maps form a complex : @xmath38 where @xmath39 , @xmath40 and for @xmath32 , @xmath41 are the spaces from the notations above . also , @xmath42 and @xmath43 for @xmath32 . we have @xmath44 . so @xmath45 is well defined . also , @xmath46 for any @xmath47 , as @xmath48 . so , indeed we have a complex . @xmath10 is @xmath0formal iff @xmath49 for every @xmath6 @xmath50 is surjective iff @xmath51 there exists @xmath52 such that @xmath53 . we have @xmath54 and @xmath55 . so we get @xmath56 which give us @xmath57 . in this example we will discuss @xcite , example 5.1 . , in terms of the homology of the above complex . we must specify that all the computations are already done in @xcite , and we are just translating into topological language . @xmath10 is a real essential arrangement of rank 4 consisting of 10 hyperplanes , defined by the vanishing of the following linear forms : @xmath58 . so @xmath59 , @xmath60 and the map @xmath61 is just the map @xmath62 and has rank 4 . therefore @xmath63 has dimension @xmath64 . we have 7 nondegenerate rank 2 elements in @xmath65 and each is an intersection of exactly 3 hyperplanes . so we have 7 relations of length 3 : @xmath66 . therefore @xmath67 . the matrix of the map @xmath68 is exactly the matrix in @xcite , page 61 @xmath69 and it has rank 6 . so @xmath70 and @xmath71 . also in @xcite we have listed all the elements of rank 3 from @xmath65 : @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , @xmath77 , @xmath78 , @xmath79 , @xmath80 , @xmath81 , @xmath82 , @xmath83 . if @xmath84 is such an element ( with @xmath85 ) , then @xmath86 means that there is at least a relation among the relations of length 3 of elements of rank 2 in @xmath87 . the nondegenerate rank 2 elements in @xmath87 are nondegenerate rank 2 elements in @xmath65 and these are listed above . it is not difficult to check which are the relations of length 3 for each rank 3 element in @xmath10 . for reference , these are listed in the chart on page 62 in @xcite . also , there is no problem to check that for each @xmath85 , the length 3 relations are linearly independent . therefore we conclude that @xmath88 . so the complex we get is : @xmath89 with homology : @xmath90 and @xmath91 . so @xmath10 is formal , but not 3-formal . .2 in graphic arrangements are a class of arrangements possessing many nice properties ( see , for example , @xcite , @xcite , @xcite , @xcite ) , and for this class there is a pleasant combinatorial interpretation of lema 2.5 . in the next section , @xmath1 denotes a connected graph with no loops or multiple edges . for the graphic arrangement @xmath4 , we will identify the complex above with the chain complex of the flag complex of @xmath1 . then , with lemma 2.5 . , the statement in the abstract will be natural . let @xmath1 be a connected graph on vertices @xmath92=\{1,\ldots , n\}$ ] with no loops or multiple edges . flag(clique ) complex _ @xmath93 is the simplicial complex with : for @xmath98 , let @xmath99 be the number of @xmath95faces of @xmath2 . we have the natural chain complex of @xmath2 : @xmath100 where @xmath101 and @xmath102 is the usual differential given in terms of generators : @xmath103 ) = \sum_{j=1}^{i+1 } ( -1)^{j-1}[n_1,\ldots,\hat{n_j},\ldots , n_{i+1}]$ ] . by definition , the graphic arrangement associated to @xmath1 is @xmath104 . note that @xmath10 is an arrangement in @xmath105 of rank @xmath106 ( if @xmath1 is connected ) and consists of @xmath107 (= the number of edges in @xmath1 ) hyperplanes . notice that from the beginning we fixed the defining forms @xmath108 . to be consistent with notation , @xmath109 will be the symbols in @xmath110 ( i.e. , @xmath111 ) . with these , we can identify @xmath40 with @xmath112 by @xmath113 $ ] for @xmath114 . if we fix the form of the elements in the basis of @xmath115 s and with proper notations of those , the correspondence between the two complexes will become natural . the next lemma will do this , but before we state and prove it here is the flavor of it : we have @xmath118 . suppose for an @xmath119 we have @xmath120 . this means that we must have a dependency ( relation ) among some of the linear forms corresponding to some edges in @xmath117 . but this translates in the fact that @xmath117 contains a cycle . if the length of this cycle is @xmath121 , then the linear forms corresponding to 3 consecutive edges in the cycle are linearly independent . this contradicts the fact that @xmath122 . so @xmath117 contains a triangle . if we have an extra edge in @xmath117 , beside those from the triangle , then the linear form of this extra edge and the linear forms associated to two of the edges of the triangle are linearly independent . again we get a contradiction with the fact that @xmath122 . so @xmath123 a triangle . so each nonzero summand of @xmath124 corresponds to a triangle in @xmath1 . the converse of this statement is obvious . ( the recursive identification lemma ) let @xmath116 with @xmath125 . then @xmath126 iff @xmath117 is a @xmath127 subgraph of @xmath1 . more , @xmath128 and if @xmath129 , i_1<i_2<\cdots i_{l+1}$ ] , then we can pick a special basis element of @xmath130 to be the relation on the special elements corresponding to the @xmath131 subgraphs of @xmath117 : @xmath132 . this element is denoted with @xmath133 . if @xmath117 has just one @xmath131 subgraph , then @xmath147 . but @xmath148 is a sum of inclusions , and in this particular case it will be exactly an inclusion . so we get that @xmath149 , which is a contradiction . let s take two of them @xmath150 and @xmath151 , and first suppose they do not share any vertex . let @xmath152 and @xmath153 be two vertices of @xmath117 . through @xmath154 pass exactly @xmath155 edges and the corresponding linear forms @xmath156 are linearly independent . let s take two edges @xmath157 $ ] and @xmath158 $ ] of @xmath151 and let @xmath159 and @xmath160 be the corresponding linear forms . then , @xmath161 are linearly dependent if at least one of the vertices @xmath162 is a vertex in @xmath150 . therefore , @xmath163 are linearly independent . but this will contradict @xmath164 . hence , @xmath150 and @xmath151 have at least a common vertex @xmath154 . suppose @xmath165 are two vertices of @xmath151 but not of @xmath150 . then , through @xmath154 pass at least @xmath166 edges : @xmath155 from @xmath150 and @xmath167 $ ] , @xmath168 $ ] from @xmath151 . the corresponding linear forms are linearly independent and again we obtain a contradiction with the fact that @xmath164 . suppose @xmath117 has exactly two @xmath131 subgraphs : @xmath170 $ ] and @xmath171 $ ] . let @xmath172 . then @xmath173 for some @xmath174 . we have @xmath175 in @xmath176 . so we get a relation on the special basis elements of @xmath176 : @xmath177 @xmath178 observe that this equation is impossible . so @xmath117 has at least three distinct @xmath131 subgraphs : @xmath179 $ ] , @xmath180 $ ] and @xmath181 . if both @xmath134 and @xmath166 are vertices in @xmath181 , then @xmath134 and @xmath166 are connected in @xmath117 , so @xmath117 contains a @xmath127 subgraph . if , for example , @xmath182 , then from @xmath169 and since @xmath183 are distinct we get that @xmath184 $ ] , for some other vertex @xmath185 in @xmath117 . observe that through the vertex @xmath186 pass at least @xmath166 edges of @xmath117 : @xmath187,[1,3],\ldots,[1,l-1],[1,l],[1,l+1 ] , [ 1,l+2]$ ] . the corresponding linear forms of these edges are linearly independent so we get a contradiction with the fact that @xmath164 . we can conclude that @xmath117 contains a @xmath127 subgraph . now , if there exists an extra edge of @xmath117 not on this @xmath127 , then the corresponding linear form of this edge together with the corresponding linear forms of the edges passing through any vertex of the @xmath127 subgraph will form a linearly independent set of @xmath166 elements . again we get a contradiction with the fact that @xmath164 . so @xmath117 is a @xmath127 . with this , @xmath117 has exactly @xmath166 @xmath131 subgraphs . these subgraphs will give us the special elements of @xmath188 : @xmath189 . the only relation on these elements is exactly the special element in @xmath130 : @xmath190 we denote this element with @xmath191 and he is forming the basis for @xmath130 . with this lemma we can identify easily the two complexes . the way we pick the special basis elements will give us the same matrices for the differentials of the two complexes and , hence , with lemma 2.5 . , we have proved the following proposition : the associated flag complex @xmath2 is the boundary complex of an octahedron on the same vertices and edges . the associated chain complex of @xmath2 is : @xmath193 where , if we order the basis lexicographically we have : every 4-cycle in @xmath1 is a linear combination of 3-cycles . so @xmath4 is formal ( 2-formal ) . by the proposition above , @xmath199 and with this we get @xmath200 . therefore , @xmath201 . so we get @xmath202 . hence @xmath4 is not 3-formal .	we prove a criterion for @xmath0formality of arrangements , using a complex constructed from vector spaces introduced in @xcite . as an application , we give a simple description of @xmath0formality of graphic arrangements : let @xmath1 be a connected graph with no loops or multiple edges . let @xmath2 be the flag ( clique ) complex of @xmath1 and let @xmath3 be the homology of the chain complex of @xmath2 . if @xmath4 is the graphic arrangement associated to @xmath1 , we will show that @xmath4 is @xmath0formal if and only if @xmath5 for every @xmath6 . [ section ] [ defn0]proposition [ defn0]conjecture [ defn0]theorem [ defn0]lemma [ defn0]corollary [ defn0]example partially supported by nsf dms 03 - 11142 and atp 010366 - 0103 . ]
the intermediate silicon tracker ( ist ) is a silicon based particle detector installed at brookhaven national laboratory ( bnl ) in the relativistic heavy ion collider ( rhic ) . the detector is located in the central region of the solenoidal tracker at rhic ( star ) experiment and makes up the third tracking layer in a 4-layer vertex detector upgrade . it is comprised 24 staves , each of which is a flexible pcb wrapped around a carbon fiber core . attached to each stave are 6 silicon sensors which can detect energetic particles which pass through them . these silicon sensors are read out by 36 analog front - end chips which are connected to an external data acquisition system . each stave holds the electrical substrate known as a `` hybrid . '' the hybrid is the flexible circuit onto which the silicon sensors , analog front - end chips , and passive components are mounted . the hybrid is permanently bonded to the stave during the stave manufacturing . the hybrid is broken up into 2 areas , the sensor area and the connector area . in the sensor area we desire as little mass as possible to reduce interactions with passing particles . for this reason , the sensor area , which makes up most of the hybrid , is made of 2 layers of copper on a kapton substrate . the connector area is made of the same 2 layers of copper but with an additional copper layer and kapton layer . over both areas is a kapton coverlay attached with of adhesive . this coverlay has no cutouts on the bottom layer and acts as insulation between the copper and the structural material in the stave . on the top layer large areas of this cover lay are cut out around any area with bonding pads or component pads . conventional solder mask is applied to this area as shown in figure [ fig : hybrid ] . the hybrids are finished with enepig for compatibility with both wire bonding and conventional soldering . in order to keep the copper weight down , all vias are selectively plated through . this selective plating can be seen in figure [ fig : passive ] . , width=480 ] each hybrid provides the electrical connection from the apv chips to the connector site on one end of the stave . electrically the hybrid is broken up into three separate circuits which share a common ground . each circuit serves 12 apv chips and 2 sensors . the apv chips require power , analog output , and control signals to be routed to each chip . the sensors require a bias line which must withstand up to . in the sensor area we require as little mass as possible . because of this , the sensor area has only two routing layers , so careful routing was needed to optimize the power net distribution and signal distribution . the power nets are distributed as three large busses per section for , , and ground . the power supplies for the staves operate on a remote regulation scheme , so a sense net from each bus is routed locally away from the apv chips in each section and back to the connectors . for the analog output signals , a differential pair from each apv chip is routed to the connectors . the analog output is a differential current mode signal . source termination is provided at the board which plugs into the stave . in addition , the apv chips use i2c for slow controls to set various parameters of the chip . these nets are also routed as a bus to each of the chips in a section . finally , two differential pairs provide a clock and a trigger to each apv chip . these are routed as a bus to each of the chips in a section . the hybrid has 36 apv placement sites where apv chips are attached and 6 sensor placement sites where sensors are attached . the apv chips generate nearly all of the heat in the stave so an aluminum cooling tube is embedded in the stave directly beneath the apv chips as shown in figure [ fig : cooling_tube ] . to reduce the thermal resistance , 14 vias are placed on the apv placement sites to conduct heat through the kapton core . the 6 sensors do not produce any appreciable thermal load . , title="fig:",height=163 ] , title="fig:",height=163 ] each hybrid was assembled into a stave at the lawrence berkeley national laboratory ( lbnl ) composites shop in berkeley , california . two carbon fiber cover sheets were co - cured to a hybrid and then a carbon fiber honeycomb was glued to one side . a block of carbon foam with a channel cut in it for cooling was also glued on . the carbon foam has a lower radiation length but is homogeneous unlike the carbon fiber honeycomb . this uniformity makes it more suitable for wire bonding , so the carbon foam is located underneath any area where wire bonding will be performed . each stave nominally generates about of heat . cooling is provided by flowing a synthetic heat transfer fluid , 3m^tm^ novec^tm^ 7200 engineered fluid , through the aluminum tube which runs through the carbon foam . because the ist is installed in the central region of star , we needed a cooling fluid which would not damage the other detectors surrounding it in the event of a leak . novec has a very low vapor pressure , leaves no residue when it evaporates and has a very high resistivity , making it a good candidate . novec was chosen because it has a lower ozone - depletion potential than other engineered fluids with similar properties . other detectors using a similar design have observed galvanic corrosion between the carbon foam and the aluminum tube @xcite . in order to prevent this , the carbon foam channel was coated with epoxy before the tube was laid in to provide a barrier . the target running temperature is which is above the dew point in the experimental hall ; this will prevent moisture from condensing on the detector which could facilitate galvanic corrosion . in addition , all parts of the stave and cooling system are electrically grounded . two end pieces called `` closeouts '' were then attached on either end of the stave . one is made of carbon filled peek and the other of aluminum . the closeouts have holes and slots to provide a way to mount the stave to the finished detector . the aluminum closeout is electrically bonded to the carbon face sheets , the carbon fiber honeycomb , and the carbon foam . this provides a means to ground the materials in the stave since they are electrically isolated from the hybrid . once everything was bonded to one of the face sheets , glue was applied to the other side and it was folded over to complete the stave . figure [ fig : stave_assmb ] shows the stave right before this step . fully cured staves were inspected and then sent to mit for further assembly . , width=528 ] the ist sensors were designed by mit and manufactured by hamamatsu . the sensors are silicon with 2 metallization layers . the backside has a sputtered aluminum coating to provide the bias voltage . this voltage reverse biases the pn junctions in the sensor which form the individual sensing elements ( see figure [ fig : sensor_sch ] . particles passing through the sensor will interact with the silicon lattice and produce charge pairs which will drift apart . a capacitor above each sensing element ac couples the signal and routes it to a bonding pad located on one edge of the chip . this is a proven , standard technology for particle detection . , width=384 ] each sensor has 6 groups of sensing elements with 128 sensing elements in each group . a bond pad for each one of the sensing elements is grouped at one edge of the sensor and is arranged in a pattern which mirrors the bonding pattern for the apv chip ( see figure [ fig : bonding_pads ] ) . this allows for much simpler bonding between the apv and the sensor , especially useful because the bonding pitch is so small . the sensing elements are arranged in a grid at a pitch of @xmath0 in 64 rows and 12 columns . the bonding pads are arranged in 2 rows of 64 pads . the pads are @xmath0 with between each pad and between the 2 rows . , width=288 ] the apv25-s1 chip is the readout and pre - amplifier asic for the sensors . it has 128 channels each with a charge sensitive pre - amplifier , shaper , and long pipeline . events are read into the pipeline at . events in the pipeline are selected by triggers and marked for readout . a single differential pair per chip reads out each of the 128 channels in series for a selected event . , height=192 ] this asic was designed by the imperial college in london for the compact muon solenoid ( cms ) running at the large hadron collider ( lhc ) @xcite . it was designed for the high radiation environment present in the middle of a physics experiment and used an ibm radiation hard process . detailed information about the apv chip can be found in @xcite and @xcite . figure [ fig : apv ] shows the apv chip bonded to a stave . in addition to the sensors and apvs , a few passive components are needed on the stave . each stave has 195 additional components , mostly 0402 sized capacitors for bypassing the apv chips . in addition , there is a small temperature sensor , termination resistors for the clock and trigger lines , protection resistors for the sensor bias lines , and a connector for each section . the connectors used were the samtec tem and sem series . they provide a low mating height and high pin density which is ideal for our application . the pins are extremely fragile , however , so extreme care had to be taken during testing and assembly to reduce the number of mating cycles and prevent any connectors from being damaged . the temperature sensor used was a texas instruments tmp102 which shares the same i2c bus as the apv chips in one of the sections . a computational fluid dynamics ( cfd ) study was done to determine the location of the hottest spot on a stave assuming nominal operating conditions ( see figure [ fig : cfd ] ) . the temperature sensor is placed in this location . , width=288 ] the passives were soldered onto the hybrids after stave assembly . because the hybrid was already attached to the carbon fiber parts of the stave , it could not be put into a solder oven and had to be hand - assembled . proxy manufacturing in methuen , massachusetts assembled all of the staves . the assembly process was further complicated because there is no silk screen legend for most of the components on the board . because the bypass capacitors are located so close to the apv chips there was no space for a silk screen legend ( see figure [ fig : passive ] ) . instead , the pattern of component placement was made the same for all apv placement sites and detailed documentation was produced to prevent any parts from being incorrectly placed . a visual inspection at mit after the passives were attached to the staves provided an additional quality check for such an error . , width=288 ] during production one stave was found to have a short between two power nets on the hybrid , away from any solder areas . using a thermal camera we were able to determine the location of the short ( see figure [ fig : thermal ] ) . until that point in production , the staves were not electrically tested after receipt from lbnl or after passives were attached . after this incident , the power nets were electrically tested for shorts . we found no additional problems during production . , width=288 ] during production we performed separate assembly steps for the apv chips and the silicon sensors . our production volume was limited by the silicon sensors as they were the single most expensive component to purchase and also had the longest lead time . running out of sensors late in production would have delayed installation of the full detector . to increase the overall yield , the apv chips were installed on the staves first , wire bonded , and tested to make sure they were fully functional . only after all apv chips on a stave were verified to be working properly were silicon sensors attached . the apv chips were glued directly to the hybrid using a conductive epoxy , tra - duct 2902 . a pneumatic dispenser was used to apply a uniform amount of glue for each apv chip . the glue is both thermally and electrically conductive and acts as a means to connect the back side to vss and as a means to conduct heat away from the chip . it is important that the alignment of the apv chips is precise . since there are bonding pads on all sides of the chip , both `` x '' and `` y '' alignment are important . to achieve good alignment we made custom tooling which would allow the technician to align each chip manually . the tooling provides an edge which all apv chips are pressed against to set the `` y '' alignment . then the technician aligns the bonding pads on the apv chip with copper features on the hybrid to ensure a precise alignment in `` x '' . the tooling used for this is shown in figure [ fig : apv_attach ] . this alignment step was most important because the mounted apv chips would then have to match exactly the location of the output pad sites on the silicon sensors . misalignment would make the bonding process more difficult and increase the likelihood of shorts between wire bonds . all of the alignment was done by hand under a microscope by a skilled technician . , width=480 ] staves with apv chips attached were sent to the instrumentation division at brookhaven national laboratory for wire bonding and testing . wire bonding was done on a hesse & knipps bond jet 815 bonding machine using aluminum wire . a program for the bonding machine was written to wire bond each apv chip automatically , but inspection was done by a bonding technician during the bonding process to ensure the operation went smoothly . example images from the inspection are shown in figure [ fig : apv_bond ] . during this operation 31 control nets were bonded from the hybrid to the apv chip and 21 power nets are bonded from the hybrid to the apv chip . there were a total of 1872 wire bonds for this operation . the stave was held down using only the stock vacuum chuck on the bonding machine . no additional fixture was needed . , title="fig:",width=240 ] , title="fig:",width=240 ] bonded apv chips had to be tested before silicon sensors could be attached . in an experimental assembly building located next to the star detector we set up a clean room with a testing station for the ist . the testing station consists of a rack to hold detectors with a cosmic ray trigger inside a light - tight box ( ambient light will create noise on the sensors due to the photoelectric effect . ) the staves were attached to a data acquisition system ( daq ) ( see @xcite for more information ) which provided power and control signals and digitized the data from the apv chips . to test each stave , it was attached to the daq and operated . noise and pedestal levels were recorded over many events and then analyzed . all staves were then sent back to mit for further work . staves that had apv chips on them which did not pass the noise and pedestal level tests had those chips removed and replaced and then sent back to bnl . the replaced apv chips were bonded and then re - tested before they were sent back to mit . staves with fully functioning apv chips then had the silicon sensors installed . the silicon sensor attachment was the final assembly step done at mit . before the process began , all tooling and staves were cleaned with isopropyl alcohol to ensure that no contaminants were transferred onto the sensors . staves were placed in mit - made tooling which held the stave in place . an adjustable blade at the top edge of the stave was moved until it marked the top edge of where each sensor would sit . the blade was fastened in position and then the sensor attach began . the sensors are held down with two different types of epoxy as shown in figure [ fig : si_glue ] . the first is a conductive epoxy needed to make a good contact between the aluminized back side of the sensor and the traces below , which carry the bias voltage . we used the same tra - duct 2902 adhesive for this . the second epoxy is epon 828 resin with versamid 140 hardener . we used this epoxy for mechanical strength and stability since the conductive epoxy is not very strong . a single line of conductive epoxy was put down and then several lines of conventional epoxy were applied to make a strong bond . we found that the working time after applying the glue was about 30 minutes after which it would be much more difficult to get the sensor to lie flat . , width=576 ] the silicon sensors were aligned in a similar way to the apv chips . the tooling provided a stop in the `` y '' direction so the technician needed only focus on alignment in the `` x '' direction . under a microscope , a technician would align the sensor bonding pads with the apv bonding pads checking all 6 apv chips for each sensor . an example view through the microscope during this alignment procedure is shown in figure [ fig : si_align ] . after all sensors were aligned , weights were applied to the sensors . in early prototypes we found that the sensors would drift during the curing process . adding weights to the tops of the sensors prevented the sensors from drifting . after the weights were applied , the alignment was rechecked and adjusted if necessary . staves would be allowed to cure for 24 hours before they would be packaged and transported . , width=192 ] the bonding of the silicon sensors took place at the same instrumentation division at bnl on the same bonding equipment . again aluminum wire was used . an automatic bonding program was used but we required close supervision by the technician . because of the size and pitch of the bonding pads it was important to catch any error and repair it before bonding continued . the bonding process put down 128 bonds between each apv chip and each silicon sensor for a total of 4608 wire bonds . in addition , 24 bonds were put down between each sensor and the hybrid as part of the biasing circuitry for a total of 144 bonds . depending on the amount of technician intervention , this process would take between 1 and 2 hours . the end result can be seen in figure [ fig : si_bonded ] and figure [ fig : full_stave ] . , width=624 ] fully bonded staves were brought back into the ist clean room and tested again . using the daq system , we were able to identify any shorted bonding wires and determine if it was possible to repair them . in addition , baseline noise and pedestal information was again recorded . we found very few errors with the bonding done at bnl and nearly all of the staves produced are in use in the final detector . fully functioning staves then had an encapsulant applied to the bonding wires to protect them during handling and installation . the first prototypes used dymax 9001-e - v3.1 cured with a dymax 5000-ec series ultraviolet curing lamp . we found that , after exposure to the uv light source , some apv chips would no longer function or have higher input noise . while the exact cause of this was unknown , we suspect that the silicon was damaged by the uv treatment . for production staves , dow corning sylgard 186 elastomer was used . this encapsulant has a room temperature cure and did not affect the performance of the apv chips . a fully completed stave , ready for installation , is shown in figure [ fig : full_stave ] . , width=624 ] after the encapsulation the staves were tested extensively and optically surveyed before being installed . the structure which the staves are mounted to has a mechanical alignment accuracy of about . however , once installed we require knowing the position of the detector to a higher degree of accuracy . we are able to do this by aligning tracks which pass through the ist as well as other detectors . we can reduce the number of tracks needed to do this alignment by knowing very precisely how the sensors on a single stave are located relative to each other . before installation each of the staves was placed on an optical survey station and measured . the survey station , a nikon vm-150 , was used to measure each of the staves . we estimate the accuracy of these measurements to be about in the plane of the detector and out of plane . we combine this information along with the `` rough '' position of each detector inside the experiment and will be able to align the detector from tracks generated in this and other detectors ( see figure [ fig : survey ] ) . , title="fig:",height=192 ] , title="fig:",height=192 ] fully functional staves were installed on a carbon fiber support structure arranged to have full azimuthal coverage . as seen in figure [ fig : full ] the silicon components are facing into the support structure which protects the silicon from damage during the full installation process . electrical connections were laid on the tube and connected to each of the detectors . we were able to run the detectors in place to ensure that all electrical connections were sound . data was taken to establish the baseline noise and pedestal data before final installation . finally , cooling lines were attached to the cooling tubes running through the staves and were tested with a helium leak checker . two staves were found to have leaks inside the cooling tube . these staves were not installed and the cause of the leaks is currently being investigated . , width=480 ] once the entire ist was assembled the support structure was mated with a larger detector support system and inserted into the central region of star . the detector was then hooked up to electrical and cooling services and run in place again to ensure that all detectors are working properly . the ist for star is made of 24 staves , each of which carry 6 silicon sensors read out by 36 apv chips . the detectors are designed to operate in a high radiation environment with as little mass as possible to reduce the chance of interactions with particles . the staves were designed and manufactured at mit and lbnl . passive assembly as well as apv readout chip and silicon sensor attachment took place at mit while bonding , testing and installation took place at brookhaven national laboratory . the finished detector was installed into star and is currently taking data with 95 percent of channels working properly . the fabrication of the ist was split between mit and university of illinois at chicago . zhenyu ye at uic coordinated with the silicon detector lab at fermi national accelerator laboratory to complete all steps in production after the passive attach . their assembly process was very similar to the assembly process used at mit but not identical ; it is outside the scope of this paper to detail the two separate assembly processes . 999 g. visser _ et al . _ , `` a readout system utilizing the apv25 asic for the forward gem tracker in star , '' ieee real time conference record , berkeley , ca , 2012 . + l. jones , `` apv25-s1 user guide version 2.2 , '' royal academy london , 5 september , 2001 . + m. l. brooks _ et al . _ , `` the phenix forward silicon vertex detector , '' arxiv:1311.3594v2 [ physics.ins-det ] , 14 february 2014 . + m. j. french , _ et al . _ , `` design and results from the apv25 , a deep sub - micron cmos front - end chip for the cms tracker , '' nuclear instruments and methods in physics research a 466 ( 2001 ) 359 - 365 .	we present the design of a detector used as a particle tracking device in the star experiment at the rhic collider of brookhaven national laboratories . the `` stave , '' 24 of which make up the completed detector , is a highly mechanically integrated design comprised of 6 custom silicon sensors mounted on a kapton substrate . 4608 wire bonds connect these sensors to 36 analog front - end chips which are mounted on the same substrate . power and signal connectivity from the hybrid to the front - end chips is provided by wire bonds . the entire circuit is mounted on a carbon fiber base co - cured to the kapton substrate . we present the unique design challenges for this detector and some novel techniques for overcoming them . electrical and mechanical integration ; hybrid substrate ; nuclear physics ; carbon fiber reinforced polymer

Subsets and Splits

No saved queries yet

Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.